Naturalism and Realism in Kants Ethics

In this comprehensive assessment of Kants metaethics, Frederick Rauscher

shows that Kant is a moral idealist rather than a moral realist and argues that
Kants ethics does not require metaphysical commitments that go beyond
nature. Rauscher frames the argument in the context of Kants nonnaturalistic
philosophical method and the character of practical reason as action-oriented.
Reason operates entirely within nature, and apparently nonnatural claims
God, free choice, and value are shown to be heuristic and to reect reasons
ordering of nature. The book shows how Kant hesitates between a transcendental moral idealism with an empirical moral realism and a complete moral
idealism. Examining every aspect of Kants ethics, from the categorical imperative to freedom and value, this volume argues that Kants focus on human
moral agency explains morality as a part of nature. It will appeal to academic
researchers and advanced students of Kant, German idealism, and intellectual
history.
frederick rauscher is Professor of Philosophy at Michigan State
University. He is the editor and co-translator of Kant: Lectures and Drafts
on Political Philosophy (with Kenneth R. Westphal, Cambridge, 2015),
co-translator of Notes and Fragments (with Paul Guyer and Curtis Bowman,
Cambridge, 2005), and editor of Kant in Brazil (2012).

Naturalism and Realism in

Kants Ethics
Frederick Rauscher

University Printing House, Cambridge CB2 8BS, United Kingdom

Cambridge University Press is part of the University of Cambridge.
It furthers the Universitys mission by disseminating knowledge in the pursuit of
education, learning and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781107088801
Frederick Rauscher 2015
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2015
A catalogue record for this publication is available from the British Library
Library of Congress Cataloging in Publication data
Rauscher, Frederick, 1961
Naturalism and realism in Kants ethics / Frederick Rauscher. 1 [edition].
pages cm
Includes bibliographical references and index.
ISBN 978-1-107-08880-1 (Hardback : alk. paper)
1. Kant, Immanuel, 17241804. 2. Ethics. 3. Naturalism. 4. Realism. I. Title.
B2799.E8R38 2015
170.92dc23 2015020982
ISBN 978-1-107-08880-1 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy
of URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.

Contents

Acknowledgments
Citations of Kants writings

page vi
viii

Introduction

Part I

Laying the ground

Moral realism and naturalism

The place of ethics in Kants philosophy

Part II

Practical reason in nature

The priority of the practical and the fact of reason

The transcendental status of empirical reason

Part III

Morality beyond nature?

50
79
81
107
147

God without God: the status of the postulates

149

From many to one to none: nonnatural free choice

180

Value and the inexplicability of the practical

209

Postscript: Kants naturalist moral idealism

241

Works cited
Index

250
258

Acknowledgments

I thank rst and more than anyone else my spouse Delores and my children
Konrad, Bennett, and Audrey for their love and support as I spent long days
working on this book over too many years. Without their patience this book
would never have been nished. This book is for them.
Numerous colleagues and students have helped me to shape my ideas and
I here want to particularly thank several people who most generously commented on this book as it was being drafted. Darlei DallAgnol pushed and
prodded me into better explaining and defending my views both in conversations and through his published criticisms. I greatly beneted from conversations and exchanges with Oliver Sensen and Patrick Kain. All three of them,
along with Paul Guyer, also kindly read parts of the nal manuscript. Robert
Louden and an anonymous reviewer for Cambridge University Press provided
extensive comments that helped me to improve my presentation and arguments. I am deeply grateful to them all, as well as to those who helped me in
earlier years when some of this material was published in independent papers.
I am also indebted to Cambridge University Press editors Hilary Gaskin and
Rosemary Crawley for their guidance in this project.
I am grateful to Michigan State University for a research leave and Intermural Research Grant that allowed initial work on the book, and a sabbatical
that allowed me to complete it. In between, I was grateful to have many opportunities to present my work, resulting in quite a different book than originally
conceived. The Federal University of Santa Catarina, the Federal University of
Rio Grande do Norte, the Federal University of Pelotas, the Federal University
of Pernambuco, the Pontical Catholic University in Rio de Janeiro, the Pontical Catholic University of Parana, the State University of Campinas, and
the University of So Paulo in Brazil, Pisa University in Italy, the Philipps
University in Marburg and the University of Siegen in Germany, the University of Illinois, University of Pennsylvania, Brown University, Western
Michigan University, and Michigan State University in the United States all
provided opportunities either directly or by hosting conferences for me to
present this work in various stages of development. I am indebted to the many
vi

Acknowledgments

vii

contributors to discussions at these events who have stimulated my thought

and caused me to deepen my interpretation.
I wish also to thank the original publishers of material reproduced here.
With the exception of God Without God, which forms the content of
Chapter 5 with little alteration, these papers have been excerpted and/or
modied. In most cases, only parts of the paper are included in the book, even
scattered over several chapters.
Part of Chapter 4: Freedom and Reason in Groundwork III in
Kants Groundwork of the Metaphysics of Morals: A Critical
Guide, edited by Jens Timmermann (Cambridge University Press,
2009), pp. 20323.
Part of Chapter 2: Why Kants Ethics is A Priori and Why It
Matters, in Recht und Frieden in der Philosophie Kants: Akten
des X Internationalen Kant-Kongresses, Band 3, hrsg. Valerio
Rohden, Ricardo Terra, Guido de Almeida, and Margit Rufng
(Berlin: Walter deGruyter, 2008), pp. 34757.
Chapter 5: God Without God: Kants Postulate Kant e-Prints
Srie 2, v. 2, n. 1, jan.jun., (2007), pp. 2762 [www.cle.unicamp.
br/kant-e-prints/]
Part of Chapter 4: Reason as a Natural Cause, in Moralische Motivation. Kant und die Alternativen, edited by Heiner F. Klemme,
Manfred Khn, and Dieter Schnecker. Reihe Kant Forschungen
Band 16. (Hamburg: Felix Meiner Verlag, 2006), pp.97110.
Parts of Chapters 1 and 3: Kants Moral Anti-Realism, Journal of
the History of Philosophy 40 (2002): 47799.
Parts of Chapters 2 and 3: Kants Two Priorities of Practical
Reason, British Journal for the History of Philosophy 6 (3)
(1998): 397419.
Parts of Chapters 2 and 4 were originally published in Portuguese as
Razo prtica pura como uma faculdade natural [Pure Practical
Reason as a Natural Faculty], translated by Milene Consenso
Tonetto, Ethic@ 5 (2006), pp. 173192 [https://periodicos.ufsc.
br/index.php/ethic/issue/view/1453]
Part of Chapter 7 was originally published in Portuguese as Os
limites externos da losoa prtica e as limitaes da Deduo na
Fundamentao III, translated by Kariel Giarolo, Studia Kantiana
14 (2013): 12741 and in German as Die uerste Grenze aller
praktischen Philosophie und die Einschrnkungen der Deduktion in
Grundlegung III, in Kants Begrndung von Freiheit und Moral in
Grundlegung III: Neue Interpretationen edited by Dieter Schnecker,
(Mnster: Mentis Verlag, 2015), pp. 21731.

Citations of Kants writings

Quotations from Kant use the Cambridge Edition of the Works of Immanuel
Kant when available. I have modied the translation in some cases. When the
material is not included in the Cambridge Edition, the translations are my own.
All references will be given parenthetically in the text. References to Kants
writings except the Critique of Pure Reason will be to their volume and page
number in Kants gesammelte Schriften, 29 Vols. (Berlin: Walter de Gruyter
[and predecessors], 1902). References to Critique of Pure Reason are taken
from the edition Kritik der reinen Vernunft edited by Jens Timmermann
(Hamburg: Felix Meiner, 1998) and identied by page numbers for the rst
and/or second editions, cited as A and B. These page numbers are given in
the margins of the Cambridge translation.

viii

Introduction

Among the rst conceptions I had of Kants philosophy, provided second- or

third-hand when I was in college before I had studied him enough to be able to
judge it, was the tale that Kant had divided the world into two separate realms,
that of appearances in space and time, constituting nature, and that of things in
themselves eerily existing not in space and time, constituting morality. This
neat division seemed like a tidy way of dealing with the potential conict
between our moral lives and the scientic world view. As I learned more about
Kant, it became clear that the ontological division was not that simple and that
whatever it was, it did not map onto the nature/morality division. The proper
understanding of these issues became a lasting puzzle.
This book is an attempt to solve that puzzle by showing what morality is and
just how nature and morality relate to one another in Kant. I have two main
goals. The rst is to show that Kants ethics is fully compatible with a
metaphysical naturalism, meaning that no property or entity outside of empirically real nature in space and time is needed. The second is to determine the
extent to which Kant is a moral realist, which can be decided only through a
detailed look at the nature of Kants ethics and its specic elements. I will
conclude that the most plausible interpretation is that Kant is a moral idealist
(the term I prefer to anti-realist or constructivist) rather than a realist,
although given the resources in his philosophy, he could have been a realist in
a limited sense.
Part One, Laying the Ground, sets the stage for the detailed assessment by
providing denitions of metaphysical naturalism and moral realism and by
showing how ethics ts into Kants philosophical project as a whole. This part
is crucial for the overall project because it assesses the inadequacy of some
ways of approaching realism and naturalism, and provides an explanation for
the particular approach I take. The review of Kants philosophical project
importantly shows the way in which transcendental philosophy allows for a
metaphysical naturalism and the way in which the claims of practical reason
are prima facie not ontological. While some readers are familiar with these
basic positions, the details and conclusions I draw here bear specically on the
particular topics in later chapters and form the foundation of their arguments.
1

Introduction

Chapter 1 provides a denition of realism in terms of the independence of some

moral principles, properties, or objects from the moral agent. I argue that this is a
better denition for use in assessing Kant than one focused on the truth of moral
claims because the real issue between realists and idealists is not whether Kants
morality claims objective validity but how to understand Kants a priori moral
law, the nature of practical reason, and autonomy in relation to moral agency.
I treat moral agency in two ways: as actual agents in nature (empirical) and as the
necessary conditions for the possibility of any moral agent at all (that I label
transcendental in reference to the transcendental method of justication).
This distinction allows for Kant to be both an idealist and a realist at different
levels, and provides the complexity necessary to resolve the multi-faceted issue.
I then turn to naturalism, which is divided into methodological and metaphysical
naturalism. The former would claim that the only proper methodologies for
nding knowledge are those of the natural sciences; Kant rejects this in light of
his use of a priori concepts and his transcendental method. I explain metaphysical
naturalism, which claims that the only entities that exist are those determined by
the natural sciences, in relation to Kants own conception of nature as consisting
of matter studied by physics (and less strictly also by chemistry and biology) and
thinking nature studied by empirical psychology (and related disciplines).
A metaphysically naturalistic Kantian ethics would hold that nothing beyond
the entities in space and time, physical and mental, is needed for morality. The
chapter concludes with a list of the eight elements of Kants ethics that need to be
assessed as realist or idealist: particular ends, particular duties, absolute value, the
highest good and the postulates, moral obligation, the moral law itself, pure
practical reason, and free choice. Thus, the task of the remainder of the book is
to assess these elements in terms of the transcendental and empirical levels of
realism and the possible limitation to nature in space and time.
Chapter 2 examines the nature of Kants philosophy overall in order to place
morality in its proper context and show how that framework affects realism
and naturalism. I note that Kant insists that philosophy aims at systematically
organized cognitions, with the overall aim at the essential ends of human
reason. Each part requires a domain and further subdivision into parts in
accordance with a system. I look at several broad divisions of philosophy in
Kant critique/system, pure/empirical cognition, and theoretical/practical, and
note how they interrelate. I show how the method of transcendental argument
provides Kant with a general way to defend ethical claims without resorting to
a non-natural metaphysics. Transcendental argument justies the use of a
priori cognitions by cognizers in nature as part of their empirical cognitive
systems. In looking at various ways to construe the theoretical/practical distinction, I show that the domain for practical philosophy is free acts understood
from the perspective of agents engaged in deliberation. With these foundational issues settled, the detailed work can begin.

Introduction

Part Two, Reason in Nature, focuses on practical reason to show that

reason can exist within nature, and within the empirical minds of empirical
moral agents. The elements of Kants ethics that are tied directly to reasons
systematization of free acts through the moral law, I show, are fully explainable within nature. Practical reason functions in action to guide behavior and to
that extent, operates independently of any concerns for ontological commitments. Human beings experience the obligatory force of the moral law for
nite rational beings in the guise of the categorical imperative when engaged
in the actual process of deliberation before action. Practical reason is itself part
of nature as an empirical faculty of mind that would be validated through
transcendental argument, and in this way, empirical reason is the timeless
transcendentally free faculty that causes agents actions. This aspect of Kants
ethics would allow for transcendental idealism and empirical realism.
In Chapter 3, I note that Kant worked with two different conceptions of the
priority of practical reason. His direct discussion of primacy gives a strong role
to theoretical reason in conrming metaphysical demands of the practical in
what I call the priority of belief. In many other places, however, Kant also
insists on an independence of the moral agent in practice from any theoretical
concerns. I call this independence of moral action from theoretical concerns the
priority of action. I take this latter to express the genuine priority of practical
reason as guiding moral agents in deliberation for free acts what ought to be
without concern for ontological claims what is. These points together culminate in my interpretation of the fact of reason as the actual awareness of the
categorical imperative in empirical agents, providing a ground for the practical
perspective that would need to be defended using transcendental analysis. I see
the fact of reason more as an act of reason in prescribing the moral law to agents
during deliberation as a categorical imperative. This provides a metaphysically
naturalistic aspect of morality that is also strongly idealist.
I regard Chapter 4 as the key chapter in the book. I show that pure practical
reason, as the source of the moral law, is the most important element in Kants
ethics. I take seriously Kants claims that reason is a cause, and even a natural
cause, by identifying an empirical faculty of reason that would be part of the
natural ontology provided by a robust empirical psychology. This faculty of
reason is also a timeless transcendentally free cause, as Kant stresses in the
Third Antinomy, in virtue of its unchanging structure that produces the
categorical imperative to systematize free acts, themselves understood as
decisions by the empirical power of choice determined within nature by
empirical reason. Such a view is transcendentally idealist, since the validity
of the moral law is justied as a product of reason, which itself constitutes the
very possibility of any moral agent at all, and simultaneously, an empirically
real view, since reason and its laws would be valid independent of each and
every actual empirical agent. I note that Kant falls short of endorsing this view

Introduction

when he denies that the moral law can be subject to a transcendental deduction
and when he limits his claim to the practical point of view. Kant would then be
an empirical moral idealist about reason and the status of the moral law.
The nal part of the book, Morality Beyond Nature?, looks at God, free
choice, and absolute value. These elements of Kants ethics represent apparently
non-natural entities or properties and cannot be understood in the same metaphysically naturalistic way that reason itself can, that is, they cannot be an actual
part of nature operating as the empirically real manifestation of the transcendental conditions of moral agency. Instead, I show that God and free choice have
a role through the postulates that comprehends them as concepts created by
reason without reference to the purported non-natural object or property but
only with an immanent reference as aids to moral action. I argue that value for
Kant cannot be an independent property of objects either within or outside
nature, but is merely a part of the order imposed on nature by practical reason.
I then link that order back to the status of practical reason itself.
Chapter 5 argues that Kants conception of a postulate has much more in
common with the general nature of ideas of reason than it does with any claim
to existence. I trace Kants development of the notion of a postulate along with
the similar notions of transcendental hypothesis, idea of reason, and belief.
I show that the idea of the highest good has no ontological implications of its
own and focus on the postulate of God, since the same solution will also hold
of the postulate of immortality. The postulate of God, I show, is supposed to
have immanent reference, that is, to empirical agents moral lives, rather than
transcendent reference, that is, to a being in itself. In the practical point of
view, empirical moral agents operate with the concept of God for certain
purposes but relate it only to the ought, not to the is.
Chapter 6 assesses the most difcult problem for a naturalist interpretation
of Kant ethics: freedom of the will as the freedom of the power of choice in
making a decision uncaused by anything in nature. I have two strategies in this
chapter. I rst show that Kant insists that free choice is needed for two reasons
related to ought-implies-can and moral responsibility. Both reasons require
only one non-natural choice, not a multitude as some commentators prefer.
This single timeless choice of the entirety of ones phenomenal character is the
best interpretation of his claims about freedom, timelessness, distinct causality,
and the intelligible character in relation to free choice. While still unnatural, it
is at least the minimal non-natural interpretation. Second, I argue that stressing
the status of freedom as a postulate, which Kant himself does not, allows for a
naturalistic interpretation of this choice as a concept merely playing a heuristic
role in moral life.
The nal chapter moves from value in particular to the status of the practical
point of view, or moral experience more broadly, in general. I reject the
possibility of any non-natural, intrinsic value property and instead show that

Introduction

the value of humanity as an end in itself is nothing more than the highest rank
in the order of ends that reason imposes on nature through the categorical
imperative. The intelligible order of things is not an order of intelligible things
but a rational ordering of natural things. I draw together Kants various
discussions of this direct application of the moral law to experience as a moral
world. This is the way that practical reason applies to nature within the
practical point of view. Culminating this chapter is a look at the limits to
practical reason that Kant reveals in Groundwork III, where he admits that
reasons own structure that requires both systematic connection and unconditional explanation is responsible for the claim that there is a necessary moral
law, and holds, in language similar to that of the Third Antinomy, that reason,
the source of morality, is itself ultimately incomprehensible.
In a postscript, I review the particular assessments I made regarding the
eight elements of Kants moral theory laid out in Chapter 1. I pull together the
features of my interpretation of Kant as a metaphysical naturalist. The various
claims about transcendental and empirical realism and idealism are arranged
into their basic sets and the core interpretive points that ground the main
disagreement set out. I have identied a Kantian transcendental moral idealism
that is also an empirical realism, thus dissolving some of the realist/constructivist disagreement. I show, however, that Kant is himself hesitant to endorse
this transcendental validity for morality and, particularly in light of the priority
of the practical point of view as an agent-perspective rational ordering of
nature with no ontological claims of its own, that the more appropriate
conclusion is that Kant was an empirical moral idealist.
A brief word about my methodological approach. I am not providing a
strictly exegetical work. There are passages in Kants writings that are inconsistent with elements of my interpretation, but I would argue that the same is
true of all interpretations of Kant given his own inconsistent use of terminology and the diverse contexts in which he applies the same terms. The
purpose of this book is not merely historical but is aimed at assessing Kants
ethics in light of twenty-rst century concerns about naturalism and realism.
My work is reconstructive in that I pursue the philosophical implications of
Kants positions to sometimes make connections that Kant himself does not
explicitly make. I believe that all of my claims are consistent with Kants
general philosophical aims, methods, arguments, and conclusions, and nearly
always with his particular stated positions. Given the novelty of Kants
philosophy, the complexity of the issues he raises, the vast range of his project,
the transformations of some of his positions over time, and above all, the
requirement of the interpreter to go beyond merely quoting texts in order to
explain a great philosophers work in ways relevant to the present, no comprehensive interpretation can offer more than that.

Part I

Laying the ground

Moral realism and naturalism

My study aims at providing an interpretation of Kants ethics that cuts across

the issues of realism and naturalism. I see these two issues as closely connected. One cannot resolve the question of whether and to what extent Kant
was a moral realist without resolving questions about the metaphysical status
of the elements of Kants ethics. These elements, such as the status of reason
itself, the nature of value, and freedom of the will, relate to the conception
Kant has of the limits of human experience and legitimate claims that go
beyond experience. This concern in turn raises the issue of nature as a limit for
experience. Whether ethics goes beyond nature is in this way tied to the issue
of realism. The reverse is also true: When asking about whether Kants ethics
is compatible with naturalism, one has to start with a conception of nature in
Kant and then turn to the particular elements. One would expect that anything
in nature would count as real for Kant, and on the empirical level, that is true of
objects and of mental states and faculties. Questions arise, however, about how
to place in nature the distinctive perspective of the practical and its distinctive
role in determining what ought to be rather than what is. And since Kant also
invokes a transcendental level of analysis, some elements of the experience
moral beings have in nature might be ideal in the same sense that space and
time and the categories are ideal. A detailed understanding of what might be
real and what ideal, and in what senses, will help to resolve the issue about
whether morality requires more than is available in nature.
Both the term naturalism and the term moral realism need to be
specied and adapted to the peculiarities of Kants philosophy. Contemporary
philosophical work on these issues does not always approach these issues in a
way amenable to Kants critical philosophy. Kants different conceptions of
the sciences and his transcendental idealism inform his understanding of
nature. His use of the terms realism and idealism in both empirical and
transcendental senses, and the very nature of his critical philosophy as largely
subject-centered, do not easily map onto discussions of realism in terms of
objective facts or empirical properties of human nature.
My aim in this chapter is twofold. First, I will provide an analysis of
denitions of moral realism and nonrealism with a focus on how well they
9

10

Laying the ground

can account for the peculiarities of Kants general philosophical approach.

I use the term nonrealism to capture all alternatives to moral realism until
their particular characteristics are dened in this chapter. The analysis of
realism I adopt stresses the independence of elements of morality from moral
agency as such; I contrast realism with idealism. I examine the transcendental
and empirical levels of realism and idealism in Kant and show where my
approach differs from the typical emphases in constructivism.
Second, I will examine the meaning of naturalism. I stress that my aim is
limited to showing that Kants ethics is compatible with a metaphysical
naturalism, understood in Kants own terms as including both the physical
and the mental. He rejects methodological naturalism. In fact, his methodological antinaturalism will play an important role in my interpretation.
Third, I will identify the particular elements of Kants moral theory that can
be interpreted as real or ideal, natural or nonnatural. This taxonomy will
illuminate the point that inquiring whether Kant is a moral realist or naturalist
is not a simple yes-or-no question. On different levels and about different
elements, Kant holds realist as well as idealist views. Further, by highlighting
particular issues, this taxonomy will push the debate beyond mere terminology
to the concrete differences among Kant interpreters.
Dening moral realism and moral idealism
Philosophy over the past three decades has included extensive discussion of
the nature of moral realism and its alternatives.1 The issue entered the Kantian
literature with John Rawls John Dewey Lectures in April 1980, Kantian
Constructivism in Moral Theory,2 which also introduced the term constructivism into the literature on Kants practical philosophy.3 Rawls intention was
not simply to provide an interpretation of Kants ethics along constructivist
lines but mainly to present Kantian constructivism as a general moral theory
alongside utilitarianism, intuitionism, and perfectionism.4 Kant commentators

2
3

Some of the earliest contributions to the debate are collected in (Sayre-McCord 1988a). My
discussion will invoke contributions to the debate about moral realism in general only when
doing so is useful for understanding the particular debate within Kant circles.
The Dewey Lectures were published in (Rawls 1980) and reprinted in (Rawls 1999b,
pp. 30358).
Larry Krasnoff traces the rst use of the term constructive in moral theory to a review by
Ronald Dworkin in 1973 (reprinted in [Dworkin 1978, pp. 15083]) of John Rawls book
A Theory of Justice. Rawls did not use the term constructivism in A Theory of Justice but
soon adopted it to describe his own theory. See (Krasnoff 1999).
Since constructivism was introduced by Rawls, one might think that it can be applied in political
philosophy in addition to moral theory. In this book, I exclude directly political principles in
favor of moral ones, broadly speaking. For work on the political use of constructivism, which
draws more directly on Rawls own work in A Theory of Justice and Political Liberalism rather

Moral realism and naturalism

11

soon took up this suggestion with greater or lesser attention to historical claims
about Kants own theory.5 Since that time, Kant commentators have almost
exclusively characterized the discussion of the question of moral realism as a
choice between the alternatives of moral realism and moral constructivism,
effectively lumping together all possible nonrealist interpretations of Kant as
constructivist. While doing so, these same commentators have generally
bemoaned the lack of specicity of the term constructionism and have
attempted to dene or rene it using various taxonomies they present. Rarely
do these taxonomies match precisely; still more rarely do they cross-reference
or incorporate one another. The result is that much philosophically interesting
work on the issue of moral realism in Kant is scattered in individual articles in
isolation from one another.
Moral realism is one of those issues about which philosophers spend much
of their time simply dening the terms of the debate. There is no clear
consensus on what realism actually means, and many acknowledge that that
the meaning of realism has shifted along with the philosophical tide.6
According to this last view, some theories which would have counted as
antirealist fty or hundred years ago would count as realist today. Even this
admission seems too optimistic since it assumes that there is a general consensus at any given time.
A popular denition
Two problems beset the task of dening moral realism. One is that various
metaethical theories differ in their interpretation of key terms such as truth
and validity, objective and subjective, obligatory and permissible.
Disagreement about these key terms allows various different theories to claim
to present moral realist positions although they vary widely with one another.
The debate over moral realism also suffers from the connotation of one of the
central terms of the debate: antirealism. As an anti, the latter connotes that
one is opposed to some positive claim rather than that one is giving a positive
claim of ones own. In ethics, this tendency is exaggerated by the moral import
of the terms involved. To deny moral realism seems to imply a lessening of the
claims or values of morality itself. For this reason, I and others prefer the more

than his particular interpretation of Kant, see (ONeill 2002), (ONeill 2003), (Bagnoli 2014),
and (Kaufman 2012).
Two of the earliest Kant commentators to employ the language and method of constructivism are
Thomas Hill (Hill 1989) and Onora ONeill (ONeill 1989). They were soon followed by the
most inuential constructivist after Rawls, Christine Korsgaard (Korsgaard 1996b). The intense
debate about moral realism in Kant began in the wake of Korsgaards book.
See, for example, (Railton 1996).

12

Laying the ground

positive-sounding term idealism and still other theorists, constructivism

over antirealism.
These two problems are highlighted by a famous illustration of this bias
toward supporting moral realism. Geoffrey Sayre-McCord suggests that one
might want to be able to say that the Nazis were really bad, but might feel
unable to make such a strong claim without a theory of moral realism behind
one.7 Moral idealism is then saddled with the burden of being implicitly
equated with a rejection of any kind of validity to moral claims. An analogous
claim regarding objects would illustrate the fallacy in this attitude:
A traditional early modern realist about objects might be one who holds to
the existence of objects independent of human perception of them; a traditional
idealist about objects might hold that objects are nothing more than collections
of sense data. Someone who rejected idealism about objects on the grounds
that it really cannot be about objects would simply be begging the question
about the denition of object, as if the idealist were denying the existence of
objects entirely rather than identifying them as collections of sense data. In the
same way, someone who thinks that saying that the Nazis were really bad
requires moral realism is begging the question about the denition of bad.
A moral idealist can easily make essentially the same claim as the Nazis were
really bad when the claim is stated in some non-question begging sense such
as bad without qualication.
Sayre-McCord himself offers a denition of moral realism that has gained
traction. He claims that realism involves embracing just two theses: (1) the
claims in question, when literally construed, are literally true or false (cognitivism), and (2) some are literally true.8 Many recent Kant commentators
writing on the issue of moral realism either use his denition or assess one
much like it, generally critically.9 According to his denition, a theory is not
realist if it either (a) denies cognitivism, in which case it is some kind of noncognitivist theory, or (b) denies that there are any moral truths, in which case it
is some kind of error theory.
As these commentators note, Sayre-McCords denition does not adequately
capture the crux of the debate regarding Kants metaethics. Prima facie, Kant is
neither a noncognitivist nor an error theorist. Constructivists and other nonrealists allow that in Kant there are moral claims that bear truth or falsity.
Under this denition, they would count as realist theories. Sayre-McCords
7

8
9

(Sayre-McCord 1988b, p.1) credits the illustration to Philip Gasper. Sayre-McCord does not take
the illustration to be more than a caricature and subsequently discusses ways that nonrealism can
capture moral condemnation. It is still useful as a caricature.
(Sayre-McCord 1988b, p. 5).
Paul Formosa (Formosa 2013) and I (Rauscher 2002) cite Sayre-McCords denition directly.
Others who also note that this sort of denition does not really settle the issue of realism and
idealism include Jochen Bojanowski (Bojanowski 2012), and Sharon Street (Street 2010).

Moral realism and naturalism

13

denition casts too wide a net. It makes every single possible moral theory
that allows for moral truth into a realist theory. It co-opts the use of truth claims
in a moral theory so that any kind of nonrealist theory is seen to deny validity
to morality. Using the analogy with realist and idealist denitions of objects
again, both the traditional realist and idealist hold that there are truths about
objects, yet they have quite different conceptions of what those objects are.
Similarly, a moral realist and a moral nonrealist can hold that there are moral
truths, yet they can have quite different conceptions of what constitutes the
bearer of those truths.
Roughly speaking, the kind of denition Sayre-McCord provides equates
real with true and places the fault line between moral realism and moral
idealism precisely atop the fault line between acceptance and denial of moral
truth. Kantians should be skeptical that this perfect correspondence does
justice to Kants theory, which, incorporates transcendental idealism yet
afrms moral truth.
Sharon Street phrases this point with regard to constructivism in general:
if we understand realism this way, then metaethical constructivism counts as
a brand of realism as indeed do an extremely wide range of views, including
even a simple subjectivism according to which whats good for a person is
whatever that person thinks is good.10 She is right to stress the overextension
of the term realism to any moral theory that offers some criterion for truth. In
particular, if Kant has a theory of moral truth that can be understood as
subject-dependent, perhaps even only dependent upon a certain kind of activity
of subjects, then it would be peculiar to count his theory as realist although it
would contain a criterion for judging certain moral claims as true or false. The
basis of the truth or falsity would seem to make a great deal of difference in our
view of whether a theory should count as real.
This kind of denition of realism centered on truth claims is more suited to
consequentialist than Kantian moral theories. Because consequentialists hold
to the priority of the good over the right, they tend to take moral facts primarily
as facts about some good to be attained rather than some formal laws of right;
for a consequentialist, the content of laws that express right is derived from the
conception of the good to be attained. One can picture how accepting the truth
value of claims about the good can make one a realist in a more metaphysical
sense by noting that goods are taken to be independent of the belief that they
are good; those who deny this independence are expressivists and those who
deny that there are any goods are error theorists. Right, on the other hand, has a
more ambiguous connection to belief. It is possible to conceive of moral agents
who are so constituted that, in a roughly Kantian way, their practical cognitive

10

(Street 2010, p. 370).

14

Laying the ground

faculty both produces the moral law as rational and produces belief in it. The
truth value of the moral facts, here the criterion for right, would not be
independent of rational belief in moral agents. This kind of Kantian view
would open up space between the denial of the truth of some moral facts
(expressivism and error theory) and the afrmation of moral facts on some
independent basis. Thus, consequentialists might be more inclined to be
satised with the kind of denition that Sayre-McCord offered while Kantians
ought to be skeptical of its value.
A better denition
In order to move beyond this problem, a conception of the proper division
between realism and nonrealism more appropriate for Kantians should include
a metaphysical dimension, as some Kant interpreters recognize.11 I offer the
following denitions that are more appropriate for Kantian ethics:
Moral realism: The moral principles, properties, or objects of the world
are independent of the transcendental or empirical moral agent.
Moral nonrealism: The moral principles, properties, or objects of the
world are dependent upon the transcendental or empirical moral
agent.12
11

12

Paul Formosa labels realism that excludes only noncognitivists and error theorists a weak sense
of moral realism that amounts to nothing more than a claim about the truth of some moral
judgments in contrast to a strong realism that holds that what makes the moral judgments true
or false is an independent moral order (Formosa 2013, p. 172). The weak sense of moral
realism accords with the kind I have just criticized. Formosa is correct in noting the need for a
stronger, more meaningful sense of moral realism. This strong sense requires ontological
independence. Richard Boyd offers a denition that identies a requirement for independence
from our moral opinions, theories, etc. and also claims that ordinary moral reasoning is a
reliable method for obtaining moral knowledge (Boyd 1988, p. 182). The difculty with Boyds
denition is the vagueness included in his etc.. Allen Wood alludes to Boyds denition in his
rst book on Kants ethics as the most agreed-upon sense of the term moral realism (Wood
1999, pp. 157, 374, n. 4) but claims (quite emphatically) in his second book on Kants ethics that
he does not endorse Boyds own type of Cornell Realism as the proper way to construe
realism in conformity with that denition (Wood 2008, p. 295, n. 8).
I offered a similar set of denitions in (Rauscher 2002, p. 482). There I used the term moral
idealism while here I use moral nonrealism; the latter term broadens the denitions scope to
include anyone who rejects moral truth such as error theorists and expressivists. I will focus my
argument on idealism as I proceed. I also used the term moral characteristic while here I am
clearer in using moral principle, property, or object, the original intent of the former term;
there is no change in meaning accompanying this clarication in terminology. Finally,
I formerly used the term human mind while here I say transcendental or empirical moral
agent; this change is intended to allow for the different levels of realism or nonrealism at the
empirical and transcendental levels and reects a shift in my analysis over the past decade. I am
very grateful to Darlei DallAgnol for his criticisms of my earlier approach and for many
discussions in which we have attempted to nd the best Kantian position on moral realism.
Some of his concerns are in (DallAgnol 2012b) and are discussed in my reply (Rauscher 2012).

Moral realism and naturalism

15

These denitions dene the difference as metaphysical since both the term
moral principle, property, or object and the relation of independence are
metaphysical. Very roughly, they are intended to put the onus on the moral
nonrealist to prove that there are no essentially moral principles, properties, or
objects, or if one wants to phrase it differently, no moral truth that would
correspond to any principle, property, or object, such that the principle,
property, or object would be a part of the cosmos were moral agents qua
moral subjects not to exist. By nonrealism I mean all possible alternatives to
moral realism, including expressivism, error theory, the constructivist views
that are not realist, and idealism, to be explained later in this chapter.
A moral principle, property, or object is not the same as a principle,
property, or object required for morality. The difference between a moral
principle, property, or object and a principle, property, or object required for
morality is that the latter consists of principles, properties, or objects which are
not solely moral, the former of principles, properties, or objects which are
solely moral. Moral principles, properties, or objects are to be understood as
exclusively involving moral normativity or value. For example, the human
mind may be an object required for morality, but because it can play a role in
other areas not pertaining to moral normativity or value, such as theories of
qualia, it is not a moral object. An individuals being good, however, is a moral
property, since this characterization can play no role except in situations
involving moral normativity or value or those derivative upon it such as
descriptions of morally good persons. For denitional purposes, moral properties and objects are not limited to a particular metaphysics. As examples of
moral principles, properties, or objects, consider the following: Good, evil,
bad, rightness, wrongness, justice, value, moral law. As examples of principles, properties, or objects required for morality, consider the following: The
existence of minds, pragmatic or technical meansends principles, and so on.
Of course, particular moral theories might differ regarding which specic
principles, properties, or objects are solely moral rather than more general
principles, properties, or objects, and whether the solely moral principles,
properties, or objects are separable from the more general principles, properties, or objects. In Kant, we might legitimately question whether freedom of
the will, agent causality, or the existence of God count as moral properties or
objects or as simply properties or objects required for morality. I will treat
them as moral properties or objects and include them in my arguments in this
book.
The term independent invokes a metaphysical separation between the
moral agent qua agent and whatever moral principles, properties, or objects
are at issue. I do not intend to hold that the moral principles, properties, or
objects are to be independent of all moral agents in existence but rather
independent of the individual moral agent or particular type of moral agent

16

Laying the ground

(to be specied) as a moral subject. There is no quick route to idealism through

a claim that morality depends upon the existence of any moral beings at all.
Thus, if moral agents themselves were to have independent intrinsic value as
objects of consideration for a moral agent who is the subject facing a moral
decision, that value would be real because it would be independent of the
moral agent qua subject. Similarly, if moral value were to reside in satisfaction
of desires for moral agents in general and that moral value had a justication
beyond simply moral agents considering satisfaction of their own desires to be
good, the value of desire satisfaction would be real. Because moral agents are
not passive, dependence on mental activity is included.13 This emphasis on
moral agents qua subjects rather than objects is extremely important for a
Kantian theory that places much weight on the practical point of view or the
structure of moral agency.
A divine command theory or any theory that grounded moral laws on
something independent of moral agents would be realist on this account, as
would any theory that held that there are values independent of moral agents
acts of valuation or capacity of valuation. A moral theory that depended solely
on contingent facts about particular persons, such as a theory referenced in the
quote from Street earlier that the good for a person is merely whatever that
person happens to think is good for her, would not count as realist. An error
theory would be nonrealist because it denies that there are any moral principles, properties, or objects.
I use the phrase transcendental or empirical moral agent in order to invoke
two possible levels of assessment of moral agency. The transcendental moral
agent would be the subject of an assessment of the conditions for moral agency
as such. In Kant, such an assessment would be transcendental, in other kinds of
theories conceptual. The description of the transcendental moral agent would
be valid for all possible moral agents. The empirical moral agent would be
the subject of an assessment of the particular mental structure of human
and similar beings as moral agents. This approach, which looks not at a

13

Jochen Bojanowski cites my earlier denition but claims that his moral idealism, which he
takes to be a third alternative between moral realism and moral antirealism in Kant, does not
utilize the sense of dependence he sees in my denition. The idealism I want to ascribe to Kant
holds not that the good depends on the human mind, but that its existence depends on selfaffection in human cognizers (Bojanowski 2012, p. 4). I hope that my clarication here
adequately shows that, as I had always intended and as is clear from the position I detail in
my earlier paper, the kind of dependence on the human mind can include mental activity such as
decision-making, reection, and self-affection. Bojanowski hints in the claim quoted here that
he might mean that the meaning of the term good is dependent upon something besides the
human mind while the existence of good depends upon self-affection; because he does not
utilize semantic terminology, this attribution is speculative. Later in his article, he holds that
practical cognition precedes normative facts, which implies that the meaning of the term good
is also the result of practical activity (Bojanowski 2012, p. 13).

Moral realism and naturalism

17

transcendental moral agent but at a humanlike agent in particular, is empirically informed, including facts about the existence or types of desires, the actual
capabilities of the agent such as a capacity for free choice amid various
possible determinants of action, the access the mind has to any moral facts,
and so on. This approach is both empirical and conceptual in Kant in the
application of practical principles and ideas that are not derived empirically. It
is not the equivalent of anthropology. The description of the empirical moral
agent in Kant would be valid only for moral agents sharing the particular moral
characterization at issue. Of course, at the empirical level, moral agents may
instantiate the transcendental moral agent, but whether they do, what the latter
consists of, and precisely what the relationship between them is depends upon
the nature of Kants transcendental/empirical distinction (to be discussed later
on) and the success of particular philosophical arguments.
The scope of the empirical agent in my argument is essentially restricted
to humanlike and should be taken to refer to the relevant characterization of
the moral nature of human beings, not to similarity to human beings in morally
irrelevant ways. In Kant, a humanlike moral agent could be dened roughly as
a nitely rational being capable of a pursuit of happiness, conscious of the
categorical imperative, and capable of choosing between following that
imperative or not. From now on, I will use the terms human moral agent
and empirical moral agent to refer to all humanlike beings. I stress the
human moral agent in order to emphasize that dependence upon some particular type of existing moral agent is at issue. A comparison with more general
mental features is useful. Some features of the human mind14 such as the
ability to sense only a particular portion of the electromagnetic spectrum are
certainly contingent and not a necessary feature of universal sensibility; other
organisms can see ultraviolet light or respond to magnetic elds. Kant holds
that space and time must be considered dependent in some way on the human
mind and are not conditions for experience for all possible minds. Arguably, a
similar variation is possible in theory with regard to mental processes more
closely related to morality. Supposing that various other types of mental
organization were differently instantiated in different species and here one
must imagine something like a mind without memories or one determined
immediately by emotion the particular parts of that mental organization that
were unique to one species might, on some theories, ground moral truth for

14

Human moral agents would have human minds. When I use the term mind, I do not intend to
invoke any Cartesian substance; rather, I intend to be agnostic about the true nature of minds. At
the same time, I limit this to the human mind as we generally experience it and conceive of it. If
it were to turn out that the true nature of human minds were that they are parts of Gods mind, a
resulting morality could be realist. I take this possibility as incompatible with Kants theory.

18

Laying the ground

them.15 Were there some general theory that showed that only one type of
mental organization could possibly exist, or that morality could be grounded
only in some particular core of mental functions that all possible moral agents
must exhibit, then those moral principles, properties, or objects would be
characteristic of both the empirical human and the transcendental moral agent.
To prove this identity is Kants goal.
The aforementioned discussion makes clear an error in equating realism with
objectivity. Objectivity can coincide with nonrealism provided that the minds in
question are all structured in such a way that they necessarily produce the same
ideas. An analogy can be made with the question of color perception. Simply
because our color qualia may be ideal and phenomenal colors may not inhere in
objects independent of the human mind, does not mean that the color qualia we
perceive are not objective. There might be some necessary causal connection
between our physical sensations and our mental perceptions. We might have a
particular mental faculty which provides for the objectivity. This objectivity is
nonetheless not universal for all possible sensible beings but restricted to beings
with the particular kind of physical and mental constitution we have. This
objectivity lacks a transcendental foundation. For morality, objectivity can be
preserved even when nonrealism is accepted, provided that the theory explains
how the structure of the human moral agent dictates that all humans share the
same moral properties and objects at the empirical level or how the structure of
15

As an example of this kind of variation, consider Sharon Streets ants (Street 2012, pp. 5354).
She supposes for the sake of argument that a species of intelligent, conscious ants could exist.
Because of the complex genetic relationships among ants, all female ants share seventy-ve
percent of their genetic code with one another while only one, the queen, is able to reproduce.
There are relatively few male ants. Survival of the colony relies on individual ants sacricing
themselves to protect the queen to ensure the existence of future generations. Street imagines
that a female worker ant would exhibit a value system to reect these facts, valuing the survival
of the queen above her own survival and not seeing herself as intrinsically valuable. Street
offers this as a counterexample to the Korsgaardian/Kantian claim that all reective beings who
are capable of valuing anything must value themselves as ends in themselves. I take this
example to illustrate my point that some kind of organism with a sufciently different kind
of nature could have a different kind of moral system. I would like to make two observations
about this thought-experiment. First, it is extremely unlikely that such intelligent ants, or
similarly genetically related social beings with the same behavioral patterns, would evolve.
Such complex intelligence in animals requires a great deal of investment in the rearing of
offspring and a great deal of relative mass devoted to the brain. Given the costs of producing
and maintaining a functioning intelligent adult, it would be nearly impossible for a species to
mass-produce them to such an extent that they could be easily expendable in such menial ways
as providing a live bridge for others of her kind. Still, as Street notes, the point is not the
plausibility of such a species actually evolving but the fact that the conceivability of the species
provides a counterexample to Korsgaards argument that all creatures capable of valuing must
value themselves the most. Second, it is interesting to note that human beings sometimes think
the same way that Streets ants do. Soldiers or others who willingly sacrice themselves for
their nation are a case in point. This puzzling kind of altruism, which cannot be explained either
as inclusive tness (near genetic relations such as offspring or cousins) or as reciprocal altruism
appears to have no genetic basis. For an attempt to explain such behavior, see (Kitcher 1993).

Moral realism and naturalism

19

the transcendental moral agent dictates that all possible moral agents share the
same moral properties and objects at the transcendental level.
Transcendental and empirical realism
The term realism in Kant has a vital twofold signicance largely overlooked
in the realism debate: Kants distinction between, on the one hand, transcendental idealism or realism and on the other hand, empirical idealism or realism.
This distinction does not appear in the contemporary general debate about
realism in moral theory because it applies to Kantian but not consequentialist
theories. Most of contemporary Kantian ethics does not even employ Kants
transcendental/empirical distinction, presumably taking it to be an ontological
claim about the real versus apparent nature of objects that is not relevant to
ethics beyond a discredited theory of free will. These Kantian moral theorists
appropriate Kants discussion of a practical point of view without linking it to
transcendental idealism. Interpreters of Kants own moral theory who operate
without the distinction have incomplete models of Kants moral realism or
idealism.16
In the Critique of Pure Reason, Kant differentiates between the transcendental and empirical senses of realism and idealism. The four resulting possibilities transcendental idealism, transcendental realism, empirical idealism,
and empirical realism work roughly as follows in relation to space: Transcendental idealism is the claim that all objects in space are mere appearances
and not things as they are in themselves because the space in which they exist
is only a form of our intuition. Transcendental idealism allows objects to be
empirically real, that is, directly knowable by the empirical subject in an
objective order in space. Transcendental realism holds that space is something
given in itself independent of human intuition and hence, that things as they
are in themselves are spatial. But this transcendental realism leads to empirical
idealism, that is, that empirical objects in space are not directly accessible to
the empirical mind because space exists independently of human intuition
(B6971, A36970). Kant, of course, is a transcendental idealist with regard
to time as well as space. The transcendental ideality of space and time means
that objects as they are in themselves are not spatiotemporal and cannot be
experienced by human beings, who are limited to experiences provided in
intuition. Human beings are able to know only appearances, that is, empirically
real objects. But the possibility that other objects exist that are not spatiotemporal remains.
16

This criticism holds for my earlier work as well (Rauscher 2002), but I do approach the
transcendental/empirical distinction without a comprehensive treatment in (Rauscher 2006b)
and with a better but still incomplete treatment in (Rauscher 2006a).

20

Laying the ground

Kants argument for the transcendental ideality of space is given through a

transcendental exposition of the concept of space as a basis for the a priori
truths of geometry (B4041). To be able to function as the basis for a priori
truth in geometry, space as an outer intuition has its seat merely in the
subject. Similar reasoning holds for the transcendental deduction of the pure
concepts of the understanding: They are held to have their transcendental
ground as necessary conditions for the synthetic unity of apperception
(B15960). Transcendental idealism thus takes these crucial formal features
of experience to depend upon the subject in a transcendental but not empirical
sense.
The transcendental/empirical distinction is relevant for morality because we
can take certain aspects of experience like objects in space to be dependent
upon the transcendental subject although they are in some sense independent
of the empirical subject. There is use for this distinction in practical philosophy. Just as objects in space are really independent of the subject in an
empirical sense but dependent on the subject in a transcendental sense, there
might be moral principles, properties, or objects that are really independent of
the moral agent in an empirical sense but dependent on the moral agent in a
transcendental sense, making Kant an empirical moral realist but a transcendental moral idealist. Or some moral principles, properties, or objects might be
transcendentally real, that is, entirely independent of the transcendental moral
agent, in which case, Kant would be a transcendental moral realist.17
For this overall transcendental/empirical distinction to work in practical
philosophy, the concept of empirical experience, or the everyday experience
of human beings, must be understood to have a practical dimension in addition
to the theoretical dimension provided in the analytic of the rst Critique. Given
Kants interest in anthropology as a natural level at which we can understand
human beings as moral beings, a practical dimension of experience in Kant is
at hand. Kant discusses our consciousness of ourselves as moral in relation to
our moral predisposition in the Anthropology (7:324). Even in the less obviously empirical works, Kant discusses everyday moral experience. He begins
the Groundwork by invoking common moral cognition or beliefs actually held
17

One might be further tempted to conclude that a transcendental moral realist would have to be
an empirical moral idealist, in which case, some moral principles, properties, or objects would
be not directly known in experience. As with Kants claim that with regard to objects in space, a
transcendental realist ends up as an empirical skeptical idealist, this translates into an
empirical moral skepticism. Suppose that the moral value of a rational agency were understood
to be transcendentally real, that is, independent of the transcendental moral agent qua agent. The
empirical moral agent might have no transcendental justication for a recognition of this moral
value, since it is independent of that moral agent considered transcendentally. Hence, on the
empirical level, an empirical moral agent might need another argument that would show the
actuality of those moral values. For an argument along these lines, see (Sensen 2011,
pp. 1820).

Moral realism and naturalism

21

by people (4:393); in the second Critique, he takes his project not be a

replacement of an old principle of morality with a new one but a philosophical
justication of it through a new formulation (5:8, ftn.). In other words, he is not
denying that the principle of morality is part of the actual experience of human
beings. Moral duties and values are experienced by human beings in nature,
that is, human beings as appearances. Given the possibility of empirical moral
experience, the controversial interpretive issue relevant to current purposes is
whether morality would be empirically real or empirically ideal, that is,
independent of or dependent on the empirical moral agent.
In order for this empirical/transcendental distinction to apply to ethics, we
have to be able to identify transcendental arguments for the necessary conditions for a moral agent. Kant attempts to provide a deduction of the moral law
in Groundwork III and later discusses a deduction in the Critique of Practical
Reason. Although Kant does not identify them as such, these deductions have
been treated by commentators as transcendental.18 If it is proper to consider
them as transcendental deductions, then they might be understood as the basis
for claims that the moral properties and objects are dependent on the transcendental moral agent. Further, one might be tempted to treat Kants derivation of
the value of humanity as a kind of transcendental deduction (as Korsgaard does
without using that term19), even though Kant never calls it a deduction.
Understanding these issues requires a look at the nature of Kants philosophy
in general and practical philosophy in particular. Chapter 2 will examine these
issues in detail.
The importance of this discussion of the empirical/transcendental distinction
is that there can be two levels of possible moral realism or idealism. At the
transcendental level, one might ask whether moral principles, properties, or
objects are dependent or independent of some possible transcendentally identied moral agent, that is, an idealist might argue that autonomy of the will as
pure practical reason might be a transcendentally ideal ground for the categorical imperative, or a realist might argue that humanity as an end in itself should
be understood as a transcendentally real value independent of the transcendental moral agent. There is a third option besides moral transcendental realism
and idealism: There might be no transcendental justication for morality at all.
If human moral experience is dependent upon some contingent features of the
human moral agent and cannot nd any transcendental basis, Kants ethics
would be empirically ideal but would have no transcendental status, in the

18
19

See, for example, the articles by John Rawls, Henry Allison, and Barbara Herman in (Frster
1989).
See (Korsgaard 1996a, p. 123). Recently, Robert Stern has endorsed a version of her transcendental argument (Stern 2011a).

22

Laying the ground

same way that color sensations are empirically ideal and have no transcendental status (B4445).
At the empirical level, an empirical realist might argue that the transcendentally ideal ground for the law might be compatible with the empirical reality of
duties such that those duties are not dependent upon the empirical moral agent
but only the transcendental moral agent. This distinction might prove useful in
dissolving some strong disagreements about whether the categorical imperative is independent of rational moral agents. The effect of the transcendental/
empirical distinction on the realism/nonrealism debate can come only after a
look at the nature of philosophy in Kant in Chapter 2.
Allow me to briey note the importance of the distinctions made so far.
A standard denition of moral realism has been shown to be inadequate in
capturing the elements peculiar to Kantian ethics. A denition that stresses
the role of the metaphysical independence of moral principles, properties, or
objects from either the transcendental or empirical moral agent serves better.
Kants moral theory the target of discussion in this book will be
subjected to a more appropriate denition that illuminates the genuine fault
line between Kantian realism and Kantian nonrealism in terms of independence or dependence on the moral agent. The independence or dependence
appears at two possible levels when Kants transcendental/empirical distinction is applied. If there is a transcendental dependence of all moral principles, properties, and objects upon the transcendental moral agent, Kant
would be a transcendental moral idealist; if the transcendental justication of
morality requires that some moral principles, properties, or objects be independent of the transcendental moral agent, then Kant would be a transcendental moral realist. A similar distinction would operate at the empirical level
of everyday moral experience with reference to the empirical moral agent.
The conception of the human moral agent and the status of Kants justicatory arguments for morality will have to be determined in detail in order to
resolve this dispute.
This fourfold division crossing empirical/transcendental and real/ideal is
still too stark a choice for interpreting Kant. There can be interpretations that
hold Kant to be a transcendental realist regarding some moral principles,
properties, or objects and a transcendental idealist regarding others. No one
takes Kant to be a realist about every moral principle, property, or object and
few take him to be a complete idealist. Philosophically, it is more fruitful to
avoid slapping a blanket identication on Kant as one or the other and instead
to assess all aspects of Kants moral theory in light of the issue. I will devote a
later section of this chapter to listing the various kinds of moral principles,
properties, or objects that are raised in the literature. Before that, however,
I will evaluate various conceptions of constructivism to show how they t into
the realism/nonrealism debate I have just characterized.

Moral realism and naturalism

23

Comparison with constructivism

No single denition of constructivism garners universal or even wide
acceptance. Kantian moral realists tend to treat all opponents as constructivists,
but some identify themselves otherwise.20 The positions are further complicated because some constructivists label themselves moral realists while others
seem to reject realism.21 And constructivism itself is an unfortunate term to
use when discussing Kants philosophy because he had a particular meaning
for the term in his theoretical philosophy that conicts with its use by contemporary Kantians in ethical theory.22 In this section, I will rst review two
possible general types of construction, showing which ought to be taken as
20

21

22

Robert Stern labels the debate about Kants moral realism the constructivist/realist controversy (Stern 2012, p. 119). Paul Formosas very title makes the assumption: Is Kant a Moral
Constructivist or Moral Realist? (Formosa 2013). Although I have described my position as
idealist and never as constructivist, I have been lumped with others as a constructivist several
times (Kain 2006, Stern 2011b), (but see DallAgnol 2012b for a notable exception). Karl
Ameriks, in contrast, is to be praised for his argument against two nonrealist interpretations of
Kant, those of J.B. Schneewind and Charles Larmore, in which he does not see them primarily
as constructivist (Ameriks 2003).
In part, this is due to their differing conceptions of moral realism. The most famous of the selfproclaimed constructivist realists is Christine Korsgaard, who distinguishes Kants procedural
moral realism from metaphysical realism, which she calls substantive moral realism
(Korsgaard 1996b, p.35) or objective realism (Korsgaard 1996a, pp. 278282). She claims
that procedural moral realism is the view that there are answers to moral questions; that is, that
there are right and wrong ways to answer them. Substantive (or objective) moral realism
further holds that there are moral facts or truths which account for the right answer to those
questions. This claim implies that any method of answering a question about right and wrong
counts as a procedure. She seems to conrm this a few pages later when she claims that
substantive realism, which holds that moral facts are true in virtue of something independent of
the moral agent, is a version of procedural realism. This understanding of procedure would be
itself too broad. The universality of the term procedure when taken to include any method of
answering questions makes it useless when trying to nd something distinctive about proceduralism. Her description would encompass a procedure of logical deduction from a set of
statements, which is of course one way to try to nd right or wrong answers. She later
characterizes constructivism as a form of problem solving, making it appear that all attempts
to nd answers to practical problems are constructivist, although without invoking the term
procedure (Korsgaard 2003). In these statements, Korsgaard effectively erases any meaningful distinction between nding an answer and creating an answer. Any attempt to answer moral
questions by reference to independently existing moral facts would still count as constructivist.
Hence, Korsgaard is not a good source for an account of the distinctiveness of a procedure for
construction.
In the Critique of Pure Reason, he distinguishes philosophy from mathematics when noting that
concepts are constructed in mathematics but not constructed in philosophy (A71316/
B74144). To construct a concept is to exhibit a priori the intuition corresponding to it. Only
mathematics can construct concepts because philosophical concepts cannot be represented in
intuition. A mathematically constructed concept is a specication of the universal, for example,
the properties of triangles in general, in a particular, say, a specic triangle drawn on paper or in
the imagination, because the concept already contains a pure intuition that needs to be exhibited.
Philosophical concepts, in contrast, are not constructed because they do not contain any pure
intuition but only nothing but the synthesis of possible intuitions which are not given a priori

24

Laying the ground

paradigm, before showing how my approach is situated in relation to constructivist theories.

At its most general level, a constructivist theory claims that some particular
moral principles, properties, or objects such as norms, values, or moral facts,
are derived from some other particular principles, properties, or objects,
whether purely moral, for example, a highest value, or not, say, rationality
itself, through a procedure of some sort.23 Rawls initially used the term
procedure in A Theory of Justice as a way to understand the process by
which individuals would determine principles of justice by means of the
original position.24 Constructivists who had studied with Rawls adopted the
term, and even Onora ONeill, who sharply differentiates her constructivism
from Rawls, sometimes characterizes her position this way.25 Yet the specic
procedures they invoke are not of the same type. I see two general types of
constructivism in the literature. I will suggest that the one that is closest to the
approach I have to moral idealism ought not to be considered to construct
anything.
Thomas Hill offers a Rawlsian understanding of the nature of a procedure in
what he calls a procedure of construction. In this type, moral principles (for
Hill, the target of construction) are valid just in case and because they would
be endorsed by all members of an appropriately dened initial choice

23

24
25

and hence can relate to objects in intuition only synthetically and discursively (A71920/
B74748). In the Critique of Practical Reason, Kant makes an analogous claim with regard
to the practical in a discussion of how to separate the doctrine of happiness from the doctrine of
morals. He suggests that making this distinction requires the care and precision of a geometer.
A philosopher, however, has greater difculties to contend with here (as always in rational
cognition through mere concepts without construction of them) because he cannot put any
intuition (a pure noumenon) as its basis (5:92). Presumably, Kant is making an analogy
between the pure intuition used in geometry with any pure conceptual space (noumena) that
would be used in practical philosophy. Practical philosophy cannot construct its concepts in a
pure conceptual space but must show how they affect actual human beings in their actual
decisions. To resolve this, Kant suggests that practical philosophy can use a method similar to a
chemist who adds alkali to a certain solution in order to separate the acid from the remainder of
the solution. The experiment that Kant suggests is that in a particular person awareness of the
moral law will similarly separate considerations of advantage from rational considerations in the
persons mind, allowing for a focus on moral motivation. For a thorough examination of this
particular analogy, see (Rohden 2012). Like other claims stemming from reason, for example,
the concept of a soul, the moral law is not constructed in Kants sense of the term.
Sharon Street believes that inclusion of a procedure is not a necessary element of constructivist
views (Street 2010, p. 366). Constructivism is better characterized as a metaethical theory that
stresses the practical point of view. Ultimately, I agree that theories that stress the practical point
of view are superior theories, but I hesitate to label them as constructivist and instead use the
term idealist. Street denes a constructivist position in terms of the justication of some moral
judgments by reference to another set of moral judgments (Street 2008, p. 208).
(Rawls 1999a, p. 17).
ONeill does not generally describe her position as proceduralist but has used the term (ONeill
1989, p. 216, n. 10). Korsgaard uses the term quite prominently when she characterizes her
position as procedural realism (Korsgaard 1996b, p. 35).

Moral realism and naturalism

25

situation.26 I will call this the decision procedure type of constructivism.

The crucial element of this denition is the requirement that a decision in an
initial situation, one that provides some material the agents may use while
deciding, is used to determine the result. This kind of procedure could produce
various kinds of moral principles, properties, or objects, not merely principles,
from some other set of moral or nonmoral principles, properties, or objects.
ONeills discussion of the construction of principles by agents seeking agreement is another example.27 An exemplary mainstream denition that stresses
the role of choice stemming from an idealized process of deliberation is given
by Carla Bagnoli.
As a metaethical account an account of whether there are any normative truths and,
if so, what they are like constructivism holds that there are normative truths. These
truths are not xed by facts that are independent of the practical standpoint, however
characterized; rather, they are constituted by what agents would agree to under some
specied conditions of choice.28

The crucial characteristic of the decision procedure is that it entails some

ability to make a decision, whether real or hypothetical, on the part of the
participants.29 A procedure of this kind differs from a procedure in which the
creation of the moral principles (or norms or values) conceptually precedes
any capability for agents to make decisions. In other words, this type of model
of a construction procedure requires that the agents in the hypothetical or real
situation can be conceived as capable of making a decision to endorse or reject
the outcomes of the procedure. Neither agency itself nor any prerequisites for
agency can be constructed through a decision procedure because one needs
agents in order to make decisions in the rst place.
26
29

27
28
(Hill 2012, p. 78).
(ONeill 1989) and (ONeill 2002).
(Bagnoli 2011).
Paul Formosa calls actual constructivism all the way down because it involves actual willing
by individuals either collectively as a culture or individually. Hypothetical constructivism is
not all the way down because both the content and the authority of the procedure itself are
laid out rather than the subject of some actual choice (Formosa 2013, pp. 17374). Formosa
does not provide any example of an all the way down constructivist but suggests that a model
would be someone who takes moral norms to be embodied in cultural practices but nonetheless
subject to the individual or collective act of will. This characterization is problematic because
if the norms are embodied in the cultural practices, they are not the result of any direct acts of
will at all. If he means that the acts of will are those of an agent independently endorsing the
cultural norms, then his characterization would not capture the spirit of constructivism in
particular but would be applicable to any acts of an agent who faces the question of whether
to subject himself or herself to moral norms. Formosa does not discuss actual procedures in the
remainder of his article but defends a not all the way down reading of Kant.
An example of an all the way down constructivism might be what Andrews Reath calls the
Principle of Individual Sovereignty in which each particular will is subject only to laws it
actually legislates for itself. Reath and Patrick Kain cite problems with this view on its own
grounds and as an interpretation of Kant and the two philosophers to whom they attribute this
view, Robert Paul Wolff and Rdiger Bittner, do not themselves take it to be Kants view. See
(Reath 2006, pp. 9798), (Kain 2004, pp. 26265), (Wolff 1974), and (Bittner 1989).

26

Laying the ground

A kind of constructivism that alleges to construct moral principles, properties, or objects on the basis of agency itself is thus the second kind of
constructivism in the literature. What I will call the nature of agency procedure allows for the generation of norms, laws, or values by means of the
expression of the nature of moral agents conceptually prior to their ability to
choose anything. This procedure does not involve conscious endorsement,
whether real or hypothetical, by agents but is simply the product or characteristics of a particular kind of active being. Korsgaards attempt to ground the
categorical imperative on rational beings very ability to make reective
decisions is one example of this kind of procedure. Although she has claimed
to construct agency itself, the actual procedure she offers is one in which she
takes rational agency as the starting point from which she then draws out the
categorical imperative and moral value.30 The main question asked is about
which moral norms are intrinsically tied to the nature of free agency itself.
Certainly, the agents might later reect on their situation and endorse the
results of the activity, but this reective endorsement is not the source of the
construction.
30

Korsgaards position is more ambiguous than stated here. She sometimes describes her position
as one of a constitution of agency, but when giving details, she shifts to a constitution of agents
as particular individuals who use their ability as agents to make choices that dene who they are
(Korsgaard 2008, p. 109). It seems clear that she does not construct agency as such but uses
agency to construct other elements of her moral theory.
She also sometimes argues that the categorical imperative is a result of a decision by an agent to
be a certain kind of person, for example, one committed to the kingdom of ends or to egoism
(Korsgaard 1996b, p. 101). At other times, she insists that acceptance of the categorical
imperative is a requirement stemming from the reective structure of human consciousness
and thus not a result of a decision by such an agent (Korsgaard 1996b, pp. 10304). The former
approach is what I call decision procedure and the latter nature of agency procedure.
Some opponents of constructivism claim that Korsgaards constructivism is not only not a
genuine alternative to moral realism, it is also not even a metaethical claim. Nadeem J.Z.
Hussain and Nishi Shah argue that Korsgaards constructivism makes no metaethical claims but
is better described as making claims in normative ethics or moral psychology (Hussain and
Shah 2006). Allen Wood hints that no distinctive metaethics or metaphysics of value is
implied by the procedural account of practical reason but does not offer details (Wood 2008,
pp. 28283, n.3). I nd these charges difcult to accept. Korsgaard and other constructivists
might not have offered a clear and internally consistent metaethics, but they have made some
metaethical claims. The constructivist insistence, for example, that value is itself not a real
property of objects or persons but results merely from a particular practice guided by reason is a
claim about the nature of value itself rather than about what particular things are of value, the
former being a metaethical and the latter, a normative claim. The difference between holding
that value is a real but nonnatural property of objects and holding that value is merely a
projection onto objects is metaethical since both can identify precisely the same objects as
being of value. And if expressivist theories that tie moral claims to emotions operate at a
metaethical level, why should not a theory that ties ethical claims to a particular conception of
moral agency or to the determinations of a particular mental faculty? Not only can the
metaphysical question differentiate constructivism from other metaethical theories, some have
also claimed that constructivism makes distinctive epistemological metaethical claims (Bagnoli
2012).

Moral realism and naturalism

27

I believe that the nature of agency procedure is the better constructivist

procedure. It is, however, lacking much of a procedure. It derives moral
principles, properties, or objects from the nature of a particular kind of being
or kind of activity, and such moral principles, properties, or objects can better
be viewed as intrinsic to that kind of being or components of that kind of
activity rather than something created by it. The term constructivism is
roughly applicable because the moral principles, properties, and objects are
conceptually derived from the nature of moral agency, but they are not
constructed in any procedure that involves choices.
As a nonrealist view, the nature of agency model is very promising. Recall
that the claim to moral nonrealism requires that all moral properties and objects
be dependent upon the moral agent. The nature of agency approach treats this
dependence as inherence in the structure of the moral agent. It allows the
debate to focus on the kind of mind that transcendental or empirical moral
agents have. As this book proceeds, my main argument will emphasize the
nature of the agency approach. I prefer to use the term idealism rather than
constructivism to describe this view in order to sharply differentiate it from
the decision procedure oriented kind of constructivism that relies on the
choices of agents.
The terminology to be used in the remainder of this book can now be xed.
Since no interpreters take Kant to be offering either an error theory or
expressivism, I will not use the broad term nonrealism in reference to Kant.
I will also avoid the term constructivism in my interpretation except in
reference to others who identify themselves as such since at best it is ambiguous among the two types I have identied. For my focus on the nature of
agency, I will use the term idealism. The question to be answered is whether
and to what extent Kant is a moral realist or moral idealist.
Metaphysics and naturalism
Kant himself does not emphasize naturalism as a topic of philosophical
discussion and as a result does not provide detailed explanations and justications for his rare uses of the term.31 In some ways, my examination of Kant
31

Kant rarely uses the terms naturalism or naturalist at all, and even less in ways which would
dene a philosophical program related to contemporary understandings of the term. The few
occurrences fall into three types, which correspond to metaphysical naturalism, methodological
naturalism in knowledge claims, and method in theology.
a.
The rst type metaphysical use of the term naturalism occurs most clearly in the Prolegomena in a discussion of the cosmological ideas: The cosmological ideas, through the manifest
inadequacy of all possible cognition of nature to satisfy reason in its rightful demands, serve to
deter us from naturalism, which would have it that nature is sufcient unto itself (4:363).The
idea of a naturalist as taking nature to be sufcient unto itself provides a picture which accepts

28

Laying the ground

might seem to be anachronistic because of this lack of attention. But this issue
is important in that it highlights the lack of metaphysical commitments in
Kants moral theory. By framing my interpretation in naturalistic terms, I am
able to highlight what I take to be a central characteristic of Kants metaethics,
namely, the importance of the practical standpoint as an action-oriented facet
of experience that is largely indifferent to theoretical claims about ontology.
Kant is generally taken to require some strong metaphysical commitments, or
at least commitments that lead to metaphysical claims, in his moral theory:
Freedom of the will most prominently, God and immortality, moral value
claims, a conception of pure rational agency and the resulting need for a view
of the self as other than merely appearance. Many contemporary ethicists
inspired by Kant shy away from these claims, particularly the roles of God,
immortality, and the metaphysics of persons-in-themselves. These metaphysical commitments are largely, though certainly not fully, out of favor in
contemporary philosophy. Much contemporary work does stress the practical
no entities other than those required by the natural sciences. This meaning is also implied in
Kants use of the term in one of his earliest works. The terms naturalism and naturalist occur
in the 1755 Universal Natural History four times (1:222, 1:223, 1:224, 1:239) without denition, but in context, their meaning is one who considers nature sufciently explained without
recourse to divine causes.
b.
The naturalist of pure reason is mentioned in the Prolegomena and the Critique of Pure
Reason in a way related to method, but quite different from any contemporary denition. In the
Prolegomena, the naturalist of pure reason is dened as he who trusts himself, without any
science, to decide in matters of metaphysics (4:314). The discussion in the Critique of Pure
Reason accords with the Prolegomenas denition. In the very nal page of the Critique, Kants
discussion of the history of pure reason turns to the question of method: The naturalist of pure
reason takes as his principle that through common understanding without science (which he
calls sound reason) more may be accomplished with regard to the most sublime questions that
constitute the task of metaphysics than through speculation (A855/B883). Kant apparently
offered this denition of naturalist in his lectures on logic as well; the Dohna-Wundlacken
Logic of 1792 quotes him as saying A naturalist is one who acquires cognitions that do not
constitute a system. Metaphysics and morals are the hobby horses of such people (24:783). The
methodological naturalism Kant rejects the naturalist of pure reason is far removed from
contemporary denitions of methodological naturalism because this naturalist is said to reject
rather than accept science.
c.
A few of the references concern theology. In the 1788 essay On the Use of Teleological
Principles in Philosophy, Kant notes that his explanations follow natural science so carefully
that he is assumed to be a naturalist (presumably in the rst sense) by one reviewer but rejects
that label as more properly belonging to theological discussions (8:178). The term appears in the
Religion, where Kant denes a naturalist as one who denies the reality of any supernatural
divine revelation (6:154), and in the Conict of the Faculties, where naturalism is dened in
religion as ecclesiastical faith without the Bible (7:60).
Graham Bird (Bird 1995) also canvasses Kants use of the term, but lists only the occurrences of
naturalism and not naturalist as well. John Zammito (Zammito 2008) refers to Birds paper
for the list of Kants uses of the terms but provides a different summary of their meaning.
Zammito sees Kant as associating naturalism with two impulses: First, a claim that nature is
ontologically self-sufcient, and second, the use of common sense or healthy reason in
place of abstruse philosophical reasoning (Zammito 2008, p. 545).

Moral realism and naturalism

29

standpoint, but there has not been a thorough argument that Kants ethics as he
himself presents it in full does not require any ontological commitments. The
clearest way to frame this issue is in terms of a metaphysical naturalism. The
relations between the claims about the practical standpoint and claims about
realism and idealism also make a discussion of naturalism a good topic to
pursue for the light it sheds on the nature of Kants idealism in ethics.
Naturalism is a relatively recent term in philosophy, although what the
term signies has been part of philosophy for millennia. My discussion of the
term here is limited to the goal of providing a conception of naturalism
appropriate for an assessment of whether Kant can be considered a naturalist
in his ethics. Many aspects of the contemporary debate about naturalism in
philosophy will be left out of this discussion.32 I leave it to others who come to
different conclusions about the denition or viability of naturalism to evaluate
Kant on their own terms; however, I believe that my denitions capture the
kind of naturalism that is of most concern to Kant.33
Naturalism prioritizes the sciences over philosophy or any other discipline,
and it does so in two ways, metaphysically and methodologically.34 By
metaphysical naturalism, I mean the type of naturalism which accepts as
real only entities studied by the sciences. By methodological naturalism,
I mean the type of naturalism which accepts as sound only the procedures for
acquiring knowledge used in the sciences.
Since Kant perhaps more than anyone in the history of philosophy since
Plato is responsible for the inclusion of the a priori as a central element of
philosophy, it is clear that Kant rejects methodological naturalism.35 The
transcendental method is also at odds with a methodological naturalism. My
32
33
34

35

I argue in detail that Kant can be conceived as a naturalist to a large extent, encompassing not
only physics and chemistry but also the social sciences, in (Rauscher 2009b).
For recent discussions of naturalism, see (Ritchie 2008) and (De Caro and Macarthur 2004 and
2010).
Joel Smith and Peter Sullivan see three types of naturalism: Ontological (i.e. metaphysical),
methodological, and epistemological, which holds that knowledge is a natural phenomenon, but
they admit that the last is closely related to the rst two and can be understood for the most part
in those ways (Smith and Sullivan 2011, pp. 1014). Mario de Caro and David Macarthur
initially follow the division into ontological (metaphysical) and methodological naturalism
before suggesting a third type of naturalism, semantic naturalism which claims that only
the concepts employed by the natural sciences are genuine concepts (De Caro and Macarthur
2004, pp. 37). I believe that this semantic naturalism can be understood in terms of methodological naturalism and will not pursue it as a distinct form of naturalism in relation to Kant. Kant
has been interpreted semantically by others, although not as a semantic naturalist. See (Loparic
2005) and (Hanna 2001).
Michael Friedman has characterized philosophical naturalism as a two-pronged thesis, both
prongs of which are methodological: First, the rejection of a priori knowledge, and second, the
placement of philosophy among the empirical natural sciences. Friedman devoted his
1996 Presidential Address to the American Philosophical Association to his argument that
naturalism so understood has reached the end of its useful life (Friedman 1997, p. 7).

30

Laying the ground

interpretation will argue that the methodological antinaturalism in Kants ethics

does not require a metaphysical antinaturalism. Chapter 2 will look in detail at
Kants methodology in philosophy as a whole as well as in ethics in particular.
The limits of the sciences set the terms for metaphysical naturalism. The
naturalist ontology is tied to entities studied by the sciences, in particular,
physics. In order to specify this more precisely, one must be able to identify the
sciences and explain how their own domains are determined. That requirement
in itself is a methodological issue, but it is also possible simply to take the
sciences to refer to whatever the particular accepted sciences are. Contemporary naturalism generally centers on the ontology provided by physics. This
can lead to questions about the value and possibility of pursuing other disciplines that cannot be reduced to, or at least supervene on, the entities and
properties required by physics. For my purposes, it is sufcient to identify
the precise sciences and their objects as Kant understood them and the
resulting metaphysics of nature, for my project is determining whether Kants
ethics requires any nonnatural metaphysics on his own terms. The importance
of specifying what Kant means by nature is seen when reviewing previous
work on naturalism in Kant.
Work on Kant and naturalism
Graham Bird published the paper Kant and Naturalism in 1995 to defend
Kant against the view that Kant could not accept naturalism. Although Bird
suggests that the term naturalism is unhelpful, he takes Kant to have the view
that I am calling metaphysical naturalism: It is not that there really are two
distinct realms, a natural, empirical, realm and a nonnatural, transcendent,
realm of things-in-themselves; rather there is just one realm, empirical
reality, and a conceptual apparatus which tempts us to transcend it.36
John Zammito raised the issue a dozen years later in his Kant and Naturalism
Reconsidered in a discussion that placed Kants methodological

36

Bird follows this remark with a claim that Kants approach is acceptable to a naturalist or
antinaturalist, although by those terms he is referring to the methodological approach as well in
which the scientic approach needs a philosophical supplement (Bird 1995, pp. 40608). Bird
is responding to a claim by John Skorupski that Kants transcendental idealism is an antinaturalism. Part of Birds complaint against Skorupski is that naturalism is ill-dened. Despite a
vague denition of naturalism, Skorupski sums up his own view that Kant cannot be one quite
well in this passage:
Whatever else is obscure about Kants transcendental idealism, one thing is clear it involves
the rejection of naturalism . . . The great cultural inuence of Kants Critique stems from this:
that it upstages the natural attitude the perspective of common sense and its outgrowth,
science. If Kant is right, the natural attitude cannot give the absolute truth about the world and
our situation in it. (Skorupski 1990, p. 7)
Skorupski clearly takes transcendental idealism to be a methodologically antinaturalistic thesis.

Moral realism and naturalism

31

antinaturalism, in the form of a priori reasoning, in historical context as a

reaction not only to Hume but also to the common sense theorists in Germany.
Zammito takes Kants transcendental idealism not to be metaphysically
innocent but to involve such a strong role for reasons own self-activity that
it appears to move beyond a metaphysical naturalism through determining its
objects through its own nature. There is more than a whiff of the transcendent
in Kants transcendentalism, he concludes.37 Zammito and Bird do agree that
a metaphysical naturalism for Kant would be one limited to empirical reality.38
Konstantin Pollok more recently argued that Kant cannot be a naturalist,
either methodologically or metaphysically.39 He understands metaphysical
naturalism in relation only to the mathematical sciences, thus excluding Kants
empirical psychology, and then invokes causal closure to argue that Kant
restricts causal explanation to physical explanation. Pollok focuses on the
Third Antinomy in claiming that Kant requires a different causal explanation
other than physical causality to account for the actions of the mind. The
problem with this approach is that in excluding the objects of empirical
psychology from nature, Pollok applies a restriction that Kant himself did
not use because Kant did consider inner sense to be a part of causally
determined nature even if he did not think that psychology could rise to the
level of a science, as I will explain later on in this chapter.
Work specically about naturalism in Kants ethics is rare. An article with
the promising title Kant as Ethical Naturalist from the early 1970s is disappointingly limited to the claim that specic duties cannot be derived directly
from the categorical imperative but must involve knowledge of human nature; it
offers no general denition of naturalism in Kant.40 More recently, Paul Guyer
offered a reading of Kants ethics that argued that the precritical Kant had a
naturalistic understanding of morality; Guyer does not dene naturalism at
all, but it appears that he had in mind the use of a psychological account of
action, which can include both methodological and metaphysical naturalism.41
37

38

39
40

41

(Zammito 2008, pp. 540, 55152).The difference between Birds claim that Kant is compatible
with metaphysical naturalism and Zammitos that he is not, lies in their understandings of the
role of reason and transcendental methodology, a topic I will cover in Chapter 2.
Another discussion of naturalism in Kant that cuts across the metaphysical/methodological
divide is given by Sami Pihlstrm, who presents a naturalistic pragmatic account of transcendental philosophy in (Pihlstrm 2001).
(Pollok 2013).
The article was appropriate for its time since before the 1980s, much of the literature on Kants
ethics limited itself to the Groundwork even to the extent that Kants four examples of specic
duties provided in the explanation of the categorical imperative were misunderstood as actual
derivations (Dye 1978).
Guyers essay (Guyer 2007b) offers some textual support for his claim that Kants precritical
ethics considered this naturalistic approach, but as Henry Allison notes in his comments to the
essay, it does not reect Kants more consistent precritical position that does not rely on a
psychological foundation (Allison 2007).

32

Laying the ground

Darlei DallAgnol offers an assessment of the Kantian metaphysical naturalism

that I presented in an earlier paper. He takes a metaphysical naturalist to be
one who asserts that there are no facts besides natural facts.42 Regarding my
claims about ethics, he objects that Kant insists upon the nonnatural freedom of
a noumenal self and the nonnatural fact of reason. In my particular discussions
in later chapters, I will assess these specic points, but here I will point out that
the criteria for determining what counts as a natural fact would be dependent
upon a prior understanding of nature itself: If morality is part of nature then
moral facts are natural facts. And I will show that a moral fact derived from a
priori methodology can be perfectly consistent with metaphysical naturalism
when moral facts are understood to not make ontological but only normative
claims.
The most direct discussion of the possibility that Kant is a naturalist in ethics
comes from Allen Wood in his commentary on Guyers article.43 Wood
focuses on ethics but rst provides some discussion of naturalism in general;
unfortunately, his explanations fail to pinpoint acceptable criteria for a metaphysical naturalism. He provides what he calls an acceptable pious sense of
naturalism: A naturalist is someone who rejects a supernaturalist view of the
world. By itself, this denition is unhelpful because it fails to specify any
particular criterion for delimiting the natural from the supernatural. Wood
provides some examples of what he takes to be supernatural God or gods,
Cartesian immaterial substances, Platonic Forms, incorporeal monads, and
noumenal selves that suggest an ontology, but others vague talk about
transcendence or disgusting touchy-feely spirituality appear to relate
to attitudes rather than ontology. He does rmly assert that for a naturalist,
there is only matter and the void. One might take this kind of naturalism to
preclude nonscientic approaches to knowledge claims, in particular those
which might have ontological implications. But Wood argues that since the
current sciences cannot answer and are not even close to solving certain
traditionally philosophical problems such as consciousness, normativity, and
agency, philosophical views on these matters ought not to be dismissed as
outdated superstitions.44 This allowance leaves space for a philosophical

42

43
44

DallAgnol (DallAgnol 2013) is responding to several papers of mine, in particular the paper
Kant as Metaphysical Naturalist (Rauscher 2009b) but also a paper on the postulates
(Rauscher 2007), which forms the basis of Chapter 5, and Pure Practical Reason as a Natural
Faculty which was published in Portuguese (Rauscher 2006a), parts of which appear in
Chapter 4. He identies the scope of my claim too broadly to encompass a complete metaphysical naturalism instead of one limited to ethics.
The quotations that follow are taken from (Wood 2007, pp. 47274).
This claim is puzzling in light of Woods rejection of supernatural ideas that are used in place of
natural explanations. He admits that there is some mystery in life and even allows that pious
naturalists ought to respect and tolerate supernatural ideas. He holds that philosophers ought to

Moral realism and naturalism

33

nonnatural methodology but does not in itself preclude philosophical claims

with supernatural ontologies. The lack of a specication of the scope of the
natural is a weakness in Woods approach when applied to Kant. Wood takes
noumenal selves as an example of a supernaturalist ontology, so we can
conclude that Wood takes a two-world transcendental idealism to be nonnatural. Yet, he takes the human rational will to have value as an end in itself as
part of a realist metaphysics of value that is entirely compatible with
naturalism in the pious sense, casting doubt on the claim that the metaphysics
of his naturalism is only matter and the void. I will have much more to say
about the metaphysical status of moral value in Chapter 7. At this point,
I conclude only that Woods discussion of naturalism is insufcient to specify
the parameters of naturalism in Kant.
None of the previous discussions of metaphysical naturalism in Kant are
specic enough to provide the kind of ontology needed for the assessment of a
naturalistic ethics. In what follows, I will show how metaphysical naturalism
in Kant should be specied.
Specifying Kants nature
Kantian metaphysical naturalism will be limited to entities in space and time
subject to deterministic law and whatever properties of those entities that can
be understood to depend on the entities in space and time, even if those
properties are not themselves subject to deterministic laws. This naturalism
will encompass objects of both outer and inner sense in such a way that the
human mind in empirical consciousness will be comprehensible in a naturalistic empirical psychology.45 To be very clear: I am not here arguing that Kant
holds to metaphysical naturalism in his overall philosophy, although I think
that such a reading has merit.46 The aim of this book is limited to laying out the

45
46

be interested in them, at least to the extent of seeing which can be made intellectually
respectable. Yet at the same time, Wood denounces those who hold supernatural beliefs for
which there is no proof or evidence and even claims that people who take the mystery of life to
open the way for such ideas as guilty of intellectual self-contempt. This dichotomous
approach would need to be grounded on a criterion that can be used to distinguish intellectually
respectable from nonrespectable supernatural ideas; I doubt that any methodological naturalist
could provide such a criterion. There appears to be no naturalistic way for Wood to distinguish
philosophically acceptable explanations that operate without reference to nature and philosophically unacceptable beliefs that do the same (Wood 2007, pp. 47273).
I argue at length for this position in (Rauscher 2009b). Here, I provide a very short summary of
the main points.
There is a plausible interpretation of Kant to that effect given by Graham Bird in The
Revolutionary Kant and elsewhere. Bird offers a methodological understanding of
transcendental idealism in which Kants critical project is seen as opposed to traditional
philosophy by employing a critical method to evaluate knowledge claims made in nature. Bird
stresses the empirical reality of objects in space and time and holds that Kant does argue for the

34

Laying the ground

elements of Kants ethics that under a plausible interpretation would be

compatible with a metaphysical naturalism.
Physics is the quintessential natural science for Kant (Natural Science,
4:472) and consists of cognitions of corporeal nature, that is, spatiotemporal
objects. Cognition of these spatiotemporal objects requires both concepts and
intuitions. The objects themselves are not limited to actual intuitions and
empirical concepts but are objects of possible experience. Possible experience
as the subject matter of physics cannot be equated with actual intuitions, since
possible experience includes (1) objects of possible intuitions which have not
occurred (the matter of possible experience), and (2) a priori principles of the
understanding which govern all possible intuitions (the form of possible
experience) (B147148).
The principles of the understanding are clear, at least, relative to anything
else in Kant: He clearly delineates what they are in his table of categories
(A80/B106) and corresponding table of principles (A161/B200). The scope of
possible intuitions is less clear. The way to identify possible intuitions is given
by Kants discussion of the postulate of actuality when Kant allows that
magnetic matter is actual although we can have no direct intuition of it,
where he concludes that Thus wherever perception and whatever is appended
to it in accordance with empirical laws reaches, there too reaches our cognition
of the existence of things (A226/B273). Magnetic matter is part of possible
experience because its existence is dictated by empirical laws applied to
particular perceptions of the behavior of iron lings.47

47

existence of any transcendent objects. He cites many earlier commentators with similar revolutionary views that differ in their particular interpretations, among them Henry Allison
(Allison 2004), Arthur Collins (Collins 1999), and Paul Abela (Abela 2002a) (for his complete
list, see (Bird 2006,p. 776, n 18)). Allison does not see Kant as a metaphysical naturalist, at least
with regard to agent causation (see especially [Allison 1997]), but Abela does think that Kants
ontology need not extend at all beyond the empirically real to any things in themselves, at least
for theoretical purposes (Abela 2002a, pp. 29293). Collins takes Kant to be an empirical realist
and tries to dispel the belief that Kant is an idealist of any ontological kind. Birds book does not
use the term naturalism to a great extent but does embody metaphysical naturalism nonetheless (Bird 2006, pp. 8392). Two discussions that hold that Kant provides arguments against
metaphysical naturalism in the antinomies are (Falkenburg 2004) and (Pollok 2013). Otfried
Hffe also argues that Kant provides a critique of naturalism but limits his discussion to
methodological issues (Hffe 2010, pp. 188191).
Paul Abela interprets this passage as an admission by Kant that there can be hidden truths
truths which are inquiry transcendent that would remain outside the set of truths that would
emerge even at the end of inquiry (Abela, 2002a, p. 238). He must be operating with a crucial
assumption, namely that the laws of nature which we use to posit these unobservable entities are
not themselves part of the inquiry, even the ideal end of inquiry which for Kant would equal
the rules and concepts of the understanding ideally systematized by reason using its regulative
ideas. He states this assumption in the following way It relates instead to the formal structure
of the regress itself as we extend it, counterfactually, to the desired point (Abela, 2002a, p. 239,
emphasis mine). Abela is here advocating a position in which the laws of nature applied to our
perceptions are merely a subset of some greater set of possible perceptions which could, in

Moral realism and naturalism

35

In Kants thought, the laws of physics cannot explain chemicals, organisms, human social systems, and minds. They are not susceptible to
mathematization, and some are also not susceptible to testing or mechanical
laws. All are, however, studied by some form of investigation according to
some of the other requirements for natural law such as systematization, the
use of the categories, and the inclusion of higher level concepts. Proper
scientic causal laws are limited to those of physics and other sciences that
could in the future be mathematizable such as chemistry (4:47071). The
entities of physics in space (moving in time) provide the ontological foundation for other sciences that view the properties or arrangements of matter
from different perspectives, for example, biology describes organisms using
teleological principles but without claiming that organisms are more than
material entities.
The discipline of history which can be used to exemplify the social
sciences also relies on teleological laws. This is conrmed in the Critique
of the Power of Judgment when Kant devotes part of the discussion of the
teleological power of judgment to the study of organisms which exhibit
internal purposiveness (biology) and part to the study of beings who exhibit
external purposiveness (history) (5:425f).That the social science disciplines
also include teleological reasoning does not affect the ontological status of
their objects in nature any more than it would for biology. Further, although
Kant denies them the status of a natural science (4:471), he does not deny that
the subject matter of these social sciences is both determinate and part of
empirical nature.
Empirical psychology and, more broadly, anthropology (for simplicity,
I will focus on empirical psychology), are particularly important for this
book since morality involves the mental faculties of will, choice, and reason.
Unlike the social sciences, which rely on outer behavior, empirical psychology describes that part of nature that is revealed only through inner
intuition. While Kant does not deny that mental states could in principle be
instantiated in physical states of the brain, he does not think that such
reductionism is useful or even possible in explanation (7:119, B420). Still,
mental states as such, like material states, are subject to deterministic relations even though we are unable to know any deterministic laws describing
them. This is clear regarding empirical psychology in several places. In the
rst Critique, Kant divides what he calls immanent physiology, or the
study of nature as the sum total of the objects of the senses, into physics
and psychology, the objects of outer and inner sense, respectively (A846/
B874), and adds that empirical psychology can use the a priori principles
formal terms, connect to the laws of nature but which in fact do not, at least from the perspective
of the end of human inquiry.

36

Laying the ground

derived from pure philosophy (A848/B876).48 The quasi-scientic status of

empirical psychology is also shown in Kants discussion of the regulative use
of the ideas of reason in which he uses sensation, consciousness, imagination, memory, wit, the power to distinguish, pleasure, desire . . . even
understanding and reason as examples of the way that reasons regulative
principles can aid the cognitions of the understanding by relating and perhaps
uniting causal powers (A64849/B67677). This use importantly allows that
mental faculties themselves are proper subjects for empirical psychology
since the faculties of will, choice, and reason function in ethics as well as
empirical psychology. The determinism among mental phenomena is most
clearly stated in the second Critique when he notes that causal determinations
may be internal and they may have psychological instead of mechanical
causality, that is, produce actions by means of representations and not by
bodily movements; they are always determining grounds of the causality of a
being insofar as its existence is determinable in time and therefore under the
necessitating conditions of past time (5:96).
This ontology is important for assessing whether ethics could be metaphysically naturalistic. To the extent that ontological requirements of morality are
explainable through empirical psychology or any of the other disciplines, they
are metaphysically naturalistic. The following chapters will examine particular
aspects of Kants ethics to show that their ontology ts within these bounds of
nature. In some cases, such as practical freedom, the requirements of morality
need no more than empirical psychology. In other cases, such as the justication of the categorical imperative, the validity of empirical reason is defended
transcendentally in a way that has no ontological implications beyond empirical psychology. In still further cases, the apparent invocation of nonnatural
entities such as God or properties such as freedom of choice will be shown not
to assert their actual existence outside of nature under certain interpretations. In
all these cases, a crucial factor is that morality is a part of practical philosophy
rather than theoretical philosophy and is subject to a different philosophical
perspective that avoids direct existence claims.
The elements of Kants ethics
When I dened moral realism in terms of the independence of moral principles, properties, or objects from the moral agent, I noted that it is possible to
claim that Kant was a realist about every moral principle, property, or object
(although no interpreters actually do make that claim) or about none. Many
variations lie between these two extremes. Here I will identify and discuss the
48

For a discussion of this relation between the transcendental psychology of the rst Critique and
empirical psychology, see (Schmidt 2008).

Moral realism and naturalism

37

particular moral principles, properties, or objects in Kants theory that constitute the material for the realism/idealism debate; they will also serve as the
elements to be incorporated into a metaphysical naturalism. This analysis will
contribute two things to setting the stage for my later arguments: First, sharpen
the focus of the debate between realism and idealism on the one hand and
naturalism and antinaturalism on the other hand in order to unearth the specic
interpretive points in Kant at issue, and second, reveal some of the particular
grounds for both realist and nonrealist claims in order to begin assessing their
strength.
A.
B.
C.
D.
E.
F.
G.
H.

The value of particular chosen ends

Particular moral duties or norms
Value of humanity as end in itself
The highest good and the postulates of practical reason
Moral obligation
The moral law itself
Practical reason itself
Freedom of choice

A. The value of particular chosen ends should be uncontroversially understood as idealist. Subjective ends such as drinking wine, listening to jazz,
running a marathon, and pursuing a career have value only because some
valuing beings confer value upon them. Kant lays this relation out in the
Groundwork when he says The ends that a rational being intends at his
discretion as effects of his actions (material ends) are all only relative; for
only their mere relation to a specially constituted faculty of desire on the
part of the subject gives them their value (4:427, modied). Moral realism
about the value of particular ends would require that the value is independent of the agent, whether it be the agents choice, set of beliefs, or nature.
Hence the value of these particular ends must be empirically idealist. At the
transcendental level, one must abstract from the particular content of the
chosen ends and simply treat them as possible contingent choices of agents
capable of setting and pursuing ends. The particular ends operate at the
empirical level and clearly fall within the ontology of nature as understood
through empirical psychology. The ability of agents freely to set and
pursue ends is a further matter treated in (H). This idealism is of little
direct import in the debate about Kant and moral realism since no Kant
interpreter understands the value of particular ends as real. But it does offer
a toehold for the idealist, who can point to Kants justication the relation
to something in the subject to emphasize the role of the subject in value.
B. Particular duties or norms are distinguished from the categorical
imperative. No one doubts that Kant intends for particular maxims to be
tested using the rst formula and subsequent formulas that include the term

38

Laying the ground

maxims such as the formula of autonomy (at 4:432) and the formula of
the kingdom of ends (at 43536), both of which can be understood to carry
the emphasis on maxims over from the formula of universal law. The
formula of humanity can also provide a basis for the adoption of particular
duties or norms. They can be abstracted from their source in the categorical
imperative itself (F). The duties are distinct from the moral law or categorical imperative itself for a very good reason: They depend in part on
empirical information about the nature of human beings as a particular
kind of nite rational being.49 Each of the four examples Kant provides in
the Groundwork requires some empirical information about human nature:
Human beings are susceptible to death, cannot read one anothers minds,
have latent talents that require development and practice, and lack complete self-sufciency. Many particular principles and duties are generated
through an application of purely rational considerations from the moral law
combined with empirical information. Kant has room in his theory for
different particular kinds of nite rational beings to have different particular duties based upon different empirical natures.
A realist about these particular duties would have to admit that they
are not themselves a priori in that they rest on information that is itself
independent of the a priori categorical imperative.50 But because they rest
on human nature in general, the realist might argue that these duties are
empirically real even though transcendentally ideal. Thomas Hill and
Andrews Reath view these particular duties as constructed because they
can be said to result from a decision procedure to determine which particular duties human beings would themselves derive (hypothetically) using
the categorical imperative. Hill provides a constructivist account of the

49

50

Kants work in anthropology is in part aimed at determining these particular duties on both the
species and the cultural levels. Robert Louden and Allen Wood both devote signicant attention
to the application of anthropology to the derivation of particular duties in Kant (Wood 1999,
pp. 193320) (Louden 2000).
The Kantian moral realist Allen Wood acknowledges that our actual duties as provided in the
published Metaphysics of Morals are not entirely a priori but are derived from the categorical
imperative as applied to human nature in general (Wood 1999, pp. 19596). In order to act in
accordance with the value of humanity, for example, we would need to know what particular
kinds of human ends there are that might or might not accord with the more abstract value of
humanity. He thus allows a distinction between two levels of moral law: The pure moral law
itself and the set of particular duties that result from the application of that moral law to
empirical information about human nature. He does note that this position of Kants differs from
that provided in the Preface to the Groundwork, where a metaphysics of morals was limited to
the pure moral law, and all empirical information about human nature was part of anthropology.
See also (Wood 2008, p. 60), where he holds that the particular rules and duties result when
the pure law and values are interpreted in the light of human nature and even particular
cultural or historical circumstances and are not the result of a rigorous deductive procedure so
not contained in the moral law itself.

Moral realism and naturalism

39

derivation of particular moral rules or norms from the categorical imperative treated as a given.51 Similarly, Reath offers an interpretation of
particular duties as constructed while the categorical imperative itself is
not.52 This is the decision procedure type of constructivism in which the
duties depend upon some particular, in this case, hypothetical activity of
empirical moral agents. It is wrong to understand these particular duties as
optional, as a decision procedure sometimes suggests. In both Hills and
Reaths accounts, the hypothetical agents are not given leeway in determining the particular principles or duties at stake once the particular information about human nature, historical or cultural setting, and the like are
introduced as the basis for the decision in addition to the categorical
imperative. Since those things are empirically real, the duties could be
empirically real as well. But since for both Hill and Reath they depend
upon a hypothetical decision of the agent, they would be empirically ideal.
Under the nature of agency interpretation that I take, these duties would be
empirically real because they would follow directly from the nature of
empirical moral agents. Their empirical reality, however, is crucially
dependent on the empirical reality of the moral law itself (F). If it were
to turn out that the moral law itself is empirically ideal, then these particular
duties would be as well because they are derived from it. Similarly, these
duties could be comprehended within a metaphysical naturalism if the
categorical imperative is as well.
C. The value of necessary ends, in contrast to particular contingent ends
discussed in (A), is a matter of great controversy in the debate about moral
realism in Kant. There are two kinds of necessary ends: The ends that are
also duties presented in the Doctrine of Virtue, namely the happiness of
others and ones own perfection (6:38586), and humanity as an end in
itself as rst explained in the Groundwork (4:4289).
The ends that are also duties of virtue can be understood in two ways.
On the one hand, to the extent that the necessary ends of virtue are specic
ways in which Kant eshes out the value of humanity as an end in itself,
they can be assumed to have the same realist or idealist status as the latter.

51

52

Hill details his constructivism in his chapter Kantian Constructivism as Normative Ethics
(Hill 2012, pp. 7192). His version of constructivism centers on Kants Doctrine of Virtue
rather than the Groundwork. He takes the Doctrine of Virtue to begin with a general conception
of the value of persons and to lay out principles that human beings would adopt in order to
create a society that incorporates that value. Only the principles, not the value of persons, is
constructed.
Reath presents his position in (Reath 1994). Patrick Kain correctly argues that any position such
as Reaths does not entail that the particular duties are in any sense optional to empirical agents
in (Kain 2004). The duties are based on human nature rather than on any kind of preferences
any particular agent or all particular agents may have within the decision procedure situation.

40

Laying the ground

Kant makes this connection in the Introduction to the Doctrine of Virtue

when he connects ends that are also duties with the necessary end corresponding to the categorical imperative (6:385). On the other hand, to the
extent that one treats the particular application of these necessary ends of
virtue as a shorthand for the particular ends that one chooses to actualize or
particular duties of virtue, they would have the same status as the duties
and ends just discussed in (A) and (B), and thus real or ideal in the same
sense. Some have argued that the necessary ends of virtue ought to be seen
as having this status rather than being extensions of the value of humanity
as an end in itself, but I believe that it is more appropriate to consider them
as having the same status as the value of humanity as an end in itself.53
The value of humanity as an end in itself serves as one of the main
points of disagreement between realists and their opponents. I will devote a
large part of Chapter 7 to this topic but here would like to make a general

53

Formosa and Wood both take this option, for different reasons. Formosa classies these two
necessary ends of virtue along with the particular ends that human beings choose as their own,
calling all of them ends with conditional and nonabsolute worth in contrast to humanity as an
end in itself which has unconditional and absolute value (Formosa 2012, p.9). On his account,
these two necessary duties of virtue have the status of chosen ends as in (B) and would be
treated as constructed in the same way. His reason is that the necessary duties of virtue are wide
duties and thereby do not have to be the basis of every act. But this reason is insufcient to
make these ends conditional. Formosa is conating the happiness of others and ones own
perfection as necessary ends to adopt in general with the particular ends that we may adopt in
particular actions that will implement those necessary ends in various ways. This is the import
of calling them wide duties. The categorical imperative demands that the two ends be adopted
and to that extent they are of unconditional and absolute value. But Kant recognizes that many
particular maxims can honor these ends. A duty of wide obligation does not allow human
beings to reject the necessary end but only to act on some maxim or other that might further
ones own perfection or the happiness of others: a wide duty is not to be taken as permission to
make exceptions to the maxim of actions but only as permission to limit ones maxim of duty by
another (6:390).Thus these two necessary ends of virtue are not conditional and of nonabsolute
value. Wood holds that empirical information about human nature is used to determine which
ends we ought to set as moral beings and identies them as the ends which are also duties,
although elsewhere he allows that ones own perfection and the happiness of others are ends of
pure reason as opposed to simply ends of reason; Wood also connects the concept of perfection
to happiness in that part of perfection is being able to use ones developed talents (Wood 1999,
pp. 195, 32628). Wood is not clear about the status of the general ends that are also duties. He
appears to treat them as kinds of ends rather than generalized ends that we ought to adopt (Wood
2008, p. 167). In adopting any particular end that is, for example, of the kind that promotes the
happiness of others, we would need to use empirical information as Wood says. On Woods
account, the necessary ends of virtue would have the same status as particular duties that require
empirical information as in (A) and be subject to the kind of constructivism that Hill suggests.
But Kant holds in the second Critique that the concept of happiness not the conception, or
particular content of what things count as happiness, which is clearly empirical is an a priori
concept relevant for all nite rational beings and not dependent upon human nature (5:25).
Since the concept of happiness is a priori, the two necessary ends of virtue are also a priori, do
not rely on information about human nature, and can be understood as having the same status as
humanity as an end in itself.

Moral realism and naturalism

41

assessment of the relation the value of humanity has to the debate about
realism and naturalism. Kantian realists generally hold that the value of
humanity as an end in itself is independent of and conceptually prior to the
categorical imperative and even serves as the ground of the categorical
imperative. Idealists and some constructivists in contrast hold that the value
of humanity is derived directly from the categorical imperative, in which
case, the reality or ideality of the value of humanity depends upon the
status of the moral law itself (F). Alternatively, some constructivists claim
that the value of humanity is dependent upon a conception of agency
necessarily adopted by agents as such; without the existence of that
particular kind of agency there would be no need for the value of humanity
or any value at all.54 This view sees a parallel between the construction of
moral value and the construction of the categorical imperative itself rather
than seeing the former as derived from the latter. Its success or failure
would be tied to the success or failure of the parallel claim about the
categorical imperative to be discussed in (E) later on.
Value is often understood as a property of objects; in the case of the
value of humanity, a property of nite rational beings, but more broadly
speaking, objects or mental states could possess intrinsic value. A value
property could be understood at either the transcendental or empirical level.
If value were known to be an intrinsic property of some beings or things-inthemselves independently of the nature of moral agency, then value would
be transcendentally real. An example would be if God were to have created
some beings with intrinsic value. This value would be a constraint on the
conception of rational agency itself. In this case, the value would ground
the very possibility of a moral law. Alternately, if absolute moral value
were known only as a requirement of moral agency, namely, that in order
for moral experience to be possible there must be something of value
independent of the particular moral agents, then it could be empirically
real but transcendentally ideal. This conception of value would be limited
to the kind of thing that is compatible with the demands of the categorical
imperative, namely, humanity. Rational beings possessing humanity would
have value as an intrinsic property. It would still, however, not t into a
naturalist framework because empirically there would be a value property
of some beings in nature that does not gure in Kants conception of
nature. A third option is a transcendental and empirical ideality of value

54

I say that particular kind of agency in order to allow that a constructivist could hold either of
two alternatives. First, a constructivist might hold that there is only one possible kind of agency,
free rational agency, as Korsgaard does (Korsgaard 1996b). Second, a constructivism might
hold that it is possible for other kinds of agencies to exist which might not require the same
moral values or the moral law.

42

Laying the ground

that would be fully compatible with metaphysical naturalism. The empirical ideality of the value of humanity, I will argue, consists of an objective
ordering of ends by reason such that humanity is always to be treated as a
limiting end for maxims, whatever subjective contingent end the empirical
moral agent would be pursuing. This ordering by reason would be based on
reasons own needs and not on any intrinsic value property in human
beings and so would be compatible with a metaphysical naturalism and
an empirical ideality of value.
D. The highest good is an object of pure practical reason that, Kant argues,
requires the postulates of God and immortality for its realization. Kant
understands the highest good as an idea of reason that serves as an
archetype of a perfectly just and virtuous world to be attained. As such,
it is not an object that exists empirically or as a thing-in-itself (except to the
extent that it is actualized by moral agents through their actions). Realists
and idealists would agree that the idea of the highest good is dependent
upon the existence of nite, rational moral agents for whom the highest
good serves to satisfy both their obligation to be moral and their desire for
happiness. Perfectly rational agents are not in need of the highest good
because they have no independent desire for happiness. As an idea, the
highest good is dependent upon the moral agent. The realization of the
highest good itself would be brought about through the actions of moral
agents and of God. Actions of moral agents depend empirically upon them
and are thus empirically ideal.55 God would, in contrast, seem to have a
transcendentally real status.
Two of the three postulates of practical reason immortality of the soul
and the existence of God are postulated because they are needed for
realization of the highest good. (Freedom, the third postulate, underlies the
categorical imperative itself and has no relation to the highest good.) These
postulates are what I called earlier principles, properties, or objects required
for morality rather than purely moral principles, properties, or objects; but
because they play a role in Kants moral theory, they have been incorporated
into the realism/idealism debate, with discussion focusing on the existence
of God. The realist holds that the postulate requires that God exist, while an
idealist can hold that human beings are required to believe in God as a
distinct being but that God need not exist. The most extreme idealist would
hold that the postulate not only does not require that God exist, proper
comprehension of the function of the postulate shows that it does not involve
human beings believing that God as a distinct being exist but that the idea of
55

Treating the highest good as an object to be actualized by human beings is constructivist in the
different sense of bringing an idea to fulllment. For a discussion of the highest good as
something for human beings to create over the course of history, see (Reath 1988).

Moral realism and naturalism

43

God be used for immanent purposes. This last is the position I argue for in
Chapter 5. A metaphysical naturalism, of course, would preclude any role in
morality for the immortality of the soul or the existence of God.
E. Moral obligation is the fth area of Kants ethics at issue. One can
distinguish the obligation from the other elements of a duty because one
might be able to identify a value or acknowledge a moral law without being
obligated to uphold or follow them. Two examples would be an amoralist
and a cultural relativist: Each might recognize that there are values or laws
without accepting any obligation. (I am not concerned with the plausibility
of these positions.) The realism or idealism of moral obligation in Kant
depends upon whether it is the result of some procedure or in some other way
can be separated from the moral law, value, or practical reason itself.56 If it
cannot be separated, then obligation is inherent in them and the realism or
idealism of obligation would depend upon the real or ideal status of the moral
law or practical reason.57 In particular, if the nite will can be conceived in a
way that allows it to obligate itself to the categorical imperative, then
obligation would be idealist. Some constructivists hold that obligation is a
matter of an autonomous act of will in giving the law to itself.
One might tie obligation to the nature of the nite rational will. This is
successful in presenting an idealist conception of obligation only if the
nature of the nite rational will is understood as idealist, since it grounds
obligation in the nature of our nite will. Realists and idealists can agree
that obligation is tied to the nature of the human nite rational will; the
dispute is only about whether a Kantian theory should take the nite
rational will as real or ideal. And this question cuts to the heart of the
nature of practical reason and practical philosophy itself. Whether practical
reason itself is realist or idealist is the deeper question and will be noted in a
few paragraphs under (G).
Obligation is not considered to be a natural relation, making it difcult
to see how a metaphysical naturalism would work. Unlike value as a
property that purported to exist within objects or rational agents, obligation
is a relation between an agent and a set of duties and ends. The duties and
ends themselves, both necessary and contingent, could lie within nature as

56

57

Robert Stern offers a way to distinguish obligation from the moral law. Stern, who takes a
realist position regarding value and law, argues that moral obligation is idealist because we
give the content of morality its obligatory form, in so far as this depends on our limitedness as
nite creatures. (Stern 2011, p. 90)
Patrick Kain argues that because obligation is tied to our rational will, it must be understood as
real rather than as constructed (Kain 2004). In Chapter 3, I argue that obligation is similarly a
product of reason, but in the guise of the commanding nature of the categorical imperative as a
fact of reason. Obligation for either of these views is tied to the status of the faculty of reason.
I will show in Chapters 4 and 7 that reason should be understood as not transcendentally real.

44

Laying the ground

explained in (A)(D). Kant recognizes that the obligatoriness of duties and

ends is a practical rather than a theoretical matter. To see how obligation
can be understood within nature, one must understand the fundamental
difference between theoretical and practical philosophy as concerning what
is and what ought to be. Once the lack of ontological signicance of this
practical relation is understood, particularly in reference to the fact of
reason, a metaphysically naturalistic interpretation of obligation is not only
plausible but also demanded by the nature of Kants philosophy. I will
make that argument in Chapters 2 and 3.
F. The moral law itself, abstracting from obligation tied to it, is connected on
the one side to the categorical imperative and the fact of reason and on the
other to the nature of pure practical reason, which will be treated independently in (G). Kant allows that a rational being can act in accordance with
the representation of laws (4:412). Human beings represent the moral law
as a categorical imperative, and their immediate awareness of the categorical imperative is the fact of reason. A metaphysical naturalism would stress
this representation of the moral law within nature understood using empirical psychology. A nonnaturalism would insist that the moral law is more
than simply the representation of it in nite moral agents and would look to
an understanding of law that would make it valid with a broader scope
outside of nature. If God legislates the moral law not by creating it, since it
would stem from pure practical reason, but by promulgating it, then the
moral law would have a nonnatural basis. This option refers back to the
postulate of God (D). Another option for the nonnaturalist would be to
claim that in a more limited way, pure practical reason itself must be
understood as independent of nature, a subject for (G). At the very least,
however, the awareness of the categorical imperative in empirical consciousness provides a basis for a metaphysical naturalism, and only further
conditions for the categorical imperative would make it nonnatural.
A transcendental realist would also refer to God or to some way in which
pure practical reason would exist or be valid independently of the transcendental conceptions for moral agency. Realists about the moral law point out
that Kant denies that the moral law has any author but is instead part of the
nature of reason itself, or even the nature of things in general.58 The
transcendental idealist position would allow that reason is the source of
the moral law but claim that reason is limited to a condition for the
possibility of moral agency. This latter topic is a subject for (G).59
58
59

(Kain 2004).
It is also worth noting that there are some constructivist positions that give a different origin to
the moral law than the faculty of pure reason. One strand of Christine Korsgaards constructivism centers on the construction of the moral law itself. (Korsgaards position has shifted over

Moral realism and naturalism

45

The categorical imperative in empirical consciousness could be understood as empirically real or ideal. As a fact of reason, the awareness of the
categorical imperative is in the empirical mind of the actual moral agent. If
there is also a transcendental justication for the categorical imperative,
then it would be empirically real because it is valid independent of particular moral agents. But if it has no transcendental justication, then it would
be empirically ideal. The subjective nature of obligation, the enumeration
of particular duties, and in general, the experience of morality would
remain the same. I will examine the reasons in favor and against this
interpretation in Chapters 3 and 7.
G. Practical reason itself, the source of the moral law, is perhaps the core
topic for this book. A realist about practical reason would hold that it
discloses to moral agents some facts about the world itself independent of
those moral agents. A transcendentally real reason would disclose something about the nature of things in general because things in general would
be fundamentally rational. Morality would then be transcendentally real
independent of the transcendental moral agent. But if, as I argue Kant
holds, practical reason itself is constitutive of and restricted to moral
agency, then a transcendental idealist position about pure practical reason
is warranted. This position limits the existence of the faculty of reason as
well as its scope to only those beings capable of moral agency. I will
examine the basis for moral agency throughout the book, but the discussion
of the nature of practical philosophy in Chapter 2 will show that the
domain of practical reason is free acts, so that practical reason is valid
only for those beings.
On the empirical level, practical reason would be empirically real, that
is, independent of the empirical moral agent, if the empirical agent can
know that she meets the conditions of transcendental moral agency. Kant
provides reasons to support this identication when talking about the
priority of the practical over theoretical concerns, the inevitability of the
fact of reason, and his arguments for the categorical imperative. But he also
hesitates to allow that a transcendental deduction for morality is possible,

time. I am referring here to her arguments in The Sources of Normativity (Korsgaard 1996b).)
Her approach, as discussed here, takes as its basis a view of agency. If an agent asks not merely
what she should do on the basis of her desires but the broader question of what she should do
overall, the agent is driven by a process of reection to nd the answer that she ought to act only
in ways that could count as laws, and hence she has formulated for herself the categorical
imperative. Korsgaard claims that rational agency itself requires the categorical imperative but
not that it stems directly from reason, as might, for example, the law of the excluded middle.
Without the active reection of agents, there would be no categorical imperative. In my
argument in the following chapters, I will show that the moral law is a direct product of
reason, so I reject Korsgaards type of construction of the moral law.

46

Laying the ground

and he generally speaks of the practical as a point of view or even as

consisting of ideas. I will examine these issues in most of the chapters that
follow, but Chapters 4 and 7 provide the most direct discussion of the
empirical realism/idealism issue.
Perhaps the most extreme form of idealism would hold that reason
itself is constructed through human activity and has no correlation with any
independent structure or preexisting truths. Onora ONeill offers this type
of account.60 When a plurality of individuals are faced with the problem of
interacting with one another in an attempt to reach an agreement in the
absence of some preestablished authority, they must construct some
authority to guide them. Any authority that they choose that is based upon
some imposition of standards upon some members of the group by others
does not reect the agreement of each. To allow each to have a say in their
search for agreement, they would have to set a standard that allows for the
free agreement of each person. This would be reason itself whose results
are, as Kant says in the Discipline of Pure Reason, never anything more
than the agreement of free citizens (A738/B766). The formalization of the
idea of free agreement in turn is expressed in the categorical imperative that
demands universality of treatment respecting the voice of each. ONeill
holds that authority of reason is constructed in this way because it is
derived from the choice of agents simply to seek agreement without
reliance on any preexisting authority.61 Since Kant has connected reason
with the agreement of free citizens, there is no prior conception of reason.
To derive reason from freedom in this way, one must have a libertarian conception of freedom in Kant. Kant presents two competing conceptions of freedom, one libertarian and one rational. The libertarian
conception, in which the free will is undetermined by anything, features
prominently in his discussions of responsibility for moral decisions; its
clearest formulation comes in the Religion within the Boundaries of Mere
Reason in the discussion of the choice of a fundamental maxim (6:2125).
The rational conception, in which a free will is a will that is determined by
the moral law, appears prominently in Kants arguments in the second
Critique that a free will and the moral law as a determining ground
reciprocally imply one another (5:2829). Kant interpreters disagree about
the relation between these two conceptions of freedom and the nature of
60
61

ONeill makes this argument particularly in (ONeill 1989, pp. 350) and (ONeill 2002).
Another aspect of her argument is that any kind of agreement at all that does not accept some
kind of preexistent authority is governed by the categorical imperative, making it the supreme
principle not merely of practical philosophy but also of theoretical philosophy. I take this to
mean that practical reason has primacy over theoretical reason not merely in being authorized to
accept certain claims that cannot be proven by theoretical reason (the postulates) but to trump
claims of theoretical reason in action, as discussed in Chapter 3.

Moral realism and naturalism

47

each, but here it is important to note that a claim that practical reason is
constructed by the free actions of human beings requires the libertarian
conception. I will argue that a rst-person freedom of deliberation for Kant
is vital to the foundation of practical philosophy but that at the same time it
is not transcendentally justied. Kants conception of freedom is multifaceted, and the transcendental justication of freedom holds for reason and
rational action rather than for free choice, although he does allow for a
postulate of freedom related to free choice. Still, this freedom of choice
does not ground actual deliberation; empirically, human beings can consider themselves rational with a greater justication than they can attribute
to themselves a free choice.
To say that reason itself is ideal is to make a claim about the nature of
practical reason and, further, the nature of philosophy as a whole. It delves
into the heart of Kants claims about philosophical method itself. ONeills
interpretation, although incorrect as an account of the nature of reason,
identies the core of Kants philosophical project as a human-centered
endeavor, not merely in the familiar sense in which Kant is delineating the
basis of human knowledge and its limits but also in the broader sense in
which Kant comprehends that philosophical project itself as a part of the
active life of human beings who must confront the world from their
perspective. To be able to judge whether one can claim that reason itself
is constructed, or real, or ideal, one must examine the nature of philosophy
and reason in Kant.
H. Freedom of choice is the nal controversial element to be examined.
Kants moral theory emphasizes freedom. I differentiate the freedom of
the faculty of reason to provide the moral law from the freedom of the
faculty of choice to make decisions as the specic way in which I discuss
Kants Wille/Willkr distinction. The freedom of reason is a part of the
discussion of the nature of practical reason, but the freedom of the faculty
of choice is a distinct issue. The second kind of freedom is much more
directly at issue in metaphysical naturalism and it plays little part in the
realist/idealist debate. My Chapter 6 will examine the place of the freedom of the faculty of choice in Kants overall moral theory and provide a
metaphysically naturalistic interpretation that stresses its status as a
postulate. Whatever its specic metaphysical status, the freedom of choice
ought to be considered as ideal rather than real because it is a property of
moral agency itself. Moral idealism is perfectly compatible with a libertarian conception of agency. A realist might argue that actual freedom of
choice would have to be true of all moral agents and would thus fall
outside the boundaries of moral idealism understood as prioritizing the
moral agent as subject. A moral realist might further argue that this
freedom is the ground of the value of moral agents as objects that must

48

Laying the ground

be respected by moral agents as subjects. If this point is conceded, then

the issue would be settled by a determination of whether freedom of
choice is an actual nonnatural property of moral agents or whether it has
some lesser status that precludes it from being an actual property of moral
agents. Precisely that lesser status, I will argue, is given to freedom of
choice as a postulate of practical reason.
These eight aspects of morality the value of particular chosen ends,
particular duties, the value of humanity, the postulates, moral obligation, the
categorical imperative, practical reason, and free choice are each topics in the
debate about Kants realism and naturalism. All except the rst two can be
understood at either the transcendental or empirical level. Answering the
question about whether or not Kant is a moral realist or metaphysical naturalist
turns out to be so multifaceted that no one participant in the debate so far has
provided a comprehensive answer.
Concluding remarks
In this chapter, I have set out the parameters for determining whether Kant is or
is not a moral realist or a naturalist. I rejected a denition of moral realism that
encompasses every moral theory that allows for moral truth as inappropriate
for Kant interpretation. The denition of moral realism I advocated is one that
draws the line between realism and nonrealism in terms of the independence of
moral principles, properties, or objects from the moral agent. I introduced the
term moral idealism to cover those theories that claim that there is moral
truth but that truth is dependent upon the moral agent. Moral agency must be
assessed at both the transcendental and empirical levels. The general nature of
moral agency itself, if justied by a kind of transcendental argument that moral
experience is possible only with this kind of moral agent, operates at the
transcendental level. Any moral principles, properties, or objects that would
be independent of the transcendental moral agent would be transcendentally
real; the others transcendentally ideal. Transcendental ideality would still
allow for empirical reality, so independent evaluation of the nature of empirical moral agency would be required to resolve the debate at the empirical
level. I then assessed two different models of constructivism, the decision
procedure view that constructs moral principles, properties, or objects from a
choice made in some kind of ideal situation, whether hypothetical or actual,
which assumes that agents are already conceived in such a way that they can
make choices, and the nature of agency view that derives moral principles,
properties, or objects from a conception of agency prior to and as a ground for
the capacity of agents to make choices. I take this latter approach but label it
idealist as a more accurate term than constructivist.

Moral realism and naturalism

49

I then turned to naturalism, dividing methodological from metaphysical.

Kant clearly rejects the former but, I will argue, allows for the latter in his
moral theory. After reviewing some work on naturalism in Kant, I dene a
Kantian naturalism as limited to entities and properties in space and time, both
physical and mental. Proper scientic causal laws are limited to those of
physics and other sciences that could be mathematizable such as chemistry.
The entities of physics in space (moving through time) provide the ontological
foundation for other sciences that view the properties or arrangements of
matter from different perspectives, for example, biology and history both use
teleological explanation. Empirical psychology falls short of being a proper
natural science but does nonetheless describe the part of nature that is revealed
through inner intuition. Mental states, like material states, are subject to
deterministic relations even though we are unable to know any deterministic
laws describing them. A robust empirical psychology will ground the ontology
needed for the metaphysically naturalistic morality I will advocate.
Finally, I looked at a series of particular elements of Kants moral theory
that might be susceptible to realist or idealist interpretations and that need to be
addressed in relation to metaphysical naturalism. They point to several main
points of focus for the argument I will present. The value of particular chosen
ends has already been shown to be empirically ideal in this chapter. Particular
duties have already been shown in this chapter to be transcendentally ideal,
with their empirical reality standing or falling depending upon the general
empirical status of the moral law. The absolute value of humanity and the
related issue of the nature of the moral world posited by reason will be the
subject of Chapter 7. The existence of God as a postulate of practical reason
used for immanent purposes is the focus of Chapter 5. Moral obligation in
relation to the fact of reason and the nature of the primacy of practical reason
are the key parts to the idealistic interpretation given in Chapter 3. The origin
of the moral law in practical reason, the transcendental status of the faculty of
pure reason itself, and its place in nature form the content of Chapter 4. The
conception of practical reason itself in Kant is given in Chapters 2 and 3 but
runs through most of the chapters. Freedom of choice is a difculty for
metaphysical naturalism and secondarily for idealism and is the subject of
Chapter 6.

The place of ethics in Kants philosophy

The work of the previous chapter on realism/idealism and metaphysical naturalism point toward the need to examine the nature and methodology of Kants
philosophy in order to see how it might determine the ontology required for
ethics. The empirical/transcendental distinction, the nature of moral agency
and the status of reason itself, and the ontological status of moral principles,
properties, and objects will all depend on their place in Kants philosophy as a
whole. This chapter will show that the very broad strokes of Kants philosophy the topic and structure of philosophy as a whole, the transcendental
method, and the practical/theoretical distinction provide a key basis for
resolving the particular issues about naturalism and realism.
The broadest questions that one might ask about Kants ethics (indeed,
about any ethical system) are these: what is ethics? Why does existence
have a normative component? What is the nature and scope of morality?
What kinds of beings partake of it? Why is there any such thing as ethics at
all? This last is a twofold question. It is not only asking about the existence
and nature of ethics, that is, whether there is good or right or value. The
question what is ethics also asks about the philosophical study of those
aspects of the world such as good and right and value. Why should the
study of ethics have a place within philosophy instead of some other
discipline? How does it differ from other elements of philosophy? Is there
a distinct methodology for ethics? To what kind of philosophical treatment
is it susceptible: justication, application, explanation? The broadest question about Kants philosophical ethics is really about what kind of inquiry
ethics is and whether such inquiry is possible. The particular subject matter
of ethics will in part determine and in part be determined by ethics as
philosophical inquiry.
To ask about the very possibility of a philosophical study of something is
itself the hallmark of Kantian philosophy. Kants Critique of Pure Reason is
devoted to determining whether and how metaphysics itself is even possible
(Axii); the Prolegomena is even clearer in making this question a centerpiece of the structure of the book, devoting the entire third part and the
conclusion to the question of how metaphysics is possible (4:327, 4:365).
50

The place of ethics in Kants philosophy

51

Regarding practical philosophy, Kant is less clear. The Preface to the

Critique of Practical Reason, where Kant famously claims that the question
is not whether or how pure practical reason is possible but whether there is
such a thing (5:3), seems to put practical philosophy on a footing different
than that of the theoretical philosophy of the rst Critique and Prolegomena. As I will show here, however, this apparent difference is a result of
the very nature of ethics as a philosophical study. This chapter will look at
the place that practical philosophy has in Kants conception of philosophy
as a whole.
The chapter will rst examine the basic nature and subject matter
of philosophy in the rst section in order to frame the closer examination
of practical philosophy. This rst section will show that Kant saw philosophy
as ultimately an activity of reason geared toward reasons own ends.
Philosophys structure is multifaceted, and the most important distinctions
within philosophy will each be treated in turn in the other sections. The
second section will show that the critical task of philosophy is to be
distinguished from the more positive system of metaphysical cognitions
and principles. The special feature of critical philosophy is the use of
transcendental deduction as a method in discerning legitimate uses of
reason and other faculties. The transcendental method will be shown to
have minimal ontological import. The third section sifts through Kants
various explanations of the difference between theoretical and practical
philosophy in order to determine the proper subject matter of ethics. The
contrast between theoretical and practical will be shown to have some
ontological implications in itself. Free acts, not any particular set of ends,
constitute the domain of the practical. The ontology of their transcendental
basis will be determined in the following chapters.
What is philosophy?
Kant discusses the nature of philosophy itself in the Architectonic of Pure
Reason in the rst Critique. Philosophy is the system of all philosophical
cognition (A838/B866). Two terms are key to this denition: rst, cognition, and second, system. Cognition can be either historical, from what
is given, or rational, from principles. If one merely learns about a philosophical system from a textbook, Kant says, one does not have rational cognition.
For rational cognition, one must comprehend the cognitions from the universal sources of reason, namely, principles (A8367/B8645). Instead of
the rote learning of historical cognition of philosophy, rational cognition
involves the active creative work of reason itself. Among rational cognitions,
philosophical cognition is dened as cognition from concepts in contrast to
mathematical cognition from construction of concepts (A837/B865).

52

Laying the ground

Philosophy, then, is cognition drawn from concepts. Philosophical cognition

is the result of this activity of philosophizing.1
A system of cognition is a distinct characteristic of philosophical method.
Philosophy cannot merely be a collection of cognitions; it must be an organized arrangement of cognitions. Kant lays out two characteristics of such a
systematic organization:
Under the government of reason our cognitions cannot at all constitute a rhapsody but
must constitute a system, in which alone they can support and advance its essential
ends. I understand by a system, however, the unity of the manifold cognitions under one
idea. This is the rational concept of the form of a whole, insofar as through this the
domain of the manifold as well as the position of the parts with respect to each other is
determined a priori. (A832/B860)

The systematic organization of cognitions must conform to an idea which

determines rst, the domain2 of that system where it ts externally in
comparison to other systematic organizations of cognitions and what might
be beyond any possible cognition and second, the internal composition of
that system how the particular cognitions which constitute that domain are to
be arranged vis--vis one another. This twofold organization is done in
accordance with an idea, which itself must stem from reason; any other kind
of organization would have to be empirically based and contingent and would
constitute a technical but not an architectonic unity (A833/B861).3
Kant clearly held that theoretical cognition is subject to this method even at
the critical level. The very rst paragraph of the Transcendental Analytic in the
rst Critique uses precisely the language of an idea of the whole that leads to
a division of concepts that then have connection in a system (A6465/
B8990, Kants emphasis). Yet, one might think that this organizational
structure does not apply to ethics, for in the section of the Critique of Practical
1

The contents of the Architectonic chapter parallel the rst pages of Kants lecture on Philosophical Encyclopedia (29:545), with the exception of the denition of philosophy in relation to the
highest ends of human beings. This material does not appear in the textbook which Kant used for
the course (Feder 1769). Information on Kants textbooks is taken from (Naragon 2006).
I use the English term domain for various German terms that Kant uses with the same
meaning: Umfang in the rst Critique (A832/B860) and Gebiet in the third (5:17476). In the
discussion of the latter, Kant contrasts a domain determined by our cognitive faculties with a
broader territory (Boden) that includes all areas where cognition is possible for us, including
empirical cognition. Although Kant does not use the term idea here in establishing a domain,
he does claim that the domain is determined by the legislative power of the concepts and
corresponding faculty of cognition (reason or understanding).
Here, I follow Paula Manchester (Manchester 2008), who shows how Kants discussion of
architectonic is intended primarily to stress the active nature of reason itself as providing the
structure for its own investigations. This active organization replaces the passive scholastic
organization that stems from historical cognition. Manchester also provides a thorough discussion of previous views of the Architectonic that associate it more with merely providing an
architecture or structure within which reason would operate.

The place of ethics in Kants philosophy

53

Reason suggestively entitled Doctrine of the Method of Pure Practical

Reason, he states that method for practical reason differs from method in
theoretical reason in that the former concerns application of the moral laws to
human beings.
The doctrine of the method of pure practical reason cannot be understood as the way to
proceed (in reection as well as in exposition) with pure practical principles with a view
to scientic cognition of them, which alone is properly called method elsewhere, in the
theoretical (for popular cognition needs a manner but science a method, i.e., a procedure in accordance with principles of reason by which alone the manifold of a cognition
can become a system). Here the doctrine of method is understood, instead, as the way in
which one can provide the laws of pure practical reason with access to the human mind
and inuence on its maxims, that is, the way in which one can make objectively
practical reason subjectively practical as well. (5:151)

In this paragraph, Kant holds that the method for pure practical reason
concerns what is done with the pure practical principles in motivating human
beings to action and providing means for adoption of the correct maxims. But
the paragraph goes further than this to suggest that there is, in fact, no method
for collecting together practical principles, a claim which would entail that
there is no system of pure practical reason but only an aggregate of particular
moral principles. This further claim, however, does not reect Kants actual
practice, given, for example, that he devotes his Metaphysics of Morals to the
task of delimiting a system of ethics based upon the categorical imperative. As
this chapter develops, I will show that Kant does apply the requirements for a
systematic organization to ethics.
The domain of philosophy
Philosophy as a whole is a system of rational cognition from concepts, and
thus the systematic organization of rational cognition must be organized by
reason in such a way that it has both a domain and an internal structure. The
domain would be related to philosophys guiding idea, which has often, Kant
hints, been mistakenly neglected in place of mere systematic unity of rational
cognition as such; this is the scholastic concept of philosophy that has no
further purpose or end. The true ideal concept of philosophy, which Kant
calls the cosmopolitan concept, is the science of the relation of all cognition
to the essential ends of human reason of which the highest is the nal end
(Endzweck) or vocation of human beings (A83940/B86768).4
4

As an ideal, Kants denition of philosophy has a limitation: there is as yet no such complete
science. Kant claims that philosophy ought to be understood as the archetype against which
attempts to philosophize are compared. Actual philosophical practice ought to approach this
ideal philosophical cognition. (A838/B866)

54

Laying the ground

This guiding thread for philosophy provides a broad domain for philosophy
insofar as virtually all cognitions can be related to the nal end of human
beings, given that human practical deliberation and its resulting actions can
benet from knowledge and skill related to any possible action. One might
think that it is problematic that some of these cognitions would be empirical
even though reason itself operates a priori. Kant does in fact allow for
reasoning based upon empirical principles as part of empirical philosophy
in contrast to cognition from pure reason (A840/B868). Cognition by reason
from empirical principles is still cognition by reason. The domain for philosophy is identied as all cognition related to the essential ends of human reason,
provided that cognition is derived from principles.
In only one other published work does Kant provide such an allencompassing denition of philosophy.5 The Jsche Logic, which was based
on Kants course lecture notes and not written personally by Kant, uses the
same language as the Architectonic in distinguishing the scholastic concept
from the cosmopolitan concept of philosophy, the latter of which is the
science of the nal ends of human reason (9:23), elaborated as the science
of the relation of all cognition and of all use of reason to the ultimate end of
human reason, to which, as the highest, all other ends are subordinated, and
in which they must all unite to form a unity (9:24).6 But since choice of ends
and thereby, the activity of reason are involved, Kant also calls philosophy
a science of the highest maxim for the use of our reason (9:24, see also
Metaphysic L2 28:533).
The ultimate end of human reason should not be confused with the ultimate
end (Endzweck) of the existence of the world, a topic that Kant discusses in
light of his teleology in the Critique of the Power of Judgment (5:43436). It is
possible for human reason to have one end and the existence of the world
another. Further, reason as an activity can possess its own ends or goals that

Kant does offer less sweeping denitions of philosophy. In the Introduction to the Critique of
the Power of Judgment, he denes philosophy excluding logic as principles of rational cognition of things by means of concepts (logic is said to consist of principles of the form of thinking
in general) (5:171); the First Introduction similarly calls it the system of rational cognition
through concepts(20:195). The Groundwork, accepting the ancient Greek division of philosophy into logic, physics, and ethics, simply treats philosophy as rational cognition (4:387). In
Kants lectures on ethics, he generally started with a simple division of philosophy into two parts
without dening philosophy as such (e.g., Collins 27:243), but the notes from his winter
1793 course do offer a denition similar to that of the Groundwork and Judgment (Vigilantius,
27:479).
Kants course lectures on logic generally contained the same distinctions, although without the
connection to the nal ends of human beings; the textbook Kant used matches the absence of
nal ends ((Meier 1752), which is available in Band 16 of Kants gesammelte Schriften). The
Vienna Logic (24:797800), for example, does not provide discussion of system or the denition
of philosophy.

The place of ethics in Kants philosophy

55

are intrinsic to that activity7, whereas any nal end of the existence of the
world depends upon the supposition that the world has a supreme cause that
acts for ends. Without that kind of teleological view, there could be no ends for
things that are not themselves active as human reason is. A mere mechanical
explanation of the world does not invoke any ends at all. The activity of reason
is self-sufcient for producing ends while the existence of the world is not.
Finally, reason itself with its ends might provide human beings with the
conception of the ultimate end of the existence of the world for them to use
in interpreting their experience. The ultimate end would then be a product of
reason and not something in any way inherent in the world itself.
The identication of the domain of philosophy as the ultimate end of human
reason does not provide the content one might hope for, namely, an identication of the ultimate end of human reason. Kant identies this end in the Canon
of the Critique of Pure Reason as the determination of what is to be done
(A800/B828).8 The ultimate end of reason is thus moral. Since philosophy also
has the study of morality as one of its two main topical divisions, this is
another way of saying that the practical has primacy over the theoretical. I will
examine the nature of that priority in Chapter 3. Before that must come the
identication of precisely what counts as practical philosophy later in this
chapter.
The structure of philosophy
In addition to the domain, philosophy would have to have an internal structure
which also stems from the idea of a nal end for human beings. Kant offers
ve ways of dividing philosophy as a whole: logic and object-directed philosophy, critique and metaphysics, theoretical and practical, pure and empirical,
and the three (or four) main questions of reason.
A. Logic and object-directed philosophy: Logic is separated from the remainder of philosophy because, although it consists of cognition from principles, it has no object of its own in contrast to nature as the object of
theoretical metaphysics and freedom as the object of practical metaphysics.
Logic is instead a study of the rules of thought or the formal aspects of
cognition. The Preface to the Groundwork is clear in calling logic a science
7

The issue of any goals or interests of reason is more complicated than I am presenting it here.
Pauline Kleingeld argues that Kants claims about interests of reason cannot be taken merely
metaphorically (Pauline Kleingeld 1998). Axel Hutter stresses the centrality of interests in Kants
entire philosophical project (Hutter 2003).
Kant initially identies the ultimate end of human reason with three objects: freedom of the will,
the immortality of the soul, and the existence of God (A798/B826). He then subordinates them to
their more remote aim, which is the determination of duties. I will discuss the relation between
God and moral duties in Chapter 6 about the postulate of God.

56

Laying the ground

of the universal rules of thinking in general, without distinction of

objects (4:387), and the Jsche Logic declares logic to be the science of
the necessary laws of thought regarding all objects in general and thereby
to be a science of the correct use of the understanding and reason in
general (9:16). This division does not itself have any implications for
understanding the methodology of practical philosophy and will not receive
particular treatment as I proceed in the analysis of Kants methodology.
B. Critique and metaphysics: Putting logic aside, pure philosophy, which
excludes the empirical, is divided into the propaedeutic task of critique
and the systematic presentation of pure cognitions in metaphysics (A841/
B869).9 The critical task is intended only to defend the possibility of
certain a priori cognitions or principles and to exclude the possibility of
others. Any actual cognitions or principles thus defended are part of the
system of philosophy. This division is clear in the contrast between Kants
three Critiques and two systematic metaphysical works, the Metaphysical
Foundations of Natural Science and the Metaphysics of Morals (itself
divided into Metaphysical Foundations of the Doctrine of Right and
Metaphysical Foundations of the Doctrine of Virtue, thus paralleling the
volume on natural science). At the end of the Preface to the third Critique,
Kant declares with this I bring my entire critical enterprise to an end and
points to the work on metaphysics as his other task (5:170).10 The division
between critique and metaphysics is crucial for understanding Kants
metaethics, since critique provides the transcendental framework for ethics
and helps to establish what ethics in general is.
C. Theoretical and practical: Again putting logic aside, one can divide philosophy into theoretical and practical (Judgment, 5:171) or speculative and
practical (A841/B869) or, as the Greeks called them, physics and ethics
(Groundwork 4:387). One might think that Kant intended this division to
apply only to the systematic metaphysics, but he applies them to the task
of the Critiques as well as in the Preface to the Critique of Practical
Reason (5:3). Kants most famous division is also deceptively complex.
9
10

For another examination of the Architectonic chapter that stresses the difference between
Critique and system, see (Kinnaman 2008).
Kants division is problematic in a few other aspects: the Groundwork appears to hover in a no
mans land between critique and systematic metaphysics. The titles of its sections indicate
transitions from and to various approaches to morality, including philosophical moral rational
cognition, metaphysics of morals, and critique of pure practical reason (4:393, 446).
Although commentators generally place the book as a foundational part of the metaphysics
rather than a critique, the third section is nonetheless often treated as if it were a kind of critique
in its discussion of the limits of practical philosophy and its inclusion of a deduction, which is
compared to the parts of the second critique that replace a deduction with the fact of reason.
Karl Ameriks, for example, features extensive discussion of the Groundwork in a line with the
Critiques in his book Interpreting Kants Critiques (Ameriks 2003). The Prolegomena presents
the results of the Critique of Pure Reason in an analytic rather than synthetic form and thus, has
a tenuous critical identity.

The place of ethics in Kants philosophy

57

He offers four different characterizations of the difference between theoretical and practical philosophy. Because these characterizations reveal
important methodological considerations concerning the specic domain
for the practical, I will devote extensive space to them later in this chapter
by comparing them in detail.
D. Pure and empirical: Another division of philosophy is between cognition
from pure reason and rational cognition from principles derived empirically
(A840/B868). Philosophy had a wide scope in the centuries leading up to
Kant and included what today are considered natural and social sciences, so to
identify part of philosophy as based on empirical sources was not unusual.
To count as philosophy the sciences must still use principles that reason can
identify and work with. Empirical information from which no rational principles could be derived would not count as philosophy. Empirical philosophy,
then, is still rational philosophy rather than an exclusively empirical study.
E. The questions of reason: In two publications, Kant divides the interests of
reason into a set of questions: What can I know?; What ought I to do?;
What can I hope?; and What is the human being? In the Canon, Kant
includes only the rst three questions and then quickly assigns the rst to
speculative (theoretical) and the second to practical reason, with the third
consisting of both (A805/B833). In effect, then, this division reduces to
the theoretical/practical division. In the Jsche Logic two decades later, he
claims that the rst question belongs to metaphysics, the second to morals,
the third to religion, and the fourth to anthropology, which itself can also
be seen to subsume the rst three questions as well (9:25). He appears to
correlate the rst three questions with three tasks for the philosopher:
(a) determining the sources of knowledge, which would presumably correlate with theoretical metaphysics, (b) determining the use of knowledge,
which would correlate with morals, and (c) determining the boundaries of
reason, which might correlate with a philosophical approach to religion (the
term boundary (Grenze) invokes Kants Religion within the Boundaries
of Mere Reason11), or with the project of critique in general, or more
narrowly with the transcendental dialectic and teleology as they relate to
religious questions of the soul, free will, and God. One might also be
tempted to align the rst three questions with the three Critiques and the
last with the published Anthropology; however, the topics covered by the
questions extend to other works as well. Further, if anthropology is to
encompass the three prior questions, then it is not a distinct subdivision
within the structure of philosophy. The explanation of the four-question
division is in the end too vague to be helpful in dividing philosophys
domain, while the three-question division simply reduces to the theoretical/
11

In a letter to Carl Friedrich Studlin on May 4, 1793, Kant repeats these four questions and
explicitly identies the Religion as completing the answer to the third question (11:429).

58

Laying the ground

practical dichotomy.12 In the end, despite its fame, this division of philosophy is unhelpful, and I will not discuss it further in this book.
These ve ways of providing an initial division of philosophy do not
correspond to one another. Leaving aside the questions that summarize philosophy, the others work variously in parallel or perpendicular to one another.
How do these various ways of distinguishing the tasks of philosophy provide
its internal structure, particularly in relation to ethics? The justication for this
structure rests on the requirements for eshing out the demands of morality.
First, the separation of logic from the remainder of philosophy provides the
basis on which any investigation by reason may progress. Second, the work of
critical philosophy is needed to develop the particular limits and foundations for
further systematic work. Third, the division into practical and theoretical is
necessary because the practical must be informed by the theoretical in order to
know how to implement practical ends. Finally, to determine the particular
moral duties that form the vocation of human beings, the pure must be separated
from the empirical so that reason can properly identify pure duties that will
ground all particular empirically-informed duties. These explanations of the
structure of philosophy, excepting the rst about logic, will receive more
attention in direct relation to the practical in each of the remaining two sections
of this chapter. One section will cover both the critique/system and pure/
empirical divisions, the second, the practical/theoretical.
The nature of the transcendental method and ontology
The divisions of philosophy into critique and system on the one hand and pure
and empirical on the other hand are related. Kant is particularly concerned
about the division between pure and empirical philosophy as this passage from
the Architectonic shows.
It is of the utmost importance to isolate cognitions that differ from one another in their
species and origin, and carefully to avoid mixing them together with others with which
they are usually connected in their use. What chemists do in analyzing materials, what
mathematicians do in their pure theory of magnitude, the philosopher is even more
obliged to do, so that he can securely determine the proper value and inuence of the
advantage that a special kind of cognition has over the aimless use of the understanding.
(A842/B870)
12

Axel Hutter suggests that another difference between the two sets of questions is that the task of
uniting the questions together belongs to the third question in the Critique of Pure Reason but
the fourth in the Logic. He claims that since the unication by anthropology in the Logic is
weaker than that of the Critique, Kant loses genuine systematic connection among the questions. I believe that Hutter is reading too much into the short presentation in the Logic and that
Kants fourth question does not exclude the third from combining the rst and second (Hutter
2003, pp. 8890).

The place of ethics in Kants philosophy

59

The philosopher must be careful to distinguish the a priori from the a posteriori.
Kants rejection of any empirical basis for philosophy, any philosophy based
solely on a posteriori cognitions, is not a complete rejection of empirically
derived knowledge but a separation of empirically derived knowledge from a
priori. The study of nature and the study of morals will both have their pure and
empirical parts. The pure parts, metaphysics, are the distinctive contribution of
philosophy as the cognition from concepts which have an a priori origin.
The critical transcendental method is used in the determination of pure as
opposed to empirical cognitions. In what follows, I will summarize the transcendental method in a way that I hope is not controversial. Then I will show
how it can relate to a metaphysical naturalism and the empirical/transcendental
levels of moral realism.
Transcendental method in general
That philosophy requires a priori claims is central to Kants philosophical
framework and needs little elaboration. A priori cognition is dened by Kant
as that which occurs absolutely independently of all experience (B3) and
which involves necessity and strict universality (B4). A merely empirical
approach, one which draws all its cognitions from experience, could not
conclude that the cognitions are true necessarily and universally. But knowledge from concepts alone must be true necessarily and universally for the
domain governed by those concepts, in this case experience. When Kant
distinguished pure and empirical philosophy, he was distinguishing principles
known a priori from those known a posteriori. He makes this same kind of
distinction using different terminology when he contrasts historical and
rational cognitions, the former from what is given, the latter from principles
(A8356/B8634). Only rational cognition is such that it could have arisen
from reason alone. Kant puts it this way in the late essay What Real Progress
has Metaphysics Made in Germany?: If all knowledge is of empirical origin,
then . . . transcendental philosophy is itself an absurdity. (20:275). Critical
philosophical method must focus on a priori rather than a posteriori cognitions.
This role for the a priori lies at the basis of Kants critical project. Since
philosophy as a discipline is characterized by the use of reason, and any claims
by reason to knowledge of its own would be a priori, Kant holds that philosophy is in need of a critique in order to determine the possibility, validity, and
extent of reasons a priori claims. Critique exhibits a unique method: the
transcendental argument.13
13

The method of a transcendental argument is controversial inside and outside Kant scholarship.
I will not enter into the details of this work but will only offer a relatively non-controversial
account of its nature and role. Within Kant studies, see (Frster 1989). In contemporary

60

Laying the ground

Transcendental philosophy is taken to be the study of the ability of reason or

the understanding to have pure cognitions rather than any direct claim about
the nature of objects themselves. Kant is clear about this in the Mrongovius
lectures on metaphysics given in the winter semester of 1782: Transcendental
philosophy . . . does not say something a priori of objects, but rather investigates the capacity of the understanding or of reason to cognize something a
priori (29:784, modied). Since it focuses on the faculties of understanding
and reason rather than objects, transcendental philosophy does not itself make
any direct claims about objects. Only the pure cognitions themselves are
claims about objects. A few pages later, Kant expands this to clarify that the
pure cognitions of reason or understanding are themselves not transcendental
and are to be distinguished from empirical cognitions not by being transcendental but by being a priori (29:786). The transcendental is dened in terms of
determining the possibility of certain a priori cognitions. The pure cognitions
themselves, presumably, may be offered for this investigation regardless of
their own particular metaphysical status or origin, that is, their actuality. They
are cognitions which human beings can possess already in experience, or
which are embodied in experience. The transcendental investigation of these
concepts asks not whence human beings possess these concepts but rather on
what basis the concepts are valid a priori.
Kant makes this point in the Critique of Pure Reason when he stresses the
difference between the task of tracing the actual generation or source of a
concept from the justication of its validity (A8587/B117119). He praises
Locke for having paved the way for the investigation of the origin of concepts
from experience. This origin might serve to explain the possession of empirical
concepts, but one cannot nd the justication for a priori concepts this way
for in regard to their future use, which should be entirely independent of
experience, an entirely different birth certicate than that of an ancestry from
experiences must be produced (A86/B119). Empirically derived concepts can
have no a priori validity.
The possession of a concept is here argued to lie on an entirely different
plane from the justication of that concept. Whatever the occasional causes of
their generation, concepts do not receive their validity from those causes.
Rather, the validity of concepts is a result of considerations of an entirely
different type. Kant summarizes these two types of considerations of concepts:
I therefore call the explanation of the way in which concepts can relate to
objects a priori their transcendental deduction, and distinguish this from the
empirical deduction, which shows how a concept is acquired through
philosophy, a good collection is (Stern 2003). A seminal modern paper on the topic is (Stroud
1968). For an argument that there is nothing distinctive about the transcendental method, see
(Pihlstrm 2004).

The place of ethics in Kants philosophy

61

experience and reection on it, and therefore concerns not the lawfulness but
the fact from which the possession has arisen (A85/B117). Transcendental
philosophy abstracts from any empirical manifestation or source of a concept
to ask instead about its justication. The transcendental deduction of an a priori
cognition or principle, then, will show that it is necessary for experience yet
not derived from experience.
The method of transcendental deduction is not restricted to concepts necessary for the very possibility of experience in general. A transcendental deduction is one which assesses and justies the conditions necessary for the
possibility of some specic kind of experience, using the term broadly.
Although he does not identify them as transcendental, this is how Kant
understands a deduction when he discusses the role of deductions for the moral
law (4:447,454, 5:467), for aesthetic judgments (5:27980, 5:28990), and
for property (6:249250). The experience at issue in the rst Critiques transcendental deduction is the cognition of objects of experience, or as described
in the Prolegomena, the possibility of natural science (4:275 and 4:29496).
The range of other sorts of experiences whose conditions of possibility could
be the subject of a transcendental deduction is not explicitly limited by Kant. In
the rst Critique, a priori concepts which can relate to objects of experience are
specied, but this stricture is violated in the discussion of a transcendental
deduction of the moral law in the second Critique and other deductions. We
can assume that Kant restricts transcendental deductions to what can be known
absolutely a priori, whether it is a concept for an object, a moral law, a manner
of judgment, or a property relation.
Regardless of the kind of pure a priori cognition or principle, Kant is clear
that they cannot be derived empirically. A transcendental justication cannot
be based upon derivation from experience, but it can justify the validity of
cognitions or principles of the understanding or reason as they are used in
experience. Just as space and time are empirically real but transcendentally
ideal, a priori cognitions and principles can be understood as transcendentally
ideal (dependent upon the faculties of the subject) yet empirically real (formal
elements of experience governing empirical objects).
Take the pure concepts of the understanding, in particular causation, as
paradigmatic. Kant argues in the transcendental deduction that a necessary
condition for experience is the transcendental unity of apperception, that is,
that all intuitions be synthesized into one whole; cause and effect is one of the
ways in which this synthesis is exhibited. But the concept of cause and effect is
one that we can meet with in experience. The physical sciences make synthetic
a priori causal claims; to that extent, we are in possession of universal and
necessary principles such as causation (Prolegomena 4:295). Kants transcendental justication of the causal relation is intended to show how pure natural
science is possible. He does not deprive physics of these claims in order to shift

62

Laying the ground

them to philosophy when providing their transcendental justication; he

instead conrms their a priori validity. Since appearances in space and time
are empirically real, as we have already seen, the causal relations among
appearances are also empirically real. They are transcendentally ideal in the
sense that they receive their justication not by being derived from experience
but from being proven as necessary elements of an objective, empirically real
experience of nature. As Kant puts it in the Mrongovius lectures: consideration by pure reason of the nature and possibility of such a pure cognition of
reason is transcendental, e.g., the concept of cause and effect is pure but not
transcendental, but the consideration of the possibility of such a concept is
transcendental(29:786).
Since the pure concepts are valid in space and time, they t in with an
ontological naturalism. They are cognitions which human beings can possess
already in nature, or which are embodied in nature. The transcendental investigation of these concepts asks not whether human beings possess these concepts but rather on what basis the concepts are valid a priori. Human beings in
nature may possess these concepts empirically as part of the empirical mind
studied by psychology. That is, the actual empirically real agent will possess
and employ a cognitive system that contains these pure concepts. In order for a
cognitive system to be able to represent to itself an objective nature, it must be
able to unite intuitions in accordance with causal relationships. This requirement can be understood as the possession of a concept, as in natural sciences
study of nature, or as simply the functioning of the cognitive system itself in its
uniting of various sensations under causal laws. These are two sides of the
same coin since Kant himself identied concepts with functions (A68/B93).
The pure cognitions themselves and the faculty that employs them can be
part of nature. In contemporary terms, we can say that humans can have
evolved these faculties in response to their natural environment. The passages
I just quoted are consistent with the a priori cognitions of reason corresponding
to the actual, pre-existing structure of nature itself. The a priori necessary
cognitions can stem from (or be embodied as functions in) human cognitive
faculties that have themselves evolved partly in response to the matching
objective ontology of nature. Nature can itself, for example, be governed by
necessary causal relations on its own, independent of the existence of any
being who might evolve. And beings who evolve to cognize causal relations in
nature certainly evolve this ability in part, at least, in response to the actual
causal relations in nature. There are obvious evolutionary advantages to
cognizing an existing structural feature of the environment.
That these cognitive faculties have evolved seems to make them empirical,
and contingent, rather than transcendental and necessary. But this is Kants key
insight: these transcendental arguments show the necessary conditions for any
possible organism that might evolve an ability to represent an objective

The place of ethics in Kants philosophy

63

experience. The a priori necessity of representing nature as having causal

relations is not dependent upon nature having causal relations but is an
independent necessary condition of any being representing any objective
experience. It is true that nature itself necessarily has causal relations, but that
necessity neither results from nor grounds the subjects a priori necessary
representation of causal relations; at most, the one parallels the other in that
they are the mirror images of the results of the transcendental deduction of the
conditions of experience. The transcendental arguments reveal that certain
structures of mind are necessary for certain types of organisms. The actual
evolution of these cognitive structures is an empirical matter; their transcendental necessity is not. The transcendental level justies the necessity of the
concept for the cognitive structure of a being representing any possible experience; its empirical manifestation is a product of the actualized cognitive
structure. Even though that cognitive structure evolved in response to
the causal structure of nature, it is still pure because the transcendental justication shows that no other possible cognitive structure could evolve to provide
such objective experience.
Barry Stroud has presented a view of transcendental argumentation that
rejects the kind of claim I am making. Roughly, my view of the ontology of
transcendental argumentation is like that of P.F. Strawson in his inuential The
Bounds of Sense, in which he provides a two-world interpretation of transcendental idealism, rejects it as disastrous, and instead distinguishes the positive
results of Kants transcendental considerations as a metaphysics of experience.14 Stroud, in reaction to the use of transcendental arguments by Strawson
(although not in reference to The Bounds of Sense) and others, contended that
transcendental arguments require transcendental idealism: since transcendental
arguments concern the necessary requirements for the subject of experience to
be able to represent an objective experience, these a priori requirements can be
shown valid only as requirements for experiences of that particular kind of
subject, not as requirements for objects as such.15 I do not want to enter into a
detailed debate about his position since others have offered criticisms of Stroud
on this matter.16 I have three brief responses to any objection that could be put to
my claims about transcendental argumentation and naturalistic metaphysics on
this basis. First, it is abundantly clear from the passages I quoted from Kant that
14
15

16

Strawson calls transcendental idealism disastrous at (Strawson 1966, p. 21).

Stroud phrased his conclusion in terms of the beliefs of agents that x is a feature of the world
they experience; I am phrasing it in terms of the structure of the subjects cognitive system
(Stroud 1968).
Robert Stern provides a thorough assessment of Strouds objections, as presented both by
Stroud and by others, in (Stern 2000, pp. 4365). Stern concludes by accepting some of
Strouds objections but holding that some forms of transcendental argumentation is acceptable.

64

Laying the ground

he did hold to an empirical realism about transcendental claims, so there is reason

to believe that he would accept the account I have given in terms of evolutionary
development of the cognitive systems in question within the empirical realm of
nature in space and time. Second, as I discussed in Chapter 1, I follow Graham
Bird in his claim that transcendental philosophy is a set of claims about the
empirical world that is compatible with metaphysical naturalism.17 Third, even if
Stroud were correct and my account could not hold of objects in nature, it would
still be applicable to the use of transcendental arguments in ethics, since (as will
be shown later in this and also in subsequent chapters) there are no independent
objects to which moral claims would refer.
Allow me to put these matters into the terminology I introduced in the
previous chapter. The concept of cause is transcendentally ideal because it is
transcendentally justied as necessary for a certain kind of subject of experience, namely experience of an objective order of objects in any possible
nature. There is no requirement that causality be transcendentally real, that
is, a necessary feature of all possible objects. Only objects that are to be able to
form the basis for an objective experience must be governed by causal relations. On the empirical level, causation is empirically real as a relation among
objects in space and time. The empirical subject must employ causal functions
and concepts in order to be able to represent those empirically real objects as
experience. The actual cognitive faculties that employ causality are then
comprehensible through empirical psychology as a part of the ontology of
nature. The transcendentally ideal concept of causality is both empirically real
and comprehended within metaphysical naturalism.
Transcendental method and ethics
Now the divisions in philosophy between critique and system on the one hand
and pure and empirical, on the other hand are clearer and their application to
ethics can be considered. In ethics, the critical approach attempts to justify the

17

(Bird 1995). In contrast, David Bell understands transcendental idealism methodologically and
argues that it is a re-description of reality in a way that construes objects in accordance with
the awareness we may have of them, that is, their appearance (Bell 2001). Empirical realism, in
contrast, accepts that objects exist independent of any subjective awareness of them, and
naturalism would be a claim that only empirically real objects exist. Bell argues that appearances are incompatible with naturalism because as phenomena they are not items in the natural
world or subject at all to the laws of nature. Phenomena are instead a primitive. Bell
problematically separates appearances the subject of the transcendental analysis from
empirical reality as if the transcendental analysis did not provide the ground for the scientic
analysis of empirical objects. The problem with his conception is apparent when considering the
transcendental ideality and empirical reality of space and time in which the transcendental
arguments for the necessity of space and time as pure forms of intuition are to guarantee their
empirical reality with regard to all empirical objects.

The place of ethics in Kants philosophy

65

pure principles of morality on the basis of their relation to the transcendental

moral agent (the specics of which have yet to be determined). To the extent
that moral principles, properties, and objects are dependent upon the moral
agent, they can be said to be transcendentally ideal. Those arguments will be
carefully examined in subsequent chapters. At the same time, these moral
principles, properties, and objects might be considered empirically real to the
extent that they are taken to be independent of the empirical moral subject.
The theoretical conceptions of the subject of experience at the transcendental level and the empirical reality of objects and their properties in space and
time, one must keep in mind, will not be directly transferable to morality.
Morality is not a physical relation among spatiotemporal objects so the same
sense of empirical reality is impossible. The equivalent conception of empirical moral experience will be determined. And the use of critical transcendental
method is different in the moral case: Kant irts with a transcendental deduction in the Groundwork but then denies its possibility in the Critique of
Practical Reason. The subject of morality at the transcendental level will also
need to be determined. Thus, Kants application of the critical transcendental
method will be modied for morality.
Pure and empirical moral principles seem much easier to distinguish. Kant
decries moral theories that base the determination of right on happiness, which
can be specied only empirically. Pure moral principles must be a priori. That
practical reason possesses and uses both pure and empirical practical principles
in moral experience is a parallel that follows from the comparison but is not
obvious to readers of Kant. The pure and empirical divide is based upon whether
the principles used by reason are derived from reason itself, in which case, they
are pure, or derived empirically and assessed by reason, in which case, they are
empirical. My goal is to show that the methodological divide between pure and
empirical principles does not correspond to any ontological divide between
things-in-themselves and appearances in nature; rather, morality is experienced
by human beings in nature in the form of awareness of particular duties and the
general categorical imperative as well as the implementation of moral duties
through action in space and time.
I earlier noted that Kant uses an analogy to a chemical experiment to explain
that the task of critique in practical philosophy is to separate the pure moral
elements from the empirical (e.g., 5:9293). Merely identifying some
moral principles as pure does not preclude them from being part of empirical
moral experience and at the same time, justied transcendentally. He makes the
same analogy to chemistry in the nal section of the second Critique, the
Conclusion in which he invokes his famous image of the starry heavens
above me and the moral law within me (5:161). He offers the hope that morality
can advance by using a method based upon the method of the recent sciences,
the maxim of carefully reecting beforehand so that particular events in

66

Laying the ground

nature, such as the fall of a stone, are resolved into their elements and the forces
manifested in them (5:163). This is applied to ethics by a similar reection on
the moral dispositions of our nature encountered in experience:
We have at hand examples of reason judging morally. We can analyze them into their
elementary concepts and, in default of mathematics, adopt a procedure similar to that
of chemistry the separation, by repeated experiments on common human understanding, of the empirical from the rational that may be found in them and come to know
both of them pure and what each can accomplish of itself; and in this way we can
prevent on the one hand the errors of a still crude, unpracticed appraisal and on the other
hand (what is far more necessary) the leaps of genius by which, as happens with the
adepts of the philosophers stone, without any methodical study or knowledge of nature
visionary treasures are promised and true ones are thrown away. In a word, science
(critically sought and methodically directed) is the narrow gate that leads to the doctrine
of wisdom. (5:163)

Although the description of the proper method is here extremely sketchy, it

requires that we already have pure moral principles in our moral experience. In
the Preface, Kant explicitly allows that humans can encounter a pure cognition
in experience: We say that we cognize something by reason only when we are
aware that we could have known it even if it had not presented itself to us as it
did in experience; hence rational cognition and cognition a priori are one and
the same (5:12).
These passages are supported by others in which Kant claims that we are
conscious of morality, most directly when in the second Critique , he tries to
work out the starting point for our cognition of the unconditionally practical
(5:29) given that freedom and the moral law reciprocally imply one another.
He rejects freedom because we can neither be immediately conscious of it nor
derive it from experience. Our immediate consciousness of the moral law is
instead the entry point to the practical. He then asks:
But how is consciousness of that moral law possible? We can become aware of pure
practical laws just as we are aware of pure theoretical principles, by attending to the
necessity with which reason prescribes them to us and to the setting aside of all
empirical conditions to which reason directs us. (5:30)

Kant holds that we have consciousness (Bewutsein) and are aware (bewut)
of the moral law because reason prescribes it to us. This claim is central to
understanding Kants fact of reason, which will be covered in Chapter 3.
Empirical consciousness of the moral law and the general awareness of
moral principles in moral experience do not yet make them empirically real. If
the moral duties are products of the empirical moral agent, they would have to
be considered empirically ideal, that is, dependent upon the human mind. But
if they are grounded in something other than the empirical moral agent, they
could be considered empirically real. Kants division between pure and empirical as two kinds of principles with different grounds points to the possible

The place of ethics in Kants philosophy

67

ground for an empirical realism of moral principles. If the moral principles are
understood as pure because they are grounded in reason, then the status of
reason itself will determine whether these principles are empirically real or
ideal. If reason can be shown to have its independence from empirical moral
agents, then moral principles would have empirical reality. The possible
transcendental ideality of moral principles would similarly depend upon
whether reason can be conceived as independent of the transcendental
moral agent.
The importance of these observations lies in their ontological signicance.
Our awareness of the moral law and moral principles is in empirical consciousness. Empirical consciousness is the subject matter of empirical psychology,
which is part of the theoretical study of nature. Kant believes that causal
determinism is true of empirical psychology, even if we are unable to know
any of the causal laws. Moral principles should, then, be available for use in
causal explanations of human beings in nature. Further, if the moral principles
are available in empirical psychology, the faculty that causes us to be aware of
them by prescribing them to us could be included in empirical psychology. If
so then there is a strong basis for a claim that reason as well as moral principles
are empirically real and subsequently for a robust conception of moral experience that includes the role of practical reason in actual agent deliberation and
causal determination. In addition to these systematic connections, this picture
is supported by some direct textual evidence such as the identication of
a causality of reason in the determination of the will as one of the natural
causes (A803/B831) and the claim that a rational being has the capacity to
act in accordance with the representation of laws which would have to be in
the empirical mind (4:412). If moral principles are available to moral agents
through moral experience, there is reason to call them empirically real. In
addition, if the agents moral actions are caused by reason in a way that can be
understood in empirical psychology, then the property of rational causation of
moral agents would be empirically real as well. In both cases, they could be
understood as transcendentally ideal as well if the transcendental conception of
moral agency also required them.
Kant also provides reasons for casting some doubt on this robust understanding of moral experience as an empirical moral realism. He claims that we
are never aware of the actual causes of our actions (Groundwork, 4:407). Also,
he sharply separates a genuine self that acts on the basis of reason from an
appearance of the self that is determined by desires and inclinations, suggesting that reason can play no part in appearance (Groundwork, 4:457). And he
strongly rejects any attempt to explain freedom using psychological determining grounds in time (Critique of Practical Reason, 5:96). In these cases, Kant
appears to be denying any kind of empirical realism for moral explanations of
action. Kant need not be denying empirical realism for our awareness of moral

68

Laying the ground

principles in our moral experience; he is denying that freedom can be part of

the empirical world.
The difference between these two pictures reects the tension in Kant
regarding the ontological implications of his moral theory. Kant is pulled in
two directions by different moral concerns. On the one hand, he must account
for the everyday experience of morality that human beings, qua natural beings,
experience. On the other hand, some of Kants moral commitments, particularly but not exclusively freedom of choice, seem incomprehensible as part of
the causal determinism in nature. The task of the remaining chapters in this
book is to show how all of Kants moral commitments are explainable in a
naturalist ontology.
The practical/theoretical distinction and the domain for ethics
This chapter began with the claim that the question what is ethics? is the
most basic question one can ask about any moral theory. Kants answer as
developed so far showed that ethics can be understood as a part of philosophy,
which itself has as its domain: rational cognition aimed at the ultimate end of
reason itself. It is important to keep this basic approach in mind, since
philosophy as the activity of reason in pursuit of reasons own ends emphasizes subjectivity, and the dispute between moral realism and moral idealism
rests on the relation of moral principles, properties, and objects as dependent
on the subject. In the preceding section, I showed that through its critical
method, philosophy focuses on the faculties of reason and the understanding to
defend the possibility of pure rational cognition using transcendental arguments. I suggested that the ontology of the subject who employs pure rational
cognition could be understood in some way as empirically real. Making out the
practical analogue to the theoretical empirical reality of objects in space and
time is left to be developed in detail.
A big step in that direction comes from the division into theoretical and
practical as the most important distinction in philosophy. In this section, I will
show that Kant has several different characterizations of the distinction that can
have different connotations but essentially the same denotations. The characterization of practical philosophy will also identify its domain as free acts.
The practical/theoretical division
One must be aware that Kant uses the term speculative and theoretical
nearly interchangeably and even on the rare occasions when he distinguishes
them, he is not consistent. Kant contrasts the practical in different places with
either the speculative or the theoretical or both. I will treat speculative and
theoretical reason as interchangeable, usually using the term theoretical.

The place of ethics in Kants philosophy

69

Similarly, Kant uses the language of theoretical and practical reason in making
the division between theoretical and practical philosophy. Since philosophy is
the activity of reason, discussing them together is unproblematic.
Further, theoretical reason and practical reason are, for Kant, not two separate
faculties but one and the same faculty of reason. Kant uses the terms theoretical
reason and practical reason as short for the different uses by one faculty. He
makes this clear in the second Critique when he says it is still only one and the
same reason which, whether from a theoretical or practical perspective, judges
according to a priori principles (5:121; see also 4:391). In the many different
passages referring to theoretical reason, speculative reason, and practical reason,
Kant clearly means to refer to those uses of one faculty of reason. Any mention of
practical, speculative, or theoretical reason ought to be taken to mean these uses
or aims.
Given these clarications regarding the apparent identity of theoretical and
speculative reason, and the real meaning of the different faculties of reason as
different uses of one faculty, the contrast between practical reason and speculative reason becomes approachable. Kant appears to present four different explanations of the contrast.
A. Is/ought: Perhaps most famous contrast between the two is that the speculative aim of reason is to discover what is, while the practical aim of reason
is to tell us what ought to be (A633/B661; A802/B830; A806/B834; A845/
B873).
B. Nature/morality: Reason in its speculative use is said to investigate nature,
while reason in its practical use investigates morality. Kant presents such a
contrast when identifying a metaphysics of nature and a metaphysics of
morals (A841/B869; 4:3878; 5:545; 5:170).
C. Given/creating: The third contrast between speculative and practical reason
lies in their relation to whatever objects they consider, not necessarily in
the objects themselves. Specically, practical reason is said to create its
objects, while theoretical reason concerns objects somehow given to it
(A550/B578, 5:15; 5:5455; 5:89).
D. Knowledge/action: Speculative reason concerns knowledge while practical
reason concerns the determination of the will to action. Not the objects but
the method and purpose of application of reasons powers and principles is
the difference (A327/B384; A365; A475/B503; A803/B831; A841/B869;
5:1920; 5:456, 5:89).
Kants rst way of contrasting speculative and practical (A), using the is/ought
dichotomy, fails to clarify his intended meaning and actual use of the contrast.
The fame of this denition is rivaled only by its obscurity. The meaning of the
is is fairly straightforward. It is important to keep in mind that whatever the
practical turns out to be, it will not include ontological claims. But what does it

70

Laying the ground

mean to say that something ought to be? What is the nature of that which
ought to be? In one way, this contrast falls under (C), for Kant says that the
object which ought to be is the moral world which is created by practical
reason through human actions (A808/B836). This moral world which ought to
be is created only to the extent that practical reason is actualized in the moral
agents acts. In another way, the is/ought contrast is like (D) because practical
reason is to determine our actions in accordance with that which ought to be
(A802/B830). The ought in this sense has meaning only as an imperative
commanding the will to action. Because of this ambiguity, I will not treat
Kants references to is and ought to be as comprising a distinct meaning of
the difference between the practical and speculative, although it remains
important as a summary of their fundamental difference.
The contrast between the sets of objects of nature and morality (B) is more
successful as a characterization of Kants intended meaning and use of the
contrast but by itself, insufcient to explain the unique characteristics of
the practical. This contrast can be construed in two ways related to different
characterizations of object, rst as sets of things and second as subject matter.
In the rst, one would assign the theoretical to one set of substantial objects,
perhaps appearances, and the practical to another, perhaps things in themselves.
Besides being limited by relying on a two-world interpretation of transcendental
idealism, this simple kind of contrast also does not t Kants overall theory. The
two uses of reason cannot have completely different sets of objects because they
share concern with at least some of the same objects. On the one hand, God,
immortality of the soul, and freedom of the will, can be approached, Kant says,
both speculatively and practically (A796/B824; 5:45), and, on the other hand,
the human power of choice is treated, as he says in the First Introduction to the
third Critique, as free in practical philosophy and as a cause in theoretical
philosophy as cognition of nature . . . the practical part of a philosophy of
nature (20:197). Further, Kant himself stated that speculative and practical
reason concern the same set of objects of possible experience. As I noted in
the rst section of this chapter, in the Critique of the Power of Judgment, he
distinguished the concepts of domain and territory: the domains of speculative
and practical differ in that they each legislate different sets of concepts, but their
territories are identical because both sets of concepts concern the objects of
possible experience as appearances (5:174). The difference between speculative
and practical cannot lie in distinguishing any corresponding sets of things,
because these sets are coextensive or at least overlap.
Alternatively, the set of objects in (B) can be understood as different domains
construed not as substantial objects but as merely different subject matters. In
the Collins lectures on metaphysics, Kant insists that theoretical and practical
have distinct objects: Practical philosophy is practical not by form, but by the
object, and this object is free acts and free behavior (27:243, Kant uses the

The place of ethics in Kants philosophy

71

latinate Object rather than Gegenstand). The object of practical philosophy is not
any kind of substantial object but is instead said here to be free acts, which as
noted in the previous paragraph have their effect in nature and can be understood
in relation to natural causes. The free acts, however, are a different subject matter
than objects in nature. This interpretation is the better way to construe (B), but it
also turns out by itself not to be specic enough to be helpful since it merely says
that they have different domains, one nature and the other morality, without
characterizing the basis on which to separate them. The quote identies morality
as free acts and free behavior which will be identied as the domain after more
assessment of the nature of the practical.
The third contrast (C) does not itself specify a domain but only a relation
between reason and its objects that can ground a division of subject matter.
Theoretical reason is given its objects by intuition while practical reason
creates its objects. They could both concern the same things in appearance
provided that those objects are subject to both kinds of relations. Practical
reason can be said to create objects in two ways: in the rst way, practical
reason does not actualize its objects; that is, it does not create those objects as
empirical objects, but creates the concept of that object (in some cases, these
are the ideas of reason18); in the second way, practical reason does actualize its
objects in the empirical world. Only in the second way can practical and
theoretical reason relate to the same objects in different ways. In the Dialectic
of the second Critique, Kant identies the objects created by practical reason
as actions in the sensible world and the highest good, which is the nal end
of our right acts (5:134). In particular, the performance of actions by the will
creates objects in the realm of appearances that are then subject to theoretical
reason, which is given the effects of the will in appearances to integrate into
knowledge of nature. Regarding both objects of practical reason our right
acts and the highest good practical reason can be said to create the concepts
of those objects. Through the categorical imperative, moral agents conceive of
the right acts they ought to perform. Likewise, in conceiving of the highest
good, practical reason creates the concept of happiness necessarily connected
with virtue (5:110111). But using these concepts, practical reason seeks to
actualize right acts and the highest good in the world; these activities create
practical objects corresponding to those concepts. Particular right acts of
course occur in the empirical world and are thereby actualized. The highest good,
18

I use the term concepts to discuss that which reason can create because the particular actions
guided by practical reason would be understood in conceptual terms. Certainly, any empirical
ends that would be the goals of actions in accordance with practical reason would be conceptualized as well. When Kant labels the concepts of reason as ideas in the Critique of Pure
Reason (A311/B368), he is not considering practical reason and relations to actions at all. Some
of the concepts of objects to be created by practical reason are ideas, in particular, the
highest good.

72

Laying the ground

Kant says, is practical, that is, to be made real through our will (5:113).
Practical reason creates not only the concepts of right acts and the highest good,
it also then uses them as a guide in specic actions in the sensible world which
perhaps work toward but do not fully actualize the highest good.
The claim that practical reason creates its own objects resembles the claim
of constructivists that various principles, properties, or objects are constructed.
Of the specics mentioned by Kant, the actualization of the idea of the highest
good would have to be constructed in an obvious and unimportant sense of
being the result of actions which themselves are of course created as empirical
events in another non-controversial sense. The concept of the highest good is
created by reason when nite rational agents look for a way to reconcile their
moral obligations with their desire for happiness. As such, it stems from the
nature of moral agency rather than any decision procedure and is not constructed. Besides the idea of the highest good, the postulates of practical reason
are ideas created by practical reason (irrespective of the status of their objects).
The fourth contrast (D), between knowledge and action, is not about objects
or relations to objects but about two different ways that reason can comport
itself. This contrast is related to the previous one because reasons creation is
done in part through action and reasons being given its objects through
intuitions provides reason with material to comprehend. The comparison is
not perfect since knowledge can involve creation by reason of speculative
ideas in order to aid the understandings cognitions (A702/B731) and Kant
sometimes talks of practical cognition (5:20). The key difference is that in
both knowing and action, reason has principles or ideas such as the moral law
or the idea of God, but it uses them for different purposes. The two purposes
relate to the is/ought dichotomy already discussed.
This contrast between the theoretical and practical ts best with the domain
of philosophy as a whole and also best justies the internal division within
philosophy that provides its basic structure. Philosophy is identied as the
science of the relation of all cognition to the essential ends of human reason,
of which the highest is the nal end of human beings (A83940/B86768). To
reach those ends, reason must prescribe actions, but in order to act rationally,
reason must know the world in which it acts. Theoretical philosophy provides
knowledge essential to the actions governed by practical philosophy. The basic
division of philosophy into theoretical as knowledge and practical as action
stems from the domain of philosophy as a whole. This result meets Kants
requirement for a rational system of knowledge in which one idea determines
the position of the parts a priori (A832/B860).
I will show here that this fourth contrast also best ts with the identication of
the specic domain of practical philosophy as free acts. Action is the fundamental
characteristic of practical philosophy. Still, the other three contrasts, each also
reect aspects of the practical even if they are not fundamental. In focusing on

The place of ethics in Kants philosophy

73

action, practical reason creates certain concepts as well as rules for free action.
Actions are viewed as created rather than given and so not initially identied as
part of nature to be known; hence, they can be considered as the distinct objects
(as subject matter) of morality as opposed to given objects in nature to be
cognized. At the same time, actions do occur in nature and are secondarily
integrated into the causal series in nature as something given through experience.
Qua actions to be performed, they embody the ought rather than the is of
theoretical philosophy.
Free acts as the domain of ethics
Kant identied free acts as the domain of ethics, that which is to be systematized. He does insist upon the systematization of ends in the kingdom of ends
in Groundwork II, and the systematic relation between virtue and happiness as
part of the highest good in the Critique of Practical Reason, but neither of
these constitutes the complete domain of the practical.
The most explicit argument that Kant gives for the claim that free acts
constitute the domain of ethics is in his course lectures on ethics. He apparently
always began his lectures on practical philosophy by distinguishing the practical from the theoretical (Powalski 27:96; Collins 27:243; Kaehler 3; Mrongovius 29:597; Vigilantius 27:479). For simplicity, I will refer to the Collins
lectures of 178485, which provides the clearest formulation of the distinction
between theoretical and practical philosophy and the domain of the latter, but
the other lectures and even the Groundwork Preface make the distinction and
at least hint at the domain.
The initial denitions Kant provided for the theoretical/practical distinction
on the rst day of his Collins ethics course provides a fair summary of Kants
view of the division:19
All philosophy is either theoretical or practical. Theoretical philosophy is the rule of
knowledge, practical the rule of behavior in regard to free choice. The difference
between theoretical and practical philosophy is the object. The one has theory for its
object, and the other practice. (Collins, 27:243)

19

When reading Kants lectures, one must always be wary that Kant sometimes follows his
textbook closely, so attributing to Kant any structural and denitional passages from the lectures
without checking the text as its possible source is risky. In this case, the distinction clearly stems
from Kant himself because the textbook he was using, Alexander Baumgartens Initia
philosophia practicae primae, begins by discussing practical philosophy as a part of philosophy
without distinguishing theoretical philosophy by name; neither does Baumgarten dene the
domain of practical philosophy as Kant does: for Baumgarten, philosophy in general is the
science of our knowledge of the properties of things, and accordingly practical [philosophy]
is the science of our knowledge of the obligations of human beings (19:9).

74

Laying the ground

Kant describes the practical as the rule of behavior in regard to free choice.
This reveals the link between the nature of practical philosophy and freedom.
But the specic conception of freedom is not obvious and must be examined.
The passage mentioned here claims that theoretical and practical philosophies have distinct objects. Kant goes on to clarify that to say that they have
distinct objects contrasts with a distinction one might make in which theoretical and practical philosophy differ only according to form. That is, given a
single object of study, theoretical philosophy can relate to it in one particular
manner, namely, nding concepts of the objects, while practical philosophy
can relate to the same object in another manner, namely, exercising knowledge of the objects (27:243). Kant gives the example of theoretical and
practical medicine, the former presumably consisting in knowledge of health
and bodily conditions, the latter in application of that knowledge to maintain or
alter the state of health. Both share the same object. But this difference
according to form is not what Kant intends by his distinction. Rather, he
insists that theoretical and practical have distinct objects (Object): Practical
philosophy is practical not by form, but by the object, and this object is free
acts and free behavior. The theoretical is knowing, and the practical is behaving
(27:243).
The contrast with theoretical philosophy provides the support for a characterization of freedom at issue. Theoretical philosophy is about knowing what
exists. The practical must have a distinct object, so its object must be other
than knowing what exists. The practical must concern behavior. But since the
behavior at issue cannot be known for then, it would be the object of
theoretical philosophy the behavior in question cannot be understood empirically, as an event whose causes can be known. A study of causes of behaviors
would fall under psychology as the theoretical science or knowledge of the
mind. Practical philosophy must abstract from possible particular causes of
behavior. It must look at an act not as something to be known but as something
to be decided.
But this is the same as saying that all possible choices are open in deliberation as possible courses of action. The individual must view herself as not
constrained by any determinant of her decision-making. Kant puts it this way
in the Collins lecture:
If I abstract from the particular matter in hand, the philosophy of behavior is that which
gives a rule for the proper use of freedom, and this is the object of practical philosophy,
without regard to particulars. So practical philosophy treats the use of free choice, not in
regard to particulars, but independently of all of them. (27:243)

By looking at behavior independent of all particulars, practical philosophy is

looking at free choice independent of any and every particular object that could
determine acts. This conception of freedom serves as the domain of practical

The place of ethics in Kants philosophy

75

philosophy. Here the direction Kants argument takes can be seen; I had earlier
noted that the domain of a science could be an empirical concept (as matter for
physics), but now that option is precluded for the domain of practical philosophy.
The specic meaning of free choice must still be claried. As a starting point
for ethics, one cannot assume that it can be dependent on theoretical philosophy, so the nature of free choice must be internal to practical considerations.
This view of freedom is the nature of the free power of choice facing decisions
for actions from a perspective abstracted from any consideration of determination in nature.
An outline of an argument in a Reection most likely from the mid-1770s that
Kant labels Metaphysical Concept of Morality begins with this same conception of freedom of the power of choice:
We do not abstract the concept of freedom from experience. When we want to act, we
consider the prospective action as completely problematic with regard to the present
moment, and the ought is a condition of the consensus of the prospective action with
reason, which is thus not at all in a pre-determined interconnection with appearances,
i.e., with nature. (R6854; 19:180)

Freedom is here considered as the apparent indeterminacy of the outcome of

reection and decision from the point of view of the reecting individual.
Kants language is reminiscent of his argument from the Collins lecture in
which he held that human beings must see themselves as free from determination by any particular empirical ground in deliberation. This conception of
freedom is also similar to the famous claim in the Groundwork that to every
rational being having a will we must necessarily lend the idea of freedom also,
under which alone he acts (4:448). This is an agent-perspective conception of
freedom. This agent-perspective conception of freedom is independent of any
metaphysical claims regarding a proof of freedom. It merely functions as a
description of the agents viewpoint in decision-making, roughly that the
process of reecting and deciding is open-ended and within the control of
the agent. Without this assumption, agents would not experience their deliberation as deliberation. As Kant puts it, One cannot possibly think of a reason
that would consciously receive direction from any other quarter with respect to
its judgments, since the subject would then attribute the determination of his
judgment not to his reason but to an impulse (4:448).
There is an extremely important consequence of this reliance on the agentperspective conception of freedom for comprehending the nature of practical
philosophy as a whole, and subsequently for resolving some of the issues about
moral realism in Kant. First-person freedom does not entail actual independence
from causal necessitation. Actual independence would have to be shown by a
different kind of argument, namely, a theoretical argument. The relation between
rst-person freedom, which might be illusory but is inescapable as a practical

76

Laying the ground

self-ascription, and actual freedom, which would need independent verication,

is crucial for understanding the nature of practical philosophy.
Human beings, then, experience their decision-making deliberation as rstperson freedom. A natural fact about human beings is that they face decisions
about what they ought to do and that their own conscious deliberation appears
to them to determine their actions. Causal explanations in anthropology or
psychology do not substitute for deliberation. Nothing empirical can be understood by the agent as controlling her deliberation. These factors allow Kant to
derive the requirement that any moral law to be used in that deliberation must
be a priori.
Given that the domain of practical philosophy is free acts understood as free,
deliberative decisions seen from the perspective of the agent, some aspects of
Kants methodology in ethics and its relation to metaphysics are clear. First,
free acts understood from the agent perspective have no ontological status.
I had said earlier that the difference between theoretical and practical as a
difference between what is and what ought to be is vague, but that it was clear
that there would be no ontological claims given by the practical. With the
domain of ethics understood as agent-perspective free action, the lack of
ontological conclusions is seen as twofold: not only is there no object involved
that can be said to exist, there is also a priority of the agents perspective over
any theoretical understanding even of the nature of action. This priority will
form the core of Chapter 3.
Second, the relation between the systematization of the contents of this
domain and the transcendental method can be specied. I earlier argued that
Kants transcendental method would have a place in ethics. Now I can say
that the transcendental argumentation in practical philosophy is directed
toward nding the conditions under which free actions will be possible
considered in an agent-perspective manner. The transcendental condition for
free action in the deliberation of agents will be the availability of some nonempirical basis upon which the agent can make a choice without having to
consider that agency as determined by empirical desires and inclinations.
That non-empirical basis will be pure practical reason itself, the subject of
Chapter 4.
Concluding remarks
At the beginning of this chapter, I noted that the broadest questions one might
ask about any ethical theory include why is there any such thing as ethics at
all? Kants answer can now be seen. Only because free acts that must be
understood in a rst-person perspective exist must there be a science that
attempts to systematize that domain. If no such free beings existed, there
would be no ethics and correspondingly no philosophical study of ethics.

The place of ethics in Kants philosophy

77

In order to identify ethics, I looked at the even broader question of the

methodology and subject matter of philosophy as a whole. Philosophy requires
systematic organization of cognitions related to a particular subject matter or
domain. Philosophys own domain is the essential ends of human reason. That
domain will be systematically organized, and I identied ve possible ways
that an initial division of philosophy will begin that systematicity. I discussed
three of those in detail. The division between critique and system on the on
hand and the division between pure and empirical cognition on the other hand
worked together, as the transcendental method used in critique separated
the pure from the empirical, laying the ground for the system of pure cognitions.
I showed that this transcendental method is compatible with metaphysical
naturalism. The third division I examined was that of theoretical and practical
philosophy. The main meaning of that divide is the difference between knowledge and action, and a closer look at Kants identication of the specic domain
of the practical as free acts seen from the agents perspective t with that
characterization. The coming chapters will now employ the specic subject
matter of ethics and its relation to transcendental method in order to answer in
more detail the main questions of this book.

Part II

Practical reason in nature

The priority of the practical and the fact of reason

The practical possesses so much more importance than the speculative or

theoretical for Immanuel Kant that he claims in the Critique of the Power of
Judgment that the true value of all the efforts of the latter pursuits lie in their
usefulness as guardians of morality as the nal end of the world (5:4423). We
have also seen that in the Critique of Pure Reason, he identies the domain of
philosophy with a practical aim. His related famous claim in the Critique of
Practical Reason that practical reason has primacy over speculative reason
(5:119121) declares that, while the two uses of reason cannot reach conclusions that contradict one another, the former may be entitled to make metaphysical claims that the latter must accept. I will show in this chapter that
Kants thought contains two distinct conceptions of the priority of the practical, one explicit and one implicit, and that attending to this distinction will
reveal an important point about any possible practical ontology. Once I have
shown the value of this distinction I will apply it to an interpretation of the fact
of reason. I will show how the fact of reason, which I identify as the experience
of the commanding nature of the categorical imperative, arises in actual
situations in which moral agents face moral choices. The fact of reason ts
in this way with the nature of practical reason as concerned with free acts
understood from the agent-perspective. I will apply this understanding of the
fact of reason to the two kinds of priorities for practical reason.
Two priorities of practical reason
A close analysis of Kants use of the priority of the practical will reveal that
Kant offers the two independent conceptions of priority. The rst, his explicit
doctrine as revealed in the section On the Primacy of Pure Practical Reason
in the Critique of Practical Reason, I will call the priority of belief. The
priority of belief makes the following claim: when reecting upon the conditions of human morality, ones concern is with reconciling the metaphysical
claims, the postulates, arising from the presuppositions of the practical use of
reason with the limitations on speculative claims detailed in the critique of the
speculative use of reason. To say that the metaphysical claims of the practical
81

82

Practical reason in nature

have priority is to say that reason may insist on these postulates although
speculative knowledge can neither conrm nor deny them. The second, implicit conception that emerges when Kant discusses particular examples of
practical action, I call the priority of action. The priority of action involves
the following claim: when acting, humans are confronted with the categorical
imperative as a command obligating them to act in certain ways. To say that
the practical has priority is to say that speculative metaphysics is irrelevant to
practically oriented action, so that even a speculative denial of the postulates
would not affect practical reason. Kants discussions of the relation between
practical and speculative in fact contain these two conceptions. I will argue that
the former and not the latter conception actually relies on a higher priority of
theoretical reason. I will also argue that the latter conception, Kants implicit
view, is the only one that genuinely asserts the priority of the practical and best
ts with his conception of the nature of practical reason as discussed in my
previous chapter.
The priority of belief
In analyzing the concept of priority, one must rst note that Kant never directly
claims that practical and speculative reason contradict one another. The nature
of priority cannot lie in holding some proposition P (asserted as true by
practical reason) to be true and holding some proposition not-P (asserted as
true by speculative reason) to be false. Kant holds that a condition of the
existence of reason is that it never contradicts itself, and since speculative and
practical uses of reason are one and the same faculty of reason, one can never
be in a position to choose between rival claims to the truths of P and not-P
(5:120).
The basis for a need for priority lies elsewhere in what Kant calls the
interests of practical and speculative reason. Kant denes interest of a
faculty of the mind as a principle that contains the condition under which
alone its exercise is promoted (5:119). He says that the interest of the
speculative use of reason consists in the cognition of the object up to the
highest a priori principles while the interest of the practical use of reason
consists in the determination of the will with respect to the nal and
complete end (5:120). These interests match the contrast between speculative and practical as knowing and acting (the determination of the will).
As principles, these interests are known through concepts1 and would
constitute a priori goals of the uses of reason. These interests as goals mean
that speculative reason is furthered only when working toward the highest

See A300/B3567 for Kants denition of Prinzip.

The priority of the practical and the fact of reason

83

possible knowledge of objects and practical reason is furthered only when

actually determining the will to its proper actions.
These interests themselves do not conict with one another, Kant says, but
do make different assertions about certain theoretical positions (5:120),
namely, the reality of the objects freedom, immortality, and God. (Here
I simply discuss the postulates without specifying the precise status that they
have; in Chapter 5, I will present an argument that they make no ontological
claims about actual objects.) Speculative reason can neither prove nor disprove
the reality of these objects; practical reason requires their reality as a condition
of the possibility of morality itself. If speculative reason has priority, then
practical reasons claims can be ignored by our theoretical faculty and not
integrated into our set of beliefs about our experience. If practical reason has
priority, then speculative reason must attempt to integrate belief in these
objects into our greater set of beliefs about our experience.2
The priority of practical reason ought then to consist of practical reasons
interest in determining the will trumping speculative reasons interest in
knowledge of objects. What exactly is the nature of practical reasons claim
about certain objects that requires its interest to take priority over speculative
reason? For there to be a need for priority, it appears that one of the two
following options holds: either practical reason creates objects not already
considered by speculative reason that speculative reason must not ignore but
integrate into its set of cognitions about the world, or practical reason must
require of speculative reason agreement with some assertion not already made
by speculative reason about certain objects already considered by speculative
reason.
First, determination of the will to action could create new objects that then
would themselves be subject to the speculative use of reason. This option,
however, does not appear possible in Kants account. There is of course the
2

Paul Abela argues that the proper understanding of the primacy of the practical is one that avoids
what he calls the integrationist model in which one of speculative and practical reason
whichever is said to have primacy is required to integrate the relevant products of the other
(Abela 2002b). He discusses the paper on which this chapter is partly based, so I will respond
here. On my account, the priority of action is not integrationist in this sense, and I take it to be
just as much Kants view as the priority of belief. Nonetheless, Abela chooses to discuss my
account of the priority of belief including the higher priority of theoretical reason it implies as
representative of my view because in that article, I label it as Kants ofcial position (In this
book, I drop that label and use only explicit and implicit for the two views). The priority of
belief is integrationist, as Abela notes. He fails to stress that the integrationist need in my
interpretation is based on the need of theoretical reason to know. In the process of trying to know
what comes before it, it must try to nd consistency among the principles and concepts and
objects because consistency is a mark of a set of true propositional claims. I do not think that all
of Kants discussions of practical reason include the integrationist claim but only those that
include such a role for theoretical reason, and only because of the nature of theoretical reason.
The priority of action is not integrationist.

84

Practical reason in nature

obvious sense in which reason creates new objects through particular acts of
will that result in actions and alterations in the sensible world, but they are
unproblematically incorporated into the knowledge of nature as events in space
and time. The priority of practical reason concerns its postulation of freedom,
immortality, and God. For practical reason to require priority, it must be the
case that freedom, immortality, and God do not already fall under the interest
of speculative reason; otherwise, practical reason cannot be said to create these
objects. But Kant himself admits in many places that freedom, immortality,
and God are already objects of speculative reason (B395n; A798/B826; 5:34;
5:134). Indeed, the Dialectic of the rst Critique concerns these objects
approached theoretically, as the triad soul, world, and God. It cannot be the
case, therefore, that the priority of the interest of practical reason lies in its
creation of some objects not already subject to speculative reason.
The second possible conict between the interests of the two uses of reason
is that determination of the will to action presupposes some speculative
assertion neither presupposed nor proved by the speculative use of reason,
yet pertaining to objects of speculative reason. In this case, freedom, immortality, and God are to be subject to the interest of speculative reason; practical
reason adds only the assertion that we as moral agents must believe in these
objects. Speculative reason must then integrate this belief into its system of
cognitions.
This latter understanding of the conict between the two interests of reason
is Kants explicit position. Kant concludes that speculative reason must yield
to the demands of practical reason and take up the assertion of certain
postulates freedom, immortality, and God for integration with other beliefs
and cognitions (5:121).
The priority of belief is thus Kants explicit position given in the section of
the second Critique on the primacy of practical reason. This conception of
priority, however, seems at odds with Kants discussion of the nature of the
distinction between speculative and practical. Determination of the will to
action, which is the interest of the practical use of reason, seems tenuously
connected to the notion of belief. Further, Kant does not describe the priority
of the practical in terms of the creation of new objects. Hence, Kant cannot
have in mind the contrast between practical and theoretical as either given/
creating or knowledge/action.
Kants explicit doctrine must presuppose a higher priority of speculative
reason as reasons interest in knowing about given objects.3 Kant is operating
with an unstated assumption that reason must have interest in certain objects,
3

Marcus Willaschek has interpreted the primacy of the practical and the theoretical more narrowly
(Willaschek 2010, pp. 17375). Drawing on the material in Kants second paragraph of the
section on primacy, he understands the primacy of speculative reason only in terms of setting the

The priority of the practical and the fact of reason

85

namely, the objects of the postulates, not for determining right acts but for
understanding the relation between right acts and the rest of our experience.
Hence the postulates appear not as determinants of right acts in the process of
deliberation but only when reecting on the nature of morality, the fact that
right acts do occur, and the fact that agents believe those acts ought to occur.
And for this reection, Kant must be assuming that speculative not practical
reason is at work, and hence that there is a higher priority of speculative
reason. This assumption is a theoretical, not practical, assumption. Moral
agents require knowledge of the non-impossibility of the existence of God,
freedom, and immortality, in order to understand how their right acts can be
related to their experience as a whole. The acts themselves are still determined
by practical reason alone.
One might instead construe this situation as conrming the priority of
practical reason. If practical reason is the use of reason that demands acceptance of its postulates by speculative reason, then one can claim that practical
reason has priority, because speculative reason is constrained to modify its set
of beliefs in order to incorporate the postulates of practical reason. In this way,
practical reason would have priority even concerning beliefs. I believe, however, that such a relation nonetheless reveals the priority of speculative reason
in carrying out the incorporation of all beliefs into one whole. First, these
objects (the postulates) are themselves also objects of speculative reason in the
Dialectic of the rst Critique. Thus, speculative reason is forced to deal with
their possibility as well as their possible integration into a complete set of
beliefs already on purely theoretical grounds, so practical reason does not have
priority in solely insisting that speculative reason consider these objects.
Second, practical reason cannot demand that speculative reason accept beliefs
that the latter has already determined to be false, since then integration of all
beliefs would fail as the postulates turn out to be inconsistent with other results
of speculative reason. Since it is speculative reason that does the required
integrating of beliefs and that could presumably reject beliefs stemming from

limits to what the practical can assume for its purposes. This ts Kants text at that point but does
encompass the larger issue that arises after the speculative is constrained to incorporate practical
reasons beliefs. The larger priority of the speculative lies in its interest in knowing everything,
which includes the relation between the beliefs of practical reason and the knowledge speculative
reason already has at hand. The primacy of practical reason does not entail a surrender by
speculative reason but only that speculative reason must incorporate non-speculative content.
Nathan Rotenstreich also notes as I do that Kants view of the priority of practical reason seems
to rely on a higher priority of theoretical reason (Rotenstreich 1965, p. 130). He argues that the
postulates involve the validity of existence claims and hence must be theoretical. He does not,
however, specically link this claim to the interest of speculative reason in knowledge of objects
but focuses on his claim that validity is a theoretical category. Rotenstreich also uses his analysis
to reject Kants claims about priority of the practical rather than to search for Kants other
conception of priority as I have.

86

Practical reason in nature

practical reason should they have proven false on purely speculative grounds,
speculative reason can be said to have priority.
This priority of speculative reason is exhibited in other passages in Kants
works on ethics, revealing that Kant must have understood that speculative
reason has interests in many issues that arise from practical activity. Kant notes
the interests of the theoretical in some seemingly practical issues such as
freedom in several places, although nowhere does he emphasize it. In the
Groundwork, he quite explicitly contrasts the interest of speculative reason
with the lack (in this case) of interest of practical reason in the question of
freedom: The settlement of that controversy [about freedom] does not belong
to it; instead it only requires of speculative reason that it put an end to the
discord in which it entangles itself in theoretical questions, so that practical
reason may have tranquility and security from the external attacks that could
make the land on which it wants to build a matter of dispute (4:4567). Kant
implies here that the question of freedom needs to be answered for indirect
practical purposes, namely, the defense of morality from outside attacks, rather
than the direct practical interest, namely, any use of the idea of freedom to
determine the will to right acts. And the type of reason that is to answer the
question of freedom is speculative not practical. Speculative reasons interest
in knowledge includes put[ting] an end to the discord in which it entangles
itself regarding the possibility of freedom. But the role of speculative reasons
interest in knowledge about issues such as freedom raised in the investigation
of practical philosophy is not limited to serving indirectly any practical
purpose of defending morality from outside attacks. Rather, Kant says in the
Preface to the Groundwork, speculative reason has its own interest in knowing
what it can about human morality: A metaphysics of morals is therefore
indispensably necessary, not merely because of a motive of speculation for
investigating the source of the practical principles that lie a priori in our
reason but also because morals themselves remain subject to all sorts of
corruption as long as we are without that clue and supreme norm by which to
appraise them correctly (4:38990, modied, emphasis mine). Notice that
speculative reason itself is said to have interest in investigating issues, in this
case, the source of the moral law, which we might otherwise assume belong
exclusively to practical reason.
The second Critique contains some passages making similar points. Kant
notes, for example, that speculative reason concerns itself with certain concepts raised by the practical: it is practical reason which rst poses to
speculative reason, with this concept [of freedom], the most-insoluble problem
so as to put it in the greatest perplexity (5:30). Speculative reason is put in
difculty because it must integrate beliefs required by practical reason with its
other beliefs. This integration is a speculative not a practical issue, a problem
of comprehensive and coherent beliefs not of direct determination to action.

The priority of the practical and the fact of reason

87

Because it involves a higher priority of speculation, the explicit position of

the priority of belief has the following benet: it supports Kant against the
traditional complaint by his critics that the postulates cannot be required as
presuppositions of right action without making them into part of an agents set
of motives and thus requiring the heteronomous motive of an expectation of
happiness.4 Although this objection does not have much traction for Kant
interpreters today, it is illuminating to examine it in detail because it reveals
the role of the theoretical in the postulates. The objection centers on Kants
claim that belief in freedom, immortality, and God is required by the interest of
practical reason in order to support belief in the possibility of the highest good.
The interest of practical reason is identied by Kant as determination of the will
to action. Hence, one seems forced to conclude that belief in freedom, immortality, and God are required for determination of the will to action because they
are required for the possibility of the establishment of the highest good. The
highest good, of course, involves providing each moral agent with the amount
of happiness she justly deserves. The highest good is the state that combines
supreme virtue with supreme happiness. If a moral agent requires, as a requirement for the determination of her will to action, belief in the highest good as an
actual future state brought about by freedom, immortality, and God, then that
moral agents will is being determined, at least in part, by the motivation of the
happiness that is guaranteed in the highest good. Since this belief is part of the
required set of expectations of the agent, according to the objection, it can be
said to be part of the agents set of motives. In other words, the objection states
that priority of belief in the postulates and the highest good violates Kants
demand that moral agents must act from the motive of duty alone.
My analysis of the priority of belief points to the best response to this
objection. Given the explicit view of the priority of the practical as actually
requiring a higher priority of the speculative, Kant can claim that belief in the
actuality of freedom, God, and immortality is not required by practical reason
for its own interest of determining the will to action but only required by
speculative reason in attempting to integrate right acts into its knowledge of
the world to t its interest in knowledge. This point lies behind Kants otherwise
abstruse explanation of the moral necessity of belief in the existence of God:
It is well to note here that this moral necessity is subjective, that is, a need, and not
objective, that is, itself a duty; for, there can be no duty to assume the existence of
anything (since this concerns only the theoretical use of reason). Moreover, it is not to
be understood by this that it is necessary to assume the existence of God as a ground of
all obligation in general (for this rests, as has been sufciently shown, solely on the
autonomy of reason itself). What belongs to duty here is only the striving to produce
and promote the highest good in the world, the possibility of which can therefore be
4

See, for example, (Beck 1960, pp. 24244) and (Wood 1970, pp. 3868).

88

Practical reason in nature

postulated, while our reason nds this thinkable only on the presupposition of a
supreme intelligence; to assume the existence of this supreme intelligence is thus
connected with the consciousness of our duty, although this assumption itself belongs
to theoretical reason (5:1256).

To assume the existence of God is not a duty because one cannot have duties
to believe theoretical matters but only to perform actions. The only duties
one has are those that follow directly from the autonomy of reason, namely,
the duties required by the categorical imperative (5:3033). Since these duties
involve working toward the realization of the highest good, one can be said
to have a duty to further the highest good. But any belief in an actual realization of the highest good involves the theoretical claim that God exists. Practical
action implies belief in God but does not conrm the existence of God. One
can be said only implicitly to believe that God exists because ones actions,
when analyzed, are seen to require the existence of God for the realization
of their ultimate aim. But even this conclusion reveals the higher priority of
theoretical reason because even this conclusion involves not determination
of the will to right acts but reection on the determination of the will to right
acts. What Kant calls the moral necessity of belief is a product of theoretical
reection not a prerequisite to action. Since it is not a prerequisite to action,
belief in the postulates is not a motive to action. Kant can thus avoid the
problem some see in his linking the highest good to moral action because the
link is theoretical rather than practical and thus independent of the determination of the will to right acts.
The priority of belief, Kants explicit version of the primacy of practical
reason, thus actually involves a higher priority of speculative reason. Only
when reecting on the circumstances and aims of moral action does one ask
about the prerequisites for realization of the highest good. This reection, itself
speculative, nds reason integrating concepts related to human moral action
with other theoretical matters.
The priority of action
Since the priority of belief actually involves a higher priority of the theoretical, genuine priority of practical reason over speculative reason can occur
only in the other conception of priority, the priority of action. The priority
of action better captures the interest of practical reason in determining the
will to action because only the priority of action involves use of the
categorical imperative. In acting, humans are aware of the categorical
imperative as a command obliging them to act in certain ways by adopting
certain maxims. To say that the practical has priority is to say that the
results of speculative reason are irrelevant to the determinations to action of
practical reason.

The priority of the practical and the fact of reason

89

The priority of action, although never directly stated by Kant, seems to lie
behind many passages he uses to support various arguments. At times, he
separates the practical approach to action from any theoretical issues arising
from an attempt to integrate a conception of action with other beliefs and
cognitions. At other times, when he offers examples of right or wrong acts, he
consistently treats them independently of the theoretical considerations
involved in the priority of belief. When acting, all these passages seem to
say, a moral agent simply does not need to know whether or not speculative
reason precludes the possibility of freedom, God, and immortality. A review of
these passages will show that Kants thought did possess the essential elements
of the priority of action.
Kant hints at the priority of action in several places. In the Antinomy chapter
of the Critique of Pure Reason, he describes the irrelevance of theoretical
considerations for the practical task of doing and forbearing:
If a human being could renounce all [architectonic] interests . . . such a person would be
in a state of ceaseless vacillation. Today it would strike him as convincing that the
human will is free; tomorrow, when he considers the indissoluble chain of nature, he
would side with the view that freedom is nothing but self-deception and that everything
is mere nature. But now if it came to be a matter of doing or acting, then this play of
merely speculative reason would disappear like the phantom images of a dream, and he
would choose his principles merely according to practical interest (A475/B503).

The practical purposes of action trump speculative considerations of the possibility of freedom. When faced with a moral decision, the agent makes his
choice only according to the practical interest of determining the will to action
and not according to any concepts or beliefs of speculative reason. Kant does
not say here that an agent would stop reversing positions by settling on one
position within speculative reason; rather, it appears that the agent stops
speculating at all. A similar passage in the Canon states that the question of
freedom of the will does not concern the practical sphere . . . it is rather a
merely speculative question, which we can set aside as long as our aim is
directed to action or omission (A803/B831; see also 5:172).
His opening paragraph in the Critique of Practical Reason also seems to
assert this priority of the action: If as pure reason it is really practical, it proves
its reality and that of its concepts through the act, and all subtle reasoning against
the possibility of its being practical is futile (5:3, modied). The reality of pure
practical reason is proved not theoretically but in instances of its use. Practical
reason is here given priority because its actual use in action trumps any
speculation regarding its possibility or impossibility. Theoretical rationalizing
is futile when practical reason asserts its priority in action.
Note, however, that in these passages, Kant does not replace a theoretical
basis for inquiry into the possibility of freedom with a practical demand for
belief in freedom as the basis for theoretical inquiry. Had these passages taken

90

Practical reason in nature

that approach, they would accord with Kants explicit doctrine of the priority
of belief. Rather, in both cases, Kant asserts that the theoretical questions
become irrelevant from the standpoint of the moral agent faced with a decision. The agent would choose his principles merely according to practical
interest, and practical reason proves its reality and that of its concepts
through the act independent of futile speculation. The agent would decide
on an action taking into account the dictates of the categorical imperative, not
any beliefs she might hold, because the categorical imperative and not any set
of theoretical concepts is the means by which practical reason determines the
will to action.
The priority of the practical as a priority of action also lies behind Kants
famous claim from the third section of the Groundwork that Every being that
cannot act otherwise than under the idea of freedom is just because of that
really free in a practical respect, that is, all laws that are inseparably bound up
with freedom hold for him as if his will had been validly pronounced free also
in itself and in theoretical philosophy (4:448). In this passage, unlike the two
quoted earlier, Kant attributes to the moral agent, an assumption of freedom:
when a rational being faces moral actions, she must assume that her decision is
free rather than determined by conditions beyond her control. But, Kant says,
her assumption that she is free sufces for the practical reality of her freedom.
As long as she considers herself as free, she is justied in treating herself as if
she were free in her decision. She is not, however, postulating freedom in any
theoretical sense. Kant intends here to distinguish freedom as valid for theoretical philosophy from freedom as sufcient for the practical interest.
Several examples Kant uses to illustrate various aspects of his moral theory
also illustrate the priority of action. In the Critique of Practical Reason, after
arguing that freedom and the moral law reciprocally imply one another, Kant
offers the example of a man commanded by his sovereign to provide a false
deposition who recognizes his freedom to refuse this immoral command. He
judges, therefore, that he can do something because he is aware that he ought
to do it and cognizes freedom within him, which, without the moral law, would
have remained unknown to him (5:30). This person recognizes his freedom
without hesitation when confronted with his sovereigns command to provide a false deposition. And later, after reviewing the possibility of a noumenally free decision underlying the entirety of ones phenomenal character, Kant
turns to conscience to show that our moral experience of placing blame on
ourselves is in agreement with this conclusion; since conscience involves no
theoretical reection but is in perfect agreement with the results of theoretical reection on the possibility of freedom (5:98), conscience must be connected with practical reason rather than theoretical reason. But it appears that
conscience does not require any conrmation of the possibility of freedom
by the theoretical reason, rather, it independently conrms the results of

The priority of the practical and the fact of reason

91

theoretical reection. The practical interest of conscience in determining

responsibility for acts has priority over theoretical reection about our practice
of determining responsibility for acts.5
Given these examples of the priority of practical reason as an independence
of determination of the will from theoretical considerations, one wonders
whether the Critique of Practical Reason and Groundwork of the Metaphysics
of Morals are, in fact, works of practical reason. In my previous chapter,
I noted that the theoretical/practical divide is not limited to philosophys
systematic metaphysics of nature and of moral but can be seen to extend to
the critical level as well.
Admittedly, Kant himself is not consistent in whether he treats critical
philosophy as itself divisible into theoretical and practical. He sometimes
claimed the Critiques to be critical but not theoretical or practical, particularly
in the First Introduction to the Critique of the Power of Judgment (20:195).
Certainly, the third Critique does not t into that division except in the very
broad sense that all philosophy ultimately has a practical aim. The Critique of
Practical Reason, then, might be critical only and neither practical nor theoretical. But such a conclusion would confuse two different concepts: reason and
philosophy. When in these passages, Kant contrasts critique on the one hand
with the theoretical and practical on the other hand, he explicitly distinguishes
critique from the system of philosophy that is divided into theoretical and
practical philosophy, the metaphysics of nature and metaphysics of morals,
respectively. He claims that his critical works are works of reason, which itself
has only speculative and practical uses. In the Critique of Pure Reason, Kant is
clear in his insistence that critique is performed by reason itself (Axi-xii).
Reason as a faculty still has speculative and practical interests, which result in
theoretical philosophy, practical philosophy, and critique. In this sense, the
critical works, as I will show, can be seen as works of speculative reason. To
the extent that a critical work is aimed at knowledge, it is speculative. To the
extent the work is aimed at determining free acts, it is practical. Kant offers
compelling reasons to view the second Critique and the Groundwork as
themselves partly theoretical.
Kant did sometimes describe his works in moral theory as meeting a
practical need. To be practical, the second Critique and the Groundwork would
5

An anonymous reviewer of this book suggests that one might take these examples as evidence
for the theoretical claim of freedom, under the principle that ought implies can, and that thus,
these examples are not limited to action. I discuss the kind of freedom associated with ought
implies can in Chapter 6. Here I will point out that the kind of evidence in these examples
remains at the rst-person practical level. The agent judges that he can do the required deed. The
agents practical belief that he ought to do x does not provide evidence for a theoretical claim that
he can; rather, the principle of ought implies can would only lead the agent to believe that he can.
Any inference from ought to can from the agents perspective must retain that practical status.

92

Practical reason in nature

have to contribute to the determination of the will to action. Kant provides

precisely this account of the aims of the Groundwork when he says at the end
of its rst chapter that the common human reason, while generally accurate in
its moral reasoning, can sometimes err due to the mixture of inclinations with
rational motives. In this way common human reason is impelled, not by some
need of speculation (which never touches it as long as it is content to be mere
sound reason) but on practical grounds themselves, to go out of its sphere and
to take a step into the eld of practical philosophy (4:405). The passage from
the Preface to the Groundwork that I quoted earlier also conrms this practical
role alongside a speculative role (4:38990). At least the Groundwork, then, is
to clarify the source of the moral law in order to sharpen the ability of common
human reason to allow itself to be determined by practical reason.
The Critique of Practical Reason, however, is a different type of book that
does not concentrate on detailing the moral law itself but rather focuses on
theoretical issues that arise from the fact of human morality. The theoretical
nature of the second Critique itself is manifest in two ways: rst in the
occasional particular references to its theoretical nature, and second in Kants
general pronouncements about the soundness of common moral practice.
One example of the rst case appears after Kant reviews the possibility of a
noumenally free decision underlying the entirety of ones phenomenal character. As I have already noted, Kant immediately turns to conscience to show that
our moral experience of placing blame on ourselves is in agreement with this
conclusion (5:98). Kant rst provides a theoretical argument based upon
concepts provided by the practical use of reason, then turns to conscience,
which independently agrees with the results of these theoretical considerations.
A similar, although much more concise, approach is given in the rst paragraph of the Preface that I also quoted earlier and in which Kant calls on
experience to conrm the existence of pure practical reason (5:3). Note that in
both of these examples, the moral agent immediately recognizes her freedom
without any need for theoretical reection. Kant does not rst present the
examples as providing puzzles that cause the moral agent to reect upon the
possibility of freedom, then provide his theoretical argument about the conditions for freedom. He instead rst presents a theoretical argument with its own
conclusions, then presents examples drawn from experience in which the same
conclusions are independently demanded by practical reason in action. This
shows that the main argument in the second Critique is theoretical, albeit
concerning the practical use of reason, rather than practical in the sense of
action-guiding.
The discussion of the moral necessity of freedom also provides support for
the claim that the Critique of Practical Reason is a work of speculative reason
(5:30). We are given moral laws by pure practical reason, and this moral law
then reveals freedom to us. The problem of freedom is given to the speculative

The priority of the practical and the fact of reason

93

by the practical: it is practical reason which rst poses to speculative reason,

with this concept [of freedom], the most-insoluble problem so as to put it in the
greatest perplexity (5:30). Noting that a large part of the second Critique as
well as the third section of the Groundwork involve discussions of precisely
the issue of freedom, one seems justied in labeling both of them works of
speculative philosophy. Further support of this view appears when Kant points
out in the Preface to the second Critique that his readers can understand some
perplexing and problematic concepts from the rst Critique, which reappear in
the second, by comparing their old with their new employment (5:68). The
concept of freedom in particular, he says, is comprehended this way; Kant
even advises the reader to pay close attention to his discussion of freedom in
the nal section of the Analytic (Critical Elucidation of the Analytic of Pure
Practical Reason 5:89106). These discussions by Kant of how the reader can
comprehend the proper relation between the practical and the theoretical are
themselves theoretical, based on a need to integrate beliefs and cognitions,
rather than practical, designed to determine the will to action.
The second support for the claim that the Critique of Practical Reason is
itself a work in theoretical not practical philosophy comes from Kants discussion of the relationship between his work on ethics and common moral
practice. In many places, Kant praises the sufciency of the common view
of morality (Bxxxiiiii; 5:35). Human moral agents do not require any new
principle of morality and they are not ignorant of their duty as it stands (5:8n).
Moral agents need very little help from philosophy in realizing the practical
interest of determining their wills to right acts. Kants primary purpose in
writing his works on ethics, then, is not to further that practical interest but to
explore the theoretical issues related to ethics.
Despite the theoretical nature of much of Kants moral theory, some of the
objects treated in those works, namely right acts and the highest good, remain
objects of practical reason. The postulates of freedom, immortality, and God
are in contrast, rst objects of speculative reason and are so treated in the
Critique of Pure Reason. Recall that one difference between theoretical and
practical reason lies in that theoretical reason is concerned with given objects
while practical reason creates its own objects. This emphasis on the relation
means that it is possible for theoretical reason and practical reason to relate to
the same objects. Now recall that practical reason demands that certain actions
occur through the obligatory force of the categorical imperative. The actions
thus created and the goal of those actions, the highest good, are created by
practical reason. But it does not follow that these objects are of concern to
practical reason alone. The most obvious case is that once the actions have
been performed, they are part of nature to be understood theoretically in
accordance with natural laws. And once practical reason has created the idea
of the highest good, it is within the scope of theoretical reason to attempt to

94

Practical reason in nature

explain its possible relation to other objects. Hence the concept of the highest
good, while created by practical reason, could be considered an object for
theoretical reason. In fact, I claim, this is precisely what occurs in Kants
investigation of morality and precisely the nature of the Critique of Practical
Reason. The second Critique is not demanded by the interests of practical
reason but only by the interests of theoretical reason. Practical reasons sole
concern is with the determination of actions.
At least some of the works or parts thereof that we regard as Kants practical
philosophy, therefore, are themselves works of theoretical philosophy about
the practical use of reason. Given that Kant was more a reective philosopher
than an active agent in world or local affairs, we might expect as much. The
philosophical issues surrounding a practical point of view are treated theoretically by Kant. But, ironically, one of those theoretical results may be an
understanding of the priority of the practical as the priority of action, Kants
implicit doctrine, which allows practical reason to function independently of
any results of theoretical reason.
Action and the nature of the practical
Given that Kant had two different senses of the way in which practical reason
has priority, it is legitimate to call both of them Kantian. The priority of
belief preserves the unity of reason at the cost of denying genuine priority to
the practical; the priority of action preserves genuine priority of the practical at
the cost of denying the necessity of a unity of reason. Adopting either will
therefore have both positive and negative consequences for interpreting Kant.
Both capture elements of Kants philosophy. The interpretation I will give
shows that they can be reconciled even if they do not t at all points. This
result has implications for the realism/idealism debate, given that the priority
of action places so much emphasis on the perspective of the moral agent.
The two views are reconcilable in an interpretation of Kant that stresses the
nature of the practical as concerned only with questions of free action, of
determining what moral agents ought to choose, and not with ontological
questions at all. The work in my previous chapter detailing the nature of
practical reason bears fruit here. Practical reasons focus on action rather than
knowledge shields it in practice from any ontological concerns. Practical
reason itself is exhibited only in the moment of decision. As Kant puts it in
the second Critiques Preface, If as pure reason it is really practical, it proves
its reality and that of its concepts through the act, and all subtle reasoning
against the possibility of its being practical is futile (5:3, modied, my
emphasis). The phrase through the acft translates durch die Tat, which
could also be translated as through the deed. Whether the precise act is that
of a moral agent deciding upon an action or the act of reason as a call to follow

The priority of the practical and the fact of reason

95

a certain principle (to be discussed in the next section), the key point is that
practical reason operates in agents actions. Practical reason arises when the
agent is pondering a decision to act and thus within the agent-perspective
understanding of free acts as the domain of ethics. Any and every concept or
idea that practical reason brings to bear on that decision must be understood
from the practical point of view of determination to action and not from the
theoretical view centered on knowledge of what is. The agents perspective, so
to speak, is framed by the action of practical reason. This agent-perspective lies
at the heart of moral experience.
Theoretical reason enters the practical picture only when the agent steps
back from that perspective and reects upon it. This is the step in the priority of
belief in which the demands of practical reason are given to theoretical reason
in order for the latter to incorporate them into a theoretical comprehension of
experience. What is given to theoretical reason, then, is (in addition to the
action as event in nature) the fact that agents do take up a certain perspective
that is determined at least in part by practical reason. Theoretical reason is not
given a set of objects such as God that it must then incorporate into the
ontology of the world but is given a set of principles and ideas that practical
reason uses in determining action. I will discuss this issue in relation to the
nature of the postulates directly in Chapter 5. Theoretical reason does not
obtain any ontological claims from practical reason but only the principles and
ideas that constitute the agent-perspective. There is, then, no possible clash
between the ontology provided by theoretical reason and that demanded by
practical reason because the latter demands no ontology.
At the heart of Kants moral theory, then, is the claim that practical
philosophy must be separated from theoretical philosophy. The basis of his
claim that moral theory must seek an a priori moral law surely a touchstone
of Kantian morality lies in the nature of morality as practical philosophy
concerned with a rule for free action in contrast to and without reference to
theoretical philosophy concerned with knowledge of what exists. This basis
must always be kept in mind when assessing particular claims in Kants ethical
theory. In general, one must always be skeptical about anything that looks like
an existence claim in Kants practical philosophy.
Another way to see the importance of this interpretation is that Kant
simultaneously allows one type of naturalism while rejecting another type.
He rejects, quite clearly, a methodological naturalism that would base
morality on something empirical, such as an empirically derived claim
about human nature. The agent-perspective view on decision-making precludes anything other than a priori practical principles. No particular empirical fact about human beings, or about the human environment, or about
nature at all, can serve as a basis for investigation into the ultimate principle of morality. This ultimate principle of morality, then, must be a priori

96

Practical reason in nature

rather than a posteriori, and must be a necessary moral law rather than just a
contingent practical principle.
While insisting on this methodological anti-naturalism, however, Kant also
shows that any resulting morality must not make existence claims; Kant
thereby at least allows for a metaphysical naturalism. The domain of the
practical must be sharply distinguished from that of the theoretical, and only
the theoretical is allowed to be about knowing what exists. Kantian morality,
then, must be prima facie compatible with a naturalistic metaphysics.
These rather broad conclusions are quite general and could be misinterpreted. The separation does not mean that practical philosophy never concerns
things that exist. Such a conclusion would be absurd, since practical laws
govern human beings, who exist, in their relations with each other and with
things, which exist. On the one hand, Kant himself insists that morality must
be applied to human beings as they exist in nature (this is the study of
practical anthropology, 4:388). Here, however, knowledge from theoretical
philosophy as a study of human beings in nature is brought into the practical
from outside, so to speak, in order to provide the material needed to apply the
pure a priori ethics to actual human beings in nature. On the other hand,
however, the parts of his pure moral theory that are not borrowed from an
empirical study of nature yet appear to make existential claims I am thinking
in particular of the postulates of freedom, immortality, and God must be
capable of being interpreted in a way that avoids any existence claims.
The interpretation I have just given will be eshed out in more detail. In the
remainder of this chapter, I want to show how the best interpretation of Kants
fact of reason ts with the role of the priority of action.
The fact of reason and agent-perspective
The interpretation of the fact of reason will deepen the discussion of the
difference between the empirical and transcendental agent introduced in
Chapter 1. Empirical agents exist in nature as individuals who undergo a moral
experience. As in Kants examples, they are faced with particular decisions
regarding moral (and of course non-moral) situations. From their agentperspective, they must consider themselves able to make that decision freely,
that is, they act under the idea of freedom. The priority of action holds that this
perspective is self-validating and need not concern itself with the metaphysical
question of whether the agent really is free. The transcendental moral agent is
the conception of agency that results from a transcendental deduction of
morality, the description of the a priori conditions for there being empirical
moral agents. This agent need not be a transcendent being who in some way
exists and acts independently of nature in space and time unless such a nonnatural ontology of agency is required by the specic a priori conditions of

The priority of the practical and the fact of reason

97

moral agency as such. The transcendental agent could simply be a conception

of the necessary structure of a moral agent in nature, including any requirements for the faculties of will or reason as well as the necessary structure of
nature (or of human moral experience) that would allow for such agency if
there is to be a valid, objective morality at all.
I place the fact of reason squarely in empirical moral agency. I will argue
that the fact of reason is the awareness of the categorical imperative that
expresses moral obligation by its nature as a command. It is an aspect of the
agent-perspective on free decisions.6 Thus, the fact of reason is experienced by
agents within the practical perspective and falls squarely within the connes of
the priority of action: when faced with a moral decision, an empirical moral
agent feels the obligation to act in a certain way and does not question the
nature, source, or validity of that categorical imperative from within that
perspective. Upon reection, when agents take up the questions of the nature,
source, or validity of the categorical imperative, they then consider the nature
of the moral law, freedom, and other issues that would explain that part of
moral experience. In this reection, they are in effect asking about the transcendental moral agent, working within the broad theoretical framework that,
as I argued earlier in this chapter, motivates large parts of the Critique of
Practical Reason and parts of the Groundwork. The nature of the transcendental moral agent is a practical concern in the framework of the priority of belief.
To show that this interpretation of the fact of reason in particular and the
surrounding issues in general is correct, I will rst discuss the nature of the
categorical imperative and its relation to the moral law in order to show that
Kant differentiates morality as humans experience it from a hypothesized real
morality. The difference between the moral law and the categorical imperative
will be shown to rest on the status of pure practical reason itself and its less
pure realization in human beings. Second, I will examine the nature of the fact
of reason directly. I will partly endorse the interpretation that stresses the Latin
root of the word Faktum used by Kant as a deed rather than a static fact. I will
argue that understanding the term to refer to our consciousness of the categorical imperative as an immediate command of reason properly incorporates both
the sense of fact and the sense of deed. Third, I will show the way in which this
interpretation of the fact of reason paves the way for an empirically idealist
interpretation of Kants ethics but also still allows for a transcendental level of
moral agency that would ground a transcendentally ideal interpretation. In both
cases, I suggest that a naturalist ontology would be sufcient to capture these
6

Pauline Kleingeld offers a very similar interpretation that stresses the agent-perspective, calling
it radically agent-based (Kleingeld 2010, p. 72). With some minor differences, we share this
general interpretation. The largest difference is the way that I relate the interpretation to the
empirical/transcendental division.

98

Practical reason in nature

elements of Kants ethics. Finally, I will examine the kind of freedom that the
fact of reason is said to reveal to us.
Categorical imperative and moral law
Although the fact of reason plays an extremely important role for Kant, he
never clearly dened it, forcing commentators to attempt to elucidate it in light
of different, even conicting, references.7 I will argue here that the fact of
reason consists of the categorical imperative as our consciousness of the moral
law. This denition involves Kants distinction between the moral law as such
and the categorical imperative.
Kant discusses this distinction in the Critique of Practical Reason and the
Groundwork. The moral law holds for purely rational beings who would use it
as the sole principle of action because they lack any other possible determinants for action. Purely rational beings have no sensuous impulses, so would
always act rationally. They would conceive of the moral law not as a command
but as a description of their principle of action. We, humans, however, can
guide our actions not only using laws of reason but also by incorporating
sensuous needs and desires. Unlike purely rational beings, we do not automatically guide our action according to the moral law. We humans encounter the
7

The initial assessment in English is by Lewis White Beck in his Commentary on Kants Critique
of Practical Reason (Beck 1960, p. 166f), where he quotes four passages from the text and
concludes that the fact is either consciousness of the moral law, the moral law itself, or
autonomy. Beck cites additional passages from the second Critique and further identies four
additional references to the fact of reason after the second Critique that, he says, fall in line with
the earlier ones. See also (Beck 2002). Darlei DallAgnol quotes eleven passages in which Kant
mentions the fact in the Critique of Practical Reason and concludes that it has a family of
meanings (DallAgnol 2012a, p. 115). Other commentators identify a single meaning for the fact
of reason, generally taking it to be the consciousness of the moral law. Dieter Henrich takes it to
be simply the demand of the good (Henrich 1994, p. 83); Henry Allison, the consciousness of
standing under the moral law (Allison 1990, pp. 23139); John Rawls, that in our common moral
consciousness we acknowledge the authority and direction of the moral law (Rawls 1999c,
p. 102); Guido de Almeida, the consciousness that nite rational beings have of the moral law as
a categorical imperative (De Almeida 2012, p. 147). Two quite different understandings of the
fact are given by Marcus Willaschek, to be discussed later on, and Ian Proops, who argues that
the fact is both the moral law itself and our consciousness of it, although he stresses the former
(Proops 2003). What makes Proops interpretation different is that he interprets Kants German
term Faktum in light of a legal proceeding in which the factum (the Latin term) is the thing
identied as the matter for a legal judgment, as opposed to the juris which is the legal basis for
the judgment. Kant uses this terminology in the Critique of Pure Reason, in his discussion of the
difference between the possession of a concept such as causality and the justication for that
concept, the latter being the subject of the transcendental deduction (A8487/B11619). Proops
claim is weakened when one notes that in the second Critique, Kant uses the German term rather
than the Latin term, a difference from the rst Critique and at odds with his normal practice in
discussing legal terminology. Pauline Kleingeld also rejects Proops reading on the basis that it
does not t with Kants own direct explanation of the fact and that it cannot explain the passages
in which Kant describes the fact in active terms (Kleingeld 2010, pp. 6465).

The priority of the practical and the fact of reason

99

moral law in the guise of the categorical imperative (5:32), a command to

guide our behavior according to rational form alone. In the Groundwork, Kant
uses the phrase Act only . . . when formulating the categorical imperative to
emphasize its status as a command. He sharply divides the conception of the
purely rational will, all of whose actions would always conform to the autonomy of the will, from the will that is in the world of sense whose actions are
subject to the ought (4:454). Only in the latter case is there a synthetic a
priori proposition, he says, allowing the reasonable conclusion that the former
is an analytic proposition describing a purely rational will.8 To reinforce the
point, Kant compares the distinction between the conception of a purely
rational will on the one hand, and a rational and sensible will, on the other
hand to the distinction between the pure concepts of the understanding, which
by themselves signify nothing but form of law as such, from their application
to sensible intuitions in a priori synthetic propositions about nature. Just as
there is an analytic proposition Every effect has a cause and a synthetic
proposition Every event has a cause, there is an analytic principle, Pure
rational will acts autonomously and a synthetic principle The human will
ought to act autonomously.
There are good philosophical grounds for this division between categorical
imperative and moral law as well. We experience the categorical imperative as
an obligation; it is not merely an intellectual awareness of a principle. Any
principle that describes purely rational behavior, however, would be merely
something we would understand intellectually because, as sensible rational
beings, we do not experience the decisions of our will as always determined by
reason. This phenomenological difference is reected in the difference
between categorical imperative and moral law. Put in another way, the categorical imperative is part of the agent-perspective when faced with a decision
but the descriptive principle of the behavior of a pure will is not.9
Human beings experience the moral law as a categorical imperative: this
I claim is the fact of reason.
Consciousness of this fundamental law may be called a fact of reason because one
cannot reason it out from antecedent data of reason, for example, from consciousness of
freedom (since this is not antecedently given to us) and because it instead forces itself
upon us of itself as a synthetic a priori proposition that is not based on any intuition,
either pure or empirical (5:31).

The categorical imperative, as a command, presents itself to us as obligatory.

We experience the call of moral obligation in the guise of the categorical

8
9

A similar division is given with regard to a holy will (4:439).

This way of differentiating the moral law from the categorical imperative expands on (Beck
1960, p. 121). Allison also employs this distinction, see (Allison 2011, pp. 27483).

100

Practical reason in nature

imperative. How this happens and what precise status the fact has requires
more explanation of the meaning and function of the fact of reason.
Fact and act
The precise way in which this imperative strikes us in our moral experience is
as a command. We take commands to issue from some authority, and this case
is no exception. The authority is assumed to be reason; hence, the fact of
reason is a command of reason of which we are conscious. We also understand
commands as acts, and since reason provides the command, we can ascribe this
act to reason. The fact of reason, then, can be understood as having a twofold
signicance as our actual conscious awareness of an act of command by
reason. The term that Kant uses, Faktum, allows for this meaning.
Every translation of the Critique of Practical Reason into English simply
translates the term Faktum as fact without even mentioning the availability
of an alternative translation of the term as act or deed. The German term
Faktum at the time was a germanization of the Latin term factum, and was
allowed to have two meanings (both related to the Latin term): fact and
deed.10 Kant uses the Latin term factum in the Metaphysics of Morals and
throughout his Reections and lectures on ethics and natural right as equivalent to the German Tat, which is best translated as deed (6:224), although at
6:371, he equates Factum with Tatsache (fact). In the second Critique, had
Kant meant fact in the sense of a piece of knowledge, he likely would have
used the German Tatsache instead of Faktum.
Marcus Willaschek has argued convincingly that the meaning deed or
act makes more sense than fact in many of the passages in which Kant
uses the term in the Critique of Practical Reason.11 These passages tend to
discuss reason as announcing itself (5:31), giving (5:31), forcing itself
(5:32), and determining the will in relation to dynamic laws that can
determine its causality (5:42). At least in these passages, it is better to
understand the Factum of reason as an act of reason in providing the moral
law. Particularly strong evidence comes from a quotation from the Critique of
Practical Reason in which Kant uses the terms Tat in a way equivalent to the
use of Faktum: For, if as pure reason it is really practical, it proves its reality
and that of its concepts through the act [die Tat], and all subtle reasoning
10

11

Neither the word Faktum nor Factum appears in the Grimm Brothers dictionary. Duden
indicates that the term Factum (or Faktum) was rst used in Germany in the seventeenth
century, and that it had meanings corresponding both to fact and to deed. Kleingeld cites
Zedlers Universallexikon from the mid-century in noting that deed was the preferred
meaning (Kleingeld 2010, pp. 6263).
See (Willaschek 1992, pp. 174193). The discussion of the Factum of reason appears
separately as well in (Willaschek 1991).

The priority of the practical and the fact of reason

101

against the possibility of its being practical is futile (5:3, modied). These
considerations at least ought to put the translation of Factum as fact in
doubt, and provide compelling evidence that Kant intended the term to refer in
some way to the activity of reason.
To see the entirety of Kants use of Faktum as deed, however, would leave
out the important element of our awareness in the discussion of the Faktum.
Pauline Kleingeld suggests interpreting the Faktum as consciousness of the
fundamental law (of pure practical reason), a consciousness that reason produces in agents rather than a fact as something alien confronting the moral
agent; the Faktum is a fact generated by a deed. She gives an analogy
comparing the fact of reason to a decision by a king or a painting by
Rembrandt: the act results in a thing that individuals can become aware of as
a product of that act.12 I agree that consciousness is necessary and believe that
Kleingeld has provided the best interpretation of the Faktum of reason yet, but
I think that she still separates the conscious awareness from the act of reason
too much. My distinction between the moral law and the categorical imperative
allows for a deeper connection between the senses of fact and deed.
A moral agents awareness of the categorical imperative is not the result of
an act of reason, it is an act of reason in commanding us to act a certain way.
There is no lag between the generation of the command to act so that ones
maxims can serve as practical laws and the awareness of the categorical
imperative. They are the same thing within the agent-perspective. In moral
experience, when the agent is faced with a free decision with moral implications, reason acts in pressing the agent with the categorical imperative as a
command.13 Unlike the decision of a king or painting by Rembrandt that can
be known well after the actual deciding or act of painting, the awareness is
simultaneous with the deed. A better analogy, and one that illustrates what
I take the status of the fact of reason to be as a kind of deed, would be to
watching a dance by a ballet dancer because the result that one is aware of is
inseparable from the current action of the dancer. One can still remember the
dance later, but the phenomenal experience of the memory is different from the
experience of the performance. Analogously, the experience of the obligatory
force of the categorical imperative during moral deliberation is different from
the later memory of that experience. The Faktum of reason is the direct
experience of the categorical imperative.

12
13

(Kleingeld 2010, p. 65).

David Sussman also interprets the Faktum as a deed rather than a fact in saying that the Faktum
is not merely revealing a metaphysical truth about the moral law, but that pure reason declares
the law to be binding, and, in this declaration, somehow makes it so (Sussman 2008, p. 68).
He does not identify it with the categorical imperative as I do.

102

Practical reason in nature

The best interpretation of the Faktum of reason, then, is that of the consciousness of the categorical imperative as a command being issued by reason
to an agent caught up in the process of deliberation. Seeing the Factum of
reason as awareness of this deed of reason instead of a static fact ts the
philosophical function of the Faktum. Kant introduces the Faktum after presenting and resolving the two problems of nding the property of the will
that makes it determinable by mere lawgiving form, the solution being freedom, and of nding the law to determine a free will, the solution being
lawgiving form. Kant concludes that freedom and unconditional practical
law reciprocally imply each other (5:29), a claim Henry Allison has deemed
the reciprocity thesis.14 Without an independent basis to ground either side
of the reciprocity, neither freedom nor law can be justied. Kant is faced with a
circle15 (although he does not use that term). When the Faktum is seen as the
awareness of the categorical imperative as a command in moral experience, it
is clear how Kant is able to nd a way to justify both moral law and freedom
despite the limits of their reciprocal derivation. The circle is broken by invoking the practical experience of reasons commanding us in the categorical
imperative.
This practical perspective is within the agents perspective. Given the
assessment earlier in this chapter of the nature of the practical/theoretical
divide and my claim that the priority of action emphasizes the practical point

14
15

See (Allison 1986) and (Allison 1990, pp. 201213).

In a part of the article from which part of the current argument is drawn, I even call it a vicious
circle of mutual dependence (Rauscher 2002, p. 490), prompting Patrick Kain to object that
there is no vicious circle and even if there were, my invocation of the categorical imperative as
the fact of reason would merely substitute a circle from categorical imperative to freedom and
back for the circle from moral law to freedom and back (Kain 2006, pp. 45354). Kain claims
that Kant avoids any circle by distinguishing between freedom as the ratio essendi of the moral
law and the moral law as the ratio cognoscendi of freedom (5:4 note), but that is precisely the
way in which I claim that Kant avoids the circle as well. Our conscious awareness of the
categorical imperative as a command of reason in our moral experience is immediately known,
prompting us to ask for an explanation of it, which leads us to the conclusion that only a free
lawgiving on the part of pure practical reason could be the basis for any valid moral imperative,
hence our awareness of the categorical imperative is the basis on which we make a knowledge
claim about freedom. But as part of this claim, we understand the nature of pure practical reason
as providing the principle for action of purely rational beings, that is, the moral law, which
would be the basis for the categorical imperative for us imperfect rational beings, hence we see
the freedom of pure practical reason as the basis for a possible moral law. Further, such a
solution does not deny that there is a circle if the derivation is restricted to the material Kant
provides in Problems I and II on 5:2829. Nowhere in the statement of those problems does
Kant invoke the ratio cognoscendi claim. Only in the Remark does Kant raise the additional
question asking from what our cognition of the unconditionally practical starts (5:29,
emphasis in original), asking, in effect, for an entrance to the circle and then providing it
immediately with the consciousness of the moral law soon identied as the fact of reason.
Problems I and II by themselves constitute a circle of derivation even if the addition of the fact
of reason allows for them to be justied using that independent consideration.

The priority of the practical and the fact of reason

103

of view from the agent-perspective, the fact of reason can be understood within
that perspective as an unquestioned contribution to moral deliberation. The
agent views herself as subject to the categorical imperative and her will as
having the capacity to be determined by the categorical imperative to action.
The moral agent, when faced with a situation that demands action, is simply
aware of the call of the categorical imperative. The agent at that time and qua
agent cannot doubt the validity of the imperative since it is experienced as an
undeniable command. The possibility for doubting the validity of the categorical imperative comes only when the moral agent is not faced with a situation
that demands action, and thus when the moral agent in not directly involved in
moral deliberation and is not caught up in the agent-perspective. At this time,
the agent engages in theoretical reection about moral deliberation itself, and
the experience of the categorical imperative is subject to philosophical examination. The priority of belief is applicable at this point since this reection
involves a higher priority of theoretical reason.
Our experience of the categorical imperative can cause us to ask in reection
about its ultimate source. One possible source is pure practical reason, that is,
the categorical imperative might be derived from the moral law as a particular
instantiation for nite sensuous and rational beings. But it is also possible
before completion of the philosophical reection that the experience of the
categorical imperative does not have its source in pure practical reason; it
might, for example, merely stem from empirically conditioned reason (5:15).
In the rst case, the explanation of the categorical imperative will invoke
transcendental considerations about rational agency as such. In the second
case, our awareness of the categorical imperative in our experience stands on
its own as a practical fact about empirical agency. Because the categorical
imperative might not stem from pure practical reason, Kant is impelled to
investigate its source. The status of the categorical imperative will depend
upon the status of its source, if any, discovered during the investigation
conducted in the Critique of Practical Reason and elsewhere. My next chapter
will look in some detail at the ways in which Kant attempts to explain the
nature of reason, both empirical and transcendental, in order to provide a
justication for the categorical imperative upon reection. It needs no justication in practice.
The fact of reason is ideal rather than real at the empirical level because it
has no ontological status outside of the agent-perspective. It is the agents
conscious awareness of the categorical imperative. At the transcendental level,
the fact of reason must be understood in relation to the broader understanding
of the nature of reason itself. If Kant is able to defend the claim that the
possibility of moral experience depends upon a certain kind of moral agent,
and that this agent must experience the moral law as a categorical imperative,
then the fact of reason has transcendental justication. But since the fact of

104

Practical reason in nature

reason at this level is not extended beyond the transcendental conception of

moral agency, it would be transcendentally ideal. At both levels, then, the fact
of reason as the consciousness of the categorical imperative is ideal rather
than real.
The fact of reason is also compatible with a naturalist ontology. Provided
that empirical psychology is able to include in its descriptions of a moral agent,
the consciousness of the categorical imperative, there is no non-natural ontology required. The fact of reason, of course, is not an empirical fact (5:31). It is
not based on any intuition or other empirical experience but is simply forced
upon us by reason. By this, I understand that consciousness of the moral law as
a categorical imperative cannot be understood as normative except from within
the agent-perspective. It would be in principle possible for empirical psychology to include as a matter of fact that an agent made a decision on the basis
of a certain representation of a law (4:412) without empirical psychology
including the normativity or obligatory force of that representation for the
moral agent. The next chapter will provide more detail on the relation between
empirical and pure reason to show how they are able to be comprehended in a
naturalist ontology.
Fact of reason and the status of freedom
The fact of reason leads directly to the concept of freedom (5:30). Freedom
is the ratio essendi of the moral law, the basis for the existence of the moral
law, but freedom is known only by means of the moral law (5:4). These claims
are often taken to mean that in the Critique of Practical Reason, Kant uses the
fact of reason as an epistemological basis for revealing the objectively real
truth that human beings are free.16 Freedom is known to be true on the basis of
the practical fact of reason. Kant uses language that appears to support this
interpretation. He says freedom is real (5:4), a moral agent cognizes
freedom within him (5:30), and freedom has objective and . . . undoubted
reality (5:49). In these and similar cases, however, Kant is not as rm in his
assertions as it seems at rst glance. In the three instances just quoted, each is
placed within a context that either limits the referent or stresses the practical
approach that, as has been shown, is concerned with the determination of free
acts rather than knowledge of what is. When he says that freedom is real, he
immediately identies it as this idea and earlier in the sentence, the concept
16

Karl Ameriks provides this analysis in (Ameriks 1982, pp. 203220). He couches this interpretation in terms of a relation to Groundwork III in which Kants attempt to provide a
deduction of the moral law by rst proving that we are free falls short. Kant is then said, in
the second Critique, to reverse the order of proof and begin by asserting the moral law as an
undeniable premise, from which Kant is then able to assert that we really are free as a
metaphysical claim, as though we have practical insight into what theory cannot prove.

The priority of the practical and the fact of reason

105

of freedom; the concept is also linked to the concepts of God and immortality
so that those concepts, not the objects corresponding to them, obtain objective
reality (5:34). The moral agent who cognizes freedom within himself does
so at the moment of decision when faced with the demand by his prince to
provide false testimony; the agent is caught in the agent-perspective from
which the agent must believe his decision is freely made (5:30). And it is
not freedom but the concept of freedom (my emphasis) that is said to have
objective and, though only practical, undoubted reality (5:49). This latter
sentence contains several caveats: not freedom itself but only its concept is
what is at issue; its reality is undoubted (unbezweifelte), implying that the
belief in the concept of freedom is inescapable rather than that freedom is itself
real; and most important, that the concept has only practical reality.
This quick look at three passages is by itself insufcient to show that
freedom is not an objectively real property of human beings. There is, however, also evidence that Kant intends freedom to be seen merely as an
assumption taken up for practical purposes. Of all the intelligibles, Kant says,
only freedom has reality for us only insofar as it is a presupposition inseparable from that law, which he glosses to on behalf of that law and of the use of
pure practical reason, implying that this reality of freedom is only from the
agent-perspective when confronted with the moral law when using practical
reason (5:70). Freedom is identied as one of the postulates of practical reason
(5:132, see also 5:45) and is subordinate to the moral law that is itself also
declared a practical postulate (5:46), clearly in a less technical sense than
Kant uses the term in reference to immortality, freedom, and God, but nonetheless something to be taken up only in a practical point of view.
When one takes into account the larger context of Kants practical philosophy, the freedom that is revealed by the fact of reason should be understood in
relation to the two priorities of practical reason. Under the priority of action,
the fact of reason is experienced by a moral agent from the agent-perspective
when faced with a decision. If freedom is understood in this way, it too would
be valid only from the agents perspective and not be an ontological claim.
Under the priority of belief, freedom would be tied to the fact of reason when
the agent is later reecting upon her experience. In this case, the immediacy of
the disclosure of freedom through the experience of the fact of reason is lost.
Reection upon the awareness of the categorical imperative would raise the
question of how such an imperative is possible, and freedom would be the
answer derived from analysis performed from speculative grounds. This kind
of analysis is provided by Kant in the second Critique, to be sure, but freedom
in this sense must be understood as one of the presuppositions for a
practical claim.
The fact of reason discloses freedom to us as a practical postulate. Kant
often uses the term freedom by itself, although he also identies freedom of

106

Practical reason in nature

the will (Wille) directly (e.g., 5:2829). The Wille, however, is identied as a
practical reason in the Groundwork (4:412), and practical reasons determining
the will through the moral law is what constitutes its freedom. In order to
understand freedom of the will, it is necessary to determine the status of pure
practical reason such that it is able to determine a free will.
Concluding remarks
While the previous chapter viewed practical philosophy from the perspective
of philosophy in general, this chapter has examined practical philosophy
directly. I showed that Kants famous claim that the practical has primacy
over the theoretical can be understood in two different ways. The priority of
belief understands the practical as raising certain possible beliefs, the postulates in particular, that stem from reection on our moral lives and presenting
them to theoretical reason, which in turn is tasked with integrating those beliefs
into its own claims about the world. This task reects a higher priority for
theoretical reason because part of its goal is to understand human experience;
reection on practical experience presents theoretical reason with content it
then tries to incorporate into overall comprehension of experience. I argued
that even the Critique of Practical Reason can be seen as a work of theoretical
reason examining the nature and use of practical reason. The priority of action,
in contrast, gives the practical independence from any possible theoretical
claim that might cast doubt on the practical. When faced with deliberation
about a choice for action, agents decide as if they are free and using the
categorical imperative of which they are conscious without requiring any
theoretical conrmation indeed, even if there were theoretical disconrmation, Kant sometimes allows. This sense of priority is, I argue, operative in the
discussion of the fact of reason. I take the fact of reason to be the awareness of
the categorical imperative as a command issued by practical reason during
deliberation. Agents do not question the categorical imperative when deliberating but may later, upon reection, ask themselves for explanation and
justication. At that point, the experience of the practical becomes the subject
matter for theoretical investigation. The need for freedom, for the moral law
as a principle (and the conception of pure rational agency itself), for the
highest good, and for the postulates arise through that theoretical reection
but are still practical claims because they are based on practical experience.
The priority of action also gives a general indication of how Kants moral
theory is idealist both at the transcendental and empirical levels, since practical
experience is understood from the perspective of the moral agent. The following chapters will provide more details about specic practical claims, starting
with Chapter 4 about the status of the moral law and practical reason itself.

The transcendental status of empirical reason

The previous two chapters discussed the place of the practical in Kants
philosophy as a whole and the specic nature of the practical. Practical philosophy is about free acts understood from the agent-perspective (Chapter 2).
The only method for systematizing those free acts is through an a priori moral
law. Moral duties must not be derived empirically. Yet the nature of Kants
transcendental method allowed that even pure duties could be found in empirical moral experience; their justication would be priori. A closer look at the
nature of practical reason (Chapter 3) showed that there are two main ways
of understanding it: in relation to action and in relation to belief. In the rst,
practical concerns do not need any conrmation by theoretical reason; in the
second, the task of understanding the place of the practical in experience falls
ultimately to theoretical reason. I identied the fact of reason with human
empirical consciousness of the categorical imperative in moral experience
complete with its normative commanding nature and argued that it is compatible with moral idealism at both the empirical and transcendental levels.
This picture is overly simplistic, however, since the transcendental method
of justication itself might show that morality requires some non-natural
metaphysics, or even transcendental realism. By that, I mean that some moral
principle, property, or object is independent of the transcendental moral agent.
It is possible that a conception of reason itself or God would have to exist
independent of moral agents themselves. The identication of the practical
with the agent-perspective does not by itself preclude this non-natural metaphysics. This chapter will examine this issue more closely with a focus on
practical reason.
I believe that the Wille as reason, not will (as Willkr or faculty of choice), is
the core faculty in Kants ethics, and that understanding its relation to the
transcendental and empirical levels is crucial for determining whether Kants
ethics should be considered ideal or real. Philosophy is a product of the activity
of reason in determining and pursuing its fundamental ends. The categorical
imperative is grounded in reason and duties stem from reason. The value of
humanity itself is based upon reason, either as a product of the categorical
107

108

Practical reason in nature

imperative (idealism) or as independent value of rational beings as ends in

themselves (realism). Kant calls his second Critique the Critique of Practical
Reason, not the Critique of Will or Critique of Choice. And most importantly,
when Kant discusses the requirements for a free will, he does so by invoking
reason as the source of laws for the will in the two most important discussions
of freedom prior to the second Critique, namely, the Third Antinomy and
Groundwork III.
This chapter will begin with an analysis of the Third Antinomy to show that
Kants ethical theory is centered on the defense of reason as a basis for
justication of the moral law. The dual levels of empirical experience and
transcendental moral agency become even more important when reasons
causal role in moral acts is understood. The second section will show that
the Groundwork also emphasizes the validity of reason as the main issue in
ethics. The third section assesses the nature of transcendental freedom of
reason as a faculty within nature. Finally, the fourth section argues that Kant
could have been a transcendental idealist and moral realist regarding reason
but that he holds back from such a claim.
The conceptual focus of this chapter will be the nature of transcendental
freedom of reason.1 I will argue against an ontological reading of transcendental freedom in the practical in favor of one that holds that reason is free by
virtue of its structure within the sensible world. I will use the term structural
to signify this non-ontological conception because it stresses the independence
of the structure of reasoning itself from natural causal determination.
This freedom of reason is distinct from freedom of choice (Willkr), which
is an independent conception of freedom applying to a different faculty. The
Wille/Willkr distinction has become standard in interpreting Kants ethics.

Kant had many different kinds of freedom, and for many different faculties. Lewis White Beck
identied ve concepts of freedom with their own internal variations, and he did not even
include political freedom. The ve identied by Beck are (1) an empirical, psychological,
comparative conception that provides a basic distinction between voluntary and coerced action,
(2) moral freedom, or the ability to act on the moral law, which Kant says requires transcendental
freedom, (3) freedom as spontaneity, used for imputation and implying that the decision had no
external cause, (4) transcendental freedom, or a causal force independent of causality in nature,
and (5) postulated freedom, a two-perspective view that Beck nds the most plausible Kantian
conception of freedom (Beck 1987). Beck also emphasizes the dual character of freedom in the
Wille/Willkr distinction. He correlates Wille with the lawgiving in the conception of autonomy
and Willkr with the initiation of a new causal series in the conception of spontaneity (Beck
1960, p. 177). I disagree with Becks insistence that we must never suppose that there are two
faculties related to each other in some external, coercive way. There is only one, but it has prima
facie two kinds of freedom (Beck 1960, p. 180). Even while a famous and widely cited work
on freedom in Kant Henry Allisons Kants Theory of Freedom uses the singular, its author
acknowledges the bewildering number of ways in which Kant characterizes freedom and the
variety of distinctions he draws between various kinds or senses of freedom before announcing
his focus on transcendental freedom (Allison 1990, p. 1).

The transcendental status of empirical reason

109

Translations into English of Kants works now generally render them as will
and choice or faculty of choice respectively. Unfortunately, this device
leaves the unsuspecting reader to associate what in English is generally called
the problem of freedom of the will with Kants term Wille. Since the
problem of freedom of the will is generally understood in terms of a free
choice, it is preferable to associate that problem with Kants term Willkr;
Chapter 6 will look specically at the problem of freedom of the will in
relation to the Willkr. The faculty of Wille is identied by Kant as nothing
other than practical reason (Groundwork, 4:412) and as the lawgiving capacity of rational beings (Metaphysics of Morals, 6:226). My interpretation
stresses this legislative role. Since, as I will show, Kant identies this legislative role specically with reason (Vernunft) in the Third Antinomy and the
Groundwork, I prefer to identify the legislative faculty as reason. In what
follows, I will stress reason but will continue to use the translation will when
Kant uses the term Wille and use reason only for Kants Vernunft.
Reason in nature in the Critique of Pure Reason
Despite the fact that Kants most famous work is entitled Critique of Pure
Reason, Kant does not provide any coherent denition of pure reason. Reason
is identied as the faculty of principles (A299/B356). He sometimes discusses
the pure as opposed to empirical use of reason, implying that there is one
reason applied to different subject matters (A563/B591). Other times, he contrasts empirical reason as appearance with reason in itself (A549/B577). I want
to contrast empirical and pure reason by showing that empirical reason is the
embodied faculty of reason in appearance and pure reason is the transcendentally free structure of reason considered independently of its empirical
manifestation.
The Critique of Pure Reason in general and the Third Antinomy in particular include claims that reason itself is a cause that must be conceived as part of
nature but also as transcendentally free. In this section, I will argue that the
Third Antinomys discussion of freedom centers on reason rather than on the
faculty of choice. I will show the evidence that Kant has a place for reason in
nature and offer an interpretation that best shows how reason can be both a
natural cause and have transcendental freedom.
In the solution to the Third Antinomy, Kant highlights the causal power of
reason as a faculty necessary for morality. Now that this reason has causality,
or that we can at least represent something of the sort in it, is clear from the
imperatives that we propose as rules to our powers of execution in everything
practical(A547/B575). There are in general two different possibilities for
understanding the causal relation of reason to these particular actions in nature.
First, reason could have some status that makes it independent of natural

110

Practical reason in nature

causality yet causing the particular empirical actions as appearances in nature.

Second, reason could exist within nature subject to natural causality as a
faculty of the agent in nature causing particular actions of that agent in nature,
thus describable entirely in terms of natural causality. In the rst case, the
causal relation would seem to be a non-natural relation between the thing-initself and appearance; in the second case, it would be a natural relation among
appearances. Kants solution to the Third Antinomy holds the former relation
to be transcendental freedom, and he certainly holds reason to be a transcendentally free cause, but I will suggest that even the transcendental freedom of
reason can be understood as metaphysically naturalistic although methodologically anti-naturalistic. I will argue that reason is a cause, understood as
having two types of causality, and that it is in both cases a natural faculty of
human beings in nature. Let me be clear about the point I am making: it is not
controversial that reason acts as a cause, but it is controversial that reason in
nature is that cause.
Reason and choice
Empirical reason acts through motives to cause particular decisions of the
empirical faculty of choice, or Willkr. Kant describes the relation between
motives of reason and the human faculty of choice in two well-known places in
the Critique of Pure Reason, in the Solution to the Third Antinomy (A534/
B562) and in the Canon (A802/B830). The human faculty of choice, like
animal faculty of choice, is immediately affected by sensations. Since animals
are affected only by sensations and have no other possible determinants to their
actions, the decisions of their faculty of choice are necessitated.2 Human
beings, however, are also affected by representations of reason, or motives.
These representations depend on reason and yield laws that are imperatives,
an implication that the categorical imperative is among these representations.
Kant also refers to these representations as motives that can only be represented by reason. The human faculty of choice is thus not necessitated by
sensations because it is able to act on motives instead of only on sensations.
Kant calls this sense of freedom from necessitation by sensations practical
freedom (there are, of course, other senses of the term practical freedom in
other contexts). It is important to note that while human faculty of choice is not
necessitated solely by sensations, it is not necessitated solely by motives of
reasons either, for were it so, human beings would be perfectly rational, holy
beings (5:32).

Kant sometimes discusses animals actions as determined by instinct. See, for example, R7199
(19:272).

The transcendental status of empirical reason

111

It is understandably common to interpret this sense of practical freedom as

an ability of the human faculty of choice to decide upon courses of action
without any determination by natural causes. Indeed, Kant claims as much in
the discussion in the Solution to the Third Antinomy, where the human faculty
of choice is said to have a causality to begin a series of occurrences independently of natural causes (A534/B562). But one might, without denying
the importance of transcendental freedom as an independent causality for the
human faculty of choice, instead read Kants description of the human faculty
of choice as describing the empirical causality of sensations and motives in
appearance. Such a reading is suggested by Kants remarks in the Canon where
he describes practical freedom as one of the natural causes, namely a causality
of reason in the determination of the will (A803/B831). On this reading, while
the human faculty of choice is not causally necessitated solely by sensations or
solely by motives, it is causally necessitated by the combination of sensations
and motives. By analogy, if I throw a die, it is not necessary for the resulting
number to be odd, and it is not necessary for the resulting number to be even,
but it is necessary for the resulting number to be either odd or even. My
reading of Kants claims about the determinants of the human faculty of choice
is that it is not necessarily determined by sensation, and it is not necessarily
determined by motives, but it is necessarily determined by either one or the
other. Kant says as much in this passage from the Mrongovious lectures on
Morals:
If this free will is nevertheless to be necessitated, the necessitation must be practical or
per motiva. Motiva are all representations of the understanding, or of reason, that
determine the will. They are set in opposition to the stimuli, and are called incentives
of the soul (elateres animi). If the motiva are to necessitate, the stimuli cannot do so;
only free choice can be necessitated per motiva. (29:61112, modied)

The human faculty of choice at issue, of course, would have to be the

empirical faculty of choice in space and time, and the faculty of reason
at issue would be empirical reason. These two faculties in nature would be
understood as the empirical correlates of a transcendentally free faculty of
choice and a transcendentally free pure reason. A twofold distinction arises
in this picture of Kants moral psychology: (a) the faculties of reason and
choice are two faculties necessary for morality, with reason legislating
the categorical imperative and the faculty of choice deciding particulars
in light of it, and (b) both of these faculties have empirical manifestations
in appearance but also require transcendentally free counterparts. Giving
these empirical faculties their proper place in understanding, Kants ethics
will show that there is room for a naturalistic interpretation of Kants ethics
in which the motives of empirical reason determine the empirical faculty
of choice.

112

Practical reason in nature

Reason in the Third Antinomy

That Kants Third Antinomy has freedom of the will as its content is a
common assumption that, like many common assumptions, is imposed from
without rather than being drawn from the text itself. Kant does, of course, rely
on the Third Antinomy in his defense of freedom of the will. He correlates the
Third Antinomy in fact, the antinomies in general with freedom of the will
when noting the importance of the Dialectic for practical philosophy (Bxxix
xxx, A798/B826). But this correlation does not show that the Third Antinomy
is about freedom of the will any more than it shows that the Paralogisms
chapter is about the immortality of the soul, with which the Paralogisms are
likewise identied for practical philosophy. The Third Antinomy is about
causality. Kants intention is to prove that there must be another kind of
causality besides natural, a kind of causality he calls freedom immediately
at the opening of the antinomy in the thesis statement (A444/B472). The Third
Antinomy is broader than, and opens the way for, a belief in the freedom of the
will as its practical import. There is a further reason to minimize the claim that
the Third Antinomy is about freedom of the will. Freedom of the will in the
rst Critique rests upon reason.3 This is clear even in the preface to the second
edition when Kant asks readers to draw the conclusion that transcendental
freedom is necessary for the origin of the moral law: Now suppose that
morality necessarily presupposes freedom (in the strictest sense) as a property
of our will, citing a priori as data for this freedom, certain original practical
principles lying in our reason, which would be absolutely impossible without
the presupposition of freedom (Bxxviiixxix). Kant here mentions freedom of
the will but clearly relates this freedom to reason as the source of the moral
law. Hence, the project of showing how reason can be conceived as transcendentally free supports any claim that the will could be free.4

Many of the issues I raise in my discussion of Kants critical works are treated by Henry Allison
in (Allison 1990). Unlike many commentators on Kants ethics, Allison recognizes the importance of some conception of freedom for reason itself in addition to the problem of the freedom of
the will. I will note my specic disagreements with Allison particularly regarding the nature of
transcendental freedom in some of the following notes. In general, he equates Kants use of the
term reason in the Third Antinomy with a more expansive rational behavior, so that behavior
based on maxims is seen as the causality of reason (Allison 1990, p. 33). He also takes the
freedom of reason to consist in acts of judgment made on rational grounds rather than, as I do,
the principles of reason themselves. Practical spontaneity is then assigned to judgments, not the
faculty of reason as the source of a priori principles (Allison 1990, p. 36f).
Here, I largely concur with Keith Ward when he notes that the Third Antinomy is solved by
conceiving of reason as the faculty which, as changeless and eternal, is the transcendentally
free source of free actions. Ward uses the term noumenal freedom rather than transcendental
freedom. See (Ward 1972, pp. 757). We differ in explaining the status of that changeless and
eternal reason.

The transcendental status of empirical reason

113

There are several places in the solution to the Third Antinomy that make it
clear that reason is the key faculty under discussion. When asking how human
beings are able to attribute to themselves more than just an empirical character,
he answers in regard to certain faculties, he is a merely intelligible object . . .
we call these faculties understanding and reason (A54647/B57475). Neither
will nor a faculty of choice is mentioned. Reason provides greater support for
this claim than the understanding because the understanding is linked to empirically conditioned powers while reason relates to objects only through its own
ideas. In the very next paragraph, Kant singled out imperatives as evidence that
reason must have causality.
Now that this reason has causality, or that we can at least represent something of the sort
in it, is clear from the imperatives that we propose as rules to our powers of execution in
everything practical. The ought expresses a species of necessity and a connection with
grounds which does not occur anywhere else in the whole of nature. (A547/B575)

This discussion of reason as a cause will be taken up in my next section, but for
now, the key point is that reason is specically cited for its role in practical
determination to action. Kant continues in his discussion by saying that
reason does not give in to those grounds which are empirically given
(A548/B576); it is at least possible for us to assume that reason actually does
have causality in regard to appearances (A54849/B57677) and there is
a certain causality of his reason (A549/B577); that reason is the cause of
producing actions and that the ideas of reason have actually proved their
causality in regard to the actions of human beings as appearances (A550/
B578); asks could reasons action then be called free and answers in part by
invoking the fact that pure reason, as a merely intelligible faculty, is not
subject to the form of time (A551/B579); and among many other occurrences
in the following pages, Reason is thus the persisting condition of all voluntary
[willkrlich] actions under which the human being appears . . . every action,
irrespective of the temporal relation in which it stands to other appearances, is
the immediate effect of the intelligible character of pure reason; reason therefore acts freely (A553/B581).5
Even the discussion of freedom in the solution to the Third Antinomy that
seems most focused on the decision of the agent actually reveals this causal
5

Although this last passage clearly identies reason as the freely acting intelligible, it is sometimes used in support of very different claims. Eric Watkins, for example, quotes this passage up
through the semi-colon but omits the last four words (reason therefore acts freely) and instead
concludes that Kant . . . refrains from attempting to explain which particular intelligible
character we might be able to impute to things in themselves on the basis of the empirical
characters that cause the particular actions we perceive in the sensible world (Watkins 2005,
pp. 32728). On the same page, he similarly overlooks Kants claim in R5611 that the activities
of reason are not appearances by taking the reference to more generally apply to things in
themselves.

114

Practical reason in nature

power of reason. Near the end of his discussion, Kant offers the example of the
malicious liar to show the way that the freedom he has discussed has an
empirical use. The malicious lie, we are told, can be traced back causally to
bad upbringing, a wicked natural temper, and so on. These determining causes
seem to fully explain the lie. Yet, Kant says we blame the agent despite all of
this empirical explanation. This blame attributes the cause of the lie to a lack
of causation by reason: This blame is grounded on the law of reason, which
regards reason as a cause that, regardless of all the empirical conditions just
named, could have and ought to have determined the conduct of the person to
be other than it is (A555/B583). Further, Kant says, we transfer blame from
the agent to reason: now, in the moment when he lies, it is entirely his fault;
hence reason, regardless of all empirical conditions of the deed, is fully free,
and this deed is to be attributed entirely to its failure to act. This attribution of
blame to reason for its failure to act is at odds with blaming an agent for
making the wrong decision. And the active way that reason could have and
ought to have determined the conduct of the person rather than a claim that
the agent could have chosen to act out of reason also reveals that the key
faculty under discussion in the solution to the Third Antinomy is reason rather
than free choice.
Reason as a natural cause
I will now discuss in more detail the nature of empirical reason.6 In particular,
I will show the plausibility of a robust conception of empirical reason that
allows reason to be an entirely natural cause within appearances.
A non-naturalistic interpretation of Kants discussions of reason, as well as
the faculty of the understanding and other active faculties, is that they exist
only in things in themselves and not in appearances, but that reason and the
other active faculties have effects in appearance. But there is also a place in
Kants metaphysics to count reason as among the appearances in space and
time. Things in themselves can have counterparts in the world of sense as they
appear. Certainly, if reason is to be considered to be a thing-in-itself or faculty
in something-in-itself, then it can have a counterpart as appearance or as part of
6

Empirical reason is not directly mentioned by Kant very often. This term empirical reason can
be and certainly was at times used by Kant as shorthand for empirical use of reason, as opposed
to a pure use. The earliest use of the term I have found is in R4582, dating from 177275, in
which it is listed alongside pure reason and practical reason in theological matters: The need of
reason to cognize a highest being is that of a necessary hypothesis of the employment of reason:
(1) of pure reason; (2) of empirical reason (both are speculative); of practical reason (17:601).
A slightly earlier note from 1771 asks what are the appearances of reason? (Subjectively
determined general cognitions) (R4375; 17:525). In R4582, Kant appears to be discussing the
empirical use of reason, but in R4375, he is certainly asking about the existence of reason in
appearances.

The transcendental status of empirical reason

115

some beings appearance. This is the sense of empirical reason at stake here.
Kant even describes the task of the Critique of Practical Reason as taking for
granted that there is empirical practical reason and asking whether a corresponding pure practical reason can exist and be related to empirical actions. One
of Kants notes made in the 1780s on a loose piece of paper rather than in his
textbooks, and thus presumably made in preparation for the Critique of
Practical Reason rather than in preparation for his lectures, stresses this point:
The critique of practical reason has as its basis the differentiation of empirically conditioned practical reason from the pure and yet practical reason and
asks whether there is such a thing as the latter (R7201; 19:27576).
In the Solution to the Third Antinomy, he holds that an empirical manifestation of reason is necessary in order for reason to be understood as a cause at
all. Kant identies the causal power of reason, at least in part, with the
application of a rule:
Thus every human being has an empirical character for his power of choice, which is
nothing other than a certain causality of his reason, insofar as in its effects in appearance
this reason exhibits a rule, in accordance with which one could derive the rational
grounds and the actions themselves according to their kind and degree, and estimate the
subjective principles of his power of choice. Because this empirical character itself must
be drawn from appearances as effect, and from the rule which experience provides, all
the actions of the human being in appearance are determined in accord with the order of
nature by his empirical character and the other cooperating causes. (A549/B577)

The empirical character of the human power of choice is linked to a causality

of reason by means of a rule. Kant denes the character of a cause as a law
of its causality without which it would not be a cause at all, and species that
an empirical character would be a natural law or laws that connect actions as
appearances with other appearances (A539/B567). The rule, then, that connects the decisions of the power of choice with its causes would be a law that
explains how those decisions arise in nature through other appearances. This
rule links rational grounds with the actions themselves and thereby holds
rules of reason to be empirical causes of actions.
The role of rules is important for empirical reason, but not complete. I take a
complete description of reason to require some metaphysical claim about a
faculty of reason as something over and above all the particular relations and
principles that might be called rational. A faculty of reason would explain,
through its structure or through its activities, the generation and application of
the particular relations and principles. Reason is the faculty of principles, Kant
says, and he interprets that to mean that reason also is the faculty of derivation
from principles, or what we might call today logical reasoning (A299300/
B35657). By referring to particular relations and principles, I mean things
like logical rules and axiomatic principles (if there are any), all of which I am

116

Practical reason in nature

calling the rules of reason, To speak rather loosely, a complete description of

reason would include the rules of reason and the faculty that houses them.
Kants insistence that reason has an empirical character might be nothing more
than a claim that the rules of reason must be capable of empirical manifestation; it certainly is at least that. The question I am asking regarding reason in
nature, then, could be sharpened to ask whether the faculty that houses these
rules must be in nature or in things in themselves.
The passage I just quoted about the causality of reason in appearance as a set
of rules falls short of requiring that reason as source of the rules exist in nature.
The rules themselves are exhibited in the effects of reason in appearance.
It is compatible with this claim that reason in itself is a non-natural free source
of the rules that are exhibited in appearance. Reason would be the cause of
these rule-governed appearances without itself being appearance, hence without being in nature.
In the Prolegomenas discussion of the Third Antinomy, Kant appears to
state the stronger claim that reason as the faculty containing the rules, in
addition to the rules themselves, exists in appearance: We have in us a faculty
that not only stands in connection with its subjectively determining grounds,
which are the natural causes of its actions and thus far is the faculty of a
being that itself belongs to appearances but that also is related to objective
grounds that are mere ideas, insofar as these ideas can determine this faculty, a
connection that is expressed by ought. This faculty is called reason
(4:34445). The faculty of reason is said here to stand in connection with
subjectively determining grounds as natural causes. More specically, reason
is said to be a faculty of a being in appearances. Given this positive reference
to an empirical reason, the possibility of such a faculty in Kants system must
be admitted.
While stating that reason is a faculty in appearance, here in the Prolegomena
as well as in the Critique of Pure Reason, Kant insists that reason cannot be
understood only as empirical reason. There must be an intelligible (A551/
B579), or intellectual (Prolegomena, 4:344), or pure (A551/B579) reason as
well that is not subject to the form of time and cannot be considered part of
appearance. In this passage from the Prolegomena, he characterizes it as
reason providing objective grounds expressing the ought of moral obligation. Later, I will look more closely at the status of non-empirical reason; here,
it is necessary only to note that Kant holds that reason has both empirical and
intelligible characters. But how full is the empirical character of reason? Can a
description of reason as a faculty in nature be a complete description of reason?
One reason to hold that empirical reason must be more than just the rules of
reason relates directly to empirical causal relations. Reasons empirical character has been identied at least with rules as causal laws that explain actions.
Recall that the work of empirical reason at issue is determination of the power

The transcendental status of empirical reason

117

of choice to action in lieu of determination by sensible impulses. To assume

that some rules or laws of reason can be postulated that provide causal explanation of some decisions of the power of choice is to assume that something is
the cause of those decisions according to those rules. That is, laws link causes
to effects; and natural laws link causes in nature to effects in nature. If the
decisions of the empirical power of choice are the effects in nature, what are
the causes in nature? The causes could not be sensible impulses by denition;
this means they cannot be anything originating outside the mind. Since they
cannot be sensible impulses, they must stem from within the empirical mind.
This is the equivalent of saying that these causes cannot themselves have any
sensible content but must be constituted by form alone. The categorical
imperative is an example of this sort of rule of reason. Since these causes
must be capable of being represented merely formally, as formal rules or
principles, their source must be empirical reason as a faculty in nature, in
space, and time, as studied by empirical psychology. Empirical reason would
have to be considered not only a collection of rules of reason but also some
empirical faculty capable of causally effecting change in accordance with those
rules. As empirical, it would itself be the effect of previous natural causes, but
nonetheless, it would have a place in nature as part of the set of natural causal
determinants of particular actions.
To say that empirical reason is a faculty in nature is not necessarily to equate
it with a spatiotemporal piece of material nature, as if one could identify certain
sections of the brain in a CT scan and declare their causal relations to be the
causality of reason. Kant would not rule this possibility out, but he thought that
a search for the link between thoughts and their physical location or instantiation was a hopeless quest. Instead, empirical reason should be studied by
empirical psychology without reference to its status in nature as material or
immaterial. And as I showed in my rst chapter, empirical psychology
describes part of nature.
Some might object that there can be no faculty of reason in nature because
nature is constituted by intuitions synthesized by the understanding, and only
the effects of a faculty of reason and not the faculty itself appear in inner
intuition. This objection is misguided, as explained in my rst chapter while
discussing Kants conception of nature. Admittedly, only the effects of such a
faculty would appear in inner intuition. But Kant does not limit nature as a set
of objects to intuitions; instead, nature is constituted by whatever can be linked
to intuitions using natural laws. Empirical reason is still a part of nature, as
appearance, because it would be used to explain thoughts in inner sense as part
of a natural description of the mind in empirical psychology. As a faculty,
empirical reason would be one of the powers of the mind postulated to explain
particular appearances in inner sense. Human beings would not experience
empirical reason itself but would experience the effects of empirical reason,

118

Practical reason in nature

particular thoughts that would require the existence of this faculty for consistent and comprehensive explanation of these appearances in nature.
Two other possible objections deserve discussion. In the Solution to the
Third Antinomy, just after his discussion of empirical reason, Kant appears to
hold that when reason is held to be the cause of actions, it cannot in any way be
considered part of nature but must be considered as an alternative cause to
natural causes. He contrasts the causality of ideas of reason with empirical
causes and insists that reason must be independent of the conditions of time,
thus not in nature. There are two variations of this reasoning, one focused on
practical reason, the other on reason in general. It is important to look at these
arguments to determine whether they preclude the strong sense of empirical
reason I am advocating that would include a representation of the categorical
imperative in empirical consciousness as the natural cause of some actions.
First, Kant argues that reason as practical cannot itself have an empirical
character in nature as a natural cause.
But if we consider the very same actions in relation to reason, not, to be sure, in relation
to speculative reason, in order to explain them as regards their origin, but insofar as
reason is the cause of producing them by themselves in a word, if we compare them
with reason in a practical respect then we nd a rule and order that is entirely other
than the natural order. For perhaps everything that has happened in the course of nature,
and on empirical grounds inevitably had to happen, nevertheless ought not to have
happened. At times, however, we nd, or at least believe we have found, that the ideas
of reason have actually proved their causality in regard to the actions of human beings
as appearances, and that therefore these actions have occurred not through empirical
causes, no, but because they were determined by grounds of reason. (A550/B578)

Kant can be taken to present the following argument here: Reason as cause
provides ideas that themselves reect a normative rule and order regarding
what ought to happen. This normative rule and order is different from the rule
and order of nature, in other words, nature is not what it ought to be. Because
the normative rule and order is other than the natural order, any idea that
expresses a normative rule and order cannot be a natural cause in the natural
order. The argument is ambiguous between an incompatibility of the content
(the order) with the order of nature and an incompatibility of the normativity
(the ought) with the order of nature.
Either way, this argument does not show that reason cannot be in nature.
Suppose, on the one hand, the argument relies on a non-natural content in the
idea. The content of a representation in inner sense may be incompatible with
the natural order, but the idea itself may be part of that natural order. It is
entirely conceivable to have a representation in inner sense whose content
is incompatible with the natural order (for example, a dream in which I am
speaking with the living Immanuel Kant) yet which as representation is
entirely within the natural order. Suppose, on the other hand, that the argument

The transcendental status of empirical reason

119

relies on a claim that the normativity of the idea is non-natural. But mere
normativity is not non-natural; the existence of desires shows that representations in nature can contain both preferences for certain outcomes and a moving
force beyond themselves. Thus, it appears that an idea of reason can exist in
nature even though it may represent normative preference for a non-natural
order. Empirical reason, then, can include a representation that provides a
moral order, namely, the categorical imperative. Note that this conclusion is
different from saying that empirical reason can be the only source of a
representation of the categorical imperative. To allow that the categorical
imperative can exist in empirical reason is not the same as allowing that it
can be justied by empirical reason. Justication of the categorical imperative
can still require pure practical reason.
A second objection that might be given to show that Kant has no room for
empirical reason is that he repeatedly attributes spontaneity to reason and other
faculties. Henry Allison makes this point central to his interpretation of Kants
theory of freedom7. Spontaneity of the faculties of understanding and reason
involves judgments (which Allison calls takings) and creation of ideas
(which Allison calls framings). These together imply that the faculties of
reason and understanding must be understood as involving deliberation and
justication, which in turn cannot be simply understood in natural causal
terms. This epistemic spontaneity is mirrored by a practical spontaneity, in
which decisions of the power of choice for particular actions cannot be
understood in natural causal terms but must be understood in terms of an
incorporation of ones incentive into ones maxim. The incorporation
requires rational deliberation and spontaneity of the agent. Hence, rationality
and agency can be understood only in terms of an intelligible character of the
agent as spontaneous. The requirement of spontaneity is so strong that, he says,
we cannot both deny such a character and afrm our status as rational agents,
that is, rational agency can be understood only in terms of non-natural spontaneity of reason and will. For present purposes, the important point is that
Allisons argument would preclude the possibility of an empirical reason, for
rationality could be understood only as spontaneous and different from the
causal order of nature.
Allisons position has generated much debate in the literature.8 Here, I will
focus only on one claim: namely, that Allisons interpretation reads too much
into the requirement for epistemic spontaneity. Kant does hold that reason
must be spontaneous, but this spontaneity need not involve elaborate transcendentally free deliberation independent of natural causal relations. Empirical
deliberation may proceed according to causal laws in nature precisely because
7
8

(Allison 1990, p. 35ff). The quotation later in my paragraph is from (Allison 1990, p. 41).
See some of Allisons replies in (Allison 1996, pp. 10942).

120

Practical reason in nature

some of those causal laws are psychological rules of reason. If the resulting
conclusion is correct, it is in part because the deliberation utilized rules of
reason that are themselves justied, even though the deliberation might have
been entirely determined as a series of events in nature in the mind. What is
required for the justication of those rules of reason themselves is a separate
issue. Here, transcendental freedom for reason would be required. But this
justication need not involve any extended deliberation. It may result from the
timeless nature of reason itself. I will say more about this in the next section.
Reason as a transcendentally free cause
Reason as transcendentally free in the Third Antinomy
In understanding the solution to the Third Antinomy, one must keep in mind
Kants claim that the Thesis has proved not that a single cause (Ursache) is
necessary but that a non-natural type of causality (Kausalitt) must exist
(A444/B472). Indeed, a review of the argument for the Thesis shows that
Kant does proceed this way. At least one instance of an Ursache of the proper
Kausalitt must exist to avoid the contradiction that would result were all
causality merely natural. Kant then announces that We are now permitted also
to allow that in the course of the world different series may begin on their own
as far as their causality is concerned (A450/B478). Whoever gave Kant that
permission was in error, since the argument (if successful), strictly speaking,
has shown only that one cause of that type would be necessary and one
normally cannot make an argument from a single instance to other similar
particulars. But for the sake of argument, let us allow that if there must be at
least one cause of a type of causality, then further causes of that type could be
invoked if other reasons call for them.9
Freedom as a type of causality should be understood as the causality of
reason in determining the will as power of choice to action, conceived
alongside the natural causality of empirical reason but as a different type
of causality that also exists in nature. I will show that there is no metaphysical conict with holding that reason is a cause in both types of
causality, each understood for different purposes. The requirement for
reason to be a transcendentally that is, justiable under a transcendental
argument free cause does not entail that reason is a transcendent that is,
existing outside nature cause. A transcendentally free reason can exist in

Watkins also stresses that Kant is working with a different type of causality but still gives this an
ontological reading rather than seeing both types of causality as attributed to beings in nature
(Watkins 2005).

The transcendental status of empirical reason

121

nature just as other transcendental conditions for the possibility of experience such as causal relations can exist in nature.
That the transcendentally free causality Kant discusses in the Third Antinomy is supposed to be an alternative kind of causality among objects in
nature is, in addition to tting the philosophical function of practical freedom,
supported by some direct evidence in the text. An overlooked passage in the
solution to the Fourth Antinomy describing the solution to the Third Antinomy
is very clear. After clarifying that the necessary being required by the Fourth
Antinomy would not be a member of the series of empirically conditioned
beings in the world of sense, Kant contrasts that metaphysical claim with a
claim that the transcendentally free cause in the Third Antinomy is part of the
series of conditions in nature:
Hence this way of grounding an unconditioned existence [in the Fourth Antinomy]
would be distinguished from the empirically unconditioned causality (of freedom) in
the previous article in that in the case of freedom, the thing itself as cause (substantia
phaenomenon) would nevertheless belong to the series of conditions, and only its
causality would be thought as intelligible, but here the necessary being would have to
be thought of as entirely outside the series of the world of sense (as an ens extramundanum) and merely intelligible; this is the only way of preventing it from being
subjected to the law of the contingency and dependency of all appearances. (A561/
B589, emphasis in original)

Only the causality of freedom and not the thing that possesses the causality of
freedom would be intelligible; the object possessing transcendentally free
causal power is an object in nature.10 Kant makes a similar point in a passage
in the solution to the Third Antinomy when he denes intelligible and relates
it to a different kind of causality: I call intelligible that in an object of sense
which is not itself appearance and states that one can consider the causality
of this being in two aspects, as intelligible in its action as a thing in itself, and
as sensible in the effects of that action as an appearance in the world of sense.
Of the faculty of such a subject we would accordingly form an empirical and at
the same time an intellectual concept of its causality (A538/B566, partly my
emphasis). The causality, not the being, is understood as intelligible yet as
attributable to the object of sense. Admittedly, Kant uses the term thing in
itself here, and there is some justication for the metaphysical claim that Kant
intends to discuss a non-natural being. But the evidence I presented earlier that
ties freedom to reason suggests that it is worth looking at an interpretation that
identies the transcendentally free causality as some kind of causality of
reason.
10

Henry Allison notes this passage but interprets it to mean that the intelligible character is
assigned to a being with a phenomenal side instead of recognizing that Kant intends the entire
being to be phenomenal and only the causal relation to be intelligible (Allison 2012, p. 29).

122

Practical reason in nature

In the Canon, Kant revealed this need for the possible transcendental freedom
of the faculty of reason when he admitted that practical freedom can exist within
sensible nature while transcendental freedom deals with the non-sensible status
of the faculty of reason as well as with the origin of actions:
We thus cognize practical freedom through experience, as one of the natural causes,
namely a causality of reason in determination of the will, whereas transcendental
freedom requires an independence of this reason itself (with regard to its causality
for initiating a series of appearances) from all determining causes in the world of the
senses, and to this extent seems to be contrary to the law of nature, thus to all possible
experience, and so remains a problem. (A803/B831, emphasis mine)

The question of transcendental freedom arises when the status of the faculty of
reason itself is questioned about its ability to begin a series of appearances
through its causality. And with regard to its causality not the entire faculty
of reason itself is to be independent of determining sensible causes.
Transcendentally, free causality is tied to the intelligible character of reason,
which ought not to be equated with a transcendent reason. Kant introduces the
idea of an intelligible character as opposed to the sensible character. The
character of a cause is dened as a law of its causality without which it
would not be a cause at all (A539/B567). An agents actions in appearance
are said to have an intelligible character, through which it is indeed the cause
of those actions as appearances, but which does not stand under any conditions
of sensibility and is not itself appearance. This causal relation does not stand
under conditions of time, so no action would arise or perish in it as intelligible.
As intelligible, then, it would not change, yet it would have causal effects in
the world of sense in time. Thus freedom and nature, each in its full signicance, would both be found in the same actions, simultaneously and without
any contradiction, according to whether one compares them with their intelligible or their sensible cause (A541/B569). Kant looks for a faculty that is
only intelligible, in that its determination to action never rests on empirical
conditions but on mere grounds of the understanding (A545/B573).
As I have noted, human beings count themselves as intelligences because they
experience the effects of the understanding and, above all, reason (A54647/
B57475). The understanding with its a priori concepts and reason with its ideas
appear to exhibit spontaneity because these concepts and ideas cannot be
understood as arising from sensible causation. The ideas and concepts cannot be
represented as sensible themselves. Hence, individuals must view themselves as
intelligences in order to believe that their reason is valid.11 To view oneself as an
intelligence is simply to take oneself to possess valid cognitive faculties.

11

In contrast, Ameriks takes this passage to be evidence that Kant retained pre-critical
rationalist views about freedom into the early 1780s. Freedom is taken to be a property of the

The transcendental status of empirical reason

123

The causal power of reason on the sensible will is, importantly, a timeless
cause since reason itself is timeless. Many passages in the solution to the Third
Antinomy point this out: Pure reason, as a merely intelligible faculty, is not
subject to the form of time . . . The causality of reason in the intelligible
character does not arise or start working at a certain time in producing an
effect (A551/B579). For since reason itself is not an appearance and is not
subject at all to any conditions of sensibility, no temporal sequence takes place
in it even as to its causality (A553/B581). Reason does not alter (even if its
appearances, namely the way in which it exhibits its effects, do alter)
(A55556/B58384). And reason is present to all the actions of human beings
in all conditions of time, and is one and the same, but it is not itself in time, and
never enters into any new state in which it previously was not (A556/B584).
The causal condition that lies in pure reason is a timeless, persistent condition that is able to inuence the sensible will to action. Kant does not describe
it as different for different individuals, as individual free choices certainly
would be. This general, time-independent, determinant of human sensible
choices must be the moral law itself understood as the pure structure of reason
itself applied to the circumstances of possible human free choice. Pure reason
can be a cause of actions in the sensible world by dictating that they ought to
occur. Reason as a faculty can be understood as providing a structure for
experience, specically, a structure for systematic and consistent actions.
Pure reason as timeless structure
The previous sub-section showed that in the Critique, Kant is concerned with
the transcendental freedom of reason with regard to its causality. This subsection will show the kind of transcendental structure that pure practical reason
exhibits, in particular in relation to the moral law. The following section will
discuss whether Kant is justied in attributing transcendental freedom to pure
practical reason by means of that structure.
It is rst useful to review Kants few descriptions of reason in general in the
Critique of Pure Reason. Reason combines its logical use, where reason
abstracts from content, and a real use as a faculty in which it contains the
origin of certain concepts and principles, which it derives neither from the
senses nor from the understanding and itself generates concepts (A299/
B355). The real (or transcendental) and logical uses of reason are combined
in his denition of reason as the faculty of principles (A299/B356). He
implies that this denition entails that reasons principles rest on thought alone
soul to think and act independently of determination in nature, and the faculties of reason and
understanding are evidence that human beings have immaterial souls that can think and act
freely (Ameriks 1982, p. 190).

124

Practical reason in nature

or contain in themselves universals according to concept alone. Hence, the

principles of reason do not directly apply to experience but only indirectly
provide the concepts that can unify other concepts into a system. Reason in
general, as described in the rst Critique, operates by generating principles that
can then systematize other concepts.
The importance of this denition is that Kant recognizes that the faculty of
reason generates and creates both concepts and principles, but with a systematic purpose in mind. Not just any kind of concept is generated for any
reason. Only principles or concepts that can systematize or unite other, more
particular, concepts are generated. The only way that reason, as a faculty, can
perform this particular function is if reason possesses a particular kind of
structure. The structure of reason would be the highest level functional
relations for producing systematic connection and consistency among possible concepts. Its structure generates particular principles and concepts. The
main principle of pure reason given in the rst Critique exhibits this origin:
that principle is that when the conditioned is given, then so is the whole
series of conditions subordinated one to the other, which is itself unconditioned, also given (A30708/B364). The structure of the faculty of reason is
one that connects particulars to universals, which however are considered
themselves particulars in relation to still further universals. This systematic
structure is reected in the principle that when the conditioned is given, the
unconditioned is also given. One can say that reason generates this principle
because of its structure. It is an a priori principle dependent upon no content
from experience.
Reason, then, has a structure that generates principles. Pure practical reason
is that structure of reason generating a principle for systematic considerations
of a practical kind. In Chapter 2, I showed that practical philosophy must be
organized around an idea that provides its domain and allows for a systematic
organization of that domain. The domain of practical philosophy is free acts
(understood from the agent-perspective). The systematic organization of free
acts is done through the moral law. The moral law provides the formal
structure for free actions, and this structure is embodied in rational relations
themselves and thus in reason itself.
Kant begins the Critique of Practical Reason by noting that a practical law,
one that is valid for every rational being, would be one whose ground pure
reason can contain within itself (5:19). He notes on the next page that a rule
that is to be necessary can arise [entspringen] only from reason (5:20). This
generation of a necessary rule by reason is its lawgiving or legislation: it is
requisite to reasons lawgiving that it should need to presuppose only itself,
because a rule is objectively and universally valid only when it holds without
the contingent, subjective conditions that distinguish one rational being from
another (5:21). Reason alone, considered as what is common to all rational

The transcendental status of empirical reason

125

beings apart from any of the particular conditions that distinguish these
particular rational beings, is the structure of that faculty.
This structure of pure practical reason that generates the moral law is the
autonomy of reason in morality. In the second Critique, Kant claims that the
moral law expresses the autonomy of pure practical reason because it is a
lawgiving of its own on the part of pure and, as such, practical reason (5:33).
Autonomy is the independence of practical reason, which is equivalent to the
will (4:412) from determination by anything outside itself. The moral law is
determined by the structure of reason, so in one sense, reason is not free
because it simply subsists with the structure and its corresponding moral
law; I believe that this is the sense that Kant has in mind when in the
Metaphysics of Morals, he says that Wille is neither free nor unfree. Reason
is considered independent of determination by laws of nature because, in this
sense, there is no determining ground for reason at all (6:213). It is fair to say
that for Kant, autonomy of pure practical reason in generating the moral law is
the transcendental condition for their being any morality for freely acting,
rational beings. This transcendental structure is essential for Kants ethics. It is
familiar to everyone who has studied Kants ethics in its guise as the autonomy
of the will as opposed to the heteronomy of all other foundations for ethics,
and the arguments that the moral law is based on form alone as opposed to any
content. It is expressed through Kants insistence that the moral law be a priori
and thus necessary rather than a posteriori and thus contingent. (I will say more
about autonomy in relation to moral value in Chapter 7.)
This reason must be transcendentally free in its causality, that is, an
independence of this reason itself (with regard to its causality for initiating a
series of appearances) from all determining causes of the world of the senses,
to which Kant adds that this independence seems to be contrary to the law of
nature (A803/B831). The transcendental freedom of reason itself is here
understood to be the ability of the faculty of reason to begin a new causal
series in nature independently of the prior causal series in nature. There are
many places where Kant seems to take transcendental freedom of the faculty of
reason to require an ontologically distinct realm of things in themselves
independent of appearances in nature in space and time. But a closer consideration of this issue shows that Kant does not in fact require transcendental
freedom to be understood as ontological independence of the faculty of pure
reason from nature, and that transcendental freedom can be met within a
metaphysically naturalistic ontology.
Independence from causation in nature does not mean independence from
existence in nature. There are two ways in which something (A) can be
independent from determination by something else (B). First, A may exist
apart from B in such a way that B never has any contact with A. I will call
this transcendent independence or freedom. An example of this sort of

126

Practical reason in nature

independence from determination is when an ill person is quarantined so that

other persons will not receive the infection; another example is when protective parents prevent a child from having any contact with another child they
consider to be a bad inuence. This is the idea behind transcendent freedom of
reason: reason is said to exist as a thing-in-itself (or in things in themselves)
independent of spatial and temporal nature so that spatial and temporal phenomena are incapable of, so to speak, contacting pure reason and thus incapable
of determining its principles.
The second way that A can be independent from determination by B is
that A may exist in contact with B but consist of such a structure that B is unable
to alter A. This is structural independence. A clear example lies in a sealing wax
mold used on hot wax applied to an old parchment letter. The mold itself has
contact with the wax while remaining unchanged by the wax. Because of the
relative malleability of the wax and rigidity of the mold, on the one hand,
the wax is not capable of changing the mold at all. The mold, on the other hand,
is capable of determining the shape of the wax. It itself is timeless and unchanging relative to the wax. And yet, when the mold is used to give shape to the
wax, it plays a part in the efcient causal determination in time of the shape of
the wax on the parchment. The causal explanation of the resulting wax on the
parchment requires use of the mold as one of the causes. The shape of the wax
on the parchment can also be understood independently of the efcient causality
and in terms of the form alone that the wax now has that reects the timeless
structure of the mold.
This is the idea behind the structural freedom of reason: reason is said to
exist in nature but as an unalterable structure of thinking that then processes
empirical inputs in a manner independent of and unchanged by that
empirical input. Empirical reason plays a part in the causal determination of
action as part of the efcient causality of sensible impulses and motives of
reason operating in the empirically real mind (as described by empirical
psychology). To the extent that empirical reason determines the empirical
power of choice to action, a moral agents choice is practically free. Reason
in this way acts as a natural cause.
Now this structural conception of the freedom of human cognitive faculties is entirely metaphysically naturalistic. Human cognitive faculties are
conceived as existing in the empirical self as appearance, thus immanently
and not as any transcendent entity. But their validity is defended through
transcendental arguments for the necessity of the a priori cognitions, concepts,
ideas, or judgments they provide. These faculties can be understood as having
a particular structure or even essence that provides for the actual generation of
their pure cognitions (etc.), and this structure can be seen as existing in nature
understood by empirical psychology (as part of the mind, i.e., inner sense, or
the brain). But the structure itself is necessary for any being who is going to be

The transcendental status of empirical reason

127

capable of whatever experience is at issue. This transcendental argument is

independent of any particular manifestation of these faculties, so the structure
expressed through its resulting pure concepts or principles is a priori relative to
all of experience.
Here it is important to note a possible objection to this account. In the
sealing wax analogy, it is true that the sealing wax mold is not determined by
the sealing wax. But it is determined by something else, namely, whoever
made the sealing wax mold itself. It is contingent because there is no necessity
that a sealing wax mold be used simply because sealing wax is used; one could
simply drip sealing wax onto the folded parchment. The particular mold used
in this instance, and even the very existence of molds in general for sealing
wax, is contingent. By analogy, the actual structure of reason, like the particular pattern on the mold, could have arisen as a contingent feature because it is
prima facie possible that some other faculty could have operated instead of
reason as an alternative to sensible impulses in the causal determinants of
actions. It even seems possible that reason might not have existed in nature at
all: the structure of reason does not need to exist in order for agents to act at all
because they can act on the basis of sensible impulses alone. The question may
still be raised whether that particular natural cause is anything more than
merely one efcient cause among many, as Kant puts it in the Canon, whether
that which with respect to sensory impulses is called freedom might not in turn
with regard to higher and more remote efcient causes be nature (A803/
B831). Empirical reason might itself be merely a product of natural causes
without transcendental justication and have no more special status than an
instinct. The question can be raised whether reason ought to be considered to
have a particular structure that deserves to be singled out from other empirical
causes and deemed to be a distinct kind of causality as well. I will discuss this
issue in the following sub-section. Here, let me quickly say two things: rst,
Kant claims that we view our reason as the source of a priori claims that
themselves could not originate from sensible experience but that arise from the
nature of reason itself, so isolating reason from other possible faculties as
necessary rather than contingent has a basis; and second, actual knowledge that
reasons principles are really legitimate and that reason as a faculty must exist
requires a transcendental deduction, which Kant argues is not possible in
practical philosophy; so in the end, Kant does not claim to know that reason
must exist in order for human beings to act.
Another worry must be considered here. When the transcendental structure
of reason is embodied in a particular individual, it is identical with the
particular empirical reason of that individual. All particular judgments are
subject to causal determination by prior causes in nature. Does this not mean
that the judgments are mere events in nature that, as such, lack truth? No.
Particular judgments can have causal antecedents in nature and still follow the

128

Practical reason in nature

proper structure of reason. There is no more a problem with human beings

embodying valid judgments than there is with computers calculating; in both
cases, a transcendental requirement for a particular type of logical or rational
process is embodied in the being without being dependent upon that instantiation.12 When exercising their proper mental functions, human beings can
reach true conclusions, or in ethics, make valid judgments, as natural beings
acting within the chain of natural causation. Transcendental structure of
reason, one might say, identies particular parts of the chain of natural
causation as bearers of validity. Whether a judgment is determined in accordance with laws of nature or not, what makes the judgment valid is if it follows
the proper procedure for judgment which is determined transcendentally. That
proper procedure can be embodied in nature just as easily as not.
The structural freedom of reason shows how it is independent of causation
in nature without being a thing existing outside nature. It is true that the
particular instantiation of reason in a particular individual human being in
nature is the product of natural causation, and is thus contingent. The existence
of any particular individual, and thus the existence of any particular instantiation of the faculty of reason, is thus the contingent result of causal forces in
nature. But the structure of reason itself is independent of the individual.
Reason produces the moral law because of its structure, which is determined
independently of any and every particular instantiation.
The timeless character of pure practical reason is understandable on this
structural account. Pure practical reason is not subject to time. The structure of
reason does not change, does not arise or cease, does not have any particular
duration. In particular individuals, of course, embodied in their particular
empirical reason, this structure does operate in time, but without being dependent, qua structure of pure practical reason, on any prior causes in nature.
This transcendentally free reason can in this way also be understood as
initiating a series of appearances (A803/B831) as demanded by the thesis of
12

Corinna Mieth and Jacob Rosenthal raise this issue but do not go far enough in recognizing its
import. Since, they say, an electronic calculator operates according to both the laws of nature
and the laws of mathematics, there is a clear way in which one and the same result (in this case,
a mathematical conclusion) is the result of two kinds of laws without contradiction (Mieth and
Rosenthal 2006, pp. 27273).They take Kant to reject this kind of double understanding of
processes that are ontologically in nature. They quote the following passage from the Groundwork in which Kant raises a possible contradiction between freedom and nature: It would,
however, be impossible to escape this contradiction if the subject who seems to himself free
thought of himself in the same sense or in the very same relation when he calls himself free as
when he takes himself to be subject to the law of nature with regard to the same action (4:456).
Yet precisely in that passage, Kant allows that different relations could be at work in a being
understanding itself as free and as naturally caused in the same action, and the different causal
relations of an empirically determined causality and a transcendental free causality. This is
exactly the kind of view I am attributing to Kant. Mieth and Rosenthal recognize that the view is
plausible but do not attribute it to Kant.

The transcendental status of empirical reason

129

the Third Antinomy (A50/B478). The solution to the Third Antinomy requires
that both thesis and antithesis be true, so some events are determined in the
series of natural causes and effects in time and are also understood as the
initial spontaneous beginning of a new series. In the resolution of the Third
Antinomy, Kant identies those events as decisions of the empirical power of
choice determined by reason. Because reason is timeless as a certain kind
of structural relation among concepts, it has no preceding state that would
have been different and from which it, qua reason itself, would have arisen.
(This cannot be said of individual manifestations of reason in particular
individuals which, as an empirical faculty of an organism, would have come
into existence and developed with the emergence and growth of that organism.) When a particular moral action is attributed to reason as a cause, this
timeless structure of reason that has generated the moral law as a categorical
imperative is viewed as the spontaneous cause, even as simultaneously, the
categorical imperative itself is understood in terms of the fact of reason as
existing in the empirical agents conscious mind. A different type of causality, rational causality, is at work in addition to the deterministic causality of
natural psychological causal forces. That empirical reason exists in the
empirical mind as described by empirical psychology does not prevent that
faculty of reason from being understood in terms of its structure. As Kant had
claried in his discussion of the Fourth Antinomy, in the case of freedom,
the thing itself as cause (substantia phaenomenon) would nevertheless belong
to the series of conditions, and only its causality would be thought as
intelligible (A561/B589). The transcendentally free causality of pure practical reason in providing the moral law is a timeless cause that can be
understood to begin a series of effects in nature from a practical point
of view.
Moral agency is delineated by the relation an agent has to the structure of
reason. To be a moral agent is to be capable of having the decisions of ones
empirical power of choice in nature determined by pure practical reason
through its empirical manifestation as empirical reason. The particular structure that reason imposes on decisions is the systematic determination of the
conditions for the free actions of all moral agents. The moral law as a description of the activities of purely rational agents provides the idea of such a
systematic union; the categorical imperative, as the fact of reason, is the
manner in which the human empirical power of choice experiences the obligation to conform ones actual decisions as much as possible with that moral law
by opting to resolve deliberation in favor of the principles of reason. This last
decision is understood from the agents perspective; from the theoretical
account given in empirical psychology, it would be seen as the motive of
reason rather than the impulses of sensibility determining ones power of
choice.

130

Practical reason in nature

This section has shown that pure practical reason as a faculty possesses a
timeless structure that generates the moral law. Reason as a faculty in general
is a set of formal relationships and laws to provide systematic connection
among concepts. Pure practical reason provides the formal systematic basis for
decisions, namely, that they be based on law-like principles that systematize
free actions. There is thus a transcendental basis for the moral law and a reason
to attribute transcendental freedom to pure practical reason. The transcendental
freedom of reason had been dened in the Canon as an independence of this
reason itself (with regard to its causality for initiating a series of appearances)
from all determining causes of the world of the senses (A803/B831). The
structural account of the freedom of reason that I have provided here meets all
the criteria that Kant offers for transcendentally free causation in moral actions.
The transcendental freedom of reason cannot be proved
While Kants conception of morality includes this transcendental structure, he
also denies that it can be proved to be real. This denial sets up a complicated
situation for determining the extent to which Kant is a moral realist. A later
section will assess realism and idealism about reason in detail. This section
will show why Kant insists that the picture of morality involving a transcendentally free faculty of reason providing an a priori moral law cannot be
proved.
The end of the solution to the Third Antinomy contains Kants own limitation on the conclusions to be reached. Freedom, he says, has not only not been
shown to be real; it has also not even been shown to be possible. The reasons
he gives anticipate the reasons he provides in both Groundwork III and the
second Critique to limit the status of claims about freedom. First, from
experience we can never infer something that does not have to be thought in
accord with the laws of experience and second, from mere concepts a priori
we cannot cognize anything about the possibility of any real ground or any
causality (A558/B586). The rst reason stresses that the kind of transcendental freedom shown in the Third Antinomy is part of the minimal explanation of
experience in accordance with the laws of nature. This reasoning does not
preclude transcendentally free causality from existing in nature, but as a
distinct kind of causality, it cannot be proved from any empirical data. The
second reason limits the status of transcendental freedom to a purely conceptual level, a limitation that likens it to a transcendental idea, the status of which
will be part of my next chapter on the postulates.
In the second Critique, Kant denies that the moral law is subject to a
transcendental deduction. Unlike the categories of the understanding which
are susceptible to the transcendental proof of their necessity for experience, the
moral law has no proven transcendental role in experience. He makes this

The transcendental status of empirical reason

131

argument in the section of the Critique of Practical Reason called On the

Deduction of the Principles of Pure Practical Reason, where he discusses the
moral law:
The exposition of the supreme principle of practical reason is now nished, that is, it has
been shown, rst, what it contains, that it stands of itself altogether a priori and
independent of empirical principles, and then what distinguishes it from all other
practical principles. With the deduction, that is, the justication of its objective and
universal validity and the discernment of the possibility of such a synthetic proposition
a priori, one cannot hope to get on so well as was the case with the principles of the pure
theoretical understanding. For, these referred to objects of possible experience, namely
appearances, and it could be proved that these appearances could be cognized as objects
of experience only by being brought under the categories in accordance with these laws
and consequently that all possible experience must conform to these laws. But I cannot
take such a course in the deduction of the moral law. For, the moral law is not
concerned with cognition of the constitution of objects that may be given to reason
from elsewhere but rather with a cognition insofar as it can itself become the ground of
the existence of objects and insofar as reason, by this cognition, has causality in a
rational being, that is, pure reason, which can be regarded as a faculty immediately
determining the will. (5:46)

Kant denies the possibility of a transcendental deduction of the moral law

because of its lack of possible object in experience. The transcendental deduction of the categories of the understanding proceeded with the assumption that
there are objects of possible experience to be synthesized; the categories
constitute the rules for synthesis. Kant describes this as reference to objects
(Gegenstnde) of possible experience as appearance. Since the moral law is
not concerned with objects, a deduction cannot be given in the same way. Kant
implies that the moral law instead reverses this relationship, being itself the
ground of the existence of objects, that is, the basis for decisions of the will to
perform certain actions that bring about certain ends, including the
highest good.
Kants concern here is not that the structure of reason in producing the moral
law is itself a transcendental structure, but rather the concern is that there is a
strong enough basis to prove that human beings in fact possess this reason. He
expresses this point by noting at the beginning of the paragraph that the
exposition of the supreme principle is nished. That is, if there is to be a
supreme transcendental principle of morality, the categorical imperative is it.
(He makes the same kind of argument at the end of the second section of the
Groundwork of the Metaphysics of Morals [4:44445].) What he admits he has
not shown is that this moral law is actually possessed by, that is, really valid
for, human beings.
This division between laying out what the transcendental structure provided
by a faculty would be on the one hand, and providing a transcendental

132

Practical reason in nature

deduction of that faculty on the other hand, mirrors the distinction in the
Critique of Pure Reason between the metaphysical and transcendental deductions of the categories. The metaphysical deduction provides and defends the
list of pure concepts of the understanding but does not prove that they are
necessary for experience; the transcendental deduction proves that these pure
concepts as a set are necessary for experience but does not explain their details
(B159). The categories themselves are in essence the structure of the pure
understanding, and that structure is known prior to its transcendental deduction. What the structure of the pure understanding is, is not in question in the
transcendental deduction. In the same way, the structure of reason necessary to
legislate the moral law is not questioned in the passage I quoted.
Reason in Groundwork III
The interpretation of the importance of reason and the nature of its transcendental freedom in the rst Critique holds for Groundwork III as well. In
Groundwork III, Kant is not concerned with individual choice but with the
faculty of reason. He asks whether we as empirical moral agents are also
justied in attributing to ourselves a transcendental conception of moral
agency centering on a valid faculty of reason. In this section, I will defend
this reading of Groundwork III.13
The argument Kant provides in Groundwork III is primarily a justication of
reasons ability to provide valid law and not an argument for free human moral
action in any sense other than its determination by that valid law. Once Kant
can justify this legislative power of reason, he can defend the human obligation
to will maxims that conform to the categorical imperative, and will have
secondarily defended freedom of the will as the empirical power of choice
able to follow reason. He will have defended the obligation human beings have
to try to make their actions conform to the rational dictates of morality, and to
that extent, will have shown that it is the responsibility of human beings to
attempt to make their actual world conform as closely as possible to a perfectly
rational world.

13

The interpretation of Groundwork III presented here builds on the work of several other
interpreters while not claiming the full endorsement or agreement of any of them. I am indebted
in particular to the work of Marcus Willaschek, which has been discussed in my previous
chapter in relation to the fact of reason (Willaschek 1992), Julio Estevess argument that Kant
does not invoke theoretical arguments in Groundwork III and that his claims are fully compatible with the Critique of Practical Reason (Esteves 2012), Corinna Mieth and Jacob Rosenthals
interpretation of the nature of the freedom of reason itself in relation to justication (Mieth and
Rosenthal 2006), and Jens Timmermanns assessment of freedom of the will that shows it as
merely a capacity to be determined by principles of reason rather than a free choice among open
possibilities (Timmermann 2003).

The transcendental status of empirical reason

133

The freedom of reason

The paragraph Freedom Must be Presupposed as a Property of the Will of All
Rational Beings (4:44748) sets out Kants minimal explanation of freedom
of the will. By minimal, I do not mean that in this section, Kant does not
prove freedom but merely claims that human beings presuppose or act
under the idea of freedom (although, of course, this is part of Kants
argument); rather, I mean that the freedom Kant claims all humans must
presuppose is not freedom in particular judgments themselves but merely
freedom of reason to determine its own principles that are then used in
judgments. His paragraph raises the issue that I have argued is the central
topic of the solution to the Third Antinomy. The key argument of the passage
is as follows:
Now I assert that to every rational being having a will we must necessarily lend the idea
of freedom also, under which alone he acts. For in such a being we think of a reason that
is practical, that is, has causality with respect to its objects. Now, one cannot possibly
think of a reason that would consciously receive direction from any other quarter with
respect to its judgments, since the subject would then attribute the determination of his
judgment not to his reason but to an impulse. Reason must regard itself as the author of
its principles independently of alien inuences; consequently, as practical reason or as
the will of a rational being it must be regarded of itself as free. (4:448)

Others see this as the claim that Kant is discussing freedom of judgment here,
that is, a transcendentally free particular act of the power of judgment independent of any causal determination according to the laws of nature.14 It is
thought that if my judgment is determined in accordance with the laws of
nature, then it is determined only by an impulse or some previous efcient
cause in accordance with the laws of nature. Were my judgments or choices
determined by an impulse, I would have no reason to ascribe validity to them.
I must assume, then, that the particular judgment I make is not determined by
impulses. This approach concludes that every rational being must assume she
is free in making particular judgments or moral decisions.
But in focusing on particular judgments as acts, this approach makes too
broad a jump.15 The passage does not require that Kants focus is on particular

14

15

Henrich stresses the importance of the freedom of judgment in reasoning (Henrich 1975, p. 65).
Allison discusses epistemic spontaneity as part of Kants argument for the freedom of
decision for the will in relation to a judgment (practical spontaneity) and more recently in
terms of a distinction between the space of reasons and the space of causes (Allison 1990,
pp. 22223) and (Allison 2011, pp. 30708).
Mieth and Rosenthal provide an excellent critique of the standard approach to this question.
They note that there are ve different ways that one can discuss freedom, the rst of which is
freedom as rationality. Freedom as rationality entails that ones deliberations and judgments are
in accord with reason. Mieth and Rosenthal note that that this allows these deliberations and

134

Practical reason in nature

acts of judgment; it can instead be seen as discussing the principles of the

faculty of reason as the basis for judgments. There is no gap for Kant to ll if
his argument about freedom concerns not a transcendentally free power of
choice or decision but only a transcendentally free power of reason to produce
the moral law. Kant does not attribute freedom to the particular act of judgment
but only to reason itself. The judgment is not free but is determined: the
subject would then attribute the determination of his judgment not to his reason
but to an impulse (4:448, my emphasis). Reason cannot consider a judgment
valid unless the cause of the judgment is a principle of reason. Attribution of
the determination need not exclude other causes. If I attribute a particular move
to a chess-playing computer, I can claim that it chooses its moves because of
the causal network of electronic impulses, but I attribute these moves to the
rational principles of chess strategy and tactics. There is no contradiction in
allowing both a role in the resulting move. When I then make a move,
I similarly attribute my move to those same principles of chess strategy and
tactics, but I can also, without contradiction, assume that my judgment is the
causal product of chemical and electrical events in my brain.16
Note that in this passage, Kant also claims that we think that reason . . . has
causality with respect to its objects. The objects in question are the judgments
made in accordance with the principles of reason. To attribute a judgment to
reason is to attribute a causal role to reason itself as the basis for the judgment.
Particular judgments are causally determined, not transcendentally free in the
way the standard reading assumes. Instead of the judgment being free, reason
must regard itself as the author of its principles independently of alien causes
. . . it must be regarded of itself as free. This is to say that, in making a
judgment, I used valid principles of reason. The question left at the end of this
argument is simply whether the attribution of freedom to reason as the author
of laws and principles is merely an inescapable but unwarranted assumption of
rational beings, or whether it can be independently justied. This is another

16

judgments to be simultaneously caused by impulses in accordance with the laws of nature, a

point I invoke in the earlier text. But they do not attribute this dual-causality to Kant as I do
(Mieth and Rosenthal 2006, pp. 27273).
This might seem similar to the anomalous monism interpretation of Kants theory of freedom
offered by Ralph Meerbote, drawing on the work of Donald Davidson, which treats reasons for
actions as different explanatory devices than appeal to causes (Meerbote 1984). Explanations of
human actions in this view can be made in two ways: rst, using scientic laws that describe
causal processes, and second, using descriptively different explanations offering reasons. The
second kind of description of actions is understood independent of the rst, although both are
taken to refer to the same objects and events. The explanation I am giving here differs from
Meerbotes in that I take reasons and the faculty of reason to be part of empirical psychology,
and hence part of the explanation of human actions in accordance with natural laws. The part of
the judgment that I claim must be understood as independent of causal determinism in nature is
the faculty of reason itself, which is invoked in particular judgments but which is identical in
each of those judgments.

The transcendental status of empirical reason

135

way of stating the question that Kant posed at the end of Groundwork II,
namely, whether the moral law is itself possible.
At this point in Groundwork III, Kant could simply present a further
argument for the possibility of a free reason in order to resolve the issue. But
before attempting to do that, he stresses the importance of the upcoming
argument. The rst four paragraphs of the section entitled Of the Interest
Attaching to the Ideas of Morality (4:448450) focus on the question
of obligation. Having shown that rational beings unavoidably attribute an
as-yet-unproved freedom to themselves (and hence attribute the moral law to
themselves), he recognizes that mere self-attribution is not enough to provide
obligation. But why, then, ought I to subject myself to this principle, and do
so simply as a rational being? he asks (4:449). Perfectly rational beings act
rationally without any interference from impulses. But human beings are not
perfectly rational beings, and instead nd our actions determined by sensibility. Why should human beings not simply allow their actions to be determined
by sensibility? Why take any interest in the moral law? If the moral law is
merely something human beings assume for themselves, there appears to be no
reason to take it as binding. Put another way, if human beings merely assume
that they are free, that is, possess a reason that can produce independently valid
principles, then what guarantee is there that these alleged principles of reason
really are valid? Kant concludes It seems, then, that in the idea of freedom we
have actually only presupposed the moral law, namely the principle of autonomy of the will itself, and could not prove by itself its reality and objective
necessity (4:449). On what grounds, he asks, is the moral law binding?
This consideration leads Kant to admit to a kind of circle in his argument.
The way that he describes this circle is telling: he notes that he has shifted from
talk of efcient causes to talk of the order of ends:
It must be freely admitted that a kind of circle comes to light here from which, as it
seems, there is no way to escape. We take ourselves as free in the order of efcient
causes in order to think ourselves under moral laws in the order of ends, and we
afterwards think ourselves as subject to these laws because we have ascribed to
ourselves freedom of the will: for, freedom and the wills own lawgiving are both
autonomy and hence reciprocal concepts, and for this very reason one cannot be used to
explain the other or to furnish a ground for it. (4:450)

When he says take ourselves as free in the order of efcient causes, Kant is
referring to the claim we make that our reason is independent of impulses.
When he then says that we think ourselves under moral laws in the order of
ends, he implies that from the lack of efcient impulsive cause, we assume
that we are free in the use of our reason to legislate the moral law for ourselves.
But why is this categorical imperative really valid? Merely our assumption that
we are governed by reason is insufcient. Hence, the problem that we have

136

Practical reason in nature

only so far ascribed freedom to ourselves. Our mere assumption that we are
subject to the moral law is not enough to show that we are actually subject to
the moral law. There would have to be some additional basis for the claim that
we are rational beings and can then assume that reason freely provides its own
principles.
Reason and the intelligible
The circle can be escaped only if there is an independent way to afrm that we
can legitimately take ourselves to be rational. This will afrm the validity of
the moral law, and thus afrm the reality of autonomy or freedom of the will,
its ability to act on the basis of the moral law. It will conrm that the order of
efcient causes in nature is not the source of the categorical imperative, but
that a different source of the categorical imperative is available. This is the path
Kant takes to escape the circle by invoking the intelligible world as the view
that human beings take themselves as rational and therefore as subject to the
moral law provided by reason. The circle problem is declared resolved a few
pages later when Kant says:
for we now see that when we think of ourselves as free we transfer ourselves into the
world of understanding as members of it and cognize autonomy of the will along with
its consequence, morality; but if we think of ourselves as put under obligation we regard
ourselves as belonging to the world of sense and yet at the same time to the world of
understanding. (4:453)

His solution invokes the world of understanding (identical to the intelligible

world and the intellectual world) that allows for autonomy of the will. The
crucial question to answer, then, is what Kant means by the intelligible world
and how it makes us free. This would explain how Kant takes himself to avoid
the circle and defend the validity of morality beyond a mere self-ascription.
(I will show in Chapter 7 that later in Groundwork III, Kant indicates that the
result is actually a modied kind of self-ascription.)
The main characterizations of the intelligible world in the Groundwork
concern pure activity in the human mind as revealing something about
how humans are in themselves. One is tempted to equate this idea of pure
activity with a human being as a non-natural thing in herself and claim that we
can have a direct awareness of this activity.17 But Kants language is less direct:
he claims that our experience of what appears to be pure activity which he
denes as what reaches consciousness immediately and not through affection
17

Henrich, for example, emphasizes that Kants argument for the consciousness of our freedom
depends on his crucial two-world doctrine so that the consciousness of our freedom depends
upon the claim that our will belongs to a world other than the sensible world (Henrich 1975,
p. 98).

The transcendental status of empirical reason

137

of the senses (4:451) allows us to regard ourselves as not completely

dened by our sensible nature. But it does not furnish us with any positive
characterization of what a human being would be in herself. The nature of what
the self might be independent of sensible intuitions is still incomprehensible: a
human being must count himself as belonging to the intellectual world, of
which however he has no further congnizance (4:451, my emphasis after
comma). He warns in the next paragraph against those who want to give
content to this invisible self because they can only make it an object of
intuition and hence of sensation. Given Kants warning, it is important to
focus on precisely the extent to which human beings must consider themselves
as intelligible rather than sensible.
The particular nature of the non-sensible, apparently active self is not any
active awareness of free choice but the more limited awareness of a use of the
faculties of understanding and reason. In language from Groundwork III,
nearly identical to that quoted from the Third Antinomy earlier, Kant cites
understanding and especially reason as the faculties a human being appears to
possess:
Now, a human being really nds in himself a capacity by which he distinguishes
himself from all other things, even from himself insofar as he is affected by objects,
and that is reason. This, as pure self-activity, is raised even above the understanding by
this: that though the latter is also self-activity and does not, like sense, contain merely
representations that arise when we are affected by things (and are thus passive), yet it
can produce from its activity no other concepts than those which serve merely to bring
sensible representations under rules and thereby to unite them in one consciousness,
and without which use of sensibility it would think nothing at all; but reason, on the
contrary, shows in what we call ideas a spontaneity so pure that it thereby goes far
beyond anything that sensibility can ever afford it, and proves its highest occupation in
distinguishing the world of sense and the world of understanding from each other and
thereby marking out limits for the understanding itself. (4:452)

Human beings experience of the uses of their faculties of understanding and,

especially, reason causes them to attribute self-activity to themselves. Only to
that extent can a human being attribute membership in the intelligible world,
that is, only as a being that apparently possesses the faculties of understanding
and reason that themselves provide concepts and ideas that are not comprehensible through sensation. No kind of choice or decision or judgment but only the
faculties of understanding and reason are ascribed to the intelligible world.
As in the Third Antinomy, Kant stresses reason as the source of laws for
human behavior. Given the attribution of reason to the self along with the
already accepted sensibility, a human being can view herself as belonging to
two worlds: in one sense, she recognizes that she is a sensible being and is
thus subject to the laws of nature. In another sense, she believes herself to be a
rational being and thus, subject to the constraints of rational moral law on her

138

Practical reason in nature

behavior. The intelligible world is understood in terms of the possibility that

reason itself determines the actions of a being in the sensible world.
At this point, the circle is avoided because Kant has a reason for individuals
to attribute reason to themselves. The awareness of ideas and principles of
reason provides a basis for the attribution of reason to ourselves, since there is
no other explanation of how these ideas and principles can be valid. Hence, we
are beings who may believe that they possess reason. But we cannot understand how such governance by reason in particular with regard to our own
decisions in the will is possible. We cannot even conrm independently of
the apparent use of reason that we are rational beings. Nonetheless, we now
have a reason to assume that the moral law is valid, and that it can govern our
actions. We afrm autonomy of the will its capacity to be governed by the
rules of reason and equally afrm the validity of the moral law based in
reason.
That Kant takes reason to be the key faculty whose freedom to provide a
valid moral law is at issue is also supported by some language later in
Groundwork III. The nal paragraph raises the characteristic of reason that it
always pushes its cognition until it nds necessity. In the nal paragraph of
Groundwork III, Kant applies this to the practical use of reason, which
demands absolute necessity, but only of laws of actions of a rational being
as such (4:463). The third section of the Groundwork as a whole has been a
search for this absolute necessity of the moral law, not merely the conditional
claim that if there is a moral law binding on human beings, it is the categorical
imperative, but the unconditional claim that in fact there is such a binding
moral law. In the previous section, On the Extreme Boundary of All Practical
Philosophy, we can nd a more specic explanation of relation between
freedom and the necessity of the moral law.
Thus the question, how a categorical imperative is possible, can indeed be answered to
the extent that one can furnish the sole presupposition on which alone it is possible,
namely the idea of freedom, and that one can also see the necessity of this presupposition, which is sufcient for the practical use of reason, that is for the conviction of the
validity of this imperative, and so also of the moral law. (4:461)

The validity of the categorical imperative depends upon the possibility of

freedom. The kind of freedom that concerns Kant must be one that could
provide validity to the categorical imperative. A standard approach to Groundwork III would identify this freedom as the freedom of the will as a power of
choice in determining the action of a non-natural person in herself independent
of sensible causation. A look at Kants previous paragraph, however, shows
that the standard understanding of this freedom is misplaced. In that paragraph,
Kant notes that the causal relation of freedom is not one of a choice by the will
of a person in herself to sensible effects but of a causality of reason to

The transcendental status of empirical reason

139

determine sensibility in conformity with its principles (4:460). This emphasis

on the causality of reason itself through its principles is again reminiscent of
the Third Antinomy solution. The causal relation is between reason as a cause,
through its principles, and the decision of the will as effect.18
I have provided a brief reading of Groundwork III that stresses the validation of reason rather than choice in the will. My aim has been to show that
Kants claims in this purely moral work match those in the Critique of Pure
Reason. Our experience of the ideas and principles of reason, in particular of
the categorical imperative, cannot be readily explained using the sensible
world alone. We attribute a distinct kind of justication to reason (one that is
ultimately inexplicable, as I will discuss in Chapter 7).
Reason and moral realism
The position I am advancing here would, under a certain restriction, make Kant
an empirical moral realist. If one can identify a transcendental structure of pure
practical reason that all deliberative rational beings must possess, then one has
a basis for a moral principle that would be independent of the empirical moral
agent. The structure of reason that produces the principle is independent of the
existence of any particular moral agent that embodies the structure. A similar
conception of Kants moral theory is one advanced by many moral realists, one
in which the moral law is not constructed by moral agents but that exists
independently of them because it stems from the nature of reason or of the
will.19 But there is also reason to attribute to Kant only an empirical moral
idealism. This section will lay out the interpretive options. The discussion is
limited to reason and the moral law; other moral issues such as God, freedom
of choice, and moral value will have different factors in assessing whether they
are transcendentally real or not; Part Three will examine those issues and the
postscript will tie all elements of Kants ethics into the same framework
used here.
18

19

Strictly speaking, the causal relation is between reason as a cause and the feeling of respect as
an effect. The feeling of respect can then be understood as a purely natural cause of a decision of
the will (4:460). The relation between a timeless idea and a sensible feeling operating at a
particular time in the human mind remains.
Patrick Kain is a clear example. In (Kain 2004), he lays out Kants position regarding the
authorship and legislation of the moral law. Human beings can be seen to legislate the moral
law, as can God, because to legislate means only to place someone under an obligation to obey
the law. But the law is not itself created by either individual moral agents or by God; rather, the
law itself has no author. It is given by the nature of reason itself (Kain 2004, p. 288). Allen
Wood also makes Kant a moral realist in part on the basis of an objective conception of rational
will that is independent of the existence of any particular rational agent and that creates a moral
law valid for all rational beings whether they are aware of it or accept it (Wood 1999, p. 157).
But I will show in the following text that Wood also hesitates in a way that allows for a moral
idealist interpretation.

140

Practical reason in nature

Three options
Moral realism or idealism regarding reason and the moral law could be either
transcendental or empirical. As I dened these terms in Chapter 1 on realism,
the transcendental distinction between moral realism and moral idealism
concerns whether the transcendental conception of the moral agent requires
any principles, properties, or objects independent of the conception of moral
agency itself. This transcendental conception is the result of philosophical
work examining the necessary conditions for there to be any moral agents at
all. If all the principles, properties, and objects required for morality can be
said to be dependent upon the conception of the moral agent, then morality
would be transcendentally ideal. If, however, one or more principles, properties, or objects were required for morality and had to be conceived independent
of the moral agent, then morality would be transcendentally real.
The empirical reality of morality could follow from either transcendental
reality or transcendental ideality of morality. Empirical reality considers
whether the actual principles, properties, or objects required for morality as
part of the transcendental assessment of morality would exist independent of
actually existing moral agents. An empirical moral idealism would hold that all
principles, properties, and objects required for morality are entirely dependent
upon the moral agent qua moral subject. Empirical moral idealism would
actually preclude a transcendental realism (this is one way in which these
terms are not used in the same sense that Kant uses them with regard to objects
in space, for Kant takes transcendental realism to imply empirical idealism
[B6971, A36970]).
Here the aspects of morality under discussion are the categorical imperative,
the moral law, and the faculty of reason that generates them. The issue is
whether practical reason itself must be considered transcendentally and empirically real, transcendentally ideal but empirically real, or only empirically
ideal. Since reason is the source of the moral law it, will have the same status
as reason. The categorical imperative is tied both to the fact of reason qua
experience, which is empirically ideal, and to reason if reection upon the fact
of reason can provide justication for its validity.
The strongest interpretation one can give to the claim that reason has a transcendental structure is transcendentally realist. A necessary condition for moral agency
would be that the moral agent is subject to a law that holds independent of the
structure of moral agency, and thus is independent of the transcendental moral
agent. This interpretation would entail a conception of reason that somehow exists
in the nature of things and not merely as the wills or reasons of moral agents.20

20

I take this phrase from Patrick Kain, who holds to this view (Kain 2004, p. 303).

The transcendental status of empirical reason

141

The transcendental ideality of reason is the claim that the conception of

moral agency requires that the moral agent use reason and its principles in
deliberation before decision-making but that practical reason has no place
outside of the context of moral agency. Practical reasoning is simply the a
priori formal structure of reasoning in deliberation when one attempts to
choose freely, that is, without accepting determination from non-rational
elements such as desires or sensible impulses. This transcendental idealism
still allows for an empirically real morality. An empirically real reason would
have a basis independent of the individual moral agent. My metaphysically
naturalist interpretation of reason has already shown that the faculty of reason
can exist in nature within individual minds as described by empirical psychology. The transcendental justication of that principle provided by that reason
is valid whether or not the moral agent is individually aware of it or personally
acknowledges it. The conception of reason is thus independent of the particular
moral agent and to that extent, could be called empirically real. In this way,
Kant can be interpreted as an empirical moral realist because of the conception
of reason that transcends the empirical moral agent.21
A third alternative interpretation is also available. It sees Kant as an empirical moral idealist who provides no transcendental status to reason and sees the
categorical imperative only as an idea and a sense of obligation in the empirical
agent without further validity. All moral principles, properties, and objects
(here, as earlier, restricted to reason and the products of reason) would depend
upon the empirical moral agent. In this interpretation, all of the transcendental
structure attributed to reason and the moral law that results from it would be

21

Kain admits that the conception of realism that stresses reason as the source of the moral law
independent of any particular moral agents does not rule out what he calls global constructivist
anti-realism that relies on the nature of Kants transcendental idealisms restrictions on the
validity of a priori synthetic judgments (Kain 2004, p. 261). The position of the transcendental
ideality of reason can be understood in terms of a global view. He offers only a very brief
objection to global anti-realism and takes a global anti-realism based on transcendental idealism
to require that morality would then, like the categories, have to be valid in nature in space and
time. He takes this to imply an empirical realism. To this extent, his reasoning is correct.
However, he then limits the understanding of nature in space and time to restrict all comprehension of objects in nature in moral ways. He objects that causal determinism in spatiotemporal
nature would preclude the empirical reality of autonomy. My explanation of the structural
account of transcendental freedom of reason shows that the autonomy of reason can describe an
empirically real faculty of reason. He also objects that morality cannot be limited to beings in
nature because morality is supposed to be valid for all rational beings, not simply those who
share our forms of intuition. It is true that the moral law would be valid for all rational beings,
but in the same vein, the pure concepts of the understanding would be valid for any possible
experience. In both cases, they are in practice limited to nature in space and time if, as
naturalists would claim, there are no other forms of intuition and our ontology is restricted to
nature in space and time. Validity for hypothetical beings and for hypothetical nonspatiotemporal existences does not translate to actual existence restrictions, and the global
idealist position would still be valid for all possible rational beings.

142

Practical reason in nature

only ideas held by empirical moral agents as part of a practical perspective.

They would have no actual transcendental status that is, they would not be
conrmed as transcendentally required but only pictured as such from a certain
point of view of the empirical agent, the way that Kants admission that no
deduction of the moral law is possible, the identication of the fact of reason
with a conscious experience in the empirical agent, and the nature of practical
philosophy in general as directed only at determination to action and not to
ontological conclusions, point in this direction.
There are reasons in favor and against each of these interpretations. In the
next three sections, I will evaluate each as interpretations of the role of reason
in Kants ethics, given the structural account of pure practical reason that
I argued for earlier.
Transcendental moral realism
The strongest interpretation of the role of reason in Kants ethics places it
independent even of transcendental moral agency. I think that such an interpretation would make Kant into a traditional rationalist and should be rejected.
If reason were to be understood as transcendentally real and independent of the
conception of the rational moral agent (or more generally, the conception of
rational beings), it would have to be valid in itself for all possible existences
and for all possible beings in all possible worlds. This independent and
universal reason would have to include the theoretical as well as the practical
since there is only one reason with two different uses. In other words, reason
would have to have absolutely universal applicability. The ideas of reason
would have to be understood to correspond to actually existing properties and
objects. A soul and God would have to be conrmed as actual based on the
mere ideas of them produced by reasons search for the unconditioned. All
possible worlds would be rationally structured and all ideas of reason would
have to be correct. Such use of reason would violate Kants restrictions on
reason.22 Giving reason an independent transcendentally real status violates
the strictures on reason that Kant so carefully argues for in the rst Critique.
A further difculty is ontological. If reason is supposed to exist independently of the transcendental moral agent, then under an ontological reading of
transcendental idealism, it would have to exist independently of human beings
in themselves as ultimately real beings existing ontologically prior to their
appearances in space and time, since on an ontological reading, these beings in
themselves would be the transcendentally real moral agents. Further, a deus ex
machina solution identifying God as the locus of reason independent of moral
22

I discuss the immanent rather than transcendent validity of the ideas of reason in (Rauscher
2010).

The transcendental status of empirical reason

143

agents would not work since God would also be a moral agent. I have always
been puzzled by a claim that there can be a faculty like reason independent
of the existence of beings with that faculty. There are two ways that reason
could be independent of rational agents: rst, by being embodied in the things
that exist as their rational structure and second, by existing abstractly as a kind
of relation of thoughts. The rst option, I have already rejected in the previous
paragraph because it would violate Kants critical restrictions on reason. The
second option is essentially identical to the transcendental ideality of reason as
structure and so does not work as an ontological alternative to the conception
of reason as restricted to moral agency. The transcendental conception of
moral agency with regard to reason thus points to a transcendentally ideal
morality.
Transcendental idealism and empirical moral realism
I believe that the interpretation that makes Kant a moral transcendental idealist
and empirical realist would t some of Kants and most of contemporary
Kantian intentions better, and that this version of realism regarding the faculty
of reason and its principles is quite plausible as a philosophical position and
interpretation of Kant.
The reasons that point to a transcendentally idealist and empirically realist
view begin with a look back at Kants claim that no deduction of the moral law
is possible. I believe that Kant is mistaken in denying that a transcendental
deduction can be given for the moral law. Kant had argued that there was no
empirical object in morality and so no possible empirical conrmation for any
possible deduction. It is true that the moral law has no objects of experience for
which it can serve as the transcendental basis, and thus, a transcendental
deduction could not proceed in the same manner as the theoretical transcendental deduction of the categories. But Kant does not have to deny that there is
a kind of object for which the moral law is a basis, namely the experience of a
free rational agent deliberating courses of action. No empirical object is
involved in this deliberation, only a decision about free action. The moral
law could be seen as the necessary basis for any rational deliberation at all.23
I believe that some sort of transcendental argument about the conditions for
agency is what Kantian realists must strive for. If such an argument is
successful, it still ts within a naturalistic framework. The transcendental
justication of the moral law would mean that for a particular kind of experience to be possible in this case, any kind of deliberative decision-making
23

I take Christine Korsgaard to have attempted something like a transcendental deduction in this
sense in (Korsgaard 1996b). Robert Stern provides a positive assessment for a Korsgaardian
transcendental argument for morality in (Stern 2011a).

144

Practical reason in nature

seen as free from the agent-perspective a certain foundational a priori

structure is required in this case, the moral law stemming from the structure
of reason. One could broadly understand this to provide transcendental conditions for the evolution of any species of conscious deliberative agents. The
transcendental structure itself does not depend upon the existence of any
particular species to embody it; morality is independent of the existence of
particular beings. And it would not be a mere accident that human beings
evolved to possess this structure of reason, for any organism that is to evolve
the capacity for deliberative decision-making must possess this same structure
of reason. By analogy, any organism that is to evolve with the capacity for
representing an objective experience would, according to Kants transcendental deduction of the categories, have to embody a cognitive structure that
processes causal relations. The transcendental validity of the moral law does
not depend upon the empirical minds of rational beings any more than the
validity of causality depends upon the existence of particular human beings,
although similarly, if any organism is to evolve that could have objective
experience, it would be a priori restricted to certain structures in its cognitive
system that would be able to process sensations in causal relations and thus
embody the category of causality. Moral beings and cognizers embody these
different principles and concepts but do not make them. The moral law would
retain its necessity amid the contingency in nature. It is true that the fact that a
particular individual thinks about the moral law is contingent as a result of
the evolutionary and developmental processes explained earlier. Of course, the
existence of any particular individual and even the entire species is contingent.
The contingency of these particulars does not detract from the necessity of
the moral law itself considered as a transcendental condition for moral agency.
No reective, conscious, acting, rational being could have any other moral law,
since the moral law arises due to the nature of the structure of reason itself.
This picture of Kant as a transcendental moral idealist and empirical moral
realist within the ontology of nature is quite appealing.
Yet, that conclusion would also face the following difculty as an interpretation of Kant. Rather than proving that human beings really do have pure
practical reason, he allows only the admission of the fact of reason as a fact
that human beings experience the categorical imperative. Pure practical reason,
as he says in the Preface to the second Critique, cannot be proved except by
what it does in actually determining the will; that is to say, in our practical
subjective experience of decision-making, we feel the obligatory power of the
categorical imperative as the fact of reason and interpret this experience in
light of a claim that we possess pure practical reason as the only explanation of
the source of the categorical imperative. This is not an objective proof that
would rise to the level of cognition; it is only a practical belief that remains at
the subjective level of the rst person. It forms the essence of the practical

The transcendental status of empirical reason

145

perspective that, as part of practical philosophy, has no ontological commitments but is geared only toward determining free actions. The difculty in
interpreting Kant as a transcendental moral idealist and an empirical moral
realist, then, is in his refusal to afrm that human beings can be cognized
objectively as beings with pure practical reason. We might be required to
believe that morality is transcendentally justied, but that is different from a
claim that it really is transcendentally justied.24
Empirical moral idealism
The third possible interpretation would make Kant simply an empirical moral
idealist. Practical reason and the moral law would have no transcendental
status at all and no meaning outside the empirical moral agent. We can believe
but we cannot prove that we are rational beings. Although it is clear that Kant
believes that human beings possess reason, he does not make this belief an
objective theoretical cognition. This point concerns the distinction between the
practical ought and the theoretical is in Kant. Practically, human beings
conceive themselves as possessing the transcendental structure of reason that
demands consistency, uniformity, universality, and the like. Theoretically,
however, human beings cannot prove that they are obligated to the moral
law (the failure of the deduction). We can believe that there is a moral law, that
is, that there is a transcendental structure of pure practical reason. But we
cannot prove that simply because human beings possess pure practical reason,
they are obligated to follow its dictates. Obligation is a practical rather than a
theoretical matter, a matter of ought rather than is.
This empirical moral idealism is not simply based upon the lack of success
of a deduction. The positive aspect of this view is that human beings moral
lives are simply constituted by the perspective they take when faced with
action. As an inescapable perspective, it would not be relative to individuals or
cultures but would be species wide (and extend to any other species with
similar mental structures).
As empirically ideal, the moral perspective would not be objective. With no
deduction to prove that only this point of view is legitimate for active deliberative beings, the possibility would remain that other kinds of deliberative
24

Allen Wood brushes over this important distinction. He claims that his interpretation of Kants
deduction of the moral law draws no distinction between our having to take ourselves, from a
practical standpoint, to be capable of judging according to objective reasons, and there actually
being such reasons for us to judge according to . . . [and] between taking ourselves (from a
practical standpoint) to be capable of setting ends with objective worth and there really being
objective worth for those ends to have . . . [and] our taking ourselves (from a practical
standpoint) to be responding to moral requirements that are unconditionally obligatory and
the actual existence of such categorical requirements (Wood 1999, p. 381, note 30).

146

Practical reason in nature

beings would have other possible determinants of behavior besides just desires
and reason. This possible relativism would not be itself part of the practical
view but would only become apparent when one steps aside from practice and
reects theoretically on the nature of this moral perspective. From within the
moral perspective, no deduction is necessary. I will say more about this
perspective in Chapter 7 in a discussion of the nature of the objective order
provided by reason.
I conclude these observations about the relation of reason and the moral law
to moral realism, then, with the following alternatives. If, on the one hand,
there is a way to conrm that the transcendental structure of moral agency has
more than simply the status of a set of subjectively held but unconrmable
beliefs as part of the practical point of view, then Kant must be a moral
transcendental idealist and empirical realist. The structure of reason would
be known through a transcendental argument but the existence of reason
depends upon the contingent existence of particular deliberative beings. If,
on the other hand, morality is nothing more than the practical point of view
that human (and similar) beings must adopt, and there is no guarantee that it
has any further validity, then Kant must be considered a moral empirical
idealist.
Concluding remarks
In this chapter, I have presented the core of my interpretation of Kants ethics.
I argued that reason as the source of the moral law is the most important faculty
for Kants ethics, more important than the power of decision in the power of
choice. I showed that reason has a place in nature as empirical reason. The
Third Antinomys emphasis on the free timeless causality of reason was
understood in terms of the structure of intelligible reason itself. Transcendental
freedom does not require any non-natural entities but is simply the manner in
which the structure of reason can be understood, and can function as a rational
cause, independently of determination by alien causes. Reasons role in morality, and transcendental freedom itself, can be understood within a metaphysically naturalistic framework. I concluded by arguing that the role of reason in
Kants ethics would t with a transcendental moral idealism and an empirical
moral realism best, although elements of Kants thought also point toward only
an empirical moral idealism. The next few chapters will examine other parts of
Kants moral theory to show how they t within the metaphysically naturalist
framework and moral idealism.

Part III

Morality beyond nature?

God without God: the status of the postulates

The two previous chapters looked at the lack of ontology in the primacy of
practical reason and the status of reason within nature, showing that they are
compatible with both a metaphysical naturalism and moral idealism. This
chapter will argue that the postulates of practical reason particularly belief
in God can readily be understood in a metaphysically naturalistic framework.
The argument will also thereby reject any realism about God. I will show that
Kant had two different tendencies in his discussions of the postulates, one of
which focuses on God as an object and the other on the concept of God. (In this
chapter, I will generally use the term concept in discussing the idea of God,
any transcendental hypothesis about God, and the postulate of God; I intend
this term in its broad sense in which concept is usually contrasted with object.)
The latter aspect of his discussion of the postulates, I claim, provides a basis
for the naturalistic interpretation. I also argue that Kants nal conception of
the postulates in the Opus Postumum reects this concern with concept over
object.
Along with the idea of the highest good and as a precondition of its
realization, Kant offers the postulates of immortality of the soul and the
existence of God. His arguments have often been rejected by commentators,
mainly on the basis that the kind of rational hope for happiness Kant stresses is
not really required for moral obligation, thus undermining the practical nature
of these postulates; although recently more positive attention has been paid to
them and to Kants philosophy of religion in general.1 I will not assess the
1

In Chapter 3, I showed how the priority of belief claries the way in which moral obligation is
independent of the postulates. For an early example of traditional criticism of Kants argument
for the postulates, see (Beck 1960, p. 274f.). A defense of Kants argument can be found in
(Wood 1970, p. 129f.). In the past dozen years, a plethora of work on Kants philosophy of
religion has appeared, generally stressing an interpretation that takes Kant to be religious and his
philosophy to require adherence to Christian doctrine. In general, I think that this interpretation
overemphasizes the extent to which some of Kants positions require religion. My chapter will
provide an example of the kind of language that Kant uses in his philosophy that weakens their
claim. Further, Manfred Kuehn has shown convincingly that Kant had no strong personal
religious beliefs (Kuehn 2001). For a philosophical defense of a Kantian atheism, see (Denis
2003).

149

150

Morality beyond nature?

validity of those arguments; rather, for purposes of argument, I will assume

that something in Kants moral theory does require the postulate of God.
Instead, this chapter will assess the status of the postulate of the existence of
God to see just how it is compatible with a naturalistic metaphysics. I will
argue that the postulate of the existence of God is best understood as the claim
that rational beings must hold in their minds some particular concept or other
that functions as a bridge to satisfy the systematic demands of practical reason,
yet that does not, and is not intended to, refer to any existing object.2
2

In some ways, my argument is similar to the famous as if interpretation of Kant offered by

Hans Vaihinger a century ago. Based on ideas he developed around 1875, Vaihingers Die
Philosophie des Als-Ob was very inuential; it went through at least six editions in the dozen
years following its publication in 1911, including an abridged Volksausgabe (peoples edition)
in 1923 (Vaihinger 1911) and (Vaihinger 1923). In brief, Vaihinger offered a philosophical
system in which human life is structured by a set of assumptions and ideas which are themselves
ctions but which help human beings survive and thrive. He held that many mathematical
concepts such as the point and line, physical concepts such as the atom, juridical concepts such
as rights, moral concepts such as freedom of the will, and of course the practical concept of God
are all ctions. Vaihinger discusses ctions as resulting from the teleological and organic
structure of thought (the psyche) in a world which is itself mechanical not teleologically
organized. The psyche nds it very useful to employ certain concepts which help it to full
the purpose of navigating the organism through its life; these concepts are purposeful functions
(zweckmssige Funktionen) (Vaihinger 1911, p. 2). The concepts do not themselves correspond
to actual objects in nature (or outside nature for that matter) but as part of the system of thought
they function to guide the organism successfully. Vaihinger describes this relationship as
follows: Objective events and being may be what they will, nonetheless it must be stressed that
they do not consist of logical functions [. . .] Their purposiveness manifests itself precisely in
this: the logical functions, though they operate according to their own laws, still always manage
to coincide with being (Vaihinger 1911, p. 10, 12). One might summarize this reference relation
as follows: individual ctitious concepts do not correspond with reality, the functional system
which they help to constitute does. This functional system would consist of non-ctitious
concepts glued together, as it were, by the ctitious concepts.Vaihingers theory is not presented
as a straightforward interpretation of Kant. He presents the philosophy of as if as his own,
inspired by Kant (and others). Vaihinger does, however, present what he calls historical
conrmations that focus mainly on Kant but also include Nietzsche and some other gures.
His discussion of Kant does not claim to be an objective analysis of Kants actual philosophical
statements, proving that Kant really always held the theory of as if. Rather, Vaihinger admits
that he selectively chose passages from Kant which reveal this tendency in Kant and excluded
passages which contradict this interpretation, for example, (Vaihinger 1911, p. 639). His point is
that, as a philosopher, Kant was presenting the kernel of the as if theory without himself fully
developing it, indeed even without realizing the full potential of this theory. One must keep this
in mind when considering objections to Vaihingers theory, and correspondingly to the interpretation I present. For this reason, one ought to reject the charges against Vaihinger raised by
Erich Adickes in his book-length attack on Vaihingers Kant interpretation, Kant und die AlsOb-Philosophie (Adickes 1927). For all its academic rigor and thorough review of Kants
writings, published and unpublished, Adickes argument boils down to the claim that Vaihingers
Kant is not the historical Kant (Adickes 1927, p. 291). If the standards for assessing Vaihingers
view of Kant are strict historical accuracy, then Vaihinger himself would admit that his theory is
inadequate because it does not capture Kants complete historical view. Vaihinger does, however, note that some of Kants arguments point toward a theory of ctions, even if Kant himself
did not fully develop it. Vaihinger is, as it were, looking at the Kant that might have been had the
historical Kant followed a path that he saw, and even mapped, but did not fully traverse.

God without God: the status of the postulates

151

To interpret the postulate of the existence of God merely in terms of nature

in space and time is to interpret it as not requiring attribution of the actual
existence to God. Rather, the concept God is held in the mind of a moral
agent without afrming the existence of any independent being matching that
concept. Instead, the moral agent uses the concept of God, qua concept in
relation to other concepts, for the practical purpose of furthering morality.3
Certainly, by the end of his career as evidenced by the Opus Postumum, Kant
had clearly come to such a position. For example, he writes in one of his last
notes to himself, The idea of that which human reason itself makes for the
universe is the active representation of God. Not as a particular (personality)
substance outside me but a thought in me (21:154). Human beings are said
here to create the concept of God, and we comprehend this concept only as a
thought in us, not as a separate substance outside us. This chapter will show
that this conception of the postulate of God is implicit in Kants earlier
discussions of the postulates of practical reason. I argue that it is the position
Kant was working toward throughout his philosophical development, and that
it best serves the purpose of the postulate.
To reach this result, I will trace Kants conception of a postulate in three
main stages: prior to the Critique of Practical Reason, in the Critique of
Practical Reason, and after the Critique of Practical Reason.
Before the Critique of Practical Reason
This rst section will show that prior to his direct work on the Critique of
Practical Reason, Kant had not rmly conceived of practical postulates as
distinct from theoretical ideas and hypotheses. It will also show that Kant had a
tendency to treat postulates as well as ideas and hypotheses as abstracting from
ontological claims in favor of their use for some particular purposes.

My interpretation bears some resemblances to Vaihingers; the differences will become apparent
as my argument proceeds. I would like to mention a few key differences here. First, unlike
Vaihinger, I stress that this theory of ctional concepts is for Kant mostly practical; I think that
Kants theoretical philosophy of nature uses ctions only in the regulative use of reason, but not
in physics or mathematics themselves. Kant is an empirical realist about both pure and empirical
concepts of the understanding. Second, I try to be clearer about the possible referential relations
between ctitious concepts and objects and the sense of immanence Kant employs. Third, I take
the ctions to be a priori and not merely part of the empirical psyche. Finally, I do not use the
term ctions but stick with Kants terminology.
Peter Byrne advocates a similar interpretation of the postulate but offers a more comprehensive
view of Kants use of the idea of God overall. He convincingly argues against the realist view of
Kants religious claims and in favor of a subjective, anti-realist view. In particular, he stresses
that Kants critical philosophy precludes any traditional metaphysical view of God and of
religious language (Byrne 2007).

152

Morality beyond nature?

Early conception of a postulate

Kants use of the term postulate rst comes in his early metaphysical works
where it applies to theoretical claims.4 Kant was certainly exposed to the term
from its use in Meiers Logic textbook, where 315 says:
Unproven judgments are either judgments after due consideration [Erwgungsurtheile]
or judgments made in the moment [bungsurtheile]. The former are axioms [Grundurtheile](axioma) but the latter are postulates [Heischeurtheile](postulatum).5

The precise German term Heischeurtheile means a judgment on demand,

which relates the term to the Latin postulatum, which means a demand or
request. Thus, the meaning of the term postulate that Kant employs is that of
an unproven claim required by a particular context. This is conrmed in Kants
own notes to his copy of the Meier textbook in the latter half of the 1770s,
where he comments on this very section:
A postulate is actually a practical immediately [crossed out: necessary] certain proposition. But one can also have theoretical postulates for the sake of practical reason,
namely, a theoretical hypothesis that is necessary from the point of view of practical
reason, such as that of the existence of God, of freedom, and of another world. Practical
propositions are objectively certain; subjectively, they can only become practical insofar
as that hypothesis serves as their ground. (R 3133, 16:673)

At this time, Kant had not yet distinguished practical postulates as such from
other kinds of postulates. He uses the terms postulate and hypothesis
interchangeably, showing that he had not nalized any special or distinct role
for what will become the postulates of practical reason. For example, in
R4113, written sometime from about 1765 to 1775, Kant notes that the idea
of God is a necessary logical ideal, the necessary hypothesis of the natural
order and the necessary hypothesis of the moral order (17:421). Hypotheses are seen as serving both theoretical and practical purposes (see also
R4582, 17:601, and R4928, 18:30).
The postulate is also construed as a concept that must be held by the mind in
order to play some sort of functional role while the existence of any object
corresponding to that concept is not claimed. However, this suggestion is made
not with regard to a practical but with regard to a theoretical proof of God;
theoretically, only a concept of God is attainable, practically, proof of the
object of that concept is also attainable. In a Reection from the early 1770s,
Kant distinguishes between the need for a concept and the need for its proof;
however, the proof in question is a moral proof necessitated by the experience
4
5

For example, see the Inaugural Dissertation, 2:402, for the postulate of pure time, and 2:418, for
the postulate that nothing material at all comes into being or passes away.
The text of Meiers Logic is reprinted in Kants gesammelten Schriften, here 16:668.

God without God: the status of the postulates

153

of moral order rather than any purely a priori reection in pure reason as part of
an explanation of the theoretical pure concept of possibility:
The transcendental concept is necessary, not the transcendental proof; in that one sees
that one cannot make heterodox claims without any ground. The necessity of the divine
being as a necessary hypothesis either of pure concepts of possibility or of experience in
this world, and the latter as hypothesis of morality. Absolute necessity cannot be
proved.
The proof of the existence of God is not apodictic but hypothetical sub hypothesi
logica und practica [under logical and practical hypotheses].
A hypothesis that is necessary in relation to the laws of pure reason is originaria
[original], a hypothesis that is necessary in relation to experience is conditionalis sive
relativa [conditional or relative]. The rst is necessary, the second necessitated; the
former originaria, the latter subsidiaria [subsidiary], e.g. suppletoria[supplementary]
for explanation of order in the world. (R4580, 177275, 17:600)

This moral proof is one that is said to be necessitated rather than necessary,
that is, required for some particular empirical circumstances rather than purely
a priori. The moral order in the world must be explained using a transcendental
hypothesis, yet that hypothesis remains conditional upon or relative to the
alleged moral order in the world. For theoretical purposes, only a transcendental concept of God is allowed, but no proof. When a proof is available, then, it
is only contingent upon a certain empirical fact.
In the Critique of Pure Reason
Even in the Critique of Pure Reason itself, the term postulate is used
theoretically, occurring most often in the name of the three modal principles
of the understanding, the postulates of empirical thought.6 He also calls
reasons search for the unconditioned in the Dialectic, a postulate of reason
(A498/B526). There is no hint of Kants later special status of practical
postulate (although there is a relation to the nature of belief in the Canon
of Pure Reason, which will be discussed later on in this chapter)
The term hypothesis, however, is restricted to theoretical claims. The
section The Discipline of Pure Reason with Regard to Hypotheses is worth
a detailed look. Hypotheses, as Kant denes them, are not merely any kind of

Paul Guyer has argued that the use of the same term postulate here and in the second Critique
is not accidental, and that the practical postulates Kant gives in the second Critique must be
understood in terms of the postulates of empirical thought, in particular that in practical
postulates, the concepts must be understood as possessing real possibility, not merely logical
possibility (Guyer 2000, pp. 33371). This claim reads too much into the use of the same term.
The different status of the postulates of practical reason and the postulates of empirical thought
are clear in their being assigned to practical reason, the practical point of view of the ought,
and empirical thought, the theoretical point of view of the is.

154

Morality beyond nature?

groundless assertion. They must be connected as a ground of explanation with

that which is actually given and consequently certain (A770/B798). That which
is actually given in experience must always conform to the conditions of possible
experience, our reason cannot conceive of the possibility of things in any other
way. Hypotheses must be connected to experience; in this way, they are like the
hypotheses and postulates Kant noted in R3133 that cannot simply be freeoating assertions but must arise in particular situations to resolve particular
problems. Still, transcendental hypotheses, like ideas of reason, do not themselves
represent objects of cognition. Here is how Kant describes the ideas of reason
before contrasting them with transcendental hypotheses:
The concepts of reason are, as we have said, mere ideas, and of course have no object in
any sort of experience, but also do not on that account designate objects that are
invented and at the same time thereby assumed to be possible. They are merely thought
problematically, in order to ground regulative principles of the systematic use of the
understanding in the eld of experience in relation to them (as heuristic ctions). If one
departs from this, they are mere thought-entities, the possibility of which is not
demonstrable, and which thus cannot be used to ground the explanation of actual
appearances through an hypothesis. (A771/B799)

Ideas of reason merely ground the regulative principles of reason as heuristic

devices. Kant provides the example of the idea of reason of the soul, into which
one cannot have any insight in concreto, yet that allows one to think the unity of
powers of the mind (A77172/B779800, compare A672/B700f). It must at
least be possible for there to be a unity of the powers of the mind. The ideas of
reason thus have a positive use. Transcendental hypotheses, however, do not
even rise to this level of positive use: they cannot provide any ground of explanation of appearances. A transcendental hypothesis is one in which a mere idea
of reason would be used [directly] for the explanation of things in nature as
opposed to the regulative use of the idea of reason, in which the idea is used only
to guide inquiry in nature and thus indirectly relate to things in nature.
The discussion of the use of transcendental hypotheses offers two important
further points. First, Kant claims that the genuine use of transcendental
hypotheses is merely negative, as a defense against unwarranted attacks on
particular philosophical positions. And second, transcendental hypotheses are
themselves never fully endorsed but only offered defensively in particular
contexts, for particular purposes, and for a particular length of time.
Suppose, rst, that one nds oneself discussing the nature of the soul and is
faced with someone who claims that experience in nature offers strong indication that our mental life is all physically based. In this context, Kant says, one
may offer the following transcendental hypothesis in defense:
That all life is really only intelligible, not subject to temporal alterations at all, and has
neither begun at birth nor will be ended through death, that this life is nothing but a
mere appearance, i.e. a sensible representation of the purely spiritual life, and the entire

God without God: the status of the postulates

155

world of the senses is a mere image, which hovers before our present kind of cognition
and, like a dream, has no objective reality in itself; that if we could intuit the things and
ourselves as they are we would see ourselves in a world of spiritual natures with which
our only true community had not begun with birth nor would not cease with bodily
death (as mere appearances), etc. (A77980/B80708)

This transcendental hypothesis, which sounds eerily like a strongly ontological

interpretation of transcendental idealism, is certainly an extreme form of
defense against a materialist theory of mind. Note that it has no direct explanatory power for anything empirical: that is, empirical experience in nature is left
phenomenally the same. The transcendental hypothesis makes no claim to
cognize any particular part of nature. This hypothesis is merely negative,
serving as a defense against an unjustied claim to knowledge.
The second and most important part of this passage is yet to come. Kant
continues immediately by noting that the person who makes this claim does so
without actually believing the hypothesis.
we do not know or seriously assert the least thing about all of this which we have here
pleaded against the attack, and it is all not even an idea of reason but merely a concept
thought up for self-defense . . . he who turns such hypothetical countermeasures against
the pretensions of his rashly negative opponent must not be considered to hold them as
his own genuine opinions. He abandons them as soon as he has nished off the
dogmatic self-conceit of his opponent. (A78081/B80809)

The hypothesis, then, is not asserted as true. It is not part of the world-view of
the person who makes this claim. It is asserted as not-impossible. The person
makes this claim only for a particular purpose without actually asserting the
truth of the claim. The particular purpose is the main point for the claim. As
Kant sums it up, Thus one sees that in the speculative use of reason hypotheses have no validity as opinions in themselves, but only relative to opposed
transcendent pretensions (A781/B809).
These two differences between ideas of reason and transcendental hypotheses
should not obscure the similarities. An important similarity to keep in mind is
that both concern concepts that we cannot have insight into in concreto
(A771/B799, modied). Kant notes this with regard to the example he gives
of a transcendental hypothesis. He says strongly We do not know or seriously
assert the least thing about all of this. He also notes that it remains wrong to
extend the principles of possible experience to the possibility of things in
general; that is, it remains wrong to apply the categories and principles of the
understanding beyond the boundaries of possible experience in nature (A781/
B809). Hence transcendental hypotheses cannot be afrmed in any detail.
A transcendental hypotheses, then, is not a reference to an object at all. If
anything, it is a placeholder, that is, it serves to highlight a lack of concrete
theoretical knowledge of objects in nature. This is to say, the transcendental
hypothesis has the function of pointing out to an opponent a lack of complete

156

Morality beyond nature?

support for the opponents claims. That is how the transcendental hypothesis is
able to be used as a defense. Since the hypothesis is never afrmed as such,
and in fact cannot ever be afrmed because not known in concreto, its meaning
is equivalent to its function. Its work is merely negative, to remind the
dogmatic opponent that he is making a claim that transgresses over the
boundaries of experience and thus over the boundaries of knowledge. No
ones ontology is extended by this claim. It is meant to remind others of the
limits of their ontology.
I think this is the best way to understand transcendental hypotheses in Kant.
I also think that this is how Kant ought to have, and eventually does, understand the practical correlate of transcendental hypotheses, namely the postulates of practical reason. But Kant does not reach this sort of conclusion right
away. Rather, he explains practical ideas in a way that makes them stronger
than theoretical ideas. Within the discussion of transcendental hypotheses,
Kant says that in regard to its practical use reason still has the right to assume
something which it would in no way be warranted in presupposing in the eld
of mere speculation without sufcient grounds of proof (A776/B 804). Thus,
the negative element of transcendental hypotheses is not at issue for practical
hypotheses. There will be further, positive reasons to afrm something about
the practical hypotheses. The positive reasons, of course, will turn out to be
related to the need for the realization of the highest good.
The canon of pure reason
The rst Critique never uses the term postulate to describe a moral claim and
seems to restrict the term hypothesis to theoretical claims. Still, moral claims
that will later be called postulates of practical reason are raised in the Canon
of Pure Reason chapter. The Canon is a rich resource for assessing the state of
Kants concerns in both ethics and systematic issues; its importance comes in
part from the fact that when Kant wrote the Critique of Pure Reason, he
planned no further critical works before turning to his metaphysics of
morals. Thus, the Canon contains what Kant considered before 1785 as the
important critical discussion of ethics in its broader relation to epistemology,
metaphysics, and his critical system.
The focus for the present topic is rather limited: the status of the concept of
God as a belief rather than a hypothesis. To what extent does Kant claim that
belief in God is a claim that the concept God refers to an actual object? And
to what extent does Kant provide good reasons for claiming any difference
between the practical and the speculative uses of the concept God.
Only reason in its practical use, Kant claims, can have a canon, or the sum
total of the a priori principles of [its] correct use (A7967/B8245). The
whole of the Critique up to this point, Kant claims, shows that speculative

God without God: the status of the postulates

157

reason has no correct use and thus cannot have a canon. (This claim goes too
far, even if one accepts that the negative use of transcendent concepts just
described does not count as a correct use; for the ideas of reason do have a
correct and positive regulative use.7) The only legitimate use of reason then
must be practical. Kant denes the practical use of reason as everything that is
possible through freedom (A800/B828). In Chapter 2, I showed that the
practical is dened as the rules for free action and free behavior, and as
determining what ought to occur as opposed to what is in existence. The
practical, then, concerns moral laws, and thus these alone belong to the
practical use of reason and permit a canon.
The concept of God, then, is permitted because of its relationship to the
moral law. Presumably this means the following: If there is a moral law
stemming from reason, then some other concepts or ideas of reason are so
tied to it that they must be afrmed. That is, one would have reason to use the
concept God if reason already provides moral laws that somehow demand
that the concept God be invoked; this relationship is exhibited in Kants
arguments in the second Critique. Kant initially does not present the problem
this way in the Canon but instead reverses their relationship: These [three
problems of freedom, God, and immortality], however, have in turn their more
remote aim, namely, what is to be done if the will is free, if there is a God, and
if there is a future world(A 800/B828). Such a statement implies that the
existence of God and immortality are to be used as determinants of the moral
law itself or our motivation to follow it.8 Kant states the proper relationship
when asking his three famous questions that summarize philosophy. What
may I hope or alternatively if I do what I should, what may I then hope? is
the link between morality and God (A805/B833). Roughly speaking, this
reverses the previous conditional. Here is the proper Kantian formulation, for
only on this basis can we start with the moral law and proceed to God.
Kant admits that the question he focuses on, what may I hope, is simultaneously practical and theoretical, so that the practical leads like a clue to a
reply to the theoretical question and, in its highest form, the speculative
question (A805/B833). One might object that the canon for practical reason
is misplaced, since it is now a canon that mixes together some elements from
the practical (namely, determination of our duties) with some elements of the
theoretical (determining causal relations and effects, in this case, happiness).
7
8

See my explanation of the positive role of reason in the Critique of Pure Reason, covering both
the Appendix to the Dialectic and the Canon of Pure Reason (Rauscher 2010).
Henry Allison points out that this is a view that Kant ultimately abandoned. Allison identies it
with the claim that the existence of God and immortality are necessary as motivations to follow
the moral law (Allison 1990, p. 67). Strictly speaking, the sentence is ambiguous between
determining ones duties (what is to be done) and looking for motivation toward those duties,
but as Allison points out, the general context favors the claim about motivation.

158

Morality beyond nature?

The relation between these two elements lies at the core of the interpretive
controversy over the postulates: determining whether the practical element
stressing action and abstracting from ontology or the theoretical element
stressing ontology should predominate.
The theoretical question, of course, is whether God exists. But even here in
the Canon, there are indications that Kant does not require there to be an
independent being God as substance. And since theoretical reason is at
work, it seems as if Kant ought to invoke some of the same concerns that
arose in the discussion of transcendental hypotheses, in particular, the way that
the concept or idea of reason would not itself be afrmed but only entertained
for the function of making some other claim, or defeating a claim made by
another.
Indeed, Kant hints at this status, although he does not clearly state it. This
comes in the following section of the Canon, On having an opinion, knowing,
and believing.9 For the most part, this section presents belief in God as belief
that there is an object matching the concept God. While denying that human
beings have any knowledge that there is a God, Kant argues that human beings
inescapably believe that there is a God. He contrasts knowing as taking
something to be true on the grounds of both objective and subjective sufciency on the one hand, with believing something to be true on the grounds of
subjective sufciency but objective insufciency on the other hand. Belief in
God, then, appears to be a taking to be true that there is a being God
matching the concept God.
But a closer look reveals some concerns on Kants part to weaken this
interpretation. To see this, it will help to focus on the denition of conviction
as subjective sufciency, which leads to the issue the nature of the practical
relation that lies at the basis of belief.
Kant explains subjective sufciency as conviction, and denes conviction this way: If it is valid for everyone merely as long as he has reason, then
its ground is objectively sufcient (A820/B848) and contrasts this with
persuasion, which is not based on reason but on the particular constitution
of the subject. Insofar as a human beings reason is the basis of the taking to
be true, then, the human being believes something on a subjectively sufcient
basis. But it does not have objective sufciency, for then it would be knowledge. Kant denes objective sufciency as certainty (for everyone) as
opposed to conviction (for myself) (A822/B850). Why would conviction,

This section of the Canon can be interpreted as either a general explanation of levels of assurance
in theoretical claims, as in (Chignell 2007), or as mainly concerned with the practical (Stevenson
2003). Given its place in the Canon, I take Kant to be using this section to explain the peculiar
status of practical beliefs.

God without God: the status of the postulates

159

based on reason shared by everyone rather than the particular constitution of

the subject, not be a taking to be true for everyone but only for myself?
The only way that validity can be universal among humans as rational beings
yet subjectively sufcient only for oneself is if the proposition in question
arises from something universally shared yet private. Reason as a faculty is, we
assume, universally shared among human beings but it cannot be their mere
possession of reason as a cognitive faculty that is private, for our understanding is another cognitive faculty that is private in this same sense, yet Kant does
not thereby claim that the cognitions of the understanding are merely subjectively sufcient. There must be some private relation between the individual
and his reason, a relation that cannot be shared with others although it can be
assumed to be experienced by others, that lies at the basis of this subjective
sufciency.
Indeed, that is exactly what Kant means. He argues that the practical
relation is the only possible basis of belief, as opposed to knowledge or
opinion (A823/B851). This practical relation is an aim for some end.
The only end that grounds belief in God, of course, is a moral end, which
Kant identies as that I full the moral law in all points, which, in combination with all ends together in a consistent system, requires that there be a
God (A828/B856). Kant is tying belief in God to the relation a person has to
obligation. He denies that this belief in God can be justied to anyone
objectively, instead placing its basis rmly in an individuals perspective:
Of course, no one will be able to boast that he knows that there is a God and a future life;
for if he knows that, then he is precisely the man I have long sought. All knowing (if it
concerns an object of reason alone) can be communicated, and I would therefore also be
able to hope to have my knowledge extended to such a wonderful degree by his
instruction. No, the conviction is not logical but moral certainty, and, since it depends
on subjective grounds (of moral disposition) I must not even say it is morally certain
that there is a God, etc., but rather I am morally certain etc. That is, the belief in a
God and another world is so interwoven with my moral disposition that I am in as little
danger of ever surrendering the former as I am worried that the latter can ever be torn
away from me. (A82829/B85657)

My moral disposition, that is the relation between my will (or choice) and the
moral law originating in my own reason, is the sole basis of belief in God. As
such, this relation is individual, depending as it does on each individuals own
choices. Kant notes in the following paragraph that an individual who was
indifferent to morality, that is, whose will did not take up a relation of
obligation to follow the law of reason, would not have this conviction, and
for such an individual, the existence of God would be a merely speculative
rather than a practical question.
Kants minimization of belief is perhaps clearest in his claim that the
effect the concept of God would have on a persons moral disposition is

160

Morality beyond nature?

what constitutes the belief. He claims that a belief has even less not
more, as one might expect weight than a transcendental hypothesis. Kant
says,
If here too I would call merely theoretically taking something to be true only an
hypothesis that I would be justied in assuming, I would thereby make myself liable
for more of a concept of the constitution of a world-cause and of another world than
I can really boast of; for of that which I even only assume as a hypothesis I must know
at least enough of its properties so that I need invent not its concept but only its
existence. The word belief, however, concerns only the direction that an idea gives
me and the subjective inuence on the advancement of my actions of reason that holds
me fast to it, even though I am not in a position to give an account of it from a
speculative point of view. (A827/B855, long emphasis mine)

Here Kant claims that a transcendental hypothesis will take a previously given
concept and invent (erdichten) only its existence. With a belief I am not in
possession of the appropriate concept to any degree of distinction, let alone am
I in a position to claim that an object exists matching that concept. Instead, the
belief is a functional relation between the idea (or concept) and the advancement of my actions. The belief has the following peculiar relation between idea
and, as it were, its referent: the idea refers not to an object but to somehow to
the individuals moral disposition. The belief functions to strengthen the moral
disposition.
That Kant offers these limitations on belief at least suggests that it is not
appropriate to interpret his conception of belief, the equivalent of a postulate in
the rst Critique, as a straightforward assumption on the practical side of what
could not be proved on the theoretical side. The nature of belief is rather
something different than simple assertion that an object exists to correspond to
a concept.
We see that there are two noteworthy elements to Kants theory of the
practical postulate of God in the period before and in the Critique of Pure
Reason. First, Kant at least sometimes assumed that theoretical and practical
hypotheses were essentially alike. He thought that both of them operated
along similar lines, falling short of proof but instead satisfying reason by
the introduction of concepts. Second, Kant, perhaps simultaneously, argued
that the practical offered a different, or stronger, kind of hypothesis that he
will later call a postulate. When he developed this difference in the rst
Critique in the Canon as a theory of belief, he hinted that beliefs were not
directed toward objects but toward the subject holding the belief. These
elements, when put together, indicate that Kant had a way to develop his
theory of postulates as concepts that do not refer to objects but that
nonetheless play a necessary functional role in our practical lives. He still
insisted, however, that the postulates did more than just that, as we will see
in the next section.

God without God: the status of the postulates

161

In the Critique of Practical Reason

The doctrine of the postulates of pure practical reason is explained in its fullest
in the Critique of Practical Reason.10 There are two contradictory tendencies
as Kant further develops his conception of the postulates. First, there is a
tendency to claim that practical reason does in fact afrm exactly the same
propositions that remain unproved by theoretical reason; in particular, there is
a stress on the existence of God as something practical reason not only may
assume for its own ends but also as something it can force theoretical reason to
accept. Second, in the other direction, there is a tendency to stress the interest
of practical reason and the function of the postulates for moral purposes along
with a minimization of their cognitive content.11
It is rst useful to note that Kant himself was worried that readers would
misunderstand the status of the postulates. He notes in the Preface to the
Critique of Practical Reason that this second Critique does not have the
problem the rst Critique had in introducing new and strange terminology.
The second Critique will even approach popularity in its topics and thus
ought to be relatively well understood (5:11). However, Kant added a footnote
to this paragraph to try to deal with what he obviously thought might be new
terminology that would be misunderstood regardless. Here is what he says
there about the term postulate:
But the expression, a postulate of pure practical reason, could most of all occasion
misinterpretation if confused with the meaning that postulates of pure mathematics
have, which bring with them apodictic certainty. The latter, however, postulate the
possibility of an action, the object of which has been previously theoretically cognized a
priori with complete certitude as possible. But the former postulate the possibility of an
object itself (God and the immortality of the soul) from apodictic practical laws, and
therefore only on behalf of a practical reason, so that this certainty of the postulated

10

11

I skip over some Reections from the period between the rst and second Critiques. In R6099
(18:45153), Kant says that a postulate is more than a hypothesis, and that this is because
morality is more than just pragmatic. R6109 (18:45657) and R6111 (18:45859) both stress
that God is a postulate not mere hypothesis, and that this is due to the inescapable moral interest
rather than just the contingent pragmatic interest. These can be taken to mean that postulates are
like hypotheses, except that instead of merely offering them temporarily for defense, they are
held in the mind as a kind of constant support for morality. See also R6283 (18:54950).
My conclusion parallels that of Markus Willaschek, whose detailed look at the postulates of
practical reason also stresses the importance of the concept over the object: as he phrases the
point, the practical extension of our cognition by the postulates does not concern the existence
of God, freedom, and immortality, but only the objective reality of our concepts of God,
freedom, and immortality (Willaschek 2010, p. 191). My chapter was originally published as
an article (Rauscher 2007) before his article was published and I am largely leaving it intact
because my analysis puts the postulates in the context of the course of Kants entire career, even
though Willaschek provides more detail about the postulates in the second Critique.

162

Morality beyond nature?

possibility is not at all theoretical, hence also not apodictic, i.e., it is not a necessity
cognized with respect to the object but is, instead, an assumption necessary with respect
to the subjects observation of its objective but practical laws, hence merely a necessary
hypothesis. I could nd no better expression for this subjective but nevertheless
unconditional rational necessity. (5:11n)

Kants attempt to clear things up, as is often the case, only makes them more
obscure. The postulates are said not to postulate an object itself but to postulate the possibility of an object. Yet possibility of objects is a theoretical
matter, not a practical determination. Further, Kant holds that the postulate is
a necessary hypothesis, thereby using a term, hypothesis, that as we have
seen, he had specically contrasted with moral belief in the rst Critique
(A827/B855), although one that he had also used to refer to both theoretical
and moral claims about God in his Reections. Kant also tries in this passage
to explain the nature of the necessity involved as not a necessity with respect to
the object but a necessity with respect to the subject, in fact with the subjects
observation of its objective but practical laws, which he calls a subjective
rational necessity. Kants attempt to clear up what he must have thought was
a confusing explanation of the postulates does not itself shed much light
on them.
The more detailed argument about the postulates comes in the Dialectic of
Pure Practical Reason, beginning with his introduction of the term in a
discussion of the immortality of the soul (5:122) and continuing through the
end of the Dialectic (5:148). I noted earlier that there are two tendencies in this
discussion. The rst is the tendency, easily recognized, to describe the postulates as claiming on behalf of practical reason the existence of objects that
theoretical reason itself could neither prove nor disprove. Kants denition of
postulate shows that the theoretical concepts appear to be at issue: a postulate is a theoretical proposition, though one not demonstrable as such, insofar
as it is attached inseparably to an a priori unconditionally valid practical law
(5:122). The bulk of section VI, On the Postulates of Pure Practical Reason in
General is devoted to showing how the postulates of immortality, freedom,
and the existence of God give objective reality to the related or identical
concepts from the Dialectic of the rst Critique (5:132133). This tendency is
also certainly quite strong in section III Primacy of Pure Practical Reason,
where Kant claims that the primacy of practical reason entails that it can insist
that speculative reason must accept these propositions and, although they are
transcendent for it, try to unite them, as a foreign possession handed over to it,
with its own concepts, implying that speculative reason must accept the
postulates as being of like kind with theoretical concepts (5:120). And of
course there is the abundant use of the term exists, both in reference to the
immortal soul and God (see, for example, 5:122 and 5:124). In fact, on face
value, it appears that this tendency to treat the postulates as the practical

God without God: the status of the postulates

163

replacement for an existence claim regarding transcendent entities is central to

Kants intentions.12
God as a subjectively posited concept
Yet Kant has another, opposite tendency in his description and analysis of the
postulates, one that limits or negates these existence claims and instead
presents the postulates as a merely subjective cast of mind that functions for
moral purposes.13 I want to begin my discussion of this other tendency with a
revealing passage that is often overlooked14, yet which is of the highest
importance in understanding the nature of the postulates. In the penultimate
section of the Dialectic in the second Critique, On Assent from a Need of
Pure Reason, Kant tries to respond to the concern that some might have that
the postulate is a command to believe something, which he says is an absurdity
(5:144). In order to show that in fact the postulate is not commanded but
merely assented to, he makes a startling concession:
I said above that in accordance with a mere course of nature in the world happiness in
exact conformity with moral value is not to be expected and is to be held impossible,
and that therefore the possibility of the highest good on this side can be granted only on
the presupposition of a moral author of the world. I deliberately postponed the restricting of this judgment to the subjective conditions of our reason so as not to make use of it
until the manner of its assent had been determined more closely. In fact, the impossibility referred to is merely subjective, that is, our reason nds it impossible for it to

12

13

14

Patrick Kain highlights many such passages in his argument that the postulates of practical
reason require a realist interpretation of Kants ethics. He acknowledges that Kant does not
prove the existence of God to any satisfactory extent but only proves the rational indispensability of believing that they are really possible and real (Kain 2006, p. 459). Kain notes that
Kants failure to prove that such supersensible objects are real does not entail that he is
committed to anti-realism about them, nor does the possibility of skepticism. In this, he is
correct. The idealist interpretation of the postulates that I am presenting does not rest upon any
fallacy of that kind, on any move from lack of proof of God to the claim that the postulates do
not require God to exist. Instead what is important is that the actual status of the nature of belief
is given immanent rather than transcendent reference, that belief is related to the content of the
concept rather than the object, and that it is only from the practical point of view that abstracts
from ontological claims that such beliefs are considered. The arguments I present in this chapter
attest to those positive considerations that buttress a naturalist, idealist interpretation of the
postulates.
Adina Davidovich relatedly notes that the function Kant assigns to the postulates would stand
even if God were not to exist. Not Gods existence but only belief in Gods existence is required
for practical purposes. She makes this claim: [E]ven if God does not actually exist, and we
only believe in the existence of God, the practical consequences for obedience to the moral law
are the same. Therefore, a postulate of practical reason does not have to be known to be true,
and probably does not even have to be true, in order to serve its practical purpose (Davidovich
1993, pp. 201).
Guyer and Willaschek note the importance of this claim (Guyer 2000, pp. 36667) and
(Willaschek 2010, p. 192).

164

Morality beyond nature?

conceive, in the mere course of nature, a connection so exactly proportioned and so

thoroughly purposive between events occurring in the world in accordance with such
different laws, although, as with everything else in nature that is purposive, it nevertheless cannot prove that is, set forth sufciently on objective grounds the impossibility
of it in accordance with universal laws of nature. (5:145, modied)

Kant himself here weakens a crucial step in his argument for the postulate of
God. The existence of God had been postulated in order to serve as the ground
of the correspondence of happiness and virtue precisely as the means to ensure
that this correspondence is possible (5:12425). Kant now concedes that it is
possible that nature itself in accordance merely with its own natural laws
would provide the correspondence (recall that these natural laws will include
laws regarding human behavior). Objective grounds for this possibility
cannot be ruled out, although human beings reason cannot fathom how this
possibility might be conceived in any detail, hence, Kant claims, there are
subjective grounds for human reason to reject the possibility.
Still, the possibility is a live option objectively even if not subjectively for
our reason. Kant calls the existence of two possible objective explanations of
the possibility of correspondence between virtue and happiness this irresolution of speculative reason. Both options are genuine. Both allow for the
possibility of the highest good. Notice how Kant describes these two options as
equal before he invokes a deciding factor in tipping the balance toward the side
of God rather than nature:
The command to promote the highest good is based objectively (in practical reason); its
possibility in general is likewise objectively based (in theoretical reason, which has
nothing against it). But as for the way we are to represent this possibility, whether in
accordance with universal laws of nature without a wise author presiding over nature or
only on the supposition of such an author, reason cannot decide this objectively. Now a
subjective condition of reason enters into this, the only way in which it is theoretically
possible for it to think the exact harmony of the realm of nature with the realm of morals
as the condition of the possibility of the highest good, and at the same time the only way
that is conducive to morality (which is subject to an objective law of reason). (5:145)

Both options are way[s] we are to represent this possibility. Thus there are
two ways that human beings can conceive in the abstract the single possibility
of the highest good: one purely in accordance with natural laws (although
without the details that are apparently beyond our reason) and one invoking
God. Either of these two possibilities will fulll the function of explaining the
possibility of the highest good. Kant claims that human beings are constrained
to represent this function by using the postulate of God because this postulate
is the only one of the two conducive to morality and because this is the only
way theoretically possible for human beings to represent this function.
Although he does not spell this reason out, presumably Kant means to invoke
something about the way that the concept of God is subjectively

God without God: the status of the postulates

165

comprehensible to reason in a way that the highest good in accordance with

mere laws of nature is not. Theoretical reason cannot provide any details on
how nature might itself bring about the highest good. The concept of God
would thus be more conducive to morality by being more amenable to
human cognitive faculties a subjective consideration not by being more
likely by any objective standards.15
This concession by Kant near the end of his discussion of the postulates
helps to illuminate the arguments earlier in his discussion and shows a
tendency to minimize the theoretical, existential claims that appear in the
postulates in favor of a functional, purely practical relation. I will look at this
with regard to two aspects: rst, the source of the concept of God as constructed by practical reason, and second, the discussion of reference to the
subject rather than the object.
Before turning to those issues, there is Kants denial of theoretical and
synthetic knowledge of the concept, and the corresponding insistence that
the concept is entirely created by practical reason. Kant denies that we have
any theoretical knowledge of the concept and denies that we have any synthetic knowledge of the object at all, except to grant that the concept somehow
does correspond to some object. In this complicated passage, Kant takes great
pains to argue that nothing about the object of the postulate is known or even
claimed for theoretical purposes besides the mere existence of something that
somehow corresponds to the concept and Kant seems reluctant to allow even
this concession:
Since nothing further is accomplished in this by practical reason than that those
concepts are real and really have their (possible) objects, but nothing is thereby given
us by way of intuition of them (which can also not be demanded), no synthetic
proposition is possible by this reality granted them . . . Now they receive objective
reality through an apodictic practical law, as necessary conditions of the possibility of
what it commands us to make an object, that is, we are instructed by it that they have
objects, although we are not able to show how their concept refers to an object, and this
is not yet cognition of these objects; for one cannot thereby judge synthetically about
them at all or determine their application theoretically . . . There was therefore no
extension of the cognition of given supersensible objects, but there was nevertheless
an extension of theoretical reason and of its cognition with respect to the supersensible
in general, inasmuch as theoretical reason was forced to grant that there are such

15

For a discussion and detailed rejection of the arguments Kant gives that an atheist would have
insurmountable difculty with moral incentives related to the highest good, see (Denis 2003).
She suggests at one point that an atheist might in fact be able to picture a non-theistic manner
for the supreme virtue to correspond to happiness, although she does not invoke Kants
distinction between subjective and objective comprehensions of possibility that Kant mentions
at 5:145 (Denis 2003, pp. 21011). For an analysis that defends Kants claim, see (Kielkopf
1997).

166

Morality beyond nature?

objects, though it cannot determine them more closely and so cannot itself extend this
cognition of the objects. (5:13435)

Kant appears in this passage to be denying that any theoretical cognition of an

object is possible, that is, that beyond a claim that some object exists that bears
some reference relation to the concept, nothing can be known about the object.
He claims even more than this, perhaps without recognizing the import of his
claim, when he asserts that no synthetic proposition is possible, for the
existence claim itself would be a synthetic proposition. If indeed no synthetic
proposition is possible, then even the existence of an object cannot be asserted
of that concept. I think it is possible that Kant intends to allow only some
existence claim regarding the supersensible in general rather than regarding
any specic supersensible object directly corresponding to the concept.16 To
put this more clearly: Kant might mean that theoretical reason must concede
only that there is some supersensible object that can fulll the function
assigned to the practical concept, without conceding that the supersensible
object must be precisely that captured in the practical concept.
God as created by practical reason
There is another side to this coin: not only is it true that theoretical reason
cannot have any synthetic cognition about an object corresponding to the
concept in the postulate, it is also true that the concept itself stems from
practical rather than theoretical reason. This point is, like many dealt with
here, unclear in Kant. We have seen how he has claimed that the postulates are
theoretical propositions, and how they correspond to the theoretical ideas from
the Dialectic of the rst Critique. These considerations point toward the claim
that practical reason borrows the concepts for the postulates from theoretical
reason. The concepts, then, in all their detail, would be those constructed by
theoretical reason for theoretical purposes. Practical reason would only have to
locate those concepts in order to postulate the reality of objects matching them.
But this understanding is misleading. Kant argues that practical reason itself
constructs the concept of God used in the postulates.
There remains for reason only one single procedure by which to arrive at this cognition
[of God], namely, as pure reason to start from the supreme principle of its pure practical
use (inasmuch as this is always directed simply to the existence of something as a result
of reason). And then, in its unavoidable problem, namely that of the necessary direction
of the will to the highest good, there is shown not only the necessity of assuming such
an original being in relation to the possibility of this good in the world but what is
16

Allen Wood argues that Kant cannot possibly mean to deny the synthetic claim of existence
(Wood 1970, p. 148f). But Woods insistence is based on an assumption that Kant must
understand the postulates to be asserting the existence of a transcendent substance. I am arguing
that this assumption is unwarranted.

God without God: the status of the postulates

167

most remarkable something that was quite lacking in the progress of reason on the
path of nature, a precisely determined concept of this original being. (5:139)

He explains that theoretical reason, by using a posteriori teleological reasoning

about God as a cause of nature, can infer a concept only of an imperfect God,
powerful but not omnipotent, wise but not omniscient, and so on. Practical
concerns, in contrast, require a concept of God with the highest perfections;
omniscience, for example, is required in order for God to know all possible
conduct and even all possible inner dispositions of moral agents (5:140).
Practical reason itself creates the concept of God, and only after creating it
does it nd a similar concept at work in theoretical reason. Kant concludes that
the concept of God, then, is one belonging originally not to physics, that is, to
speculative reason, but to morals (5:140).
It is important that the concept of God (as well as the other postulates) stems
from practical reason. Practical reason creates the concept of God for its own
purposes. The concept fulls a certain function, namely to provide, in some
detail, an explanation of the possibility of the highest good. We have already
seen that Kant admits that, objectively, it is possible that nature itself, without
recourse to anything supersensible, could be the ground of the possibility of
the highest good. Subjectively, however, reason creates the concept of God in
order to satisfy itself that the highest good is possible. The concept is not
borrowed from theoretical reason. Not only is the function of the concept
practical, the very concept itself is practical. Theoretical reason plays no direct
role in the generation of the postulates.
God as immanent not transcendent
The way in which practical reason is autonomous with regard to the postulates
leads to the second aspect of the functional nature of the postulates, namely,
Kants claims about the immanent or practical reality for the postulates,
where he even claims that reference is not to any object but to the subject. He
claims that the postulates of God and immortality are immanent rather than
transcendent in two places in the Critique of Practical Reason:17
But is our cognition really extended in this way by pure practical reason, and is what
was transcendent for speculative reason immanent in practical reason? Certainly, but
only for practical purposes. For we thereby cognize neither the nature of our souls, nor
the intelligible world, nor the supreme being as to what they are in themselves, but have
merely unied the concepts of them in the practical concept of the highest good as the
object of our will, and have done so altogether a priori through pure reason but only by
means of the moral law, and, moreover, only in reference to it, with respect to the object
it commands. But how freedom is even possible and how this kind of causality has to be
17

Earlier in the second Critique, he held that freedom is immanent practically although
transcendent theoretically (5:105).

168

Morality beyond nature?

represented theoretically and positively is not thereby seen; that there is such a causality
is only postulated by the moral law and for the sake of it. (5:133)
Theoretical reason was forced to grant that there are such objects, though it cannot
determine them more closely and so cannot itself extend this cognition of the objects
(which have now been given to it on practical grounds and, moreover, only for practical
use); for this increment, then, pure theoretical reason, for which all those ideas are
transcendent and without objects, has to thank its practical capacity only. In this they
become immanent and constitutive inasmuch as they are grounds of the possibility of
making real the necessary object of pure practical reason (the highest good), whereas
apart from this they are transcendent and merely regulative principles of speculative
reason, which do not require it to assume a new object beyond experience but only to
bring its use in experience nearer to completeness. (5:135)

In these two passages, Kant contrasts immanent with transcendent. Further, in the rst passage, Kant also holds that the postulate is only in reference
to the moral law.
Kants use of the term immanent here can be explained in part by
reference to some of the Reections from earlier periods. In several Reections
from the 1770s and into the mid-1780s, Kant uses the term immanent to
refer to nature in space and time and thus to what can be known, and
transcendent to refer to what goes beyond it and thus to what cannot be
known (R5639, 18:276, 17781788; R6154, 18:470, 178384). At one point
in a draft of the Antinomy of Pure Reason, he contrasts immanent principles,
which restrict themselves to appearances, to transcendent principles, which
stem from the spontaneity of pure reason to give a unity beyond appearances
(R4757, 17:70304, 177577). The First Critique also ts this use of immanent when, in the Dialectic, Kant says We will call the principles whose
application stays wholly and completely within the limits of possible experience immanent (A29596/B352). The meaning of immanent that can be
drawn from these Reections and applied to the second Critique is roughly as
follows: something immanent is something that refers to appearances, that is,
to nature.
Given this meaning of immanent, Kants use of the term to describe the
postulate seems puzzling. How could the concept of God ever be something
that refers to appearances? In both passages, Kant notes that the immanence
attributed to the postulate God is in its relation to highest good as the object
of practical reason that duty commands us to try to create. Our moral duties, of
course, apply to human beings as living in nature in space and time, that is, in
appearance. The postulate of God is immanent, then, in its relation to the
actions of human beings in nature as they strive to create the highest good,
keeping in mind that the possibility of this highest good is for them subjectively represented by the concept God created specically for this purpose
by practical reason itself. I think this is what Kant means by this key sentence
from one of the passages referred to earlier: For we thereby cognize neither

God without God: the status of the postulates

169

the nature of our souls, nor the intelligible world, nor the supreme being as to
what they are in themselves, but have merely unied the concepts of them in
the practical concept of the highest good as the object of our will, and have
done so altogether a priori through pure reason but only by means of the moral
law, and, moreover, only in reference to it, with respect to the object it
commands (5:133, my emphasis). We unify the concepts of God and of
the other postulates together with the concept of the highest good in order to
complete in our minds a picture of the possibility of the latter. But this is done
not by claiming that the concept of God corresponds directly to any object, but
rather that the concept is merely functioning to sustain our hopes in the highest
good as a result of our action in accordance with the moral law, or as Kant says
in reference to the moral law. The concept of God merely functions as a
support for the moral law, and thus in reference to it, and to the immanent
moral life human beings live in nature, rather than to any object that might be
God as a transcendent being.18
One might still wonder how a mere concept can function to satisfy reasons
need to conceive of the possibility of the highest good since, of course, only an
actual God and not the mere concept would be capable of bringing about the
highest good.19 But this is the same kind of question as asking in the theoretical sphere how a concept of God could create the teleological order in the
world, the parallel situation in the Appendix to the Dialectic in the Critique of
Pure Reason. Kant there denies that we suppose the existence of God in itself
but use the concept only in relation to an explanation of the world, albeit a
necessary one. And that shows clearly that the idea of that being, like all
speculative ideas, means nothing more than that reason bids us consider every
connection in the world according to principles of a systematic unity, hence
as if they had all arisen from one single all-encompassing being (A686/B714).
I am claiming that Kants postulate of God can be understood in the same way.
Reason uses the concept as a means for approaching the world in a particular
way. In the theoretical case, the approach is in regard to systematizing our
experience; in the practical case, it concerns moral action. But in both cases,
the actual existence of a being corresponding to the concept is not posited.20
Marcus Willaschek phrases this well when he says with regard to the postulates: the practical extension of our cognition by the postulates does not
concern the existence of God, freedom, and immortality, but only the
18

19
20

Here, I will note that there is yet another strain to Kants discussion in the second Critique about
God. He says, at one point, that the postulate is not a theoretical cognition of the existence of
God but a personal desire that there be a God (5:143). This strain is discussed in detail by Drew
Pierce in (Pierce 2004) where he argues that the postulate of God should be interpreted as a
claim that we ought to believe in God.
I owe this objection to Burt Louden.
I discuss this connection in detail in Rauscher 2010.

170

Morality beyond nature?

objective reality of our concepts of God, freedom, and immortality.21 The

concepts perform the function of providing a way to picture the cause of
something in human experience without endorsing that whatever would match
the concept really exists and really operates as that cause.
After the Critique of Practical Reason
The period after the Critique of Practical Reason reveals that of these two
tendencies in Kants thought about the postulates, he nally claries his preference for the functionalist, non-ontological interpretation. Nearly all commentators22 are in agreement that in the Opus Postumum, Kant afrms that the
concept God used for practical purposes does not refer to an independent
being, God. This is his nal position, one that I have tried to show is implicit in
his prior work23. But the journey from the Critique of Practical Reason to the
Opus Postumum is also marked by some of Kants most direct discussions of
religion, in particular the Religion within the Boundaries of Mere Reason, where
Kant advocates religious belief and practice to a degree that quite clearly appears
to require a commitment to the existence of God. I will show that even in the
Religion, the tendency toward making the postulates practical functions rather
than theoretical existence claims is evident.
In the 1790s
The Religion within the Boundaries of Mere Reason at the very least contains
evidence that Kant recognizes that the concept of God is constructed by
practical reason and ought to be understood to have practical and not theoretical signicance. This footnote stresses the idea of God.

21
22

23

(Willaschek 2010, p. 191).

Here, the most famous exception is Erich Adickes, who in his book against Vaihinger declares
that Kant was always a theist and that the apparent denials of the existence of God as a postulate
in the Opus Postumum concern not the existence of God but are discussions only of the origin of
the human idea of God (Adickes 1927, p. 273f).
Patrick Kain, while admitting that in the Opus Postumum, Kant takes the position that God as a
postulate is merely an idea and not an independent substance, claims in contrast to my position
that this view both conrms and qualies Kains interpretation of Kants postulate prior to
that time in which reason leads us to believe in Gods actual existence (Kain 2005, p. 137).
Kain interprets Kant as earlier insisting on the existence of God, and discounts the elements
I have highlighted. The non-ontological passages in the Opus Postumum would conrm his
reading only if it were clear that the Opus Postumum position is new, which is not supported by
that material. There is no smoking gun in which Kant either admits or denies that his position is
different from his earlier views. In contrast, I emphasize the philosophical importance of these
elements and read Kants position in the Opus Postumum as fundamentally consistent with his
earlier position, in that I take Kant to have nally resolved the tension in his original position in
favor of the postulates as not involving existence claims.

God without God: the status of the postulates

171

So far as theoretical cognition and profession of faith are concerned, no assertoric

knowledge is required in religion (even of the existence of God), since with our lack of
insight into supersensible objects any such profession can well be hypocritically
feigned; speculatively, what is required is rather only a problematic assumption
(hypothesis) concerning the supreme cause of things, whereas with respect to the object
toward which our morally legislative reason bids us work, what is presupposed is an
assertoric faith, practical and hence free, that promises a result for the nal aim of
religion; and this faith needs only the idea of God which must occur to every morally
earnest (and therefore religious) pursuit of the good, without pretending to be able to
secure objective reality for it through theoretical cognition. Subjectively, the minimum
of cognition (it is possible that there is a God) must alone sufce for what can be made
the duty of every human being. (6:15354, ftn.)

In this passage, Kant invokes the contrast between the mere idea of God and
the existence of God corresponding to that idea. He holds that only the idea is
required for moral purposes, with one small addition, namely that the minimum of cognition that it is possible for God to exist is required. Kant is still
holding on to the claim that the idea must at least possibly correspond to
reality, although he does not require the idea to be thought actually to
correspond to reality.
Even given this minimal theoretical aspect, in the Religion within the
Boundaries of Mere Reason, the practical import of the concept of God, as
well as other related religious concepts and doctrines, is stressed over any
theoretical implications. In fact, Kant even argues that the practical effect of
some doctrines requires that their theoretical details be ignored. For
example, in a long footnote regarding theoretical consideration of an afterlife, Kant comes very close to mocking the idea of hell as punishment
(6:6971). He argues that on the one hand, eternal punishment in hell would
seem by many to be too harsh a punishment for a nite amount of evil, so
evil doers might expect easy escape through last-minute conversion and
divine mercy. But on the other hand, a nite duration of punishment in hell
would seem to some as survivable and tolerable as a tradeoff for a certain
amount of pleasure obtained by evil means in this life. After discussing this
and related matters for two pages, Kant concludes that it is better to simply
ignore these issues: if, instead of [extending it to] the constitutive principles of the cognition of supersensible objects, into which we cannot in
fact have any insight, we restricted our judgment to the regulative principles, which content themselves with only their practical use, human
wisdom would be better off in a great many respects, and there would be
no breeding of would-be knowledge of something of which we fundamentally know nothing (6:71). The point Kant is making is that the detailed
tenets of religion might not survive theoretical scrutiny, and that if their
moral effect depends upon their theoretical truth, then morality will be
harmed. Instead, Kant holds, we should use religious tenets as, in essence,

172

Morality beyond nature?

regulative principles for our moral duties, principles we hold in our minds
only to the extent necessary for them to aid our moral duties.
In the Metaphysics of Morals, Kant notes that any duties we might think we
have toward God are really duties to ourselves:
This idea [of God] proceeds entirely from our own reason and we ourselves make it,
whether for the theoretical purpose of explaining to ourselves the purposiveness in the
universe as a whole or also for the purpose of serving as the incentive in our conduct.
Hence we do not have before us, in this idea, a given being to whom we would be under
obligation; for in that case its reality would rst have to be shown (disclosed) through
experience. Rather, it is a duty of the human being to himself to apply this idea, which
presents itself unavoidably to reason, to the moral law in him, where it is of the greatest
moral fruitfulness. In this (practical) sense it can therefore be said that to have religion
is a duty of the human being to himself. (6:4434)

The idea of God, rather than a given being here is said to be of importance,
and only because the idea can be applied to the moral law (Note also that Kant
puts theoretical and practical purposes on the same footing as both holding to
the idea of God without holding to Gods reality.). Later in the book, Kant
stresses that we abstract from the existence of God when utilizing the idea of
God in philosophical morals: The formal aspect of all religion, if religion is
dened as the sum of all duties as divine commands, belongs to philosophic
morals, since this denition expresses only the relation of reason to the idea of
God which reason makes for itself; and this does not yet make a duty of
religion into a duty to God, as a being existing outside our idea, since we still
abstract from his existence (6:487). In morality, we abstract from the existence of God while nonetheless utilizing the idea of God. I think this kind of
relationship can be understood only as a functional relationship in which the
concept of God is, qua concept, playing a role in human moral life.
The functional role of the postulate of God and the consequent limitation of
this postulate to the concept qua concept, is also evident in some of the
unpublished notes and drafts Kant wrote in the 1790s prior to the Opus
Postumum. There is a reintroduction of the idea of as if. This reection from
his notes for lectures on Logic, most likely from the 1790s, can be taken as
paradigmatic of this line of thought:
One cannot obtain or prove objective reality for any theoretical idea except that of
freedom alone, and only because it is the condition of the moral law whose reality is an
axiom. The reality of the idea of God can be proved only through this and thus only
with a practical purpose, i.e. act as if there is a God, thus it can be proved for this
purpose. (R2842, 16:541)

The dating of this Reection itself shows that Kant could have held these
views as early as 1776, but certainly, he held them in the 1790s because this
point is included in Kants Logic book edited by Jsche (9:93). Certainly in the
1790s, Kant has established to his own satisfaction that the idea of God is valid

God without God: the status of the postulates

173

only for the practical purpose of defending the possibility of the highest good,
and here he expresses this position with the hypothetical act as if there is a
God, not the assertoric act because there is a God.24
This position accords with the position Kant laid out in his unpublished draft
essay What real progress has metaphysics made in Germany since the time of
Leibniz and Wolff?, written sometime in between 1793 and 1795 but not
published until after his death. He claims there that the moral arguments for
God, immortality, and freedom are not proof of their truth, nor of any objective teaching as to the reality of their objects, for in regard to the supersensible
this is impossible; instead, it is merely an injunction, subjectively and indeed
practically valid, and in this respect sufcient, so to act as though we knew that
these objects were real (20:298). Regarding whether this also involves existence claims, Kant is less clear. He adds to the as if discussion a claim that we
voluntarily grant them objective reality, but only a few paragraphs later
claims that they have reality in a subjective context but are not knowledge
of the existence of the object corresponding to this form (20:299300). He
also claims that the practical proof is thus really not a proof of Gods
existence absolutely (simpliciter) but only in a certain respect (secundum
quid), namely in relation to the nal end which the moral man has and should
have, and thus with reference merely to the rationality of assuming such a
being; whereby man is then enabled to accord inuence upon his decisions to
an idea which he frames for himself, on moral principles, exactly as if he had
drawn this idea from a given object (20:305). A proof in relation to
something, only in a certain respect but not absolutely does not sound
like a proof of anything but only a subjective expression or need, and Kants
gloss on this claim that we take the idea of God that we have framed for
ourselves and allow that idea to inuence our behavior. The idea that human
beings construct the concept of God in practical reason is further discussed in
the Progress essay. Kant notes that sometimes we forget that these ideas
have been arbitrarily framed by ourselves, and are not derived from the
objects (20:300).
This review of Kants writings in the 1790s shows that before the Opus
Postumum, Kant was making claims that were major elements of the view that
we have only an idea of God and do not also afrm a being, God. Let us now
briey assess his nal position in the Opus Postumum.

24

There is a similar comment, Practical idea as regulative principle, to act as if there were a God
and another world, in a draft headed the whole of critical philosophy that Kant wrote in
November or December 1797 (R6358, 18:683), although later in that same Reection, Kant
seems to indicate that God is to be understood in terms of a metaphysics of the supersensible
(18:685).

174

Morality beyond nature?

In the Opus Postumum

Virtually all of the passages cited from the Opus Postumum in this regard are
from the years 1800 and after, appearing in what is known as the rst
Konvolut (21: 3158). They are among the very last thoughts Kant set down
in writing. The position that Kant works out in these writings is a complex
review of the nature of his transcendental philosophy. In part, he seems to be
simply toying with some of the ideas that Fichte and Schelling had published; but in part, he appears to be genuinely attempting to restate the
transcendental idealism he had earlier worked out as a unied system of
ideas. I will not attempt to do anything more here than focus on a few
aspects of the concept of God that appear in this last stage of Kants thought.
I will show that the position that I have claimed is implicit in Kants earlier
philosophy is stated clearly in the Opus Postumum. Three aspects are
treated: rst, the nature of the concept of God as a mere idea without any
claim to correspondence to a substance; second, the practical rather than
theoretical use of this concept; and third, the creation of the concept by
reason.
Kant frequently in these notes identies God with an idea or concept. In fact,
in one note, Kant rst wrote God is a personal being, then inserted the phrase
the concept of to get God is the concept of a personal being (my emphasis
on Kants insertion), and then further adds Whether such a being exists is not
asked in transcendental philosophy (21:45).
We have already seen earlier that Kant treats the concept of God as a useful
concept without an object. As he states this position clearly in the Opus
Postumum, he adds two new elements. First, he claims that all of transcendental philosophy is merely a set of ideas without existential import:
Transcendental philosophy is the system of ideas in an absolute whole (21:80) [. . .]
System of pure philosophy. First part Transcendental philosophy. what we make as
objects for ourselves. Second part what nature makes as objects for us. (21:118)

All ideas of reason are treated the way that I have interpreted the postulate of
God. This ought not to be surprising, given the earlier discussion of theoretical
hypotheses as well as the regulative use of the ideas of reason. The idea God
is but one idea that reason has created to use merely as an idea without any
assertion about Gods existence. It is true that Kant also uses terms to describe
God that might seem to have existential import. He calls God the highest cause
(21:19), a person (21:30), and a highest being (21:33). Yet, at the same time,
he says that God is a thought object and that in philosophy, it is necessary
to derive the predicates contained in that concept analytically whether or
not there may [be] such a substance (21:32). In the context of these claims,
it is clear that Kant does not intend for these predicates to have to apply to

God without God: the status of the postulates

175

any really existing God but that they have the transcendental function as part
of an idea of reason.
Kant also identies this idea of God with pure practical reason itself. Here,
Kant is thinking about the concept of God as a lawgiver. He equates God not as
a substance but as the highest moral principle in me . . . God is moral/practical
reason legislating for itself (21:1445). By equating God with the selflegislation of reason, Kant clearly abandons any conception of God as substance and instead identies God with the functioning of pure practical reason.
Such a conception conforms to the earlier claims that only the idea of God is
necessary, and that it functions within practical reason as a conrmation of the
possibility of the highest good. To say that God is pure practical reason itself
in its moving forces is to admit that the idea of God is all that is needed, and all
that is meant, when postulating God.
The second aspect I want to show in the Opus Postumum is the moral or
practical nature of the idea of God. Obviously, the identication of God with
pure practical reasons self-legislation is strong evidence in this direction. But
Kant also stresses the moral function of the idea of God:
Difference between unconditioned and conditioned duty of practical reason. The former
has God as originator God is thus not a substance to be found outside me but merely a
moral relationship in me. (21:149)

In these notes, Kant discusses God in terms of the motivating force of the
categorical imperative. We human beings create the concept of God in order to
provide ourselves with a concrete idea of the power and authority that we
accord to the moral law.
This reason for postulating God is different from the reasons that Kant gave
in the arguments in the Critique of Practical Reason.25 God in the second
Critique is needed to serve as a representation for the possibility of the highest
good. Here in the Opus Postumum (as in some of Kants other writings and his
course lectures), God is seen as the legislator posited as commanding us to
obey the moral law.26
25

26

Eckart Frster argues that Kants focus on God not as a being but a concept is made possible by
an important change in Kants conception of the highest good. In the Opus Postumum, Kant
views the highest good not as an objective relation among the totality of beings in the world but
a subjective state of a particular being. The highest good no longer is a result to be achieved
beyond human efforts, so no independent God as substance is required. I think Frster is
reading too much into Kants lack of discussion of the highest good (Frster 2000).
There is one passage in this series that does discuss the highest good: That philosophy (doctrine
of wisdom) is called worldly wisdom in German is relevant, for wisdom, the science of
wisdom, aims at the nal end (the highest good). Now this wisdom, in the strict sense, can
only be attributed to God, and such a being must be given at the same time all power; for
without this the nal end (the highest good) would be an idea without reality; so the proposition:
there is a God is an existential proposition (21:149). I admit that this passage is evidence
against my interpretation, in that Kant claims that there is existential claim regarding God

176

Morality beyond nature?

The third aspect that I wanted to show in the Opus Postumum passages is the
way that Kant discusses the concept of God as a creation of reason. This is
perhaps the most important aspect for understanding the real intent of Kants
doctrine of the postulates. For here, Kant not only shows that the postulate of
God is a concept without reference to an object, he not only shows that this
concept functions for human beings for moral purposes, but he also shows that
it is essential to the core of Kants ethics, the autonomy of reason, that the
concept of God is one created by reason for its own purposes. Here is one of
Kants clearest statements of this claim:
Transcendental philosophy is the consciousness of the capacity of being the originator
of the system of ones ideas, in theoretical as well as in practical respect [to the right of
this sentence:]. Ideas are not mere concepts but laws of thought which the subject
prescribes to itself. Autonomy. (21:93)

Reason prescribes certain laws of thought to itself. These laws are the ideas of
reason. Kant quite clearly here identies ideas of reason not with concepts
purporting to refer to objects but instead with laws that prescribe how reason
must think. Here in the Opus Postumum, we see that autonomy is understood to
extend to every aspect of the moral, including the ideas that reason requires
to support morality, namely, the postulates. The idea of God is given by reason
to itself for its own purposes. These moral purposes are intrinsic to reason, and
thus the ideas reason creates are not arbitrary but necessary. Kant stresses this
point in a passage that comments on a denition of God:
The concept (thought) of such a being is not an ideal (ctive) but a necessary [one]
emerging from reason in the highest standpoint of transcendental philosophy.
It is not a ction (arbitrarily constructed concept, conceptus factitius) but one
necessarily given by reason (datus).
There is a God. For there is a categorical imperative of duty before which every
knee shall bend in heaven as on earth etc. and whose name is holy, without the
assumption of a substance being permitted which would represent this being for the
senses. (21:64)

Reason creates the concept of God as a necessary correlate to its practical

functions. It does not choose to create this concept any more than it chooses to
legislate the categorical imperative. The nature of the transcendental philosophy is that reason has its own framework that guides human beings in their
lives without directly providing human beings with knowledge about reality.
related to the highest good. I do not think that this passage alone overturns my claim, since it is
also at odds with virtually every other passage regarding the concept of God as a mere idea in
the Opus Postumum and it can be minimized like similar claims in the second Critique. Like
me, Frster does not think that the passage I quoted is problematic for the general claim that
Kant does not treat the concept of God as involving any existential claims (Frster 2000,
pp. 14445).

God without God: the status of the postulates

177

Among Kants last written words, then, we nd the culmination of his

doctrine of the postulates of pure practical reason in a system of transcendental
philosophy. God is not seen as a substance; rather, God is an idea created by
reason for moral purposes. This position is the realization of tendencies that
were already included in Kants earliest discussion of a postulate. He always
considered some concepts or ideas to be held only for particular purposes, and
used for those purposes without afrmation of any existence claims regarding
them. Certainly in the presentation of his moral doctrine regarding the postulates, Kant appears to make existence claims. But at the same time, he insists
on the immanent, practical use of the concept of God qua concept.
Realism and God; idealism and God
At the beginning of this book, I noted that my purpose is to show that Kants
ethics is compatible with a metaphysical naturalism. By metaphysical naturalism, I mean the view that nature in space and time, understood by Kant to
range over inner as well as outer sense, provides our ontology. Philosophy has
a different methodology than the sciences, but it remains within their ontology.
In this chapter, I have shown that Kant had a naturalistic metaphysics available
to him for his doctrine of the postulates of pure practical reason, which on the
face of them appear to demand a non-natural metaphysics. But this interpretation of the postulates also shows the fault lines between moral realism and
moral idealism in Kant.
The role of God in the discussion of Kants moral realism or idealism is
generally overlooked.27 There is discussion of God and Kants theory of
religion among Kant interpreters and philosophers of religion and theologians, of course, but most of the discussion of Kants moral realism or idealism
centers on aspects of Kants theory that contemporary philosophers accept
as still operative in the contemporary debate in moral theory, and for most
participants, religious claims related to morality are not part of that conversation. For some readers, then, the issue would appear unimportant.
It is not. By showing that the postulate of God is plausibly understood in a
non-ontological way, I have shown how Kants moral idealism incorporates
properties and objects that would appear to require an independent existence
outside the moral agent. Chapter 4 on reason showed that the key faculty for
morality, pure practical reason, denes the transcendental conception of the
27

Patrick Kain is a notable exception. Kain shows how Kant regarded God as a legislator, not an
author (creator) of the moral law in (Kain 2004) and how Kant took human beings to be
required to believe in God in order to represent God as the author of the divine commands (Kain
2005). John Hare, who presents a modied divine command theory based on the weakness of
human nature, is another exception (Hare 1996) and (Hare 2001). Both Kain and Hare offer
realist accounts.

178

Morality beyond nature?

moral agent. This aspect of morality is, so to speak, within the moral agent.
Chapter 6 will continue that topic with a look at the freedom of choice of a
moral agent. Moral agency also entails a relation to things that appear to some
to be outside the moral agent herself: in particular, God in the present case and
the value of humanity in other moral agents. It is natural to assume a realist
stance with regard to these things, since God as an object is independent of the
moral agent, and the value of humanity would be a property of other moral
agents and so independent of the moral agent as subject.
When I dened moral realism and moral idealism, I distinguished between
the principles, properties, and objects required for morality and moral principles, properties, and objects. Since God could play a role in non-moral areas
such as explaining the creation of the universe or the ground of truth, God
would be an object required for morality. That is to say that the mere existence
of God does not entail moral realism because even were there no such thing as
morality at all, God would still exist. However, this assessment is incomplete
because the existence of God necessarily includes the existence of justice and
good, and other moral principles and properties as part of the nature of God.
These would be, in my classication, moral properties of the object God and
their existence independent of the moral agent as subject would entail moral
realism.
God as an existing being would be transcendentally real as independent of
the transcendental moral agent. That is, if the postulate about God entails that
God is a really existing being, then God would have to be independent of the
moral agent even if the existence of God is considered a transcendental
condition for morality. Even given my previous chapters argument that the
moral law stems from reason which in turn can be understood only as the
existing reason of particular moral agents considered to have a certain structure, the reality of God would still be needed. Since God is also considered to
be rational, Gods reason would also contain the structure that produces the
moral law and thus, the moral laws existence would not be dependent simply
upon the (nite) moral agent.
The argument I have given in this chapter provides an idealist understanding
of the postulate in which the postulates do not require anything independent of
the moral agent. This interpretation understands the postulate as about the
concept God rather than the being, God. In order to make the moral demand
of the fact of reason coherent with other aspects of the agents self-conception
as a happiness-seeking being, a moral agent is required to believe that the
highest good is possible. In order to picture how the highest good is possible, a
moral agent creates the concept God as, as it were, the placeholder for that
which makes the highest good possible, recognizing however that the highest
good might be possible in accordance with nature alone. This concept is not
intended to refer to any actually existing being. Rather, the moral agent uses

God without God: the status of the postulates

179

the concept only immanently as part of the practical purpose of determining

the agents free actions.
This idealism holds both transcendentally and empirically. The transcendental conception of the moral agent includes the set of concepts that such a moral
agent would employ as part of the determination of the agents actions. As a
concept alone, God would not exist independent of the transcendental moral
agent and no actual God is required. Empirically, the situation would also hold
and no actual God is required. Note that in neither of these cases am I denying
that God exists. The actual existence or non-existence of God is outside the
scope of the postulates, which instead must be understood merely in terms of
the concepts employed.
Concluding remarks
The explanation of the postulates that I have advanced in this chapter supports
both a metaphysically naturalist and a morally idealist interpretation of Kants
ethics. I have traced the role of postulates, hypotheses, and ideas in Kants
philosophy from the pre-critical works through the Opus Postumum. He
always had a conception of an immanent use of certain kinds of concepts that
would purport to have non-natural objects. The immanent use of ideas of
reason and transcendental hypotheses in the Critique of Pure Reason provide
the pattern that, I claim, lies behind his discussion of the postulates in the
Critique of Practical Reason. The text contains passages that appear to require
that moral agents must hold that God exists, but other passages stress that the
only reference for the postulate is immanent rather than transcendent. Certainly, the uses Kant later makes of the idea of God in his moral theory
conform to the stress on immanent use and lack of transcendent reference.
The Opus Postumum most clearly includes the kind of view I take to have been
at least implicit in the rest of his corpus. The argument I have given to show
that Kant does not require an actual God as referent for the idea would hold as
well for claims to the immortality of the soul. Chapter 6 will tackle the more
complicated issue of how to understand Kants insistence on freedom of choice
(or will) in a metaphysically naturalistic and morally idealistic way.

From many to one to none: non-natural free choice

This chapter will assess Kants metaphysics of choice in order to see the extent
to which what is usually referred to as freedom of the will is compatible with
a metaphysical naturalism. The previous two chapters already showed that the
faculty of reason and the postulates can be interpreted within a metaphysically
naturalistic ontology; the following chapter will do the same for moral value.
My approach to freedom of the will as choice mirrors my approach to
reason. Regarding reason, I showed that there is an empirical faculty of reason
in human beings in nature, one studied by the science of empirical psychology.
I then considered Kants transcendental arguments that pure practical reason
must be transcendentally free in order to legislate the pure moral law without
the inuence or determination from any causal sources in nature. I argued that
for this requirement, Kant does not need to postulate that pure practical reason
exists as a transcendent thing in itself (or a property of a thing in itself)
ontologically distinct from nature; rather I showed that Kants transcendental
method allows him to identify pure practical reason with the structure of reason
determined philosophically but existing within nature. The transcendental
freedom of reason is not a nonnatural freedom. This current chapter asks the
questions: is there a counterpart to the distinction between pure and empirical
reason in Kants conception of the power of choice? To what extent would a
nonnatural power of choice be needed for ethics? Is there a way to understand
Kants views on freedom of the will without recourse to nonnatural objects or
abilities?
This chapter has two goals. First, abstracting from my naturalist interpretation, I want to show that Kants conception of free choice is not as metaphysically demanding as many assume. I call this part of my argument many to
one because it shows that the metaphysics of free choice requires only one
instance of a nonnatural free act that explains each and every empirical
instance of an agent making a moral decision, instead of many particular acts,
each corresponding to a particular empirical decision. I will show that even
when Kant is insistent that the freedom of choice of the moral agent be free in a
nonnatural way, he requires only one nonnatural free act rather than a multiplicity. Individual choices of the empirical power of choice in nature are
180

From many to one to none: non-natural free choice

181

deemed free not as distinct occurrences of a nonnatural causality but in

collective reference to only one timeless free choice. The single timeless free
choice is nonnatural and would explain how all of the particular free choices in
nature, themselves determined in accordance with natural causality, can be
deemed free in a transcendental sense.
The second goal is to reduce from one to none the number of actual
nonnaturally free acts required by Kants ethics, thereby providing a completely metaphysically naturalist interpretation of free choice. I do this by
invoking my interpretation of the postulates from Chapter 5.
Although for my overall interpretive project, only the latter goal is necessary, the rst goal plays an important role in understanding the nature of
freedom in Kant. It shows that Kants moral theory does not have extensive
metaphysical commitments precisely in the area most often thought to require
commitment to an ontological interpretation of transcendental idealism, or
even a nonontological incompatibilist conception of moral agency in which
we conceive of every particular act as independently free. To that extent my
rst goal does contribute to my project.
This chapter will rst lay out the land regarding the power of choice. I will
defend the power of choice (Willkr) as a focus for the discussion of free will,
for there are passages in which Kant uses the term Wille for will understood
as a power of choosing, or at least beginning, actions. Further, both will and
choice must be compared to reason, both pure and empirical. I will show that
Kant has a compatibilist conception of freedom in nature which counts the
power of choice as free when it is determined by concepts of reason.
I then look at Kants reasons for thinking that a transcendentally free choice
is needed. These two reasons concern the relation between ought and can, and
the conditions for moral responsibility. I argue that in both, only one free act is
needed rather than many. I then discuss the resulting metaphysical requirements for a free power of choice and their naturalist interpretation. I will take
as central to his position the picture found in the Critique of Practical Reason
and further explained in the Religion within the Boundaries of Mere Reason
that there is only one free choice independent of nature to show how this one
nonnatural act grounds the freedom of each empirical act in nature.
The nal, and much shorter, section will argue that even this one nonnatural
free choice can be reduced to zero when it is understood in terms of the merely
heuristic role of the postulates. This argument is not as simple as it may seem
because Kant in fact rarely identies freedom as a postulate and clearly gives it
a different status than that of immortality and God. Freedom is said to be a
prerequisite for the moral law, whereas immortality and God are prerequisites
for the highest good and are not directly required for action from duty. I argue
that by carefully distinguishing the priority of action from that of belief,
freedom can also be understood as heuristic like the other two postulates.

182

Morality beyond nature?

Will and choice in nature

In Chapter 4 on reason I briey discussed the difference between the faculty of
practical reason (Wille) and of choice or decision (Willkr). Here I will review
that distinction and apply it more particularly to choice. I will also explain the
difference between what I call the many and the one in enumerating free
acts of choice.
Even with the distinction between will and choice at hand in the English
translations and corresponding distinct terms stemming from the Latin voluntas and arbitrium in the Romance language translations, there is still some
philosophical confusion because Kant often appears to assign the same function, making decisions which begin the process of action, to both faculties.
This is indicated in part by noting that he sometimes uses just one or the other
of these terms for sustained discussion of freedom in morality. For example, on
the one hand, the Third Antinomy uses the term Wille neither in the exposition of the thesis and antithesis nor in the solution to the Third Antinomy but
instead uses Willkr almost a dozen times to indicate an ability to choose.1
On the other hand, the Groundwork includes only two occurrences of Willkr (4:428 and 4:4512) while using Wille throughout to indicate the human
ability to make decisions. One would have expected both terms to be used in
roughly the same proportion in these discussions of the human moral decisionmaking framework had Kant a rm distinction between them in mind. There
are also instances where Kant denes both terms similarly. Wille is dened in
the Groundwork as a capacity to act either according to principles or subjective conditions (certain incentives) (4:41213) just as the Willkr is dened as
affected by moving causes of sensibility (A534/B562), and in both cases the
faculty in question is said to be objectively necessitated by the moral law and
subjectively contingent in its action. The human Wille is contrasted with a pure
Wille that always follows the moral law while the human Willkr is contrasted
with an animal Willkr that always follows natural law, but in both cases
whichever faculty is at issue is squarely placed in the middle as a faculty of
decision affected by both the moral law and sensible incentives yet necessitated by neither uniformly.

In Kants notation in his own copy of the A edition of the rst Critique he does mention the Wille
twice in the margins of the solution to the Third Antinomy, rst saying that the will is the faculty
of desire in a being with reason and that it is a pure will when pure reason has causality (23:41, at
A538/B566) and second in reference to the specic phrase reason has causality remarking i.e.
is the causality of its objects. This causality is called the Wille. But in transcendental philosophy
one abstracts from the Wille (23:50, at A547/B575).
And of these two, one of them (4:451) does not even appear to mean Willkr as faculty of
choice but is more closely related to willkrlich, or voluntary: Vorstellungen, die uns ohne
unsere Willkr kommen (representations which come to us involuntarily)

From many to one to none: non-natural free choice

183

While Kants own use of the terms in the 1780s does not consistently allow
for a distinction between the two terms, commentators generally agree that
Kant appears to have a rm distinction between Wille and Willkr in mind in
1797s Metaphysics of Morals.3 There he denes and applies his terminology in
the preface to dene Willkr as a faculty to do or refrain from doing as one
pleases . . . [i]nsofar as it is joined with ones consciousness of the ability to
bring about its object by ones action. Wille is dened as the faculty of desire
whose inner determining ground, hence even what pleases it, lies within the
subjects reason (6:213).
In the end, then, Kant distinguished Wille and Willkr in terms which equate
Wille with practical reason as the source of the moral law one possible
ground determining choice to action and Willkr with latitude of decisionmaking in doing or refraining as one pleases the actual determination of
action. I will not take up the details here of exactly how this distinction is
foreshadowed in Kants earlier writings, nor will I attempt to further disentangle Kants terminology. Here I assume that this distinction captures the basic
structure of Kants ethics in which Wille as reason provides a moral law for
Willkr as the power of decision. For the remainder of this paper, I will use the
English terms will, power of choice or simply choice, and reason in
my discussion to translate Wille, Willkr, and Vernunft, respectively,
and note when Kants use of Wille appears to refer to reason instead of
choice when it is otherwise not clear.
To the extent that Wille is free, it is so as reason, the freedom of which has
been shown to concern its structure independent of natural causes. In the
Metaphysics of Morals Kant says that Wille is neither free nor unfree. By this
he does not retract the picture of reason as independent of determination by
laws of nature but rather emphasizes that there is no determining ground for
reason at all (6:213) since it is merely concerned with providing law as
practical reason (6:226). This denial of freedom does have consequences for
understanding autonomy, as I discussed in Chapter 4 on reason, because the

Allison reviews the distinction in (Allison 1990, pp. 12936) but does not emphasize the dual
ontology for Willkr that gives it an empirical role in nature and a nonnatural transcendentally
free existence independent of nature. This approach forces him to gloss many passages in which
Kant talks about the empirical Willkr. For example, forgoing his normal practice of leaving the
term Willkr untranslated, he even draws attention away from Kants admission that there is a
distinct phenomenal Willkr when he quotes from Kants Metaphysics of Morals: Some have
tried to dene freedom of Willkr as the power to choose between the alternatives of acting with
or against the law (libertas indiffereneiae). But freedom of Willkr cannot be dened in this way,
although choice [Willkr!-FR] as phenomenon gives frequent examples of this in experience.
(Allison 1990, p. 133, quoting 6:226). Allison also rejects Kants empirical psychological
account of action determined in accordance with natural causality as studied by empirical
psychology, claiming that Kant rejects freedom of the empirical character. Practical freedom is
instead, for Allison, not to be understood in terms of natural causality at all.

184

Morality beyond nature?

law is something that simply stems from reason and is not freely chosen;
autonomy cannot be understood in terms of any kind of choice.
Freedom of the power of choice, the kind of freedom related to decisionmaking, is what must be analyzed here. Kant places human freedom of choice
at least in part in nature, referring to the human ability to act as beings in space
and time in a way similar to animals but not necessitated in the manner that
animals are. In order to nd out what Kant requires of freedom outside of
nature, we must rst see what he allows as (one form of) practical freedom
within nature.
One of the best known descriptions of this practical freedom is in the
solution to the Third Antinomy:
It is especially noteworthy that it is this transcendental idea of freedom on which the
practical concept of freedom is grounded, and the former constitutes the real moment of
the difculties in the latter, which have long surrounded the question of its possibility.
Freedom in the practical sense is the independence of the power of choice from
necessitation by impulses of sensibility. For a power of choice is sensible insofar as it
is pathologically affected (through moving causes of sensibility); it is called an animal
power of choice (arbitrium brutum) if it can be pathologically necessitated. The human
power of choice is indeed an arbitrium sensitivum, yet not brutum but liberum, because
sensibility does not render its action necessary, but in the human being there is a faculty
of determining oneself from oneself, independently of necessitation by sensible
impulses. (A5334/B5612)

Here the human power of choice (Willkr) is described as pathologically

affected through moving causes of sensibility but not pathologically necessitated by these moving causes as animals are. In order to be pathologically
affected, the power of choice must be empirical. Practical freedom, as opposed
to transcendental freedom, is a function of the empirical character. The denition of practical freedom hinges upon the ability humans have to avoid
necessitation by sensuous impulses by acting according to motives of reason;
it crucially does not exclude necessitation according to empirical rational
motives. Humans are practically free by virtue of a faculty that allows action
outside of sensible necessitation.
What is important for this conception of practical freedom is whether, in
saying that the power of choice must be independent of necessitation by
sensible impulses, Kant is identifying all determination according to laws of
nature, including psychological laws, as necessitation by sensible impulses.
I showed in my earlier chapters that he has available to him a robust
empirical psychology that includes an empirical reason that provides principles that can determine the decisions of the empirical power of choice.
When Kant is conceiving the power of choice as independent of determination by sensible impulses he allows it to be determined by inner motives or
principles of reason or representations of laws and still to be considered free

From many to one to none: non-natural free choice

185

in a practical sense. Take this passage from the 1792 Dohna metaphysics
lectures. Kant denes, as is typical, stimuli with sensitive impulses and
motives with intellectual.
We are affected by stimuli but not determined. Whoever is determined by motives is
free, for he acts according to the laws of his own reason according to spontaneity and
not according to receptivity. Will is the faculty of desire insofar as it is affected by
representation of a rule . . . The free power of choice is called pure if it is determined
merely by the representation of the law, by the form of conformity to the law. . . The
capacity for determining oneself through pure representations of reason is freedom.
(28:677, modied)

Here representations of a rule are taken to determine the free power of choice,
which is free when determined by motives. Representations, of course, can
occur only in the mind in inner sense. Thus, the empirical power of choice is
affected by stimuli and motives. To the extent it is determined by the motives,
it is free. For similar claims, see 29:89899 (Mrongovius, 1782), R1021
(15:45758, 177379) and R1054 (15:470, 178588).
The power of choice is, on this picture, a natural faculty. It exists in
human beings as appearance. The determination of the power of choice thus
occurs within the causal nexus of nature. But all of this still means that
human beings are determined in accordance with laws of nature as
described in empirical psychology. Kant calls this freedom because of the
role of principles and ideas. Kant uses the term freedom to refer to the
ability human beings have to act in accordance with principles and ideas as
opposed to instinct and sensible impulses. In a Reection from the mid1770s Kant even uses the term blind (blinde) to refer to the determination of the power of choice by sensible impulses: Power of choice is
either blind or free power of choice (brutum oder liberum). . . The blind
power of choice is that which is necessitated and determined through
impulses (stimulos) (R1028, 15:460). This word blinde is the same word
Kant uses in the rst Critique at A51/B75 when he says Thoughts without
content are empty, intuitions without concepts are blind. And just as Kant
distinguished animal from human power of choice, he also distinguished
animal from human cognitive faculties. Animals do not have reason or
understanding but do have the ability to connect representations of outer
sense they have no inner sense according to laws of sensibility as
Kant says in his lectures on metaphysics (28:276, see also 28:690 and
29:879). The link is this: animals do not have concepts or principles,
humans do, and these concepts or principles make knowledge and rationally
determined action possible. If human beings but not animals can act
according to concepts or principles as representations of laws, then the
human power of choice is the one that decides upon actions based upon
principles rather than directly based upon mere sensible impulses of nature.

186

Morality beyond nature?

The mere sensible impulses of nature are outer causes (R6931, 19:209) or
instincts (R7220, 19:289, 1780s) or stimulus (Powalski, 27:123) or
inclinations (R6931 19:209, 177678). But human beings are capable of
guiding their behavior according to principles rather than these mere animal
impulses.4 Decisions based upon principles are deemed free because they
stem from empirical reason, even though empirical reason is within the
causal nexus of nature. That same reason considered as a structure independent of natural causal determination is transcendentally free. Even if
Kant ultimately believes that practical freedom is insufcient for morality,
he has still at least provided a picture of human action, which completely
incorporates it into nature. This view is sufcient for the practical perspective of the moral agent (A803/B831). Moral agents are able to view their
decisions in general as determinable by reason rather than only by sensible
impulses. For mere practical purposes of determining oneself to action in
reference to a moral law, this compatibilist conception of freedom is sufcient by itself.
Kant, of course, did not believe that this sort of compatibilist freedom is
sufcient for a complete explanation of morality. He famously dismissed
Leibnizs related notion as the freedom of a turnspit (5:97). The rejection
of this compatibilist, naturalist freedom of choice must have grounds other
than that of identifying and justifying right acts. The following section will
provide those grounds and ascertain whether these grounds require there to be
many or only one nonnaturally free act. I will also examine the metaphysics of
his deeper conception of freedom for the power of decision of the faculty of
choice.
First, a word about the enumeration of the free decisions of the power of
choice. When I claim that I want to reduce the number of nonnatural free
decisions from many to one, I mean to refer only the transcendentally free
acts of the power of choice. The freedom of reason has already been
accounted for. The practical freedom of empirical choice has also already
been accounted for in a compatibilist, naturalist way. The remaining question for my naturalist interpretation is simply what the nonnatural free
decisions of the power of choice would be and how many there are. The
transcendentally free decision would be, like reason, timeless. Exactly how
that timeless choice relates to particular choices in time is the issue. Some
commentators claim that every individual decision of the power of choice is
4

My account of the empirical power of choice is relatively short and only a sketch. A more
extensive explanation of Kants empirical psychology in its relation to the human power of
choice in nature is Patrick Frierson, Kants Empirical Account of Human Action (Frierson
2005). Frierson emphasizes Kants determinism through the role of causal laws that can be
attributed to individual empirical character. See also his book (Frierson 2003) for discussion of
the empirical elements of Kants discussion of motivation.

From many to one to none: non-natural free choice

187

transcendentally free of itself, resulting in a multiplicity of nonnaturally free

acts. At the extreme this amounts to a denial of a genuinely determined
decision of the empirical power of choice in nature in favor of a view that
natural causality is to be understood in terms of an epistemic limitation, so
all decisions of the power of choice are free in a nonnatural sense.5 The
more moderate version would identify particular decisions of the empirical
power of choice determined in accordance with natural causality in some
way with particular transcendentally free decisions.6 Others have the same
picture that I am advocating of a single transcendentally free decision.7 This
single transcendentally free decision would take place independent of nature
but would lie at the base of all particular naturally determined decisions of
the empirical power of choice. This relation is possible because the timeless
decision ontologically precedes all the particular empirical decisions in time
in nature. This strongly antinaturalistic ontology is the aim of my many to
one argument. Later I will try to turn that one into none by showing how
even that kind of strong picture can be naturalized.
From many to one
In this section I will argue for my claim that only one transcendent free choice
is required. In a sense, this section provides the content of the postulate of
freedom in relation to free choice. I will rst look at the two reasons Kant gives
for insisting on free choice in morality ought implies can and moral responsibility and then examine the metaphysics of just one free act that Kant uses to
explain them both.
Many to one: ought implies can
One reason Kant is often taken to insist upon nonnatural freedom of decision in
the power of choice is that he holds that ought implies can, understood as the
claim that if human beings are morally obligated to do something, then they
5

Henry Allison is the most famous proponent of this view. I have discussed his view elsewhere in
this book and will do so later in this chapter. Regarding timelessness, he says that strictly
speaking the transcendentally free decision is not timeless but merely independent of the
conditions of time, meaning independent of causal determinism in nature (Allison 1990,
pp. 513).
Hud Hudson offers such a view that interprets Kant as offering a token-token identity between
transcendentally free acts and the corresponding natural occurrences of these acts (Hudson
1994). Ralf Meerbotes invocation of anomalous monism functions in a similar way (Meerbote
1984).
Allen Wood accepts the single, timeless decision as the basis of an intelligible character, which is
the basis of the empirical character exhibited in each particular action (Wood 1984). My account
in this chapter adopts his general approach.

188

Morality beyond nature?

have the ability to do it despite any natural causal determination that would
appear to make the action impossible.8 Kant states the basic claim that ought
implies can in several places.9 In the Critique of Practical Reason Kant
offers the example of a man threatened with execution unless he falsely
testies against an innocent person. Kant asks whether he would resist this
injustice. He judges, therefore, that he can do something because he is aware
that he ought to do it and cognizes freedom within him, which, without the
moral law, would have remained unknown to him (5:30). It appears that the
range of free actions must be quite wide if every action we ought to do is an
action we must be free to do. Consider as an extreme case a drug addict who
ought to stop ruining humanity in her own person by quitting the habit:
ought implies can would indicate that despite the irresistible urges to
continue the drug, she is free at all times to refuse to take it because she
has a power to overrule the causal determinism in nature that otherwise
would cause her to continue the drug. One could apply this to the agent
independently on each occasion on which she succumbs to her addiction,
because on each occasion her empirical Willkr is determined by the sensible
impulses caused by the addiction rather than by reason as it should be. In
order for the agent to be obligated to act against the addictive impulses, she
must be able to act against those impulses despite their necessary determination of her actual decision.
The consequences for Kants theory of freedom, however, are not as severe
as they rst appear. There are two different ways to understand the freedom
associated with can in ought implies can. First, one can hold that for every
particular action that one ought to perform, one actually has the separate and
individual freedom to perform that action in spite of all other natural causation
opposing it, including other acts and the character of the same agent. Even

Robert Stern has recently argued convincingly against a claim that ought implies can in
general and particularly that Kant has a different meaning (Stern 2004). What Stern calls the
strong interpretation of ought implies can gives priority to the can in claiming that the
ought should be restricted on the basis of the actual abilities of agents (whether particular or
considered in general). This strong interpretation would thus make the can controlling
over the ought. For example, if a study of human nature were to show that it is impossible
for an individual to be able to calculate or act upon a utilitarian standard, then no such
standard could be obligatory for human beings. Instead of this interpretation of ought
implies can, Kant is shown to hold to the interpretation that prioritizes the ought such
that the moral law is determined a priori and then applied to human beings. Whatever the
moral law dictates is understood to be within the capacities of human moral agents. This is
the proper interpretation of Kants claims, and it shows that ought implies can implies a
nonnatural ability to act in accordance with moral law regardless of the limitations of natural
causality.
Stern collects eleven passages from Kants published works in (Stern 2004, pp. 535) but does
not include the passage I cite here (he makes no claim to comprehensiveness).

From many to one to none: non-natural free choice

189

given the causal determinants in nature as it is in time before and at the

moment of decision, the agent is still able to do what she ought to at that
moment. I will call this the particular interpretation. Second, one can hold
that each person is free to have made their entire set of actions (along with the
related natural causation) different, thus changing not only the particular action
but all associated natural causal antecedents, their other acts, and their character.10 I will call this the general interpretation. I will show that Kant intends
the general rather than the particular interpretation, and later argue that this
interpretation ts with his other less expansive requirements for nonnatural
freedom.
One can understand this in terms of the individuals causal efcacy in nature
based upon character. The individuals empirical character causes certain
actions in nature. Particular actions determined by the empirical character that
violate the moral law are subject to the ought implies can principle, so the
agent must have the ability to refrain from those particular actions even though
they are determined to occur in accordance with that agents empirical character. The particular interpretation would claim that the agent has the current
and operative ability to refrain from such an action despite its determination by
the empirical character. The general interpretation would claim that the agent
has the ability to have had a different empirical character such that the
alternate empirical character would have refrained from that action. There
would be a single nonnatural decision that results in the entire empirical
character of an agent throughout her lifetime. This empirical character then
operates as part of the laws of nature to determine each particular action, but in
relation to the single nonnatural decision is understood as contingent and thus
within the agents control.
Before turning to an assessment of Kants references in the published works,
it is useful to examine a very clear presentation of this view in an exchange of
letters Kant had with Jacob Sigismund Beck. The question of whether and how
ought implies can is raised in a letter Beck wrote to Kant on May 31, 1792:
I would feel myself freed from a burden if you would kindly show me the emptiness of
this question: Cant one imagine the moral law commanding something that might
contradict its typus? In other words, cant there be activities that would be inconsistent
with a natural order but that nevertheless are prescribed by the moral law? It is a merely
problematical thought, but it has this truth as its basis: the strict necessity of the
categorical imperative is in no way dependent on the possibility of the existence of a
natural order. Yet it would be a mistake to account for the agreement of the two as
accidental. (11:340)

10

This should not be overblown to make the agent responsible for all the past causal antecedents
of her actions, which given Kants strict necessity in natural causal law would be vast. Wood
offers the correct response to objections of that sort (Wood 1984, p. 92).

190

Morality beyond nature?

Beck is asking Kant whether in fact all actions that ought to be performed, that
are commanded by the moral law, can be performed, that is, are compatible
with the natural order.
Kant replied to this question about a month later:
As for the question, Can there not be actions incompatible with the existence of a
natural order but which are yet prescribed by the moral law? I answer, Certainly! If you
mean, a denite order of nature, for example, that of the present world. A courtier, for
instance, must recognize it as a duty always to be truthful, though he would not remain a
courtier for long if he did. But there is in that typus only the form of a natural order in
general, that is, the compatibility of actions as events in accord with moral laws, and as
in accord too with natural laws, but only as regards their generality, for this in no way
concerns the special laws of any particular nature. (11:348, July 3, 1792)

Kants reply is to suggest that the can involved in ought implies can is a
general ability, that is, that such actions are not incompatible with the concept
of natural laws in general. The implication is that specic circumstances in
which those natural laws apply, and even the specic natures of particular
substances within a natural order, may actually contradict the requirements of
the moral law, even if that moral law is compatible with natural laws in
general. The example he gives of the honest courtier illustrates the point.
A courtier, which we might say is the natural concept of a permanent atterer
to royalty, is subject to the moral command not to lie. But the natural order in
which the courtier exists is one which requires lying as part of the job of
attering royalty. Without lying, there could be no courtiers. The particular
natural order, which includes courtiers, then, is inconsistent with the moral
ought. Ought in this case does not imply an empirical can, for by denition a
courtier cannot always tell the truth. There can be no honest courtiers. Nonetheless the courtier ought to be honest.
One might object that the courtier, qua courtier, may be unable consistently
to tell the truth, but qua human being is certainly capable of telling the truth.
But that would be to misunderstand Kants point. His point is that certain
particular natural laws may be inconsistent, and thus make impossible, certain
moral commands within that particular natural order. If the natural order is
changed, in this case, if the profession of the courtier is disallowed or abandoned, then of course the command may be followed. Given that allowance, it
might be incumbent upon the individual to resign his courtiership, and even to
attempt to change the particular natural order that includes courtiers, but these
presuppose that the person is being understood not qua courtier but qua human
being. There is a difference between the claim Courtiers cannot always tell
the truth, although they ought to and Human beings employed as courtiers
cannot always tell the truth, although they ought to. The rst is true because
the individual in question would cease being a courtier by denition, the
second false false, at least, on the assumption that no other special law of

From many to one to none: non-natural free choice

191

the particular natural order prevents it. However, in both cases the person as
courtier is still subject to the moral law and is still obligated to tell the truth,
because that moral command, although inconsistent with the particular natural
order, is consistent with natural order in general.
The case of the honest courtier can be generalized to include the special laws
of any particular natural order, and this is what makes Kants discussion here
so important. Moral obligation must be in accord with the general laws of
nature but need not be in accord with the particular natural order, or particular
natural laws, or particular natures of particular substances. The categorical
imperative determines what ought to be the case, which actions ought to be
performed, not by basing its assessment on the particular natural order of
things but on the general idea of consistency with laws of natural order in
general. Only then does this picture of how things ought to be get applied to
the particular natural order experienced by human beings. The can of ought
implies can here is the general can of an entire order of nature rather than a
particular can of individual actions.
This more general sense of order is used in the Critique of Pure Reason
when Kant discusses the malicious liar in the solution to the Third Antinomy.
Here he discusses the alternative in which the liar refrains from lying in terms
of an entirely different empirical character caused by a different intelligible
character (although Kant refers to this in terms of reasons intelligible
character):
[One can ask] only: Why has it [reason] not determined appearances otherwise
through its causality? But no answer to this is possible. For another intelligible
character would have given another empirical one; and if we say that regardless of
the entire course of life he has led up to that point, the agent could still have refrained
from the lie, then this signies only that it stands immediately under the power of
reason, and in its causality reason is not subject to any conditions of appearance or of
the temporal series; the difference in time might be a chief difference in appearances
respecting their relations to one another, since these are not things in themselves and
hence not causes in themselves, but it makes no difference to action in its relation to
reason. (A556/B584)

To say that the liar could have refrained from lying is to say that if reason
had had a different intelligible character, that would have determined the
particular action in nature differently through a different empirical character.
The empirical character as a whole is seen as the result of a timeless
intelligible character, and any change in actions from the natural order would
have required a different natural order constituted in part by a different
empirical character, which could result only from a different timeless intelligible character.
To be sure there are passages in which Kant suggests that ought implies
can holds for each particular action. In the Metaphysics of Morals, for

192

Morality beyond nature?

example, he says of a moral agent: he must judge that he is capable of

resisting and conquering by reason not at some time in the future but at once
(the moment he thinks of duty): he must judge that he can do what the law
tells him unconditionally that he ought to do (6:380). Kants use of the phrase
at once indicates application to the action at hand, and extends to every particular action when it is at hand. And a similar reference to a particular action is
given in the Vigilantius lecture on metaphysics of morals: For no man is in
position to determine in advance whether, in casu dato [in the given case], he
will despise all physical evil and absolutely speak the truth; he knows only that
he ought to obey the categorical imperative; he must therefore also be able to,
and for this a ground must be present, not an immediate consciousness.
(27:503). The Latin phrase in casu dato refers to the specic particular
action in question. In both of these passages Kant insists that in the particular
case of action facing the agent, the agent ought to obey the moral law and thus
that he can do that.
Yet even faced with these passages referring to the particular action, the
general sense of can Kant uses in the passages previously mentioned is
applicable. The moral agent Kant describes must be aware that it is possible for
him to follow the moral law in the action in question. If this particular course
of nature is such that he is determined within it to violate the moral law instead,
then in one particular understanding of the term, he cannot do so. Yet it
remains possible that the particular course of nature could have been different
consistent with the general laws of a natural order. In that more general sense,
the agent could have performed the action. (Still, one wants to claim that the
agent must be responsible for the action; responsibility and imputation will be
treated in more detail in the next section, here the focus is on the type of
freedom needed for ought implies can.) One wants to claim that the agent
himself has the causal power to perform the action. If to say that the agent
can perform the action is just to say that the particular course of nature were
different consistent with the general laws of a natural order, then there must be
some sense in which the agent has the causal power to determine the particular
course of nature or even alter it so that the agent performs the action. Only then
is the agent said to have the required causal power with the ability to perform
the action he ought to perform.
For this, a nonnatural ability to determine the particular order of nature
within the general restrictions on natural order is required. The principle
ought implies can has been shown to imply not an ability to alter the
particular course of nature as it is, has developed, and will causally determine
a particular act, but instead an ability to have chosen a different course of
nature so that the particular course of nature will causally determine a particular act. This is the general interpretation rather than the particular interpretation of the nonnatural ability to freely cause actions. Somehow, then, it

From many to one to none: non-natural free choice

193

must be within the individuals power to determine her empirical character so

that this empirical character determines the actions that the individual ought to
perform.11
This section has discussed the requirement for freedom based upon the
claim of ought implies can. By showing that Kant can be understood using
the general interpretation of the principle that we can do what we ought to
do, the power to determine ones empirical character can be concentrated in a
single act. The agent does not require the power to intervene in each and every
particular action in order to be able to say with regard to each and every
particular action that she can do what she ought to do. The agent may instead
have the single power to determine her agency in general, as expressed by her
empirical character, so that she can be said to have the power to determine each
and every particular action. How this nonnatural ability would relate to the
particular actions an agent does take, and how it is related to transcendental
idealism, will be discussed after the following section assesses Kants other
main claim that individuals must have nonnatural freedom of choice, namely
for moral responsibility.
Many to one: moral responsibility
Traditionally one argument for a need for freedom of the will is that praise and
blame would be inappropriate for actions not within the power of the agent to
have avoided. Surprisingly Kant is relatively silent about this aspect of freedom. He certainly does sometimes invoke the language of praise and blame
when discussing transcendental freedom, for example in the solution to the
Third Antinomy when a malicious lie is explained as arising from natural
causes. Now even if one believes the action to be determined by these causes,
one nonetheless blames the agent and assumes that the agent could have acted
otherwise (A55455/B58283). He offers similar language in other places
(5:30, 5:9596, 5:98). But considering the vast amount of writing Kant
devoted to the issue of freedom, he pays very little attention to the issue of
responsibility.
11

Eric Watkins has suggested something akin to this picture. In his reading, Kants concept of
causality involves causal powers of objects, or as he puts it, the natures of individual things, that
then are codiable into laws of nature. Individual agents choose their own natures: to say that
personal agents freely choose their own natures is simply another way of saying that personal
agents are responsible for their noumenal [intelligible] and empirical characters, which is
consistent with what common sense dictates, namely that we be at least partially responsible
for our characters (Watkins 2005, p. 336). The nature that an individual chooses determines
her empirical character, the law of her own causality in nature. Because individuals are thus
responsible for creating some of the laws of nature, they are capable of doing what they ought to
do to the extent that such capability is dependent upon their own individual natures (they cannot
violate other laws of nature).

194

Morality beyond nature?

Several important points must be noted about these discussions of moral

responsibility. First, Kant provides an empirical account of moral responsibility for actions. Despite using the term blame (tadeln), Kant does not
link imputation to the intelligible character alone. He adds a footnote to his
discussion in the Third Antinomy in which he claims that imputations can be
referred only to the empirical character and will never be completely fair on
this account because the true cause of our actions remains entirely hidden
from us (A551/B579). His explicit discussion of punishment and reward are,
rather, based on empirical considerations such as whether a person was drunk,
or in a rage, or a child, and distinguish legal from ethical imputation. A good
example is in the Moral Collins lectures of 178485 (27:28898) where
nothing like transcendental freedom is mentioned. Kant discusses the degree
of freedom (27:291, see also R6812, 19:169) detailing the subjective conditions for degrees of imputation including knowledge and ability to act. Blame
in morality has several levels, and Kant appears to be satised with an
empirical level of imputation of responsibility for actions, at least as one
coherent and justiable practice of reward and punishment.
A second point is that Kants discussions of individuals becoming conscious
of their own freedom because they blame themselves for actions always
involve an individual who has done something wrong, or is faced with the
prospect of doing something wrong, becoming aware that he could have done
the right thing instead (A5545/B58283, 5:30, 5:9596, 5:98). Kant, never to
my knowledge, provides an example of a person who performs the right action
praising himself because he recognizes that it is possible that he could have
performed the wrong action instead. Always the action performed by the moral
agent is the wrong action, and the moral agent realizes that he could have
instead performed the right action. The implication is that the agent is free
when performing the wrong action. This pattern is inconsistent with Kants
usual claim that only actions caused by reason are free, but as I will show it is
consistent with his claims about the nonnatural power of choice.
A third important point is that Kant sometimes places responsibility not on
the agents choice or anything else specic about the agent but instead on
reason itself. Kants emphasis on reason as the ground of free actions lies at the
root of this tendency, as I evaluated at length in Chapter 4. This emphasis on
reason is most prominent in the Third Antinomy. Reason is said to have
causality with respect to appearances (A547/B575, A5489/B5767, A551/
B579) through its imperatives that it imposes on individuals. One might
interpret this minimally, as I did, as simply the faculty of reason possessing
transcendental freedom to legislate the moral law, which individual moral
agents then become aware of in their deliberative processes. Reason is seen
as having a structure that generates the moral law independently of any causal
determination in nature. But in these passages Kant appears to attribute an

From many to one to none: non-natural free choice

195

active causal power to reason as if it were a substance itself. The passage

I quoted at the beginning of this section from the solution to the Third
Antinomy about the liar recognizing that he could have performed the right
action instead of the wrong one places the responsibility on reason not on
himself:
This blame is grounded on the law of reason, which regards reason as a cause that,
regardless of all the empirical conditions just named, could have and ought to have
determined the conduct of the person to be other than it is. And indeed one regards the
causality of reason not as a mere concurrence with other causes, but as complete in
itself, even if sensuous incentives were not for it but were indeed entirely against it; the
action is ascribed to the agents intelligible character: now, in the moment when he lies,
it is entirely his fault; hence reason, regardless of all empirical conditions of the deed, is
fully free, and this deed is to be attributed entirely to its failure to act. (A555/B583,
emphasis mine)

The act of the agent is described as an act of reason, which could have
determined the agents act differently. Reason is a cause, which is fully
free and this deed is to be attributed entirely to its failure to act. The agent is
responsible for the lie because reason could have instead seen to it that the
agent tells the truth; ultimately, then, reason is responsible.
These passages stressing the action of reason and reason as cause make
sense only if they are seen as taking reason not as an independent being in
itself but as an empirical faculty that does actually cause the empirical decisions of an agents faculty of choice.12 Reason could cause some of the actions
of rational beings, but does not always determine the actions of moral agents.
This conception would, however, not solve the problem of blaming particular
agents. For if blame is placed on reason rather than the agent, the agent ought
not be punished because the transcendental power of reason is beyond her

12

Allison considers but rejects such a view of reason as a cause. First, he correctly recognizes that
Kant does not advocate the choice of an intelligible character in the rst Critique (the intelligible character explains the empirical character, but nothing explains the intelligible character),
so rejects interpretations of the Third Antinomy that attribute such a view to Kant (Allison 1990,
p. 51). However, second, he goes on to interpret the Third Antinomy in terms of his own
Incorporation Thesis, which is based primarily upon the position Allison draws from the
Religion. Allison uses this later reference to incorporation to overrule the plain statements of
Kants Third Antinomy that reason itself is the cause. Allison states that he wants to construe
all of this in such a way that it does not commit Kant to the view that reason is literally a causal
power and to the conception of a timeless noumenal agency, which seems to be inseparable
from this commitment, so he tries to avoid anything that would make reason an efcient cause
(despite such language from Kant even outside the Third Antinomy, for example, A317/B374).
In order to do that Allison interprets Kants discussion of reason simply as the legislating of the
laws to a power of choice that itself either incorporates them into its maxim or does not in a
transcendentally free manner. The freedom of choice is spontaneous and the genuine causal
factor of an action. This interpretation itself cannot be squared with the text of the Third
Antinomy in which Kant discusses intelligible causality only of reason.

196

Morality beyond nature?

control. Evil actions are freely decided upon only because reason could have
instead caused the right action to have occurred. There is no corresponding
claim that good actions are freely decided upon because some transcendental
cause could have caused the wrong action to be taken.
As I have interpreted the Third Antinomys emphasis on reason in Chapter 4,
this causal power of reason is understood in a nonontological way. Reason is
an efcient cause in nature as part of the set of psychological determinants of
an individuals empirical power of choice. The empirical faculty of reason is
the empirical character of reason. Transcendentally reason functions as a
spontaneous cause because the very structure of reason itself, which is the
intelligible character of reason, is the ground of the moral law, the representation of which in the empirical mind is reason as a natural cause. The topic in
this chapter provides a different aspect of that relation. If reason as intelligible
cause is the ground of reason as empirical cause, Kant appears to be arguing,
then any transcendental responsibility for actions determined by empirical
reason lies in the structure of reason itself (empirical imputation, as already
mentioned, is a complex of empirical factors). This conclusion raises the
question of the sense in which reason can be responsible for anything.
Indeed, it is inexplicable why reason acts as it does and not in some lesser or
greater way. Kant offers this admission of a fatalism about the course of nature
and agnosticism about reason in the paragraph immediately following the
malicious lie discussion. The key is that we cannot ask why reason is what it
is or why it causes only what it causes: one cannot ask: Why has reason not
determined itself otherwise? But only: Why has it not determined appearances
otherwise through its causality? But no answer to this is possible (A55657/
B58485). With this admission, Kant appears to have abandoned any attempt
to explain transcendental moral responsibility for actions. Both right and
wrong actions occur in the empirical world not because the agent chose to
do those actions but because reason has determined the empirical course of
nature in such a way that they occur. But then moral responsibility seems
entirely divorced from the agent. Kant might have wanted to take this route in
order to avoid difculties with assigning an atemporal nature to moral agents
for purposes of assigning moral responsibility. The result, however, is that he
has avoided moral agents as well.
This is an aspect of the well-known Sidgwick problem, which dates back to
Kants time as an objection raised against Kant by Karl Leonard Reinhold.13
I will not add anything to this discussion beyond noting that this problem for
moral responsibility arises because Kant appears to hold that the only transcendentally free cause which can be postulated is reason, not simply that the
13

The details of Reinholds and Sidgwicks objection and a critique of the standard manner in
which Kant is defended are discussed by (Fugate 2012).

From many to one to none: non-natural free choice

197

will is free only when it is acting rationally. In order properly to attribute moral
responsibility, one wants instead there to be a transcendentally free choice by
the agent, not a transcendentally free causal power of reason.
Perhaps because Kant realized this problem, or perhaps because of other
reasons unknown to us, Kant does offer such a conception of a transcendentally free power of choice. I will discuss this transcendent freedom of the agent
as presented in the Critique of Practical Reason, which I take to exemplify this
conception of freedom for moral responsibility. He introduces this passage
with the claim one must recall what was said in the Critique of Pure Reason
or follows from it as a solution to the clash between freedom and natural
necessity (5:97). Indeed, Kant invokes the Third Antinomy to the extent that it
explains how it is possible for there to be such a thing as transcendental
freedom independent of natural causation. This insistence by Kant appears to
require an ontological interpretation of the Third Antinomy in which the moral
agents power of choice, and not simply the faculty of reason, in some way has
an existence independent of her existence in nature.
Kant goes so far as to call empirical, compatibilist freedom a wretched
subterfuge (5:96) and at bottom . . . nothing better than the freedom of a
turnspit, which, when once it is wound up, also accomplishes its movements of
itself (5:97). After arguing for the insufciency of empirical practical
freedom, Kant determines that genuine freedom requires independence from
the mechanism of nature. Here instead of claiming that reason is free from the
mechanism of nature and thus that reason is the free cause of moral actions,
Kant stresses the existence of the moral agent as a thing in herself independent
of nature. The free cause which is independent of nature is not reason but is the
moral agent as a thing in itself who views his existence insofar as it does
not stand under conditions of time. He goes on:
in this existence of his nothing is, for him, antecedent to the determination of his will,
but every action and in general every determination of his existence changing
conformably with inner sense, even the whole sequence of his existence as a sensible
being is to be regarded in the consciousness of his intelligible existence as nothing but
the consequence and never as the determining ground of his causality as a noumenon.
So considered, a rational being can now rightly say of every unlawful action he
performed that he could have omitted it even though as appearance it is sufciently
determined in the past and, so far, is inevitably necessary, for this action, with all the
past which determines it, belongs to a single phenomenon of his character, which he
gives to himself and in accordance with which he imputes to himself, as a cause
independent of all sensibility, the causality of those appearances. (5:9798)

Here Kant limits this freedom to rational beings, but not to their reason (or to
reason itself independent of them). The moral agent is said to give himself his
character, which appears as his phenomenal existence in nature. His creation of
his character is the basis of the legitimacy of imputation of actions to him in

198

Morality beyond nature?

nature. Kant stresses this point in the following paragraphs of this section of
the second Critique, even admitting that this noumenal character might be evil:
[Some people] show from childhood such early wickedness and progress in it so
continuously into their adulthood that they are taken to be born villains and quite
incapable of improvement as far as their cast of mind is concerned; and nevertheless
they are so judged for what they do or leave undone that they are censured as guilty of
their crimes . . . This could not happen if we did not suppose that whatever arises from
ones choice (as every action intentionally performed undoubtedly does) has as its basis
a free causality, which from early youth expresses its character in its appearances
(actions); these actions, on account of the uniformity of conduct, make knowable a
natural connection that does not, however, make the vicious constitution of the will
necessary but is instead the consequence of the evil and unchangeable principles freely
adopted, which make it only more culpable and deserving of punishment. (5:99100)

Here Kant argues that moral responsibility is attributed to the agent as a thing
independent of nature in space and time. Reason is not seen as constituting the
nonnatural free cause of moral actions; in fact, the free nonnatural cause mentioned is the evil and unchangeable principles freely adopted that result in the
single phenomenon of the persons empirical character in nature. The transcendent free adoption of these principles is the ultimate basis of moral attribution.
Despite his invoking empirical considerations or the causality of reason,
moral attribution after all does require for Kant the postulation of a free
nonnatural ability to decide upon principles for action. But as with the claim
about the need for a transcendent causal power to serve as the can in the
principle ought implies can, this freedom need not be distinct for each moral
action. The free transcendent choice that Kant thinks is involved in moral
attribution is the choice not of actions but of principles, presumably maxims.
One principle or maxim can of course serve as the basis for a plurality of
particular actions. So here again it appears that the transcendentally free power
of choice need not be understood as a plurality of transcendent free decisions,
but just one. Transcendental freedom of choice in Kant is not a series of
independent free transcendent decisions but a single free transcendent decision
of the power of choice that has as its effect the entirety of the empirical
character of the individual. This conception reduces the nonnaturalistic
requirements of Kants theory considerably. The next section will explain
how Kant ultimately describes this single free transcendent choice and how
it can be understood in a metaphysically naturalistic way.
The one transcendentally free decision
The previous sections looked at the reasons Kant gives for requiring a free
choice. This section will argue that, as a metaphysical claim on its own, it is
best not to interpret transcendent freedom of choice as allowing that every

From many to one to none: non-natural free choice

199

individual decision of the empirical power of choice can also be nonnaturally

free, as if there were a multitude of independent free choices underlying the
actions of the self in appearance. Rather, only one nonnaturally free decision is
possible because the free transcendent decision is said to be timeless.
I take Kants discussion of the power of choice in the Religion within the
Boundaries of Mere Reason to reveal the way in which Kant intended the one
free transcendent decision by an agent to relate to many empirically determined decisions. His topic in the rst section of that book is the universal
human propensity to evil, his own take on the Christian concept of original sin
viewed within a philosophical conception of religion; however, the metaphysical relations at work can apply equally to the choice of ones entire phenomenal character as Kant describes it in the second Critique. He explains a
propensity by relating two senses of the term deed: By the nature of a
human being we only understand here the subjective ground wherever it
may lie of the exercise of the human beings freedom in general (under
objective moral laws) antecedent to every deed [Tat] that falls within the scope
of the senses. But this subjective determining ground must, in turn, itself
always be a deed [actus] of freedom (for otherwise the use or abuse of the
human beings power of choice with respect to the moral law could not be
imputed to him, nor could the good or evil in him be called moral) (6:21).
Kant takes there to be two different levels of choice at work. First is the level
within the scope of the senses and the second is the subjective determining
ground of each of those empirical deeds. The second, he goes on to say, is
antecedent to every use of freedom in experience from the earliest youth
throughout ones life (6:22). The relation between Tat and Actus maps clearly
onto the empirical or phenomenal character on the one side and the single act
of freedom that creates that character from the second Critique as described in
my previous section on the other side.
The transcendentally free human power of choice grounds all particular
decisions by means of a maxim. Each empirically determined decision of the
power of choice, of course, would have a maxim as a subjective principle
operating in the empirical mind. The ultimate subjective determining ground
would be, in a sense, a maxim of maxims. This Kant calls the disposition, the
rst subjective ground of the adoption of the maxims [which] can only be a
single one, and it applies to the entire use of freedom universally (6:25). The
relationship between these two levels of maxim is hierarchical. Each particular
empirical maxim is adopted in reference to the transcendentally freely chosen
maxim: Now, the term deed [Tat] can in general apply just as well to the use
of freedom through which the supreme maxim (either in favor of, or against,
the law) is adopted in the power of choice, as to the use by which the actions
themselves (materially considered, i.e., as regards the objects of the power of
choice) are performed in accordance with that maxim. . . The former is an

200

Morality beyond nature?

intelligible deed, cognizable through reason alone apart from any temporal
condition; the latter is sensible, empirical, given in time (factum phenomenon)
(6:31). The single supreme maxim is freely chosen in a transcendentally free
way, and each particular empirical deed in time that reects that maxim can be
understood as free ultimately only as a reection of the supreme maxim.
Kant argues, in essence, that the power of choice must be understood as
choosing actions in nature and as having chosen a general course of action
independent of nature. Kant states this problem in terms of the source of human
propensity to evil. The propensity to evil, that is, the consistent placing of selfinterest above the moral law, must have a free origin. Physical propensity, that
is, the consistent decisions of the sensible power of choice in nature, is not free.
Thus the propensity to evil must have some other free source. The free source
would have to be a decision independent of nature. There are two corresponding
understandings of the term deed. There is only one deed independent of
nature, and that is the fundamental choice to follow the moral law or to instead
prioritize self-interest. This deed is the decision which formulates ones highest
maxim. Then there is the understanding of the word deed that applies it to the
particular application of that highest maxim to actions. These deeds in appearance are referred to as free, although presumably determined according to laws
of nature, because they can be traced to a nonempirical origin in the single
transcendent deed which formulates ones highest maxim.14
Combining the views presented in the Critique of Practical Reason and the
Religion results in a plausible account of how Kant can understand the freedom
of the power of choice at two levels in such a way that it requires only one
nonnaturally free choice. The single act of choosing ones character reverberates, as it were, throughout all of the appearances of ones self in nature. The
single phenomenon of ones empirical character can be traced to a single
nonnatural free decision. This nonnatural freedom is not needed at every
particular instance as an independent explanation of each particular action.
Rather, one such free act sufces to serve as the ground for all particular actions
in nature. The transcendent self need not be conceived as involved in a multitude
of free decisions that correspond with a multitude of empirical decisions.
14

Later in the Religion Kant speaks of a revolution in the disposition of a moral agent, referring
to an alteration of the fundamental stance one takes toward morality, the choice to change ones
fundamental disposition from self-love to duty (6:47). If understood as a genuinely independent
second free transcendent choice this account appears to be incompatible with my interpretation.
It is also incompatible with Kants account in the second Critique when he talks about the single
phenomenon of ones character and Kants account earlier in the Religion of a single inexplicable fundamental maxim that must ground all others. If one does allow a revolution in ones
fundamental maxim, then ones unity of character is also threatened in the way I discuss later.
For these reasons I do not think that Kants position on a revolution in character can be taken to
be any more than an idea of reason we hold as support for our efforts to improve ourselves
morally. I do not think it forms part of his metaphysics of free choice.

From many to one to none: non-natural free choice

201

There is a more directly philosophical reason for holding that only one free
transcendent choice is possible for Kant. The unity of identity of an individual
dictates that only one ultimate choice be posited per individual. A moral agents
empirical character is a single phenomenon, and hence has a unity. The unity
of that character must stem from a single cause, that is, there must be a single
decision that results in the single result. Other kinds of causes, such as the concurring causal power of God in creating nature as a whole, would of course be
allowed. But the relevant cause, the decision of the power of choice, can be only
one. To see why, assume for a minute that there is more than one such cause.
Transcendent choices of this kind are decisions that have no explicable basis,
that is, they are themselves the ultimate explanation and cannot be traced to
any other ground. They are, in effect, unrelated to one another, for if they were
related to one another, that relation would be their ground. If there is more than
one such ultimately inexplicable decision, any unity of character would be
merely random since the independent decisions might conict with one another;
were every particular empirical action to have such a corresponding nonnaturally
free intelligible choice, the odds of lack of unity approach certainty. Even if they
did not conict, there would only be a contingent unity of character since if there
were something that necessitated the consistency of these decisions, it would be
a ground of the decisions, but no grounds are allowed for them. Kants conception of the intelligible character as a transcendent cause of the empirical character
seems to require a unity that does not come about merely contingently. Thus, in
order to have a unity of character, only one transcendent free decision is allowed.
This single nonnatural decision is by itself incompatible with metaphysical
naturalism; however, it does at least show that the strong sense of empirical
reality can be given to nature in space and time as a causal system can be
preserved even with the admission of nonnatural freedom. One need not
conceive of nature as mere appearance, existing only as representations, with
ontological priority given to things in themselves. There is no need to assume
that every act in nature has a distinct transcendent cause, or that transcendently
free beings reect upon all the particular circumstances of particular acts in
nature. The freedom that Kant requires is provided by a general decision that is
conceived to occur independently of and in some sense prior to all particular
acts. One is reminded of Platos Myth of Er in which individuals are said to
have an opportunity before birth to choose a life, but after they choose that life
they were condemned to live it out without change.15 Just as Platos myth

15

Henry Allison also notes that this interpretation of Kants conception of freedom resembles the
Myth of Er (Allison 1990, p. 139). He then offers a nonmetaphysical interpretation that he
thinks will avoid the problem. My approach is different because when I maintain the ontological
priority of nature in space and time, I see no room for a transcendental freedom of the faculty of
choice in nature.

202

Morality beyond nature?

functions to provide a way to picture that which is in itself inexplicable, Kants

invocation of the single transcendent free decision functions as a way to
picture the inexplicable ultimate free source of moral decisions. One might
conceive of Kants single nonnatural act of the power of choice in this way
without diminishing the reality of nature, just as one can conceive of God
existing independent of and prior to nature without diminishing its reality. The
metaphysical cost for a naturalist, one who wants to limit ontology to nature, is
large but less than if one had to deny the independent reality of nature and its
causality in space and time.
This account satises the two arguments reviewed in the previous section
that nonnatural freedom was necessary with a minimum of nonnatural
ontology. First, the principle of ought implies can can be defended in
relation to a single free choice in allowing that each individual moral agent
can be taken to have complete control over her empirical character in nature
through the single free intelligible choice. This single intelligible choice is
made entirely independent of nature, abstracted not only from the conditions of time but considered to be itself timeless as part of a being in
herself. Since this intelligible choice creates the intelligible character the
fundamental maxim of that individual and the intelligible character is then
reected in the empirical character the particular maxims that individual
chooses using her empirical power of choice and since the empirical
character is part of the causal chain in nature that determines that she will
perform certain actions rather than others, this intelligible choice can be
understood to have determined the particular empirical course of nature for
that individual. The nonnatural, timeless decision to adopt a fundamental
maxim is subject to the ought and is conceived to be entirely determined by
the agent: the agent can, conceivably, choose any particular fundamental
maxim (although since Kants discussion is framed in terms of a universal
human propensity to evil, he takes all to have a fundamentally evil maxim).
In our actual experience, of course, when an agent is faced with a particular
decision, the ought applies to the actual maxim she adopts. Empirically she
cannot act in any way other than as determined by nature, but the picture
allows the agent to recognize that it would have been within her power qua
intelligible, nonnatural power of choice to have determined herself in a way
that would have caused her to act otherwise, so in a transcendental sense
she could have acted as she ought.
The second argument requiring a nonnatural freedom is even more
straightforward. Agents are held responsible for empirical actions they perform although those actions are determined in accordance with laws of nature
in two ways. First on the empirical level Kant provides empirical conditions
for imputation that invoke particular kinds of natural causes. One might

From many to one to none: non-natural free choice

203

object that this empirical imputation assigns praise and blame to decisions or
actions outside the agents causal power since they are themselves determined in accordance with laws of nature. This is the empirical character,
which is itself integrated into the causal nexus of nature. Kant can justify this
imputation by reference to the single nonnatural decision that forms the
intelligible character and that is said to ground the particular decisions of
the agents empirical power of choice. Transcendentally the agent is responsible for all particular deeds in reference to this one intelligible deed even
though each particular deed is empirically determined in her power of choice
in nature.
Thus it is a valid interpretation to hold that Kant accepts that there is only
one nonnaturally free decision by the power of choice. He has a good philosophical reason to do so, given his conceptions of things in themselves in
relation to nature. The unity of the empirical character is a result of the unity of
the intelligible character, itself determined by a single timeless free decision.
For moral purposes regarding the practice of praise and blame (moral attribution) as well as for determining the application of the principle ought implies
can, we can conceive of any particular act performed by an individual as
causally determined in nature yet as a reection of the single nonnatural
choice. Every particular act performed by an individual is traced to her intelligible character, a timeless principle that is the transcendent cause of that
individuals entire existence in nature.
From one to none: freedom as a postulate
The interpretation I have given shows that the transcendental freedom of
the power of choice can be understood to consist of only one nonnatural
act of the person in herself that results in the entirety of the persons
empirical character and thus all of her actions in nature. For moral responsibility and because ought implies can, we are required, for moral purposes, to assume that somehow ontologically prior to our existence in
nature we ourselves each made a decision regarding what kind of person
we will be morally.
This short section will suggest that even this requirement is susceptible
to a metaphysical naturalism. Since freedom of the will is one of the three
postulates of pure practical reason in the Critique of Practical Reason
along with immortality of the soul and the existence of God (5:132), it is
susceptible to the same interpretation I gave of the postulate of God in
Chapter 5. Kant rarely discusses the postulate of freedom of the will and
the ambiguity of Wille allows that Kant might be referring to the freedom
of pure practical reason as source of the moral law. In this section I am

204

Morality beyond nature?

assuming that one can read Kants postulate in reference to the freedom
of choice.16 In this interpretation the postulate is a concept that is held in
the mind for practical, immanent purposes yet is not taken to have any
object as referent. The postulates are limited to concepts in the mind that
are not intended to be existential claims about independent, nonnatural
objects.
I will not repeat the detailed arguments I provided in Chapter 5. In essence
one can consider the claims postulated for practical purposes as concepts
which function to support moral life without ontological implications. Kant
explains this relation as referring to the moral law and only for its own sake in
the Critique of Practical Reason (5:133). For the issue at stake here, the single
free transcendent act of a nonnatural power of choice can be understood as a
mere concept that human beings possess in order to picture for themselves the
inexplicable claim of ultimate responsibility for action (Kant calls it inscrutable, unerforschlichen, 6:21). In the same way that the concept God
functions as a kind of placeholder for any solution for the moral requirement
that the highest good is possible, the concept of a nonnaturally free decision
functions as a placeholder for any solution to the moral requirements of ought
implies can and of moral imputation. No claim to the reality of any nonnatural
cause is made in the postulate, and thus this claim is compatible with a
metaphysical naturalism.
Freedom, however, appears to have a different status than the other two
postulates of immortality of the soul and the existence of God. In the preface to
the second Critique Kant says that freedom is real and that it grounds the
possibility of the other two postulates:
Now, the concept of freedom, insofar as its reality is proved by an apodictic law of
practical reason, constitutes the keystone of the whole structure of a system of pure

16

Consider that the kind of freedom referenced in the postulates as the ground of the moral law
might not refer to the freedom of the faculty of choice but instead to the transcendental freedom
of reason. The claim that freedom functions as the cornerstone of speculative as well as practical
reason cannot refer to the freedom of choice to act for or against the law because that kind of
freedom has nothing to do with speculative reason. Instead the transcendental freedom of
reason, which Kant does at times refer to using the term Wille and that I discussed in
Chapter 4, is what makes the moral law possible by showing that the structure of reason itself
provides principles and ideas and is valid independently of laws of nature. This transcendental
freedom of reason would be able to ground both practical and speculative reason. To say that
this freedom of reason is proved real through the moral law is to say that the fact of reason as
our experience of the categorical imperative is bound up with a claim of its validity. To the
extent that the fact of reason is acknowledged, the freedom of reason is acknowledged as well.
This account is compatible not only with Kants discussion of the fact of reason in the second
Critique (and Chapter 3) but also with Kants insistence in the solution to the Third Antinomy
that it is reasons active role in producing ideas and principles that provides the basis for a claim
to the transcendentally free intelligible character of reason.

From many to one to none: non-natural free choice

205

reason, even of speculative reason; and all other concepts (those of God and immortality), which as mere ideas remain without support in the latter, now attach themselves to
this concept and with it and by means of it get stability and objective reality, that is,
their possibility is proved by this: that freedom is real, for this idea reveals itself through
the moral law. (5:34)

Kant goes on in the next paragraph to differentiate freedom from God and
immortality in that the rst is a condition for the moral law while the latter two
are conditions only for an object of a will determined by the moral law (the
highest good). Being a condition for the moral law itself means that the
requirement does not arise upon reection after the fact of a moral decision
as immortality and God do but is intrinsic to the moral decision itself.
Immortality and God arise only upon the further step of reection upon the
place of morality in the greater scheme of things, a step that lies outside the
inner context of decision making under the moral law.
These considerations give freedom a different status than the other postulates. Freedoms reality is proved though the moral law, freedom is itself a
condition for the moral law, freedom grounds the objective reality of God and
immortality, and freedom even functions as the cornerstone of pure reason as a
whole. Despite the identication of freedom as a postulate, then, it appears that
it might not be possible to interpret it merely heuristically. I think that even
with this special status freedom is best interpreted in roughly the same manner
as immortality and God.
The nature of practical philosophy itself as I showed in Chapters 2 and 3
supports such a reading with regard to freedom. The subject matter of
practical philosophy is free acts considered from the agents perspective as
free decisions. When faced with actual decisions, the moral agent in nature
does not question whether her decisions are free but must view herself as free
in the process of deliberation. The agent does not think about any kind of
freedom and in the moment of deliberation does not have any need for the
nonnaturally free power of choice. Since practical reason is supposed to
concern the ought rather than the is, there need be no ontological claim
at work.
That point, however, is valid only under the priority of action aspect of the
primacy of practical reason. The primacy of practical reason, I showed in
Chapter 3, has two different kinds of meaning in Kants philosophy. The
priority of action stresses the role of practical reason in determining the agent
to right acts. I take this to be operative in Kants examples of particular agents,
and in relation to freedom it is exemplied by the agent-perspective on
deliberation. When deliberating, moral agents always act under the idea of
freedom and while in the grip of the activity of deliberation cannot consider
their decisions as predetermined. Acting under the idea of freedom in this way
has no ontological implications.

206

Morality beyond nature?

But the status of freedom is different under the second meaning of the
primacy of practical reason, which I call the priority of belief. The priority
of belief holds that when reecting upon the conditions of human morality
ones concern is with reconciling the claims, that is, the postulates, arising
from the practical use of reason with the limitations on speculative claims
allowed by the rst Critique. I argued that the priority of belief actually
operates under a higher priority for speculative or theoretical reason because
it attempts to incorporate the concepts raised by practical reason into a coherent world view. The agent-perspective assumption of freedom in practice has
no place here. Freedom is instead raised as a condition for the existence of the
moral law when one later reects on the nature of the moral experience and
determines its conceptual requirements.
This origin of the concept of freedom, however, is the same as that of the
other two postulates. All three would arise upon reection after the moral
experience. Since I showed in Chapter 5 that the postulates are to be considered valid not as referring to any objects but only held as concepts in
support of morality, freedom can have that status provided that it can be seen
as stemming from reection upon moral experience in a similar way.
My interpretation of the fact of reason allows for this. The fact of reason is
our experience of the categorical imperative and helps to constitute our moral
experience. When faced with it in actual decision-making we do not question
either its validity or our ability to choose to follow it this is our stance
given the priority of action. But upon reection on that moral experience we
can ask both questions: what would make the categorical imperative really
valid, and whether we could freely choose whether or not to follow it. Our
answer can recognize that the categorical imperative would not be possible
were it not for the transcendental freedom of reason and its normative force
for us would not be coherent were it not for the transcendental freedom of
choice. But since these answers are wrapped up in the question of attempting
to understand our experience of the categorical imperative, we can even say
that our claim that freedom is a condition for the moral law is postulated on
behalf of our experience of the categorical imperative. The reality of
freedom proved by the moral law is simply the practical recognition that in
order to make the experience we have of the fact of reason coherent with our
entire worldview we would need to use the concept of freedom. But the
demand to make our experience coherent still rests on the status of that
experience. And the status of that practical experience is from an agent
perspective. The transcendental freedom of the power of choice is held as a
concept for immanent purposes rather than taken to refer to an actual
property of moral agents. The nonnatural free decision of the power of
choice, then, is subject to the same naturalist interpretation as the other
postulates.

From many to one to none: non-natural free choice

207

Concluding remarks
This chapter has assessed the ontological requirements for the free power of
choice in Kant in order to offer a metaphysically naturalistic interpretation of
Kants moral theory. In addition to the compatibilist view of freedom in nature
that emphasizes the decisions of the empirical power of choice, Kant insists on
a transcendental understanding of freedom of the power of choice. I looked at
two reasons he offers the principle that ought implies can and the requirement of moral responsibility and showed how both can be interpreted to
require only one nonnatural transcendentally free decision. Instead of a metaphysics that requires a nonnatural transcendent free choice for each particular
act, Kant requires only a single act to ground the character that then determines
all the particular acts performed during ones lifetime.
I examined the interpretations of this single transcendentally free act that
best t with metaphysical naturalism. Instead of a focus on how any particular
individual action can be independent of determinism in nature, the focus on
freedom of choice for the metaphysically naturalist Kant ought to be on how
ones overall character can be understood to be independent of determinism in
nature. I think that such an emphasis makes Kants moral theory less burdened
with metaphysical difculties than alternative interpretations that focus on the
transcendental freedom of each particular act distinctly from every other act.
The single-free-act interpretation allows for a view of nature not unlike a
traditional view in which an independent, timeless God creates the whole of
nature, which then exists as real and as causally closed (other than that act of
creation). The multiple-free-act interpretations face the difculty of explaining
how the allegedly timeless transcendentally free cause of the free choice of the
agent in herself is related to the temporally situated natural causal chain of the
empirical agent in nature while still retaining actual empirical determinism as a
real property of a real nature in space and time. More particularly, they must
explain how a timeless agent in herself can be affected by temporally situated
sensations and desires in nature in a way that preserves the timelessness of the
free agent while individuating the free decision to that exact context in nature.
The interpretation of Kant as requiring only a single timeless act vastly
simplies this metaphysical relationship. One still has to picture the near
divine power of an individual agent making a single general choice of what
kind of person to be, but once that admission is made, nature in space and time
would be seen as real and causation according to laws of nature entirely valid
for all events in nature include decisions made by the empirical human power
of choice.
This single nonnatural act is itself still not compatible with metaphysical
naturalism. The naturalist has available a way to avoid such an ontology by
stressing that freedom of the will is a postulate of practical reason. The

208

Morality beyond nature?

postulate of the transcendentally free power of choice is not meant to have

ontological implications. This status for the postulates ts in with Kants
emphasis on the practical point of view. Although freedom is given a status
distinct from that of the other two postulates, I showed that the difference does
not preclude the interpretation of the postulates as immanent rather than
transcendent, and as meant only as conceptual support for moral practice
rather than as ontological claims.

Value and the inexplicability of the practical

The postulates of God, immortality, and freedom of choice have in the past two
chapters been shown not only to be fully compatible with a metaphysical
naturalism but also to be best understood using moral idealism. This chapter
will examine the nature of moral value in particular and moral experience in
general. These matters concern the ways in which morality frames our phenomenal experience as spatial and temporal beings in nature. The discussion of
value will show that value is not a property of anything independent of the
moral agent but an ordering of nature by reason. The status of this ordering,
and with it moral experience in general, will be shown to be the ultimately
inexplicable product of reason that fashions the idea of an unconditioned in
morality for its own purposes as in theoretical philosophy it fashioned the ideas
of soul and God as representations of the unconditioned.
My review of the various elements of Kants ethics that individually needed
to be assessed in terms of realism or idealism included two kinds of value. The
value of subjectively chosen ends has already been seen as fully compatible
with both metaphysical naturalism and empirical moral idealism, since the
value of these contingent ends stemmed only from the choice of the empirical
moral agent to pursue them. The value of humanity as an end in itself,
however, appears not to depend upon the empirical moral agent. The categorical imperative in the formula of humanity commands moral agents always to
treat humanity as an end in itself. Reason offers no option to moral agents
regarding the value of humanity as it does regarding the value of freely chosen
ends. I will argue that, despite the necessity and objectivity characteristic of
absolute value, the proper understanding of the value of humanity is in terms
of moral idealism, and that it is compatible with metaphysical naturalism.
This chapter will provide in the rst section a direct discussion of the nature
and status of absolute value in Kant. I will show that one version of the
argument from autonomy succeeds in showing that Kant must reject value
realism. Since value must not precede the autonomous moral law, it must not
be any intrinsic property of things. I will also argue that there is no place in
Kants philosophy for any intrinsic value properties. Moral value is both
transcendentally and empirically idealist. My second section will look in more
209

210

Morality beyond nature?

detail at the order of reason. To say that humanity has absolute value is to say
that the order that reason imposes on moral agents choices includes the
requirement that humanity is a limiting condition when choosing subjective
ends. I look at Kants terms intelligible world and moral world to show
that they mean the moral order that reason imposes as a formal ordering of the
objects of nature. Moral experience is understood in terms of the formal rather
than material view of nature. My third and nal section looks at the ultimate
limits that Kant places on the intelligible order of reason by focusing on the
topic of the outer boundary of practical philosophy given in the last sections of
Groundwork III. I conclude by summarizing the nal status we ought to
understand practical reason to have as a faculty that we human beings use as
part of the rst-person view of experience.
Realism and idealism about moral value
The highest value in Kant is the value of humanity, said to be an end in itself.
There is dispute over precisely what Kant means by humanity,1 but I will
largely forgo entering into those disputes and simply refer to humanity as of
absolute value. This section will argue that the absolute value of humanity is
not an intrinsic property but is best understood as dependent upon the order of
ends imposed by pure practical reason and so conforms to moral idealism
rather than moral realism. I will rst look at the relation between value and
autonomy before showing that moral value cannot be an intrinsic property of
objects but instead is an order imposed by reason. Finally I cap off the section
with a discussion of transcendental idealism of value in relation to empirical
realism.
Autonomy and value
A contentious argument in support of the moral idealists claim that even the
absolute value of humanity must be a product of rather than a basis for the
moral law is based on the nature of autonomy.2 Kant divided all moral theories
1

A few examples sufce to show the scope of the debate. Christine Korsgaard holds that it is
rational nature, specically the ability to dene ends for oneself (Korsgaard 1996a). Allen Wood
takes humanity to be the human capacity for pragmatic reasoning (Wood 2008) (See also
footnote 5 to this chapter ). Paul Guyer advocates freedom over rationality as the ultimate value
for Kant (Guyer 2000). Richard Dean argues that humanity is the good will as the only thing
good without qualication (Dean 2006). I believe that the argument I make about the status of
the general concept of absolute value in Kant is compatible with most of these and other
particular conceptions. In the discussion in the text I will note when my claims appear to apply
differently to different conceptions of humanity.
I made this claim very succinctly in (Rauscher 2002, p. 496). For two responses to this argument,
see (Stern 2012, pp. 12122), who nds it too sketchy, and (DallAgnol 2012b, pp. 1619), who

Value and the inexplicability of the practical

211

into the autonomous and the heteronomous depending on whether the will is a
law to itself or whether some property of the objects of volition is the basis
of the law of the will (4:440). Autonomy is the autonomy of reason considered
as a faculty that is structured so that it produces the moral law of its own.3
Because of this focus on the nature of reason considered as such, autonomy so
understood is a transcendental not an empirical conception; individual choice
with leeway would be an empirical conception of autonomy. The issue here is
whether autonomy understood in this transcendental sense can be the basis for
a claim that the absolute value of humanity must be a result of determination
by the moral law (transcendental idealism of value) or may precede it (transcendental realism of value).
The argument from autonomy for interpreting Kant as a moral idealist
stresses that any basis for morality independent of pure practical reason as
the will of a transcendental moral agent would have to count as heteronomous
and could not be Kants position.4 Since only rational agency is left as the
possible basis for the moral law, and idealism is dened as the claim that all
moral principles, properties, and objects are dependent upon the moral agent,
the proper interpretation of Kants moral theory is idealistic.

thinks it misunderstands the nature of reason (my direct response to the latter is (Rauscher
2012)). The argument is also presented specically in relation to value by Robert Johnson
(Johnson 2007, pp. 14041), who compares it to a Euthyphro-style argument. Others who stress
autonomy as precluding independent criteria for right include (Schneewind 1997), (Reath 2006),
and (Sensen 2013).
By autonomy Kant does not mean individual leeway in choosing what laws to follow or even
individual power in obligating oneself to the law. There is no individual leeway the moral law
is taken to express the structure of reason itself and obligation accompanies the categorical
imperative by means of its imperatival form as the fact of reason. Robert Stern argues that
autonomy can be understood in terms of obligation. He argues that Kant could be a moral antirealist by taking moral obligation to be autonomous. He argues that obligation is idealist and
autonomous because the concepts of duty and obligation apply only to nite, sensible moral
agents. For perfectly rational agents there is no obligation because all of their actions are
automatically determined by pure practical reason. Finite, sensible moral agents are obligated
in virtue of their nitude, not by any kind of choice whether to obligate themselves but simply by
their nature. Stern takes Kant to be a value realist but to allow for this kind of anti-realism for
obligation (Stern 2011b, pp. 8991). But this is a strange limitation for anti-realism. Human
beings are taken to have in common with purely rational agents their reason, and thus in virtue of
their rational nature all rational agents are subject to the moral law. For we nite sensible rational
agents there is simply another added aspect to our natures, namely, the matter of adding sensible
incentives and not of adding anything moral. The manner in which nite moral agents are subject
to the moral law, namely through duty and obligation, stems from their nature. And the manner
in which purely rational agents are subject to the moral law, namely as sole determinant of their
actions, also stems from their nature. Why should one kind of conception of a moral agent
ground an anti-realist claim while the other grounds a realist claim? If the nature of a particular
kind of moral agent grounds an anti-realist claim in one case, it ought to in both.
Henry Allison makes this claim as well when he analogizes autonomy and heteronomy to
transcendental idealism and transcendental realism (Allison 1990, pp. 99100).

212

Morality beyond nature?

Those who advocate this defense of an idealism about value stress that any
absolute value of humanity as an end would count as a property of the objects
of volition (4:109) for these purposes and thus be heteronomous rather than
autonomous. Realists can respond by claiming that the value of humanity as an
end in itself is not an object of this sort because it is not something to be sought
or attained by the will the way that happiness is, thus autonomy, the will being
a law to itself, is compatible with that law recognizing the intrinsic value of
humanity as a limiting value.5
The realist response recognizes that the term object of volition at stake
here can mean two different things. First, it can mean the particular object or
goal to be created or attained in action, such as cooking and consuming a
meal.6 Kant apparently has this conception of an object in mind when he
denounces heteronomous moral principles as those that embody the principle
I ought to do something because I will something else (4:441). They are all
based upon some goal, understood as itself valuable to the agent, to be attained
by adherence to a moral principle: self-love, the pleasure of moral feeling,
obedience to God, or even the attainment of individual perfection on the basis
of a (vague) concept of it in reason, as he lists them in the Groundwork
(4:44243). A second meaning of the term object of volition is that the will
would have to treat certain objects as limitations on its actions.7 Here the
object of volition is not merely what is desired but is anything that is the
5

Allen Wood goes further by claiming that autonomy is based on the absolute value of humanity,
the idea being that the value of humanity as the rational will itself is what obligates moral agents
to submit to the dictates of this rational will valued in itself: The idea of autonomy identies the
authority of the law with the objective value constituting the content of the law. It bases the law
on our esteem for the dignity of rational nature, which makes every rational being the moral
legislator (Wood 2008, p. 106). If the argument for value idealism based on autonomy is
successful, as I will advocate, then both the moderate realist position that autonomy does not
preclude an independent absolute value and the more extreme realist position of Wood that
autonomy necessitates an independent absolute value would be shown false. I will not offer a
separate refutation of Woods position. I would like simply to point out that one might doubt that
absolute value of humanity is autonomy as Wood here suggests. Elsewhere in the same book,
citing Kants Anthropology lectures and his Religion, Wood identies humanity as an end in
itself as the absolute value with pragmatic and not moral reasoning, although Wood glosses this
by arguing that humanity in this sense is coextensive with the capacity to give oneself moral laws
(Wood 2008, pp. 889, 94).
Robert Johnson separates what I am treating as one kind of object into two kinds: those to be
produced by the will and those to be realized in action. He takes humanity to be an end of the
second kind because he thinks we can realize humanity by realizing our wills as good wills, or as
he also describes it, actualizing what is only potential (Johnson 2007, pp. 14546). I do not think
these two ought to be differentiated because in both cases the ends are brought into existence
through the action decided upon by the human power of choice. The claim that humanity can be
considered an end to be realized is also irrelevant in this context because the value of humanity,
its being a good, would not be based upon the action that realizes humanity in particular
situations.
Johnson lists this as his second of three types of objects of the will and recognizes that in this
sense humanity would be an object of the will (Johnson 2007, p. 146). Johnson correctly argues

Value and the inexplicability of the practical

213

content of the deliberation by the will, including the means that might be used
to attain chosen ends. An agent might have a particular reason to avoid certain
means. A desire to full a promise to save the last piece of cake for someone
would limit an individual from using that cake to offer food to a third person.
Kant appears to have this conception of humanity in mind when in the formula
of humanity moral agents are obligated always to treat humanity in a certain
way, as an end, no matter what any other end they might be pursuing (4:429).
In this second understanding of object of volition humanity is not so much to
be promoted or enlarged as it is to be recognized and taken into account.
If the argument for moral idealism based on autonomy proceeds by using
the rst sense of object of volition, then it will not succeed. Certainly
Kant never identies the good of humanity as some good to be produced.
Kant provides what he takes to be a comprehensive list of all possible kinds
of objects as a determining ground of the will not only in the Groundwork
as cited earlier but also in the second Critique as well (5:39), and humanity as an end in itself is not a part of either of these lists.8 But since this
kind of end is not the only possible understanding of the term object of
volition, the argument fails to exclude the other kind of end as a value
independent of the categorical imperative. Further, some interpretations of
humanity that might be understood as taking humanity to be such a value
treat it as something to be promoted in the sense that our actions ought to
enlarge the existence of humanity, understood as freedom or as rational nature,
because of its value; in this way they make Kant into a kind of quasiconsequentialist.9 If humanity functioned as an object of the will in this sense
it would obviously violate the principle of autonomy because it would set
out an object to be promoted, a good, prior to the categorical imperative
in precisely the way that Kant attributes to heteronomous moral principles.

that the value of humanity understood in this sense need not precede the moral law but can issue
from a command of reason.
Oliver Sensen suggests that this lack of inclusion of the absolute value of humanity in Kants list
of ends (which Sensen identies here in terms of the good) is evidence that Kant cannot have
thought that humanity can function as a value grounding the categorical imperative (Sensen
2011, p. 16). This leap is too quick. Kants omission of the value of humanity from these lists
means that humanity would not be the type of end to be created or attained through action, but it
is still possible that humanity could be a different kind of end grounding the categorical
imperative in a different way. There are other reasons to reject the latter claim that I will
discuss.
I take Paul Guyer, who sees absolute value as freedom (Guyer 2000), and David Cummiskey,
who understands Kants valuing of rational nature to allow for a consequentialism (Cummiskey
1996), to offer interpretations of this type. Their conceptions of value are not precisely equivalent to the kind of goods that human beings seek and that Kant condemns as the foundation for
heteronomous moral principles. But there is a sense that the good they identify as humanity is to
be produced. Their conception is similar to Johnsons claim (see footnote 6 to this chapter) that
humanity might be understood as something to be realized.

214

Morality beyond nature?

I take this consideration to be a reason to reject any interpretation that makes

Kant into some kind of quasi-consequentialist.
The absolute value of humanity has to be understood as the second kind of
end, a limitation on the deliberations and decisions of moral agents.10 Realists
about value can offer this interpretation as a way to argue that humanity is an
end that does not violate autonomy, because not an end to be attained, yet is
not derived from the categorical imperative.
The argument from autonomy to moral idealism can proceed, however, by
arguing that even this kind of limiting value as the basis of the moral law
would violate autonomy. Rather than being heteronomous by being a characteristic of an object of volition as something for the will to produce or create,
the value of humanity as a limiting condition that precedes the moral law
would fail to be autonomous because it prevents the will, as pure practical
reason, from being a law to itself. If any independently existing value puts a
limitation on the formulation of the moral law, then reason fails to be autonomous. The way that value realist interpretations of Kant understand the value
of humanity, it does put a limitation on the formulation of the moral law.
Allen Wood offers precisely that kind of realist interpretation of the role of
the value of humanity in relation to the formulation of the moral law: In the
most complete or universal formula of the moral law, [the formula of humanity], this value is developed into the ground of moral legislation itself.11
Wood appears to separate this claim from a relatively innocuous claim that the
value of humanity serves as a motive to follow the categorical imperative.12
If the value of humanity is to function as a motive to follow the categorical
imperative and does not determine the content of that imperative, it would not
violate autonomy. Humanity as a ground of legislation would violate autonomy by placing something as a limit to the legislation of reason.
This is a difcult point to grasp because the value of humanity is said to be
necessary for all rational beings just as the categorical imperative is. Both are a
10

11

12

Wood makes this point as well by differentiating the ends to be produced, which are the subject
of the moral law, and the objective end of humanity, which he takes to ground the principle of
morality (Wood 2008, p. 85).
(Wood 2008, p. 85, my emphasis). A motive to obey the categorical imperative is consistent
with autonomy, just as the feeling of respect as an empirical motivation to perform right acts is
consistent with the purely formal rational origin of the moral law.
He describes this relation as the fundamental principle expresses, or in turn rests on, a
fundamental value but he also describes it in terms of a motive for the categorical imperative
(Wood 2008, p. 55). Humanity can function as that motive because it is the end that rational
beings qua rational beings have already as their subjective end subjective because it is the end
of each particular rational being rather than reason itself and rational agents are thus already
able to consider it in relation to their maxims for actions. Humanity is not an objective motive
until reason declares it to be (4:42829). In a footnote to this passage Kant defers defending his
claim that humanity has this objective status until Groundwork III (the nal section of this
chapter will examine the status of Kants claims in Groundwork III).

Value and the inexplicability of the practical

215

priori and both stem from the nature of reason. Kants argument for the
formula of humanity rests on the claim that one cannot have a categorical
imperative without a necessary end. There are, however, different ways to
understand the necessity of the end. First, necessary does not imply independent of the agent in a transcendental sense. Kants conception of the a
priori necessary and universal transcendental conditions for the possibility
of experience in the Critique of Pure Reason separates necessity and independence from the transcendental subject of experience. In parallel to this claim,
Kants requirement that there be a necessary end does not dictate that the end
exists independent of the moral agent. Second, the necessity of the end can
have two possible scopes: a necessary end of maxims and a necessary end of
the moral law itself. These two different scopes for the necessity of the end
allow for the categorical imperative itself to be merely formal (4:416) while at
the same time this formal categorical imperative can command an end for
maxims. A look at Kants summary of the formulas of the categorical imperative indicates that Kant intends the necessary end to be an end for maxims: he
says that all maxims have, namely . . . a form, which consists in
universality, . . . a matter, namely an end, and in this respect the formula says
that a rational being, as an end by its nature and hence as an end in itself, must
in every maxim serve as the limiting condition of all merely relative and
arbitrary ends (4:436, rst and last emphases are mine).
The nal result is precisely that ones maxims are limited by the value of
humanity as an end in itself. The way that the value of humanity functions is
not as the legislative basis for the formulation of humanity that a priori limits
the ability of reason to formulate the categorical imperative itself, (which
would violate autonomy). Rather, the very fact that reason is legislating a
categorical imperative raises the issue of the need for some corresponding end
that would operate in relation to maxims, and reason determines that this end is
humanity as a property of rational beings.
Another consideration in rejecting Woods claim that the independent value
is prior to the determination of the moral law is to note that Wood stresses
Kants argument for the principle of humanity, which is just one formula of the
categorical imperative, over the broader context of Kants initial argument for
the categorical imperative as such in Groundwork I and Groundwork II up
through the rst formulation at 4:421. But if one reads the argument for the
principle of humanity in light of Kants more general claim about the merely
formal nature of the categorical imperative, one can see that the value of
humanity is not understood as the basis for formulating the moral law but
only as a consequence of reasons legislation of that law. In that argument Kant
stresses that the categorical imperative concerns merely the form and not the
matter of action (4:416) and requires only conformity with law and not any
condition limiting law as such (4:412, 4:42021). It is not clear why Kant

216

Morality beyond nature?

would think that he is able to formulate the categorical imperative without any
reference to any end valuable in itself yet in a manner sufcient to derive and
state the categorical imperative if, as Wood has it, the categorical imperative is
grounded on this independent value in itself. Only after Kant has formulated
the categorical imperative does he then claim that the will requires an end valid
for all rational beings as a motivating ground for the wills selfdetermination (4:427). Kant holds that practical principles are formal if they
abstract from all subjective ends (4:427), that is to say, the categorical
imperative can still be formal if it abstracts not from every end but only from
subjective ends. The formal moral law can be compatible with an objective
motivating ground but not a subjective incentive. It is important to keep this
understanding of ground as motivating ground in mind when interpreting
Kants crucial following paragraph:
But suppose there were something the existence of which in itself has an absolute value,
something which as an end in itself could be a ground of determinate laws, then in it,
and in it alone, would lie the ground of a possible categorical imperative, that is, of a
practical law. (4:428)

When Kant uses the term ground here, it is best to read him as meaning
motivating ground of rather than basis for formulating the categorical
imperative. And in that case Kant is saying that any command to a rational
being to act in some particular way in general, to adopt a maxim that can
serve as a law requires that some end be taken into consideration. Since the
categorical imperative is a general command regarding the tness of maxims,
it requires a general end that reason can provide a priori. Humanity as a feature
of rational beings as such is this general end.
The successful argument from autonomy to the moral idealism of value rests
not upon a claim that humanity would be a heteronomous end to be produced
but upon a claim that it would be an a priori constraint on the legislative
autonomy of reason. An independent value of humanity as the basis of the
moral law would violate autonomy if that value were to be seen as shaping the
actual legislation of the categorical imperative by reason. If instead the value of
humanity is understood as objective, and so an end for all rational beings as
such, only when reason promulgates the categorical imperative and so requires
such an end as a constraint for particular maxims, then autonomy is not
threatened.
These considerations do not preclude some kind of empirical realism about
the value of humanity. The considerations Kant employs in the Groundwork
are at the transcendental level of an analysis of moral agency itself. The
autonomy of reason rules out that the value can precede the legislation of the
moral law but allows that the value is a precondition for the moral laws
existence: Kant says, If, then, there is to be a supreme practical principle and,

Value and the inexplicability of the practical

217

with respect to the human will, a categorical imperative, it must be one such
that, from the representation of what is necessarily an end for everyone
because it is an end in itself, it constitutes an objective principle of the will
and thus can serve as a universal practical law (4:428). A possible metaphysics of value compatible with autonomy would have the transcendental ideality
but empirical reality of an independent value of humanity. In the next section
I will show that the very idea of an intrinsic moral value of things, including
empirical things in nature, is inconsistent with other elements of Kants
philosophy. I also give a further reason to reject a transcendental realism about
value. Thus no empirically real value property is possible. The following
section will show that absolute value in Kant is only a formal ordering of ends
and not a property of objects.
Absolute value as a nonnatural property
Absolute value, I will show, is in an important way like the contingent value of
chosen ends and is not an intrinsic property of objects or even of moral beings.
In brief, just as contingently chosen ends have value only because particular
moral agents order their ends to reect their desires, humanity in rational
persons has value only because reason orders ends according to its moral law.
Since practical reason is what denes moral agency, the value of humanity is
dependent on moral agency and is thus transcendentally ideal and not transcendentally real. A transcendentally real absolute value would be intrinsic to
objects as they are independently of any relation to transcendental moral
agency.
Certainly some of Kants language suggests that humanity possesses value
as an intrinsic property of rational beings. Kant says that a rational being is
something the existence of which in itself has an absolute value and exists
as an end in itself (4:428, modied). But there is just as much textual evidence
that Kant understands value as dependent upon the moral law or upon reason.
He says that nothing can have a value other than that which the law determines for it (4:436, modied). That the former passages should be understood
under the rubric of the latter passage is suggested by the sentence immediately
after this last quotation, where Kant adds, But the lawgiving itself, which
determines all value, must for that very reason have a dignity, that is, an
unconditional, incomparable value (4:436, modied, my emphasis). The law
determines the value of everything, including itself as a condition for the
possibility of any other value.13
13

Johnson stresses that we should take Kant at his word that nothing means just that,
nothing and acknowledge that the value of lawmaking that determines value is a product of
the lawmaking (Johnson 2007, p. 143).

218

Morality beyond nature?

Dueling passages from Kant will not by themselves resolve the interpretive
issue.14 There are several reasons to hold that Kants understanding of the
value of humanity is not a property of objects or of beings independent of the
perspective of the moral agent. First, practical reason provides norms while
theoretical reason concerns ontology. Second, human beings would lack
knowledge of any independent value. Third, even if practical reason were to
be able to determine the existence of a value property in things, it would be
part of the practical point of view and not an independent ontology. Fourth,
even if the thing or property of absolute value were to exist independent of a
property of rational beings, the value of that thing or property would as
normative not t into the ontology of nature but would have to be part of the
practical point of view.
First, existence is prima facie not constituted by normative properties. The
divide between theoretical and practical reason as that between ontology and
normativity needs to be taken seriously as a basis for further understanding of
either. Practical reason might be able to determine that rational beings ought to
be treated as ends in themselves but cannot determine that rational beings
really are ends in themselves. This point is both basic and controversial. My
Chapter 2 on the nature of philosophy showed that the practical is concerned
with determination of free actions and not with the nature of what is.
Second, an independent absolute value could not exist in nature. Oliver
Sensen argues against value realism in part with an epistemological argument
that if value were independently real, human beings would have no access to it
and it would be able to play no part in our moral lives, but his reasoning also
works to show the ontological difculty.15 The ontology of physical nature is
constituted by the laws of nature applied to intuitions. Clearly absolute value is
not something that human beings can sense through outer intuition. Nor could
it be something that is a property of an outer object accessed through inner
intuition via feeling, for several reasons: it would lack necessity and universality as a product of feeling. Feeling also reveals to human beings more about
themselves than about the object because it can at best show that there is
something in the object that causes us to have certain feelings; it no more
reveals the relevant properties in the object any more than aesthetic feeling
reveals that objects possess beauty. Without laws of nature to connect the inner
intuition of feeling to any outer property of value we cannot include any value
property in an ontology of nature based on inner intuition. One might be able
to intuit the value of oneself through inner intuition, although Kant does not

14
15

For the assessment of relevant passages by a realist, see (Stern 2011b, pp. 2640) and by an
idealist, (Sensen 2011, pp. 213, 3951).
(Sensen 2011, 1920). In my paragraph I modify and expand his approach.

Value and the inexplicability of the practical

219

claim anything like this access to value and it would not make value an
independent property.
Kant does seem to allow for something like a direct moral awareness for
human beings in inner sense. The fact of reason is an experience of moral
obligation in the guise of the categorical imperative as the activity of reason.
This direct awareness, however, does not provide a way for moral agents to
access value properties in nature. The fact of reason itself is not an awareness
of value but of the categorical imperative. At best moral agents would be able
to reason to some independent value on the basis of their awareness of the
categorical imperative. I treat this possibility in more detail in the following
paragraphs. Here I will note that the fact of reason is also said to be the basis
for practical knowledge of freedom (freedom is indeed the ratio essendi of the
moral law, the moral law is the ratio cognoscendi of freedom (5:4)) but not
value. Further, the fact of reason, as discussed in Chapter 3, is part of the
practical point of view and is not intended by Kant to reveal any ontology. For
these reasons value cannot be an intrinsic property of empirical agents in
nature. There can be no empirical value property in nature on this basis.
Lack of value in nature does not rule out the existence of an independent
value of moral agents in themselves independent of nature, that is, transcendentally real as a nonnatural thing in itself, just as reason could in theory
determine that there must be a God. This leads to the third overall reason that
value cannot be an intrinsic property, namely that there would be no way that
moral agents could know such a property through reason and it would thus be
irrelevant for morality. Reason would have to know this property directly or
indirectly through a proof. Direct knowledge would be either active or passive.
Reason is certainly capable of direct knowledge of things that it actively
creates, such as ideas and principles, but if the reason of the transcendental
moral agent were to create the value, that value could not be an independent
intrinsic property of anything. And if value is understood as a product of the
activity of reason, it would thereby be ideal rather than real because it is
dependent on reason. Reason also has no way to passively access the independent property; any passivity on the part of the moral agent, and more
broadly on the part of cognitive beings like us, implies an activity on the part
of the thing to be known and must come to us through the mediation of our
intuition, bringing the argument back to the level of intuition and nature. So
reason would not be able to directly access any transcendentally independent
value property.
The only mode left for reason to know about an independent value property
is indirectly by proof. Kant spent a decade of his life and nearly 900 pages of
text working out his arguments in the Critique of Pure Reason that reason
cannot know anything by a priori proof. The best that reason can do is to create
ideas a priori that can be used heuristically to help to systematize our

220

Morality beyond nature?

knowledge of nature but that cannot themselves provide knowledge of any

actual objects or properties of objects. The limitations on reason appear to
forbid any indirect proof for an intrinsic value property. For Kant theoretical
reason is unable to determine any such ontology by itself because it would
extend its claims beyond possible experience.
But perhaps practical reason would be able to do this. In his discussion of
the postulates of practical reason Kant does claim that practical reason is
authorized to insist on certain claims that are beyond the authority of speculative reason (5:12021). I have already argued in Chapter 5 that the postulates
are best understood as making no ontological claims. Later in this chapter,
however, I will discuss the nature of reasons claims about the moral order in
Groundwork III and show more particularly that even if reason were to
postulate, as it were, that there was an absolute value, it would still be only
an idea of reason and not have any ontological import.
Practical reasons other possible proof is based on the requirements for the
categorical imperative itself. The nal possible way for moral agents to know
through reason that there is absolute independent value is to derive it from the
categorical imperative. This approach would not violate autonomy if it could
be shown that the categorical imperative requires that there be some independent intrinsic value of the objects of moral concern rather than the stronger
claim that the categorical imperative itself must be grounded on an independent value. The precise nature of the value that is required by the categorical
imperative will preclude this option.
This fourth and nal main reason that there can be no independent value
property centers on the precise nature of value Kant identies in his discussion
of the various formulas of the categorical imperative in Groundwork II. One
must not forget that there is a difference between a property itself and any
value of that property as a further property. Human beings are said to be of
absolute value in virtue of their humanity. Rational beings must exist, one
might say, so the value of the humanity of rational beings must lie in them. But
value and humanity are still different properties. Understood as a capacity for
practical freedom or for rational agency or for exercise of a good will,
humanity can be said to exist as part of nature explainable completely in terms
of psychological or physiological characteristics. If humanity can be specied
in these ways then it is at least empirically real. Suppose even that humanity is
also transcendentally real, that is, a property of moral agents as objects
independent of the transcendental moral subject. Such an ontology would by
itself prove nothing about any value properties. The value of humanity is an
additional property. The way that Kant species that value is to say that
humanity is an end in itself. Humanity is the characteristic of rational beings
that reason identies as of value (it is not value itself) and thus as also having
the property of being ends in themselves. Kant says, rational beings are called

Value and the inexplicability of the practical

221

persons because their nature already marks them out as an end in itself
(4:428, my emphasis).
To say that something is an end in itself is to say that it is valuable and vice
versa. This value as end in itself cannot be an intrinsic property of objects.
Being an end, like being a means, is not an intrinsic property of anything. Ends
are only determined by beings who are able to set ends. Empirical moral agents
set ends and by doing so mark them out as valuable. To say that a moral agent
values something is to say that the moral agent holds it as an end (or a means to
the end). Even when a moral agent recognizes herself as an end in itself she
does so qua moral subject in relation to herself qua moral object. The same
relation works at the transcendental level. Pure practical reason as the faculty
of a transcendental moral agent holds humanity to be an end in itself; this
constitutes the value of humanity. Reason does not insist that humanity is an
end in itself because it has some other value. The value of humanity is just
being held by reason as an end. Value as end in itself is not an intrinsic
property of beings with humanity but demands a relation to practical reason.
To some realists this conclusion makes Kants identication of humanity as
having value seem arbitrary. They might say that if humanity does not already
have value, then there would be no nonarbitrary reason for the categorical
imperative to assign it the status of the most valuable end. Thus they would
reject the entire project of examining how reason can create and assign value.
I think that this worry itself is misguided and that, in fact, in two ways the
charge of arbitrariness applies better to a value realist. First, if they are
allowing that there is a reason for the thing of value to have that value, then
their claims arbitrarily assume that it is incorrect for reason to assign value to
an object based on some nonvalue properties that the object has but that it
would be correct for the value itself in the object to be based on nonvalue
properties. Kant always supplies reasons why humanity is of value: because
the nature of rational beings marks them out as an end in itself (4:428), because
humanity possesses autonomy as the ability to give laws (4:43536), and so
on. The value is based on some nonvalue properties regardless of whether
reason assigns that value or it is intrinsic to the object. Since even if value is an
intrinsic property to objects, it still rests on some other features that the object
possesses, then why not allow reason to use the same basis for its assignment
of value to the object? The value realist does not have a nonarbitrary reason to
disallow reason from building value on nonvalue properties yet allow intrinsic
value properties to rest on nonvalue properties.
But perhaps the value realist would deny that there is any basis for the value
property possessed by humanity but instead claim that value is sui generis. In
this case, a second kind of arbitrariness arises. It would be arbitrary if there
were no reason for humanity to have value. In particular if it were the case that
humanity simply was valuable without further explanation possible, then there

222

Morality beyond nature?

would be an amazing coincidence that humanity, which includes the selflegislative ability of reason, would be the value that reason is forced to
recognize as the highest value. Reason would be in the state of commanding
all rational beings to respect the highest value, which just happens to be an
activity of reason itself. The value realist position can be described this way
because the value realist in this second kind of case would be denying not only
that reason is the source of assigning value but also that there is any explanation why humanity has value. And further, when the arbitrariness of intrinsic
value is described this way, it makes it clear why reasons assigning value to
humanity is not arbitrary. If reason assigns value to humanity, which includes
the exercise of reason, then reason is assigning value to its own manifestation.
What could be more appropriate for reason to assign value to than itself? Thus
there is a nonarbitrary reason for the faculty of reason to assign value to
humanity. A charge of arbitrariness for the idealist claim about reason as the
source of value is baseless. The tables can instead be turned against the value
realist who is implicated in not having a nonarbitrary reason for the claim that
humanity is of value.
In the following section I explain in more detail what value is in terms of the
order that reason imposes upon the empirical ends of moral agents through the
transcendental order that separates ends in themselves from contingent ends.
The order of reason
The key to understanding what absolute value is lies in stressing that practical
reason itself provides a normative order to things by means of the categorical
imperative. Autonomy implies that the will, as practical reason, itself determines its own law. The law in turn determines what is of absolute value, but
that means only that the actions and ends that are set by the law are ordered by
reason. Value is merely the formal ranking of ends by reason through its
promulgation of moral law. Kant refers to this ordering by reason as an
intelligible order, the formal ordering that practical reason provides to nature
through the categorical imperative. Any absolute value has to be understood in
relation to this formal ordering of ends by reason.
To understand this moral order of reason as a merely formal aspect one
should compare it to the individual empirical moral agents ordering of
contingent ends, such as having a certain job, eating fresh papaya, playing
the piano, or helping a certain charity. Each individual moral agent chooses
and ranks ends that she desires to pursue, thereby creating a subjective order
among the objects and events in nature that constitute either the means for ends
or those ends themselves. The value of these ends is not a property of empirical
objects or events but is merely a shorthand for their place in the formal order
that the agent imposes on nature: for an end to have any subjective value at all

Value and the inexplicability of the practical

223

is for that agent to include it somewhere in her hierarchy of ends, the greater or
lesser value of ends is their place in that agents hierarchy of ends. Kant puts
the point this way in the Groundwork: The ends that a rational being proposes
at his discretion as effects of his actions (material ends) are all only relative; for
only their mere relation to a specially constituted faculty of desire on the part
of the subject gives them their value (4:429, modied). Kant understands
value in a strictly deationary way as simply the fact that an agent has that end
and ranks it against other ends.
The same conception of value works at the transcendental level for the
absolute value of humanity in relation to the transcendental moral agent.
Instead of the myriad of particular contingent ends of individuals, the transcendental moral agent must abstract from content and face two possible kinds
of ends, necessary and contingent; the latter are simply the possible contingently chosen ends of moral agents. The necessary end is deemed by reason to
be humanity. Pure practical reason imposes its own order on the objects in
nature by ranking these two kinds of ends in the abstract. The ranking is
simple: humanity is always to be ranked higher than any contingent end and it
thus has value as an end in itself. This is the import of the formula of humanity,
but one clearer place where Kant describes this is in his claim that humanity
has dignity beyond all price (4:43435). A price is xed in relation to other
objects, some of which would be considered more desirable than others and
thus command a higher price. The dignity of humanity beyond all price means
that nothing could be the equivalent or be above humanity in price.16 No
appeal to any intrinsic value property is needed. To say that human beings are
to be treated as ends in themselves is to say that when comparing various ends,
there is none higher than that of humanity on the scale provided by reason.
16

Oliver Sensen has recently argued that Kants use of the term dignity reects its traditional
meaning as standing raised up above others and does not refer to any intrinsic property (Sensen
2011, pp. 14373). Sensen makes a detailed assessment of traditional and contemporary uses of
the term dignity and shows that the contemporary understanding of dignity as an intrinsic
value possessed by human beings was not operative in Kants time and is not reected in Kants
use of the term. The traditional meaning of dignity is expressed in the sense of a particular rank
that distinguishes one from others. A dignied person might avoid behavior she feels is beneath
her dignity. Human beings were said to have dignity within nature in virtue of their possession
of certain abilities that animals did not have, such as reason. Dignity did not indicate any
intrinsic value; rather, it indicated a special status that might then make certain treatment
appropriate, although the status and elevation does not require any moral valuation. Two
specic examples of this nonmoral kind of elevation are the dignity of a teacher and the
dignity of mathematics, both of which are terms Kant used (Religion, 6:162 and Pure Reason
A464/B492). This sense of dignity is relational rather than intrinsic. Something is dignied in
relation to other things that do not have the same elevated status. Nothing possesses dignity
intrinsically. Kants use of dignity in his moral writings, Sensen shows, is in line with the
traditional meaning and so need not indicate a reference to any intrinsic moral value. (Sensen
2009). I am indebted to Sensens work on dignity and value in formulating my similar
conception.

224

Morality beyond nature?

Reason does not recognize any preexisting value of humanity but reason
assigns that value to humanity through its ordering of ends.
All value, even absolute value, is thus for Kant ideal rather than real. In a
transcendental sense all value depends upon the transcendental moral subject
as a rational being subject to the categorical imperative imposed by reason
itself. In an empirical sense, absolute value is objective in that every rational
being must recognize it, but there is no independently empirically real value
property in persons or objects in nature. This idealism of value is also fully
compatible with a metaphysical naturalism in which the order imposed by
reason is merely a formal ordering of objects and persons in nature. The order
of reason is a reection of the formal character of the categorical imperative.
Through the categorical imperative pure practical reason imposes its own
intelligible order on nature without adding any content to nature except the
actions that reason itself causes.
This order imposed by reason is the same as the order Kant cites in the Third
Antinomy when he says that reason does not give in to those grounds which
are empirically given, and it does not follow the order of things as they are
presented in intuition, but with complete spontaneity it makes its own order
according to ideas in particular declaring certain actions to be necessary and
possible through the causality of reason (A548/B576). It is the same order
mentioned in Groundwork III as the idea of another order and another
lawgiving than that of the mechanism of nature (4:458). Kant also describes
an intelligible order in the second Critique (5:42, 5:8687, and 5:106). As an
ordering of objects, this order of reason is an imposition of form onto the
content of nature. The order of value is not intrinsic to nature but is determined
only through reason and applies to nature. The intelligible order of things is
not an order of intelligible things but an order of things in our experience,
particularly in relation to rational beings.
In the Canon of Pure Reason Kant denes the moral world as the world
as it would be if it were in conformity with all moral laws (as it can be in
accordance with the freedom of rational beings and should be in accordance
with the necessary laws of morality) but which is conceived thus far merely
as an intelligible world . . . [and] is therefore a mere, yet practical, idea which
really can and should have its inuence on the sensible world, in order to make
it agree as far as possible with this idea. (A808/B836). This intelligible world
is in essence an idea of the order of reason actualized. The formal order
imposed on the content of nature would result in a fully moral world in which
all of reasons value ordering is respected and all rational beings follow the
dictates of the categorical imperative. Kant calls the intelligible world an idea.
I discussed the nature of ideas of reason in Chapter 5, but it is worth noting that
in his initial discussion of ideas in the Critique of Pure Reason Kant uses
normative examples of ideas. Platos ideas were, Kant says, preeminently

Value and the inexplicability of the practical

225

practical. The examples of virtue and of a republic show that ideas can function
as paradigms or standards for actions (A31516/B37173). Through reason
these ideas can become efcient causes (of actions and their objects), namely
in morality and thus can have real effects on the world (A317/B374). The
moral world that Kant calls an idea in the Canon must be understood along
these lines as archetype, but it is more than simply a model since it also
includes the self-conception of moral beings as members of such a world.
In the Canon the moral world is also said to include moral beings. Kant goes
on to say of this moral world:
The idea of a moral world thus has objective reality, not as if it pertained to an object of
an intelligible intuition (for we cannot even think of such a thing), but as pertaining to
the sensible world, although as an object of pure reason in its practical use and a corpus
mysticum of the rational beings in it, insofar as their free power of choice under moral
laws has thoroughgoing systematic unity in itself as well as with the freedom of
everyone else. (A808/B837)

The moral world is not merely the order of ends or results of these actions but
also includes the rational beings themselves united in a kind of mystical body.
A full idea of a moral world, however, does not picture the rational beings
outside of or independent of their places in nature but only insofar as their free
powers of choice are in a systematic connection. The connection among
rational beings in choosing their actions systematically in accordance with
moral laws is just as formal as the lawful ordering by reason itself. Thus the
conception of an intelligible world is one that also includes rational beings
acting as moral agents in the world.
It is clear that moral agents at least represent themselves as being members
of the intelligible world when considering their actions in nature. Kant even
holds that rational beings in the sensible world are also by virtue of their
obligation to the moral law already members of the intelligible world as a
moral world. In the second Critique Kants declares that We are indeed
legislative members of a kingdom of morals [Reich der Sitten] possible
through freedom and represented to us by practical reason (5:82). The moral
world can be created through free acts in the sensible world; insofar as humans
actually do follow the categorical imperative they actually bring about the
partial transformation of the sensible world into a moral world as ordered by
reason.
The Critique of Practical Reason claries this use of the intelligible world
as a partial realization of the order of reason in the sensible world by distinguishing the concepts of the archetypal and ectypal worlds. The direct application of the moral order to the sensible world provides it with the form of an
intelligible world. This law is to furnish the sensible world, as a sensible
nature (in what concerns rational beings), with the form of an intelligible

226

Morality beyond nature?

world [Verstandeswelt], that is, of a supersensible nature, though without

infringing upon the mechanism of the former . . . supersensible nature, so far
as we can make for ourselves a concept of it, is nothing else than a nature
under the autonomy of pure practical reason (5:43).
The passage continues by explaining our application of the intelligible
world to the sensible world as consisting of two stages: the archetype
(natura archetypa), the idea according to which the sensible world ought
to be shaped, and the ectype (natura ectypa), the sensible world when it
receives the form of the intelligible world. Kant refers to the archetype as the
original (urbildliche) world and the ectype as the reproduced (nachgebildete) world. Moral agents confer on the sensible world the form of a whole
of rational beings (5:43). Rational agents possess an idea of the world as it
would be if it were governed solely by the moral law. They then apply this
idea to the sensible world in order to try to make the sensible world t this
model as closely as possible. Hence the ectypal world is the intelligible world
as applied to the sensible world. The ectypal world transforms the sensible
world into a moral world as far as is possible, but this is only possible as form
and not as content.17 But it is still only a view of the sensible world under an
intelligible, formal order. There are no material moral properties at stake
here, no content that does not depend upon the formal ordering of nature
demanded by the moral law and brought into existence only through the
actions of moral agents.
I have shown that Kant intended the intelligible world in his moral theory
to apply to nature rst by providing the idea of a normative order of practical
reason to use as a guide for actions and second to be realized in nature to the
extent that moral agents are actually able to act to bring about that order.
Since the application is merely formal, no content is imposed on nature;
rather, reason through the moral law provides an intelligible order of things in
nature and not an order of intelligible things outside nature. Moral properties
such as value are part of the formal, practical view of nature and are not
found independently in nature. Just as the particular value of contingently
chosen ends is not in nature but is merely an intelligible property conceived
17

Thomas Auxter argues that the ectypal world performs the function I have assigned to the
archetypal world, namely, that of providing an idea of a morally perfect world to reason (Auxter
1982, pp. 634). His reading relies on a claim that the archetypal world as an idea of reason
cannot include any sensible content and so requires an ectypal world to apply this abstract idea
of moral perfection to sensibility in general. The ectypal world is therefore for Auxter not an
interpretation of actual experience along moral lines but only an idea of a morally perfect world.
On this reading, however, no room is left for the archetypal world as an idea of reason. The
archetypal world, as a moral idea of reason excluding all sensible content, would consist merely
of form and would be the categorical imperative itself (5:33), but would no longer be an idea of
a world. Further, Auxter quotes but does not take into account Kants remark that the ectypal
world must exist in the sensible world (5:43).

Value and the inexplicability of the practical

227

by the particular moral agent who chooses that end, the objective value of the
necessary end of humanity is not in nature but is merely an intelligible
property conceived by reason as a part of its determination of the categorical
imperative. To say that value is an intelligible property is only shorthand for
saying that an individual (for subjective value) or reason itself (for objective
value) places it above other things in the order of ends each impose on
nature.
The transcendental moral idealist can still insist that there is a kind of
empirical realism about value, although as an objective standard and not as
an independent property. The basis of this claim would lie in the status of
reason in moral agency. A successful transcendental argument for moral
agency would show that all beings with a certain ability, such as the ability
to deliberate prior to choosing to act, would have to have other certain
characteristics, such as a particular structure to the deliberative process which
for Kant would be at least agent-perspective freedom governed by reason
through the moral law as a categorical imperative. Moral agents who are
thereby necessarily governed by the faculty of reason would also thereby
necessarily share the moral experience of an intelligible world, and the
intelligible world and its value properties could be said to be empirically
real in that sense. There would be no intrinsic value or good or right in
objects, but the source of those moral properties in pure practical reason
would give them a legitimacy beyond the individual empirical moral agent,
even beyond all possible empirical moral agents, just as I said in Chapter 4,
that one might consider pure practical reason as empirically real. The key to
this transcendental structure for Kant would be that for all beings, the
capacity to have moral experience would require that the being possess an
objectively valid faculty of reason. The possession of objectively valid reason
would be a fact in nature and the requirements of reason for value
understood as the formal ordering by reason could be said to be empirically
real as part of moral experience.
That empirical reality of an objectively valid reason relies upon a successful
argument for its transcendental necessity for moral agency. But as discussed in
Chapter 4, Kant denies in the second Critique that a deduction that will prove
the reality of the moral law is possible. Even more revealing is that even when
Kant appears to have offered what he took to be a successful deduction in
Groundwork III, his own reections on it actually reveal that he neither
intended nor believed that he had done any more than shown that empirical
moral agents must operate with a certain self-conception that is itself ultimately inexplicable and, further, that it is only imposed by the subjective needs
of human reason. This concession points to an empirical idealism for value as
an objective standard as well as for other elements of moral agency, as my next
section will show.

228

Morality beyond nature?

Practical experience and the limits of reason

The nal sentence of Groundwork III ends with Kant invoking a philosophy
that strives in its principles up to the very boundary of human reason (4:463).
The long section before the nal concluding paragraph is devoted to the topic
On the Outermost [uersten] Boundary of All Practical Philosophy (4:455,
modied). Most commentators who discuss the deduction fail to extend their
analysis to these nal sections in any detail.18 They miss Kants own reections on the status and results of the deduction. I contend that the success and
scope of the attempted deduction in Groundwork III must be judged in
reference to Kants discussion about this boundary. Specically, I will argue
that there is a strong parallel between the boundary discussed here and the
boundary for legitimate knowledge in theoretical philosophy such that practical philosophy is subject to much the same limitations as theoretical philosophy, and that the limitations show that the deduction presented earlier in
Groundwork III is intended to have a weaker conclusion than is usually
assumed.
The Groundwork III discussion of the intelligible, I will show, is not a claim
about nonnatural objects but a claim about the intelligible order of things as
determined by reason and the capability of human beings to conceive of
themselves as part of this order in virtue of possessing pure practical reason.
The world of rational beings and its laws and the freedom of human beings to
be determined by pure practical reason rest not on any claims about objects
independent of nature in space and time but on claims about the validity of
pure practical reason understood as a faculty that produces its own order for
nature. The categorical imperative is a valid moral principle if pure practical
reason can be shown to be a valid faculty. But Kant refrains from claiming that
the validity of pure practical reason is proved by the deduction. He instead
invokes the ultimate inscrutability of the unconditional necessity of the categorical imperative and thus the practical standpoint overall.
My claim regarding the section of Groundwork III on the outermost boundary of practical philosophy (4:45563) is that it functions to provide a reminder
that the deduction that has just been concluded must be understood in a
limited, practical sense rather than in any kind of theoretical or speculative
sense with ontological implications. Since much of Kants language in
Groundwork III seems to invoke theoretical claims about human nature, this

18

See, for example, (Wood and Schnecker 2004, p. 198f) (Schnecker is responsible for the
material on Groundwork III). Schnecker ends his analysis at the end of the Deduction (4:455).
The notable exception is Henry Allison, whose recent commentary includes a thorough
discussion of them (Allison 2011). I will note my agreements and disagreements with his
interpretation in further notes.

Value and the inexplicability of the practical

229

caution is needed. He often asserts that human beings are members of the
world of understanding (4:453) or things in themselves (4:451), or beings
whose pure activity must be distinct from the sensible world (4:451). The basic
argument Kant provides to escape from the famous apparent circle invokes
transcendental idealism as a ground for division of the self into that which
appears and that which does not appear (4:451452). Human beings, he says,
possess a purely active faculty of reason that justies us in holding that we do
not entirely belong to the world of sense but must also belong to an independent intelligible world. Kants section How is a Categorical Imperative Possible? invokes what looks like an ontological claim: the world of
understanding contains the ground of the world of sense and so too of its
laws (4:453). All these and many other passages point to a theoretical
conclusion about the actual nature of human beings, an ontological claim
about the nature of reality. Many commentators stress these passages in their
interpretations of Groundwork III.19
The proper way to understand the deduction in Groundwork III is to see it in
context. The context comes, rst, from the initial sections of Groundwork III
that set up the problem to be resolved; these sections receive detailed attention
in the commentaries and are familiar enough to everyone to leave unexamined
here. The second part of the context is in the section on the boundary of
practical philosophy given after the deduction itself. Particularly the material
in the last part of Groundwork III shows that the deduction is not to be
understood in ontological terms but instead as an ultimately unsatisfactory
defense of an idea of reason adopted as a self-conception by moral agents,
namely, that moral agents possess an objectively valid pure practical reason.
The term boundary can refer to two different things when understood
from a certain perspective: an inner and an outer boundary. One can consider
the atmosphere of earth, for example, to have an inner boundary where it
begins as the ground and water end and an outer boundary as the atmosphere
19

Dieter Henrich sets the tone in a series of articles in which he claims that Kant seeks to provide a
justication for morality he deems moral insight which must be essentially ontological.
(Henrich 1994). His direct work on Groundwork III is (Henrich 1975). Dieter Schnecker
makes what he calls Kants onto-ethical principle the lynchpin of his ontological interpretation of the argument (Schonecker 1999) and (Wood and Schnecker 2004). Paul Guyer sees
Kant as offering a metaphysical argument (Guyer 2009), where Guyer gives a strongly
metaphysical interpretation of the argument without delving into the material regarding the
boundary of practical philosophy. In earlier work Guyer allows that Kant backs away from a
claim that he has given a theoretical argument and instead assigns freedom the status of an idea
(Guyer 2007a, p. 167). Henry Allison stands out as questioning the validity of any metaphysical
reading. Allison is careful in sifting through the various claims and terminology in Groundwork
III to separate what he sees as a metaphysical reading of transcendental idealism from what ts
his own two-aspect interpretation. Allison admits that much of Kants language appears to
invoke ontology while arguing that a more plausible interpretation is nonmetaphysical (Allison
2011, pp. 34244).

230

Morality beyond nature?

thins out and outer space begins. One can picture oneself rising from the sea,
piercing the boundary between water and air, rising through the air, and
eventually leaving air behind and entering empty space. Similarly, although
a national border delimits territorial extent the same way in every location, an
individual relates to that boundary differently when entering and when leaving
a country, as would happen to a tourist driving through one country to reach
another. The specic term used by Kant, boundary of practical philosophy
will also refer to two different things. First, it can refer to a boundary between
theoretical or speculative philosophy and practical philosophy, the conceptual
division where theoretical claims leave off and practical claims begin. This
I will call the inner boundary; Kant refers to it in paragraph 5 of the section
(4:456) when he questions where the boundary of practical philosophy begins.
Second it can refer to the extreme legitimate use of practical philosophy itself
beyond which practical philosophy cannot venture. This latter is what Kant
calls the outermost boundary. I will look at the inner boundary and its
implications for the deduction before turning to the outer boundary and Kants
nal judgment on the deduction.
The inner boundary of practical reason
The general structure of Kants section on the boundaries of practical philosophy is this: the initial paragraphs (15) lay out the distinction between the
nature of the practical and the speculative to show that the practical begins
when ontological speculation ends and practical use of the result of speculation
begins. The middle paragraphs (68) discuss the nature of the practical as
positing an order of things different from that of nature, that is, the legitimate
space for the practical beyond its beginning boundary. The nal paragraphs
(915) concern the determination of the outermost boundary and discussion of
particular questions that can never be answered because they transgress the
boundary.20
Kant provides an initial basis for the inner boundary in the rst paragraph
when he argues that there is a difference between pure concepts of the
understanding, namely here causality, and ideas of reason, namely freedom.
The basis is that the former can be conrmed in experience and the latter not.
20

In his commentary on the Groundwork, Allison holds that only the nal third of this section
concerns the boundary issue. This claim fails to take seriously the way in which the rst
paragraphs of this section begin the argument by setting out the distinction between theoretical
and practical philosophy in relation to the use of reason. The preparation both settles the inner
boundary of practical reason and lays the ground for determination of the outermost boundary
as dependent upon the nature of reason and its ideas. Allison does see the middle part of this
section as concerned with the legitimate domain of practical philosophy (Allison 2011,
p. 348).

Value and the inexplicability of the practical

231

He concludes this paragraph with the claim that freedom is only an idea of
reason, the objective reality of which is in itself doubtful (4:455). This
distinction is important for two reasons. First, it immediately raises the question regarding how, and whether, the freedom discussed in Groundwork III
and related ideas can be conrmed if not through experience. This issue is at
the heart of Kants rejection of a deduction in the second Critique, and in the
Groundwork Kant concludes that they cannot be conrmed but only ascribed
to oneself as part of our self-conception as rational agents. Second, the
identication of this freedom with an idea of reason already casts into doubt
the interpretation that assumes that Kant is making an ontological claim in his
deduction. In fact, looking back at the conclusion of Kants deduction in
Groundwork III, one sees that Kant already invoked the term idea: categorical imperatives are said to be possible because the idea of freedom not
freedom itself makes us members of an intelligible world; and that beyond
our sensibly affected will we add the idea of the same will but belonging to
the world of understanding, again not a claim that our will is actually such a
pure will in any sense (4:454, my emphasis). The ideas of reason in the
Dialectic of the Critique of Pure Reason were said not to have any reality of
their own but to be illusory, and any positive use of the ideas would come
merely from their immanent use in aiding the understandings task of cognizing our experience. We should expect, then, that Kant is working toward a
claim that the self-ascription of freedom to our wills does not itself involve
claims to reality but instead will aid in comprehending the nature of our
experience as active beings.
The distinction between sensibly conrmed concepts of understanding and
ideas of reason, which are incapable of being conrmed, is not itself the inner
boundary of the practical because speculative philosophy also uses ideas of
reason. What, then, is the proper inner boundary for practical philosophy? 21
Let us compare the roles of speculative and practical philosophy to nd it.
As already noted, Kant presents both speculative and practical reason as
ranging beyond the eld of nature itself, that is, beyond the concepts of the
understanding, to the realm of ideas. The boundary between the sensible and
intelligible is not the inner boundary of the practical.
Kant has also presented them as facing the same difculty: the clash between
freedom and natural causality. The thesis and antithesis arguments in the Third
Antinomy of the rst Critique exhibit the purely speculative origin of this
issue; the practical origin of the issue is shown in the preceding paragraphs of
Groundwork III. The appearance of this issue in both practical and speculative
philosophy shows that the issue itself cannot provide the boundary.
21

Kant uses the terms speculative philosophy and speculative reason, and correspondingly
practical philosophy and practical reason, interchangeably in Groundwork III.

232

Morality beyond nature?

Kant also notes that both practical and speculative philosophy insist on
resolving the issue, although with a difference that one might mistake for a
boundary. Neither practical nor speculative reason can simply leave the
problem unsolved. Speculative reason is said in Groundwork III to need to
resolve the problem in order to clear the way for practical philosophy.
Practical philosophy must resolve the problem in order to defend morality
against the fatalist who otherwise would chase away moral science itself
(4:456) and from similar external attacks (4:457). This discussion appears
to introduce an asymmetry that could provide the basis for the boundary:
perhaps the boundary of practical reason begins with the insistence that the
clash between freedom and nature be resolved; perhaps speculative reason
by itself has no interest in this matter. But this is not true: in the Critique of
Pure Reason Kant insisted that speculative reason resolve the antinomies to
defend reason itself from stubborn dogmatism and a hopeless skepticism
(A407/B434). Further, if the insistence on solving the problem were to lie
merely with practical philosophy, that fact alone could not constitute a
boundary but merely be an effect of whatever it is behind the inner
boundary that motivates the insistence. Even if Kant intended to claim in
the Groundwork, contra the rst Critique, that only practical reason has an
interest in resolving the conict, it would provide no more than a hint at the
actual boundary.
Kant offers another difference between practical and speculative philosophy
that points indirectly toward the inner boundary: he assigns speculative but not
practical philosophy the task of resolving this issue (4:456). This is not itself a
boundary: Kant is quite explicit that it cannot yet be said here that the
boundary of practical philosophy begins (4:456).22 By this he means that
the task of resolving the problem is not within the boundaries of practical
philosophy. One might say that this task is, however, the nal geographical
feature in the territory of speculative philosophy.
What then constitutes the inner boundary of practical philosophy? It must
concern whatever motivates practical reason to insist on a resolution to the
clash between freedom and nature and also relate to the exclusive ability of
speculative reason to try to resolve that clash. It must lie in the difference
between the types of question asked by speculative and practical reason, and
the resulting different ways that these questions can be answered. It must then
be the fact that while speculative reason considers what is, practical reason
asks what ought to be. Practical reason then has the motivation to resolve the

22

Jens Timmermann treats this paragraph as if Kant were questioning whether the outer boundary
of practical philosophy begins here, then notes that the beginning is more appropriate
(Timmermann 2007, p. 146).

Value and the inexplicability of the practical

233

issue because human beings ought to be free in order to be moral, yet it does
not have the capability of resolving the issue by itself because it cannot
determine what is the case.
Here, then, is the inner boundary of practical philosophy. The inner boundary of practical reason in Groundwork III must be the point at which mere
speculation about what is recedes and the moral ought requires the possible
truth of some claim about freedom that entails overstepping the bounds of
nature. This boundary begins where a self-ascription for moral purposes
clashes with nature. One might say that the inner boundary of practical reason
is determined by a) the line between experience as subject to concepts of the
understanding and the content of ideas of reason that go beyond experience,
and b) the requirements demanded by the ought. When one takes up the
practical perspective of the ought in a way that results in an apparent clash
with the concepts of the understanding, one has crossed the inner boundary
and entered the territory of practical philosophy.23
The specic self-ascription that Kant discusses is that of an independence
of reason from subjectively determined causation so that we may consider
ourselves as subject to the demands of reason itself (4:457). I provided
reasons for this interpretation in Chapter 4. This independence of reason
would be possible only if we could consider ourselves in a different order
of things and in a relation to determining grounds of an entirely different
kind than the order of nature and its determining grounds in desires and
sensible stimulations. This is precisely the order of things at stake in my
discussion of value and moral experience in the rst two sections of this
chapter. Since the practical self-ascription of a free will (i.e., causality by
reason) apparently clashes with causality in nature yet is inextricably linked
to the claim that we ought to have a free will, it clearly lies in the territory
of the practical.
The import of this determination of the inner boundary is that it helps to
explain the status of the conclusion of the deduction. When answering the
question How is a categorical imperative possible? (4:454), Kant explains
that the idea of freedom makes us members of the intelligible world, and that
we use the idea of the will belonging to the world of understanding. That these
are mere ideas and not actual empirically valid concepts of the understanding
places them in the realm of ideas the objective reality of which is in itself
doubtful (4:455). We human beings, for practical purposes, must think of
ourselves as endowed with a free will that responds to the dictates of pure
reason that itself gives the law. And the territory in which we adopt practical
23

The discussion about the inner boundary of practical philosophy concerns pure practical
philosophy and not empirical or anthropological practical philosophy, which can be considered
part of nature.

234

Morality beyond nature?

ideas is the territory of practical philosophy. Having seen that territory from
the perspective of its inner boundary, Kant turns to its other frontier: the outer
boundary of practical philosophy.
The outer boundary and the incomprehensibility of the practical
The inner boundary shows where practical philosophy starts and to that extent
provides part of the characterization of the nature of practical philosophy. The
outer boundary will complete that characterization by showing the limits of
practical philosophy. As limited, I will show, practical philosophy is necessarily incomplete. There are two reasons for the incompleteness of the ideas of
reason for practical justication.24 First, they cannot themselves be explained
and to that extent remain skeletal assertions rather than fully eshed out
justications. Second, because they are ideas of reason, they are subject to
the restrictions on reason itself regarding its search for the unconditioned. Both
of these reasons point to limitations for the results of the deduction in
Groundwork III.
Kant identies one aspect of the outer boundary in one sentence: By
thinking itself into a world of understanding practical reason does not at all
overstep its boundaries, but it would certainly do so if it wanted to look or feel
itself into it (4:458, modied). Although we are justied on practical grounds
in adopting the idea of a free will and pure reason, we do not know that we are
intelligences, still less do we know how we could be intelligences. To adopt
the idea of the intelligible world is not to hold to its reality. Later in the same
paragraph Kant insists that the intelligible world is only a standpoint that
reason sees itself constrained to take for practical purposes, and that the idea
of the intelligible world (the idea of another order and another lawgiving
than the natural order) is conceived under its formal condition, that is, law as
universality of a maxim (4:458). To go beyond this mere practical standpoint
as idea, and its only applicable practical use in defending the validity of reason
in commanding the will to follow its formal condition of universality, is to
pierce the outer boundary of the practical.
Specically, any attempt to explain how this reason and this will are
possible violates the boundary. Kants explanation of this restriction links
practical reason to speculative reason again, since both attempt to provide
24

Henrich identies two ways that Kant presents the outermost boundary: rst, as the incomprehensibility of how freedom is possible within the intelligible world, and second, as the
incomprehensibility of any relation between freedom in the intelligible world and an interest in
the sensible world (Henrich 1975, pp. 745). He admits that Kant treats these issues as identical
(4:45960) but takes them to be distinct because only one concerns the validity of and the other
motivation for the moral law. But this distinction is irrelevant since Kants point is that nothing
about the intelligible world is comprehensible, whether it relates to the sensible world or not.

Value and the inexplicability of the practical

235

ideas that cannot be conrmed by experience: Where determination by laws

of nature ceases, there all explanation ceases as well, and nothing is left but
defense (4:459). There is no possible experience that could correspond to the
idea of freedom, that is, the activity of pure practical reason, and therefore no
explanation of freedom is possible. The idea of freedom is only a necessary
presupposition of reason in a being that believes itself to be conscious of a
will (4:459, my emphasis). The defense consists in the assertion of ideas that
are themselves incapable of any explanation.25
This identication of the outer boundary shows the limits of the Groundwork III deduction in particular but also of practical philosophy in general.
Since no explanation of the ideas of practical reason can be given, those ideas
must be only a partial justication for our human moral practices. These ideas
can never serve to fully justify morality because there will always remain
unanswerable questions about these ideas.
The rst of the two limitations on the deduction resulting from the identication of the outer boundary, then, is the inexplicability of the idea of a pure
practical reason as a free causality. This limitation means that when the
deduction invokes the concept of an intelligible world in which we would
possess pure practical reason, the concept cannot be subjected to detailed
analysis of how pure practical reason and free will are to be understood, nor
even details about how the intelligible world could exist as such. The conclusion of the deduction is only for defense of our claim to be moral beings, not
for any positive complete proof that we are such.26
The second limitation on the deduction resulting from the outer boundary is
of even greater consequence.
In the nal paragraph of Groundwork III Kant again invokes a similarity
between practical and speculative reason: both insist on absolute necessity but
are unable to provide it. Speculative reasons concern for nature raises the idea
of the absolute necessity of some supreme cause of the world while practical
reasons concern for freedom leads to an absolute necessity, but only of laws
of actions of a rational being as such (4:463). The difference between
speculative and practical reason as a difference between is and ought is

25

26

In his extensive notes to his translation of the Groundwork into Portuguese, Guido de Almeida
similarly notes the difference between a claim that it is necessary for human beings to
presuppose freedom on the one hand and the possibility of that freedom which is presupposed
on the other. In other words, Kant allows that practical philosophy requires us to adopt ideas
that themselves refer to things that are incomprehensible and perhaps not even possible (Kant
2009, pp. 42728).
Allison describes this conclusion as an admission that Kant has shown only one necessary and
not all the sufcient conditions for the possibility of the categorical imperative (Allison 2011,
p. 359). I would stress that even this one necessary condition is only assumed as possible
because it cannot itself be sufciently explained.

236

Morality beyond nature?

apparent here, but the similarity is more important. Both insist on absolute
necessity. The import is explained by Kant as follows:
Now, it is an essential principle of every use of our reason to push its cognition to
consciousness of its necessity (for without this it would not be cognition on the part of
reason). It is, however, an equally essential limitation of this same reason that it can see
neither the necessity of what is or what happens nor the necessity of what ought to
happen unless a condition under which it is and happens or ought to happen is put at the
basis of this. (4:461)

Practical reason and speculative reason share the property of seeking a condition for everything in their explanations. In order to nd an end to this restless
search, they both create an idea of something that is unconditioned, or necessary. Kants explanation in the Critique of Pure Reason of reasons search for
the unconditioned (A32138/B37796) is treated in the Groundwork as applying to practical reason as well. All of reasons ideas are to be understood as
attempts by reason to provide an unconditioned explanation for something.27
Practical reasons idea of an absolutely necessary moral law is mentioned in
this paragraph, but it is not the only practical idea that Kant considers at the
time he wrote the Groundwork. Kant makes a similar point in the Naturrecht
Feyerabend lecture, given summer 1784 when he was writing the Groundwork. The initial fth portion of this lecture course, labeled as Introduction,
provides an overview of some of the claims of the Groundwork, in particular
showing the importance of freedom and the way in which universality and
ends-in-themselves are applied to questions of political right. At one point
early in the lecture Kant argues that in practical philosophy there must be
something necessary, the necessity of which is inconceivable. But rather than
an unconditioned necessary law as in the Groundwork, Kant here discusses an
unconditioned necessary end:
That something must exist as an end in itself and that not everything can exist merely as
a means is as necessary in the system of ends as Ens a se [a being in itself] is in the
series of efcient causes. A thing that is an end in itself is Bonum a se [a good in itself].
What can be considered merely as a means has value merely as a means when it is used
as such. Now for this there must be a being that is an end in itself. One thing in nature is
a means for another; that continues on and on, and it is necessary in the end to think of a
thing that is itself an end, otherwise the series would have no end. (27:1321)
27

Allison claims that the position Kant argues for in the concluding remarks is unique to the
Groundwork (Allison 2011, pp. 36162). Allison claims that the quest for the unconditioned in
practical philosophy does not appear in either the rst or second Critique. Allison is overlooking the key role of the limits of explanation to the intelligible character of reason in the
resolution to the Third Antinomy. As I show later on and have already discussed in Chapter 4,
while Kant does not name pure practical reason as such, he does describe free causality in terms
of the causality of reason and argues that the causality of reason is simply a given that cannot be
questioned further.

Value and the inexplicability of the practical

237

Although Kant discusses a necessary end in the system of ends rather than
a necessary law, the intent is similar.28 As in the Groundwork, Kant
compares a practical necessity with a speculative concept of a necessary
being as cause. Both are conclusions to which reason is drawn when
considering a series of conditions. In the case of morality it is a meansends series of justications; in the speculative it is the cause-effect series.
In parallel with the Groundwork we should expect Kant to invoke the
incomprehensibility of this necessity as a function of the nature of reason.
Kant does precisely that:
But how a being can be in itself just an end and never a means, is as hard to conceive as
how in the series of causes there would have to be a necessary being. And yet we must
accept both because of the needs of our reason to have everything complete. It lies in the
nature of human reason that it can never have insight into something except as
conditioned, never have insight into something without a ground, and there is no
ground for ens and bonum a se [being and good in themselves]. (27:1321, my
emphasis)

In his Feyerabend lecture, then, Kant is using the same approach as in the
Groundwork by showing how both speculative and practical reason, qua
reason, posit something necessary and unconditioned for the sake of completeness, yet remain dissatised with the result because the necessary being or end
is not fully explained through any conditions of its own.29
These passages from Feyerabend and the Groundwork are in fundamental
continuity with the nal paragraphs of the resolution to the Third Antinomy
where Kant says that there can be no further question of what causes reason to
have the intelligible character that it has. The two unanswerable questions
regarding the causality of reason are rst, why has reason not determined
itself otherwise? and why has [reason] not determined appearances
28
29

Paul Guyer also quotes from this passage from Feyerabend but interprets in in light of his claim
that freedom is the thing valued in itself (Guyer 1998, p. 33).
Without citing and probably without being aware of these passages from the Feyerabend
lecture, Christine Korsgaard attributes a similar kind of reasoning to Kant regarding the way
that pure reasons search for the unconditioned is worked out in both theoretical and
practical reason (Korsgaard 1996a, pp. 117, 119). She mistakenly, however, emphasizes a
difference in that theoretical reason inevitably runs into antinomies while practical reason
does not, which she takes to allow practical reason the liberty to use the idea of the
unconditioned positively. It is true that theoretical reason runs into antinomies when it
demands an unconditioned, but this is true only when the demand is to nd the unconditioned in experience (taken as a transcendentally real world). Reasons search for the
unconditioned also causes it to demand an absolute subject (soul) and absolute being
(God); in addition the solutions to the dynamical antinomies posit an unconditioned cause
and unconditionally necessary being. The parallel between theoretical and practical is more
important: the ideas of an unconditioned cause, unconditioned subject (soul), unconditioned
end, or unconditioned law are all created by reason as placeholders for its own satisfaction
and cannot be said to have any corresponding object in nature.

238

Morality beyond nature?

otherwise through its causality? (A556/B584). Neither question can be

answered. Kant concludes that when looking for the causality of free actions
we can only take one step, that is, attribute them to the intelligible cause, but
we cannot discover any further condition for this causality of reason. This
inquiry surpasses every faculty of our reason, indeed, it surpasses the authority of our reason even to ask it (A557/B585).30
These two other instances of Kant discussing the outer limits of the capacity
of reason to provide a completely unconditioned or necessary basis for moral
claims supplement Kants admission in the nal paragraph of the Groundwork
that reason cannot ever be completely satised with a claim that the moral law
is itself necessary and unconditioned. Reason has only been able to assume it
without any means of making it comprehensible to itself (4:463). This, then,
is the second limitation given by the outer boundary of practical philosophy to
the results of the deduction in Groundwork III. The nature of reason itself
precludes any completely successful deduction of the moral law. Since the
issue to be defended in the deduction is not empirical, it must be resolved by
reason. But as reason it always strives for more explanation that it can provide.
The ideas of reason in the rst Critique, soul, world, and God, and in the
Groundwork, freedom and its connected absolute necessity of the moral law,
and even the postulates in the second Critique are merely heuristic devices to
represent that which reason seeks. At some point in its attempts to answer
questions about the basis for a claim, reason posits that there is an answer,
represented by the idea, but can never provide a sufcient defense of that
posited answer. In the case of the moral law, reason seeks to understand
absolutely necessary obligation but can go no further than to posit that there
is an absolutely necessary moral law. Reason can offer no further explanation
for this law. Kants nal sentence admits that this is all that can fairly be
required of a philosophy that strives in its principles to the very boundary of
human reason (4:463).
30

There is one difference between this conclusion and those in Feyerabend and Groundwork. In
the latter Kant makes an analogy between the practical problem of searching for a condition for
a necessary end or necessary law and the speculative search for a being in itself or supreme
cause of the world. In the rst Critique, however, Kant makes an analogy between reason as the
unquestionable free cause and space as the outer form of human intuition. (Kant makes a similar
analogy between moral law and the forms of intuition in a note dated during the 1780s [R7201,
19:275].) In both cases, Kant says, we can give no reason why they and no others ll the role
they do. This analogy is much weaker than the others and misleads the reader. Space is not in
any sense a condition in a series whereas reason as free cause is a condition in a series of causes
and effects. One is not driven by the nature of reason to ask why human beings have the forms
of intuition that they do have; but one is so driven to ask for further conditions for any cause in a
series. The fundamental point of the analogy, however, still holds, namely that just as there is no
explanation why we have space as a form of intuition instead of some other, to us inconceivable, form, there is no explanation why we have reason as opposed to some other, to us
inconceivable, faculty.

Value and the inexplicability of the practical

239

Concluding remarks
At the end of Groundwork III, Kants nal position on his achievement in the
deduction amounts to a warning not to take the conclusion to be more than a
practical idea that human beings can use to help to justify their moral lives,
provided that they do not venture too far in their quest for that justication.
Even in the place where Kant is supposed to have provided his strongest
argument in favor of a theoretical proof of the reality of morality, he does
not conclude more than that human moral agents must conceive of themselves
as beings endowed with an objectively valid reason. Only as a self-conception,
only from within the empirical moral agents own practical point of view is this
idea even operative, and then without any ontological implications that would
conrm empirical reality to the objective rationality of other moral agents.
In very broad terms, we should expect no different from Kant than the
restriction of morality to the empirical moral agent and thus to an empirical
moral idealism. Since morality is a product of reason, we should expect that
morality would face the same boundaries of explanation that theoretical
knowledge faces given the limitations of reason. More important, we should
be prepared to accept with Kant that reason alone cannot provide knowledge of
real things, which is the same as admitting that we should not assume that the
universe is fundamentally rational. The main point of the Critique of Pure
Reason is that reason cannot divulge reality to us. We can be guaranteed to
know the structure only of the kind of world we can experience, that is, nature
in space and time. The categories of the understanding and the forms of
intuition can be known to be actual structures of our experience only because
they can be shown necessary for our kind of experience. They have no
conrmable validity beyond that limitation. Reason can add to the empirical
knowledge of the world only the systematic organization of concepts of the
understanding with the caveat that this systematic organization reects the
structure of reason and not the structure inherent in the objects of experience.
Reasons limitations are greater outside the boundary of experience. Reason
cannot provide knowledge of the nature of things in general nor prove one way
or the other whether there is a soul, a God, or a free cause. Were the universe
itself rational, were it structured by reason, then the faculty of reason would be
able to condently and correctly provide answers to these questions. But it
cannot.
Nor would the practical use of reason be able to provide this access to things
in general or even things in our experience. Unlike the concepts of the
understanding, the principles of reason do not even purport to refer to objects
in nature, and so are not amenable to any empirical conrmation or any
independent standard. If concepts of the understanding do not match nature,
nature will so to speak push back. Reasons moral principles govern actions

240

Morality beyond nature?

not objects. There is no independent standard to determine whether the

objective principle of the moral law of reason with its declaration of the
absolute value of humanity ts the circumstances; there is no way to test
whether the moral law matches reality, for there is no independent standard
against which the moral principles of reason can be judged. (Prudential
reasoning, in contrast, would have a standard, namely whether the end determined independently of reason is attained or not using the means-ends
reasoning.) Morality cannot push back against reason because it is itself merely
a product of reason.
This status of reason and its relation to reality makes for a genuine difference between those interpretations of Kant that see him as a moral realist and
those that see him as a moral idealist, even when both sides agree that the
standards of morality are determined by reason. When the relation between
reason and reality is severed, Kant cannot be properly understood as a moral
realist. Reason provides a view of experience subjectively grounded in the
nature of reason itself but not objectively grounded in any independent objects.
The moral realist, particularly a moral realist about value, would have to take
Kant to hold that reason can provide knowledge of reality. Thus when some
conclude that the moral realist and constructivist are involved in a merely
verbal dispute31 or that Kants ethics does not seem to differentiate between
morality as an assumed standpoint and morality as real,32 they are brushing
over a fundamental difference in the two interpretations. The idealist accepts as
fundamental the limitations of reason shown by Kants philosophy, which he
dened as the science of the relation of all cognition to the essential ends of
human reason (A839/B867, my emphasis).

31

32

Formosa reaches this conclusion with regard to the realists talk of the moral law stemming
from rational nature and the constructivists talk of the constitution of self-legislating beings.
Just one paragraph earlier he had identied the realist discussion of the moral law as lying
in the nature of things ([Formosa 2013, 189] quoting [Kain 2004]). When realists talk of
rational nature, they tend to mean not simply a type of being that happens to exist in nature
with some particular way of structuring its own experience but instead a being who is able
fundamentally to discern what morality is in the universe because the faculty is not in any way
arbitrary.
Allen Wood admits that his interpretation of Kants deduction of the moral law shows that for
Kant there is no difference between the actuality of the moral law and our belief in that actuality
(Wood 1999, p. 381, n. 30). This admission seems inconsistent with Woods value realism.

Postscript: Kants naturalist moral idealism

I have presented an argument for two closely related features of Kants

metaethics: metaphysical naturalism and moral idealism. In this postscript
I will collect together the various elements of this interpretation and point
out the hinge points between the realist and idealist interpretations. In my rst
chapter I identied eight elements of Kants ethics that need to be evaluated in
order to determine the extent to which Kant could be a realist; these also serve
for an assessment of naturalism. They are as follows:
A.
B.
C.
D.
E.
F.
G.
H.

The value of particular chosen ends

Particular moral duties or norms
The value of humanity as end in itself
The highest good and the postulates of practical reason
Moral obligation
The moral law itself
Practical reason itself
Freedom of choice

Kant as metaphysical naturalist in ethics

In my rst chapter, I distinguished metaphysical and methodological naturalism and limited myself to the claim that Kants ethics can be interpreted in a
metaphysically naturalistic way.
I am not claiming that Kant can be interpreted as a methodological
naturalist. By metaphysical naturalism I mean the type of naturalism which
accepts as real only entities studied by the sciences for Kant, nature in space
and time and by methodological naturalism the type of naturalism which
accepts as sound only the procedures for acquiring knowledge used in the
sciences. Kant is clearly not a methodological naturalist because his transcendental method of inquiry and stress on a priori knowledge and principles goes
beyond the more empirically constrained methods of the sciences.
I am also not taking a position on whether Kant is a metaphysical naturalist
in his theoretical philosophy. I believe that his transcendental method could be
241

242

Postscript

used in such a way, and I have used in this book the example of the
transcendental justication of causal relations in nature and the requirement
that any being who would represent that nature objectively must have a
cognitive system that processes sensations using causal relations. But there
are also signicant portions of Kants theoretical philosophy that involve
ontological claims absent in the practical philosophy, in particular the nature
of the spatiality and temporality of things. My claim is restricted to Kants
practical philosophy.
Nor am I claiming that I have presented precisely the historical Kants
position. Given the scope of Kants ambitions and the inconsistent use of
some of his terminology no interpretation can claim to comprehend all of
Kants stated claims and arguments. I offer an interpretation that encompasses
the main claims in Kants ethics and shows how Kant either did or could have,
using the resources available within his practical philosophy, presented an
entirely metaphysically naturalistic ethics. Relatedly, I do not make comprehensive claims about his philosophical development but only review the
development of some issues to draw out some material that I think has bearing
on interpreting his core critical works.
I do offer a plausible and defensible interpretation of Kants ethics along
metaphysically naturalistic lines. Given Kants distinction between physics
and psychology as two disciplines that study matter and thinking nature
respectively, his denial that the latter can rise to the level of a science, and
his rejection of a reduction of the psychological to the material, it would appear
that empirical psychology would not fall under metaphysical naturalism. But
given Kants equally strong insistence that empirical psychology is a possible
discipline short of proper science, his allowance of inner intuition as a source
of content governed by time as a form of intuition as well as the pure concepts
of the understanding, his strict empirical determinism regarding decisions
considered as appearance or phenomena, I allowed that the subject matter of
empirical psychology should be included in Kants conception of nature for
these ontological purposes despite lack of epistemic access.
With regard to entities, the elements of Kants moral theory that appear to
require an appeal to things in themselves independent of nature in space and
time are the transcendentally free faculty of pure practical reason itself, the
existence of God and immortality of the soul as postulates, and the transcendentally free faculty of choice. The elements that appear to require nonnatural
properties are the values of ends, both chosen by particular moral agents and
necessary as ends in themselves.
The transcendentally free faculty of pure practical reason (G) is not itself an
object per se but as a faculty it would have to reside in a being, and commonly
this being is understood as a person in herself independent of nature. I argue
that pure practical reason can t within nature because it is understood as the

Postscript: Kants naturalist moral idealism

243

same faculty of reason posited as part of the empirical mind in empirical

psychology. Considered as a structure for ordering concepts and principles,
reason has a transcendental justication that defends its validity in promulgating the moral law. This transcendental justication does not require any
nonnatural metaphysics. The faculty of reason acts as an efcient cause
through what Kant calls motives. These motives can also be understood within
the connes of empirical psychology. The empirical power of choice is determined in its decision either through this faculty of reason or through sensible
impulses and so too falls under empirical psychology; this is practical freedom.
The naturalistic interpretation of reason grounds a naturalistic interpretation
of several other elements as well. Pure practical reason is the source of the
moral law itself (F) and through that law also the source of particular duties
and norms (B) in combination with information about human nature. Reason is
also the source of the idea of the highest good (D). Moral obligation (E) comes
to human beings as the categorical imperative as a command of reason. My
argument that the fact of reason is empirical consciousness of the moral law
clearly places it in nature. Conrmation of the validity of that obligation would
come from the transcendental defense of pure practical reason as the source of
the fact of reason.
The postulates of God and immortality of the soul (D) purport to refer to
nonnatural entities. If Kants postulates are interpreted as ontological claims
then this part of Kants theory is not naturalizable. I showed how Kants use
of the postulate of God is more in line with his theoretical use of hypotheses
and ideas than some of his declarations would suggest. The concept God has
immanent reference to the determination of the moral agents actions rather
than transcendent reference to a supernatural being. As concepts the postulates
can be said to exist only in the empirical agents mind as part of a moral agents
practical point of view.
The transcendentally free faculty of choice (H) is the biggest stumbling
block to a naturalist interpretation. The empirical faculty of choice in nature
cannot be understood to possess transcendental freedom in the structural way
that the empirical faculty of reason can be said to possess transcendental
freedom. The empirical faculty of choice is still practically free in that it can
be necessitated in its decision by reason in accordance with the deterministic
causal relations described by empirical psychology. In order to conceive of a
power of choice that is not necessitated in its decision, one has to conceive of
an agent in herself making a timeless fundamental choice independent of nature
that determines the entirely of the agents empirical manifestation, in essence
a single transcendent choice. The naturalist can incorporate this need for the
transcendental freedom of choice because Kant identies it as a postulate,
which have been shown to have no ontological consequences, even taking into
consideration the different status freedom has relative to the other postulates.

244

Postscript

With regard to moral properties, the value of humanity as an end in itself (C)
would appear to be incompatible with Kants ontology of nature. Contingent
values of particular ends (A) depend upon empirical moral agents desires and
are thus clearly within nature because they are dependent on the empirical
subject. The value of humanity, however, is often assumed to be a necessary
property of the objects rational beings independent of the empirical subject.
Kants ontology of nature has no place for such intrinsic moral properties.
I showed that the value of humanity is dependent upon the transcendental
moral subject (a being with transcendentally justied practical reason) in a way
parallel to the dependence of contingent values on empirical subjects. To have
absolute value is only to have the highest rank in practical reasons ordering of
ends. Practical reason issues this ranking through the categorical imperative, so
the value is not inherent in objects as any intrinsic property but is only a formal
framework for experience provided by reason.
In these ways I have shown that Kants ethics is compatible with a metaphysical naturalism. I take this result to be important for two reasons. In
contemporary philosophy Kantians may defend many of the claims they derive
from Kant without worry that those claims are incompatible with nature as
now conceived. The second and more important reason is that my interpretation shows how Kants concern was to defend the moral life of human beings
as natural beings. Traditionally Kant has been seen as insisting upon supernatural abilities or entities in ethics free agents in themselves independent of
space and time relying on God and their own immortal soul to complete their
full moral lives. Even more generally, Kant is taken to deny knowledge of
supersensibles in theoretical philosophy in order to allow practical philosophy
to ll this conceptual space with the supersensible entities it requires. My interpretation reduces the claims about supersensibles greatly because I show that
the most important faculty for morality, practical reason, is comprehensible
entirely within nature, and that the other claims to supersensibles, as postulates, are not taken to have supersensible reference at all. This points to a
general interpretation of Kant as trying to comprehend the place of human
beings in the universe. And it conforms to an emphasis on the human point
of view and the limitations of reason, a topic that also grounds my claim that
Kant was a moral idealist.
Kant as moral idealist
The question whether Kant is a moral realist is more nuanced than the question
about naturalism. Realism operates on two levels: empirical, having to do with
the existence of actual moral agents, and transcendental, a consideration of the
requirements for any possible moral agent. In my rst chapter, I dened moral
realism and moral idealism in this way:

Postscript: Kants naturalist moral idealism

245

Moral realism: The moral principles, properties, or objects of the

world are independent of the transcendental or empirical moral
agent.
Moral idealism: The moral principles, properties, or objects of the
world are dependent upon the transcendental or empirical moral
agent.
The empirical level of realism is closely related to naturalism because it asks
the extent to which the moral principles, properties, or objects are entirely
dependent upon the empirical moral subject or are independently real parts of
nature. I take an empirical moral realist to be restricted to the naturalist moral
ontology I argue for. At the transcendental level the question is whether an
analysis of the conditions for moral agency shows that any of the moral
principles, properties, or objects must have an existence that is independent
of whether there is moral agency at all.
The transcendental moral agent is a rational being whose very reason, by
nature of its structure demanding systematic organization of the particulars
within any domain but also completeness to the unconditioned, generates the
moral law. Practical reason seeks to determine the free acts of rational agents
in a way that systematizes those free acts, understood in terms of agentperspective freedom. The moral law provides the systematic order for free
acts such that they can be maximally consistent among all rational agents. This
systematic order is represented as a kingdom of ends. Transcendental moral
agency also requires that the moral agent be susceptible to determination by
reason, be able to choose to pursue some particular ends, have some other kind
of freedom in making a basic moral decision that will underlie both moral
responsibility and the ought implies can doctrine, and try to reconcile the
demands of morality with her own desire for happiness.
As with naturalism, if the postulate of God (D) had a reference to a real
being, it would require a realist interpretation at both the transcendental
and empirical levels. My argument that the postulate has only immanent
reference would preclude that need. The postulate of immortality (D) and
any transcendent freedom of choice (H) might appear to require transcendental realism as well, but since they would be properties of moral agents as
subjects they conform to transcendental and empirical idealism (although
not to naturalism). Thus in all these cases Kant puts forward an empirical
idealist conception.
There should be agreement that the value of particular chosen ends (A) is
empirically idealist since there is no transcendental justication for them. Their
value is merely the fact that some empirical moral agent chooses them as ends
and is not itself independent of the empirical moral agent. Both would also
agree that some ideas created by reason such as the highest good (D) are

246

Postscript

merely concepts in the minds of empirical agents (perhaps transcendentally

justied). The realization of the highest good as part of the actualization of a
moral world is another matter.
The value of humanity as an end in itself (C) as a property of moral agents
as objects must be understood as idealist at both the transcendental and empirical levels. I defended the claim that a realism about value would violate
autonomy, but more importantly I showed that there was no place in Kants
ontology for an intrinsic value property. The absolute value of ends in themselves is simply their place in the order provided by reason through the
categorical imperative. This value is transcendentally ideal. At best it would
be empirically real not as an independent value property but only to the extent
that the order of reason could be said to be empirically real, which depends on
the status of reason itself.
The core element of my interpretation has to do with the status of practical
reason (G). The remainder of the moral elements all depend for their status on
the status of reason. If reason is transcendentally real as a cognitive faculty, the
moral law (F) is likewise transcendentally real because it is embedded in the
structure of pure practical reason. Moral obligation (E) is grounded in a valid
reason providing the objective normative order and is expressed through the
fact of reason. Particular duties or norms (B) would be empirically real to the
extent that they are required by an empirically real pure practical reason, given
the specics of human nature. And even the value of humanity understood in
terms of the order of reason (C) and other facets of the moral order as a moral
world depend upon reason.
I rejected a transcendental realism for pure practical reason, that is, the
claim that a transcendental requirement for morality is that reason must exist
and have validity independent of the moral agent, because it violates Kants
general strictures on the limits of the use of reason determined through the rst
Critique. Reason has no validity beyond experience.
I argued that Kant has a transcendental conception of moral agency that
requires the moral agent to possess the faculty of pure practical reason and
disallows validity of that reason beyond the limits of moral agency. Hence he
provides a transcendentally ideal ethics. This transcendentally ideal conception
of morality would allow for an empirical realism about the moral elements
mentioned before. If reason is transcendentally justied, then those elements
that depend upon reason for their validity would be empirically real because
independent of the existence of any or all particular moral agents. They would
not be properties or objects but principles independent of the empirical moral
agent. In this way, both Kantian realists and Kantian idealists can be correct,
each at one level of the understanding of morality.
There are two key interpretive pivot points, however, that can cast this
harmonious solution into doubt. First, Kant himself denies that there is any

Postscript: Kants naturalist moral idealism

247

successful deduction of the validity of morality. This by itself could still allow
for a view that, although we cannot prove that morality is transcendentally
valid, we must still believe that it is, and thus believe that morality is empirically real. The second point is that one can question whether Kant understands
the practical point of view to be making any claims about empirical reality in
light of its relation simply to the determination of action. On one side of these
two pivot points is the interpretation of Kant as a transcendental idealist and
empirical realist, on the other side is the interpretation of Kant as only an
empirical idealist. An analysis of these points will clarify the extent to which
Kant can be considered a realist at all.
At the rst pivot point is Kants denial of any possible transcendental
deduction of the validity of the moral law, based upon the claim that there is
no set of objects against which the moral law can be conrmed. The theoretical
transcendental deduction ranges over intuitions and so the categories can have
empirical reality as the structural framework for objects. Kant holds that there
are no corresponding independent objects to provide this conrmation for the
moral law and thus a transcendental justication for the moral law is unavailable. I argued that Kant had available to him appropriate independent objects,
namely, free acts of reason themselves. Actual empirical decisions of the
power of choice would have to implement the systematization provided by
the moral law in order for them maximally to embody freedom understood as
rationally determined decisions; systematicity among free actions would be
impossible without guidance from the moral law. I also argued that such a
view could ground an evolutionary account of the moral law in which any
possible organism that would evolve in order to make agent-perspective,
deliberate decisions would have to embody the faculty of reason to provide
this moral law. In this way Kant could have provided a transcendentally ideal
but empirically real pure practical reason. Since he claimed that he could not
provide a successful transcendental deduction of the moral law, he would have
to be interpreted as only an empirical idealist regarding reason.
Another avenue is open to the realist to preserve the transcendental ideality
of reason. The denial of a successful transcendental deduction provides only
the epistemic conclusion that empirical moral agents could not know whether
their faculty of reason was justied, but moral agents would still be free to
believe that reason is justied. This would retain the transcendental ideality
and empirical reality of practical reason as the ontology of morality even if it is
subject to some epistemic doubt. Analogously, an agnostic would admit that
Gods existence cannot be proved but would not deny that the subject of belief
would be a real God. If successful, this ontological claim would retain the
transcendental idealism and empirical realism of reason.
As an interpretation of Kant this move is questionable because of the second
pivot point: the nature of belief and the practical point of view. In a broader

248

Postscript

sense Kants denial of the possibility of a deduction of the moral law is

based on its subjectivity as guidance for action in empirical agents rather
than objectivity with reference outside moral agency. Practical reason in
contrast to theoretical reason is about determination of the empirical power
of choice in free acts and is not a determiner of ontology. This priority of
action gives morality as an empirical lived experience its own coherence
that does not rely on any kind of conrmation or disconrmation at the
transcendental level as long as the moral agent is in the grip of the practical.
When later reecting on that experience, practical reason presents theoretical reason with practical beliefs for the latter to integrate into a larger set of
beliefs, but they remain only concepts without objects, such as the postulate
of God. In this way the practical is understood not to be a set of claims
about reality such that it could be correct or incorrect in its claims about
independent ontology. There is no independent ontology beyond the practical perspective that could be true or false independent of the practical
perspective itself. The fact of reason as a command experienced by the
empirical moral agent carries obligatory force in the grip of moral decisionmaking. Practical reason provides experience with a distinct normative
order which agents actively impose on the world without any claims beyond
that experience. Morality is a perspective on the world imposed by the
empirical moral agent and hence is empirically ideal and does not operate
at the transcendental level at all.
This pivot point cuts to the heart of the difference between empirical realist
and empirical idealist interpretations. From the point of view of a realist, an
empirical idealist seems here to be denying that the categorical imperative is
true, in effect charging the empirical moral idealist with being an error theorist.
The empirical moral idealist would respond by denying that Kants understanding of the truth of morality requires anything beyond the experience of
morality itself. The moral experience is self-validating as a practical experience. Even the quest for a transcendental deduction of the moral law is a
reection of the priority of belief, which I argued is actually a higher priority
for theoretical reason and to that extent outside practical moral experience
itself.
Given these two pivot points between the interpretation of Kant as a
transcendental idealist and moral realist on the one hand and only an empirical
idealist on the other hand, I have argued in favor of empirical idealism as the
better interpretation of Kant. There are also parts of Kants theory that could
ground the transcendental idealist and empirical realist interpretation, so
regarding pure practical reason and the moral elements directly dependent
upon it, empirical realism is not unreasonable and one can even claim that
even if he was not, Kant could and should have been a transcendental idealist
and empirical realist in morality.

Postscript: Kants naturalist moral idealism

249

Kants human ethics

In the United States Kant has traditionally been portrayed as the synthesizer
of rationalism and empiricism, overcoming the limits of both while retaining
their strength. This portrayal formed the basis of the standard survey course in
early modern philosophy in which three rationalists (Descartes, Spinoza, and
Leibniz) were pitted against an equal number of empiricists (Locke, Berkeley,
Hume) culminating in Kants transformational critical account. This oversimplication bears a kernel of truth, since Kant retained the empiricist limitation of knowledge to objects of sense while insisting that the concepts provided
by the understanding are necessary for that experience. Less emphasized is the
limitation on reason itself to an aid to the understanding, unable to provide
knowledge of reality merely through a priori proofs. Knowledge of the world
is limited to the scope of human experience.
While there is no precise practical parallel to this theoretical work, the same
general stance applies to moral theory as well if one vastly oversimplies the
range of approaches. Intuitionists and rationalists insisted that reason was able
to provide knowledge of the cosmically true moral laws and values, while
empiricist moral sense theories took moral beliefs as sensible experience
without any kind of rational basis. Kant can be said to have worked out a
theory that takes our sensible experience of morality in the fact of reason and
our moral feeling of respect as a starting point but shows that it has validity in
reason. The limitation on reason is that Kant retains it within the point of view
of the rational agent and makes no claims to cosmic moral truth. The structure
of reason causes it not just to produce systematic sets of beliefs or acts but also
to seek for conditions for each of its claims, leading to the creation of ideas that
act as placeholders for the ultimate unconditioned necessity that reason seeks.
There may be no real unconditioned necessity either of being or of moral law
or of ends, but reason creates for itself ideas to satisfy this need.
Kants moral theory ought to be understood in terms of this dual nature of
reason. Practical reason both validates and limits morality for human beings.
Human moral agents nd rational constraints on their action when deliberating
through their awareness of the categorical imperative. Upon reection they ask
for the validity of reasons commands and nd it in the very structure of reason
rather than in anything independent of it. Further reection reveals that reason
not only has no need for a ground outside itself, it also cannot even provide a
complete explanation of any ground because of its very structure. Kant has
provided a moral theory for active, deliberative, and natural human beings.

Works cited

Abela, Paul. 2002a. Kants Empirical Realism. Oxford, New York: Clarendon Press;
Oxford University Press.
2002b. Kantian Walls and Bridges: Challenging the Integrationist Model of the
Relation of Theoretical and Practical Reason. British Journal for the History of
Philosophy 10(4): 591615.
Adickes, Erich. 1927. Kant Und Die Als-Ob-Philosophie. Stuttgart: Fr. Frommans
Verlag (H. Kurtz).
Allison, Henry E. 1986. Morality and Freedom: Kants Reciprocity Thesis.
The Philosophical Review 95(3): 393425.
1990. Kants Theory of Freedom. New York: Cambridge University Press.
1996. Idealism and Freedom: Essays on Kants Theoretical and Practical
Philosophy. Cambridge, New York: Cambridge University Press.
1997. We Can Act Only Under the Idea of Freedom. Proceedings and Addresses
of the American Philosophical Association 71(2): 3950.
2004. Kants Transcendental Idealism: An Interpretation and Defense. Revised and
Enlarged Edition. New Haven: Yale University Press.
2007. Comments on Guyer. Inquiry: An Interdisciplinary Journal of Philosophy
50(5): 48088.
2011. Kants Groundwork for the Metaphysics of Morals: A Commentary.
New York: Oxford University Press.
2012. Essays on Kant. 1st ed. Oxford: Oxford University Press.
De Almeida, Guido Antonio. 2012. Critique, Deduction, and Fact of Reason. In Kant
in Brazil, North American Kant Society Studies in Philosophy, edited by Frederick
Rauscher and Daniel Omar Perez. Rochester: University of Rochester Press,
pp. 12754.
Ameriks, Karl. 1982. Kants Theory of Mind: An Analysis of the Paralogisms of Pure
Reason. Oxford, New York: Clarendon Press; Oxford University Press.
2003. Interpreting Kants Critiques. Oxford, New York: Clarendon Press; Oxford
University Press.
Auxter, Thomas. 1982. Kants Moral Teleology. Macon, Ga.: Mercer University Press.
Bagnoli, Carla. 2011. Constructivism in Metaethics. In The Stanford Encyclopedia
of Philosophy, edited by Edward N. Zalta. http://plato.stanford.edu/archives/
win2011/entries/constructivism-metaethics/ (April 1, 2013).
2012. Kants Contribution to Moral Epistemology. Paradigmi: Rivista di critica
losoca, pp. 6979.
2014. Starting Points: Kantian Constructivism Reassessed. Ratio Juris 27: 31129.
250

Works cited

251

Beck, Lewis White. 1960. A Commentary on Kants Critique of Practical Reason.

Chicago: University of Chicago Press.
1987. Five Concepts of Freedom in Kant. In Philosophical Analysis and
Reconstruction: A Festschrift for Stephan Krner, edited by Jan T. J. Srzednicki.
Dordrecht: Martinus Nijhoff, pp. 3551.
2002. The Fact of Reason: An Essay on Justication in Ethics. In Selected Essays
on Kant, North American Kant Society Studies in Philosophy, edited by Hoke
Robinson. Rochester: University of Rochester Press, pp. 4556.
Bell, David. 2001. Is Empirical Realism Compatible with Transcendental Idealism?
In Idealismus als Theorie der Reprsentation?, edited by Ralph Schumacher
and Oliver Scholz. Paderborn: Mentis Verlag, pp. 16779.
Bird, Graham. 1995. Kant and Naturalism. British Journal for the History of
Philosophy 3(2): 399408.
2006. The Revolutionary Kant: A Commentary on the Critique of Pure Reason.
Chicago: Open Court.
Bittner, Rudiger. 1989. What Reason Demands. New York: Cambridge University
Press.
Bojanowski, Jochen. 2012. Is Kant a Moral Realist? In Kant and Contemporary
Moral Philosophy, Kant Yearbook, edited by Dieter Heidemann. Berlin: Walter de
Gruyter, pp. 122.
Byrne, Peter. 2007. Kant on God. Hampshire: Ashgate.
De Caro, Mario, and David Macarthur, eds. 2004. Naturalism in Question. Cambridge,
MA: Harvard University Press.
2010. Naturalism and Normativity. New York: Columbia University Press.
Chignell, Andrew. 2007. Belief in Kant. Philosophical Review 116(3): 32360.
Collins, Arthur W. 1999. Possible Experience: Understanding Kants Critique of Pure
Reason. Berkeley: University of California Press.
Cummiskey, David. 1996. Kantian Consequentialism. New York: Oxford University
Press.
DallAgnol, Darlei. 2012a. On the Factum of Reason. In Kant in Brazil,
North American Kant Society Studies in Philosophy, edited by Frederick
Rauscher and Daniel Omar Perez. Rochester: University of Rochester Press,
pp. 10926.
2012b. The Idealist Interpretation of Kants Metaethics. In Investigaes
Kantianas I: Um Debate Plural, edited by Milene Consenso Tonetto, Alessandro
Pinzani, and Darlei DallAgnol. Florianopolis, Brazil: FUNJAB, pp. 1129.
2013. Was Kant a Naturalist? Further Reections on Rauschers Idealist
Meta-ethics. Studia Kantiana 14: 14259.
Davidovich, Adina. 1993. Religion as a Province of Meaning: The Kantian
Foundations of Modern Theology. Minneapolis, MN: Augsburg Fortress.
Dean, Richard. 2006. The Value of Humanity in Kants Moral Theory. Oxford,
New York: Clarendon Press; Oxford University Press.
Denis, Lara. 2003. Kants Criticisms of Atheism. Kant-Studien: Philosophische
Zeitschrift der Kant-Gesellschaft 94(2): 198219.
Dworkin, Ronald. 1978. Taking Rights Seriously. Cambridge, MA: Harvard University
Press.
Dye, James Wayne. 1978. Kant as Ethical Naturalist. Journal of Value Inquiry 12:
11125.

252

Works cited

Esteves, Julio. 2012. The Non-Circular Deduction of the Categorical Imperative in

Groundwork III. In Kant in Brazil, North American Kant Society Studies in
Philosophy, edited by Frederick Rauscher and Daniel Omar Perez. Rochester:
University of Rochester Press, pp. 15572.
Falkenburg, Brigitte. 2004. Kants Naturalismus-Kritik. In Warum Kant Heute?,
edited by Dietmar Heidemann and Kristina Engelhard. Berlin: Walter de Gruyter,
pp. 177206.
Feder, Johann Georg Heinrich. 1769. Grundri Der Philosophischen Wissenschaften
Nebst Der Nthigen Geschichte, Zum Gebrauche Seiner Zuhrer. second. Coberg:
Findeisen.
Formosa, Paul. 2013. Is Kant a Moral Constructivist or a Moral Realist? European
Journal of Philosophy 21: 17096.
Frster, Eckart. 1989. Kants Transcendental Deductions: The Three Critiques and
the Opus Postumum. Stanford, Calif.: Stanford University Press.
2000. Kants Final Synthesis: An Essay on the Opus Postumum. Cambridge, Mass.:
Harvard University Press.
Friedman, Michael. 1997. Philosophical Naturalism. Proceedings and Addresses
of the American Philosophical Association 71: 521.
Frierson, Patrick. 2003. Freedom and Anthropology in Kants Moral Philosophy.
New York: Cambridge University Press.
2005. Kants Empirical Account of Human Action. Philosophers Imprint 5(7): 134.
Fugate, Courtney D. 2012. On a Supposed Solution to the Reinhold/Sidgwick
Problem in Kants Metaphysics of Morals. European Journal of Philosophy.
http://onlinelibrary.wiley.com/doi/10.1111/j.1468-0378.2012.00531.x/abstract
(August 12, 2013).
Guyer, Paul. 1998. The Value of Reason and the Value of Freedom. Ethics 109(1):
2235.
2000. Kant on Freedom, Law, and Happiness. Cambridge, New York: Cambridge
University Press.
2007a. Kants Groundwork for the Metaphysics of Morals: A Readers Guide.
New York: Continuum.
2007b. Naturalistic and Transcendental Moments in Kants Moral Philosophy.
Inquiry: An Interdisciplinary Journal of Philosophy 50(5): 44464.
2009. Problems with Freedom: Kants Argument in Groundwork III and Its
Subsequent Emendations. In Groundwork for the Metaphysics of Morals:
A Critical Guide, edited by Jens Timmermann. New York: Cambridge University
Press, pp.176202.
Hanna, Robert. 2001. Kant and the Foundations of Analytic Philosophy. Oxford:
New York: Clarendon Press; Oxford University Press.
Hare, John. 1996. The Moral Gap: Kantian Ethics, Human Limits, and Gods
Assistance. Oxford, New York: Clarendon Press; Oxford University Press.
2001 Gods Call: Moral Realism, Gods Commands, and Human Autonomy.
Grand Rapids: William B Eerdmans Publishing Co.
Henrich, Dieter. 1975. Die Deduktion Des Sittengesetzes: ber Die Grnde Der
Dunkelheit Des Letzten Abschnittes von Kants Grundlegung Zur Metaphysik
Der Sitten. In Denken im Schatten des Nihilismus: Festschrift fr Wilhelm
Weischedel, edited by Alexander Schwan. Darmstadt: Wissenschaftliche
Buchgesellschaft.

Works cited

253

1994. The Concept of Moral Insight and Kants Doctrine of the Fact of Reason.
In The Unity of Reason: Essays on Kants Philosophy, edited by Richard Velkley.
Cambridge, MA: Harvard University Press, pp. 5587.
Hill, Thomas E. 1989. Kantian Constructivism in Ethics. Ethics 99(4): 75270.
2012. Virtue, Rules, and Justice: Kantian Aspirations. Oxford: Oxford University
Press.
Hffe, Otfried. 2010. Kants Critique of Pure Reason: The Foundation of Modern
Philosophy. Dordrecht: Springer.
Hudson, Hud. 1994. Kants Compatibilism. Ithaca: Cornell University Press.
Hussain, Nadeem, and Nishi Shah. 2006. Misunderstanding Metaethics: Korsgaards
Rejection of Realism. Oxford Studies in Metaethics 1(2006): 26594.
Hutter, Axel. 2003. Das Interesse Der Vernunft: Kants Ursprngliche Einsichtund
Ihre Entfaltung in Den Transzendentalphilosophischen Hauptwerken. Hamburg:
F. Meiner.
Johnson, Robert N. 2007. Value and Autonomy in Kantian Ethics. In Oxford Studies
in Metaethics, edited by Russ Shafer-Landau. New York: Oxford University Press,
pp. 13348.
Kain, Patrick. 2004. Self-legislation in Kants Moral Philosophy. Archiv fr
Geschichte der Philosophie 86(3): 257306.
2005. Interpreting Kants Theory of Divine Commands. Kantian Review 9(1):
12849.
2006. Realism and Anti-Realism in Kants Second Critique. Philosophy Compass
1(5): 44965.
Kant, Immanuel. 2009. Fundamenao Da Metafsica Dos Costumes. translated and
edited by Guido Antonio de Almeida. So Paulo: Discurso Editorial.
Kaufman, Alexander. 2012. Rawls and Kantian Constructivism. Kantian Review
17(02): 22756.
Kielkopf, Charles F. 1997. A Kantian Condemnation of Atheistic Despair:
A Declaration of Dependence. New York: Peter Lang.
Kinnaman, Ted. 2008. Kants Nonfoundationalist Grounding of Reason. In Recht
und Frieden in der Philosophie Kants: Akten des X Internationalen KantKongresses. Band 5, edited by Valerio Rohden, Ricardo Terra, Guido de Almeida,
and Margit Rufng. Berlin: Walter deGruyter, pp. 34352.
Kitcher, Philip. 1993. The Evolution of Human Altruism. The Journal of Philosophy
90: 497516.
Kleingeld, Pauline. 1998. The Conative Character of Reason in Kants Philosophy.
Journal of the History of Philosophy 36(1): 7797.
2010. Moral Consciousness and the Fact of Reason. In Kants Critique of
Practical Reason: A Critical Guide, edited by Andrews Reath and Jens
Timmermann. New York: Cambridge University Press, pp. 5572.
Korsgaard, Christine. 1989. Personal Identity and the Unity of Agency: A Kantian
Response to Part. Philosophy & Public Affairs 18(2): 10132.
1996a. Creating the Kingdom of Ends. New York: Cambridge University Press.
1996b. The Sources of Normativity. New York: Cambridge University Press.
2003. Realism and Constructivism in Twentieth-century Moral Philosophy.
Journal of Philosophical Research 28: 99122.
2008. The Constitution of Agency: Essays on Practical Reason and Moral
Psychology. New York: Oxford University Press.

254

Works cited

Krasnoff, Larry. 1999. How Kantian Is Constructivism? Kant-Studien 90(4): 385409.

Kuehn, Manfred. 2001. Kant: A Biography. Cambridge; New York: Cambridge
University Press.
Loparic, Zeljko. 2005. A Semntica Transcendental de Kant. 3rd ed. Campinas, Brazil:
Coleo CLE.
Louden, Robert B. 2000. Kants Impure Ethics: From Rational Beings to Human
Beings. New York: Oxford University Press.
Manchester, Paula. 2008. Kants Conception of Architectonic in Its Philosophical
Context. Kantstudien 99: 13351.
Meerbote, Ralf. 1984. Kant on the Nondeterminate Character of Human Actions.
In Kant on Causality, Freedom, and Objectivity, edited by Ralf Meerbote and
William Harper. Minneapolis: University of Minnesota Press, pp. 13863.
Mieth, Corinna, and Jacob Rosenthal. 2006. Freedom Must Be Presupposed
as a Property of All Rational Beings. In Groundwork for the Metaphysics
of Morals, edited by Christoph Horn, Dieter Schnecker, and Corinna Mieth.
Berlin: De Gruyter, pp. 24784.
Naragon, Steve. 2006. Kant in the Classroom. www.manchester.edu/kant/Home/index.
htm (July 14, 2014).
ONeill, Onora. 1989. Constructions of Reason: Explorations of Kants Practical
Philosophy. New York: Cambridge University Press.
2002. Constructivism in Rawls and Kant. In The Cambridge Companion to Rawls,
edited by Samuel Freeman. New York: Cambridge University Press, pp. 34767.
2003. Constructivism vs. Contractualism. Ratio 16(4): 31931.
Pierce, Drew. 2004. Kant e a Justio das Crenas Morais. Impulso Piracicaba 15:
3546.
Pihlstrm, Sami. 2001. Naturalism, Transcendental Conditions, and the SelfDiscipline of Philosophical Reason. Journal of Speculative Philosophy:
A Quarterly Journal of History 15: 22850.
2004. Recent Reinterpretations of the Transcendental. Inquiry 47: 289314.
Pollok, Konstantin. 2013. Naturalism and Kants Resolution of the Third Antinomy.
In Kant und die Philosophie in weltbrgerlicher Absicht, Akten des XI KantKongresses 2010, edited by Stefano Bacin, Alfredo Ferrarin, Claudio La Rocca,
and Margit Rufng. Berlin. Band 2: 73142.
Proops, Ian. 2003. Kants Legal Metaphor and the Nature of a Deduction. Journal of
the History of Philosophy 41: 20929.
Rauscher, Frederick. 1998. Kants Two Priorities of Practical Reason. British Journal
for the History of Philosophy 6(3): 397419.
2002. Kants Moral Anti-Realism. Journal of the History of Philosophy 40:
47799.
2006a. Razo Prtica Pura Como Uma Faculdade Natural. Ethic@ https://
periodicos.ufsc.br/index.php/ethic/issue/view/1453. 5: 17392.
2006b. Reason as a Natural Cause. In Moralische Motivation. Kant und die
Alternativen, Kant Forschungen, edited by Heiner Klemme, Manfred Khn,
and Dieter Schnecker. Hamburg: Felix Meiner Verlag, pp. 97110.
2007. God Without God: Kants Postulate. Kant e-Prints 2: 2762.
2009a. Freedom and Reason in Groundwork III. In Kants Groundwork of the
Metaphysics of Morals: A Critical Guide, edited by Jens Timmermann. New York:
Cambridge University Press, pp. 20323.

Works cited

255

2009b. Kant as Metaphysical Naturalist. In Metafsica: Ontologia e Histria.

Atas do II Colquio Internacional de Metafsica, edited by Oscar Federico
Bauchwitz, Juan Bonaccini, Daniel Durante Pereira Alves, and Markus Figueira
da Silva. Natal, Brazil: Editoria da Universidade Federal do Rio Grande do Norte,
pp. 14167.
2010. The Appendix to the Dialectic and the Canon of Pure Reason: The Positive
Role of Reason In The Cambridge Companion to Kants Critique of Pure
Reason, edited by Paul Guyer. New York: Cambridge University Press,
pp. 290309.
2012. Realism and Anti-realism in Kants Ethics: A Reply to Professor
DallAgnol. In Investigaes Kantianas I: Um Debate Plural, edited by Milene
Consenso Tonetto, Alessandro Pinzani, and Darlei DallAgnol. Florianopolis,
Brazil: FUNJAB, pp. 3143.
Rawls, John. 1980. Kantian Constructivism in Moral Theory. The Journal of
Philosophy 77(9): 51572.
1999a. A Theory of Justice. revised. Cambridge, MA: Harvard University Press.
1999b. Collected Papers. edited by Samuel Freeman. Harvard University Press,
Cambridge.
1999c. Themes in Kants Moral Philosophy. In Collected Papers, edited by
Samuel Freeman. Cambridge, MA: Harvard University Press, pp. 497528.
Reath, Andrews. 1988. Two Conceptions of the Highest Good in Kant. Journal of the
History of Philosophy 26: 593619.
1994. Legislating the Moral Law. Nos 28: 43564.
2006. Agency and Autonomy in Kants Moral Theory. New York: Oxford University
Press.
Ritchie, Jack. 2008. Understanding Naturalism. Stockseld, UK: Acumen.
Rohden, Valerio. 2012. An Experiment with Practical Reason. In Kant in Brazil,
North American Kant Society Studies in Philosophy, edited by Frederick
Rauscher and Daniel Omar Perez. Rochester, NY: University of Rochester Press,
pp. 98108.
Rotenstreich, Nathan. 1965. Experience and Its Systematization; Studies in Kant.
The Hague: Martinus Nijhoff.
Sayre-McCord, Geoffrey. 1988a. Essays on Moral Realism. Ithaca, NY: Cornell
University Press.
1988b. The Many Moral Realisms. In Essays on Moral Realism, edited by
Geoffrey Sayre-McCord. Ithaca, NY: Cornell University Press, pp. 123.
Schmidt, Claudia. 2008. Kants Transcendental and Empirical Psychology of
Cognition. Studies in History and Philosophy of Science Part A 39(4): 46272.
Schneewind, J. B. 1997. The Invention of Autonomy: A History of Modern Moral
Philosophy. Cambridge; New York: Cambridge University Press.
Schonecker, Dieter. 1999. KantGrundlegung III: Die Deduktion Des Kategorischen
Imperativs. Freiburg: Alber.
Sensen, Oliver. 2009. Dignity and the Formula of Humanity. In Kants Groundwork
of the Metaphysics of Morals: A Critical Guide, edited by Jens Timmermann.
New York: Cambridge University Press, pp. 10218.
2011. Kant on Human Dignity. Berlin: De Gruyter.
2013. The Moral Importance of Autonomy. In Kant on Moral Autonomy, edited by
Oliver Sensen. New York: Cambridge University Press, pp. 26281.

256

Works cited

Skorupski, John. 1990. The Intelligibility of Scepticism. In The Analytic Tradition,

edited by David Bell and Neil Cooper. Oxford: Blackwell.
Smith, Joel, and Peter Sullivan. 2011. Introduction: Transcendental Philosophy and
Naturalism. In Transcendental Philosophy and Naturalism, edited by Joel Smith
and Peter Sullivan. New York: Oxford University Press, pp. 125.
Stern, Robert. 2000. Transcendental Arguments and Scepticism: Answering the
Question of Justication. Clarendon Press Oxford.
2003. Transcendental Arguments: Problems and Prospects. Oxford University Press.
2004. Does Ought Imply Can? And Did Kant Think It Does? Utilitas:
A Journal of Utilitarian Studies 16: 4261.
2011a. The Value of Humanity: Reection on Korsgaards Transcendental
Arguments. In Transcendental Philosophy and Naturalism, edited by Joel Smith
and Peter Sullivan. New York: Oxford University Press, pp. 7495.
2011b. Understanding Moral Obligation: Kant, Hegel, Kierkegaard. New York:
Cambridge University Press.
2012. Constructivism and the Argument from Autonomy. In Constructivism
in Practical Philosophy, edited by James Lenman and Yonatan Shemmer.
New York: Oxford University Press, pp. 11937.
Stevenson, Leslie. 2003. Opinion, Belief or Faith, and Knowledge. Kantian Review 7:
72101.
Strawson, P. F. 1966. The Bounds of Sense: An Essay on Kants Critique of Pure
Reason. London: Methuen.
Street, Sharon. 2008. Constructivism about Reasons. In Oxford Studies in Metaethics,
edited by Russ Shafer-Landau. Oxford: Oxford University Press, pp. 20745.
2010. What is Constructivism in Ethics and Metaethics? Philosophy Compass 5(5):
36384.
2012. Coming to Terms with Contingency: Human Constructivism about Practical
Reason. In Constructivism in Practical Philosophy, edited by James Lenman
and Yonatan Shemmer. Oxford: Oxford University Press, pp. 4059.
Stroud, Barry. 1968. Transcendental Arguments. Journal of Philosophy 65: 24156.
Sussman, David. 2008. From Deduction to Deed: Kants Grounding of the Moral
Law. Kantian Review 13-1: 5281.
Timmermann, Jens. 2003. Sittengesetz Und Freiheit: Untersuchungen Zu Immanuel
Kants Theorie Des Freien Willens. Berlin; New York: W. de Gruyter.
2007. Kants Groundwork of the Metaphysics of Morals: A Commentary.
Cambridge, UK; New York: Cambridge University Press.
Vaihinger, Hans. 1911. Die Philosophie Des Als Ob. Berlin: Reuther & Reichard.
1923. Die Philosophie Des Als Ob. Volksausgabe. Leipzig: Felix Meiner Verlag.
Ward, Keith. 1972. The Development of Kants View of Ethics. Oxford: Blackwell.
Watkins, Eric. 2005. Kant and the Metaphysics of Causality. New York: Cambridge
University Press.
Willaschek, Marcus. 1991. Die Tat Der Vernunft: Zur Bedeutung Der Kantischen
These Vom Factum Der Vernunft. In Akten Des Siebenten Internationalen
Kant-Kongresses: Kurfrstliches Schloss Zu Mainz 1990, edited by Manfred
Kleinschnieder and Gerhard Funke. Bonn: Bouvier, pp. 45566.
1992. Praktische Vernunft: Handlungstheorie Und Moralbegrndung Bei Kant.
Stuttgart: Verlag J.B. Metzler.

Works cited

257

2010. The Primacy of the Practical and the Idea of a Practical Postulate. In Kants
Critique of Practical Reason: A Critical Guide, edited by Andrews Reath and
Jens Timmermann. New York: Cambridge University Press, pp. 16896.
Wolff, Robert Paul. 1974. The Autonomy of Reason: A Commentary on Kants
Groundwork of the Metaphysic of Morals. New York: Harper & Row.
Wood, Allen W. 1970. Kants Moral Religion. Ithaca: Cornell University Press.
1984. Kants Compatibilism. In Self and Nature in Kants Philosophy, edited by
Allen W. Wood. Ithaca: Cornell University Press, pp. 73101.
1999. Kants Ethical Thought. New York: Cambridge University Press.
2007. Comments on Guyer. Inquiry: An Interdisciplinary Journal of Philosophy
50: 46579.
2008. Kantian Ethics. New York: Cambridge University Press.
Wood, Allen W., and Dieter Schnecker. 2004. Immanuel Kants Grundlegung
Zur Metaphysik Der Sitten: Ein Einfhrender Kommentar. 2nd ed. Paderborn:
Schningh.
Zammito, John. 2008. Kant and Naturalism Reconsidered. Inquiry 51: 53258.

Index

as if interpretation, 155
Abela, Paul, 34, 83
Adickes, Erich, 170
Allison, Henry, 21, 34, 989, 102, 108, 112,
11921, 133, 157, 183, 187, 195, 201,
211, 22830, 2356
Ameriks, Karl, 37, 56, 104, 122
amoralism, 43
anomalous monism, 134
anthropology, 17, 20, 35, 38, 578, 76, 96,
212
architectonic, 512, 56, 58
atheism, 149
autonomy, 2335
and concept of God, 176
and structure of reason, 125
and value, 21017
not arbitrary, 2212
denition of, 210
Auxter, Thomas, 226
Bagnoli, Carla, 256
Baumgarten, Alexander, 73
Beck, Jacob Sigismund, 18990
Beck, Lewis White, 87, 989, 108, 149
Bell, David, 64
biology, 35
Bird, Graham, 28, 301, 33, 64
Bittner, Rdiger, 25
Bojanowski, Jochen, 12, 16
Boyd, Richard, 14
Byrne, Peter, 151
categorical imperative, 445, See also moral
law
and empirical reason, 119
and evolution, 1434
and knowledge of value, 220
and particular duties, 37
and postulate of freedom, 206
and priority of action, 8891
and rational agency, 26

258

different from moral law, 98100

possibility of related to reason, 1359
categories, 9, 20, 34, 5964, 1302, 141, 144,
155, 239, 247
causality, 31, 36, 64, See also choice, freedom
of; determinism; reason, causality of
different from cause, 120
timeless, 12230
transcendentally free, 1203
character
denition of, 115, 122
empirical, 113, 11518, 184, 1867, 189,
1918, 2003
intelligible, 113, 116, 119, 1223, 187, 191,
1956, 2013, 237
chemistry, 24, 29, 35, 58, 656
Chignell, Andrew, 158
choice, 36, See also Wille/Willkr distinction
animal, 110, 1845
denition of, 1823
empirical faculty of, 11011, 129, 1846,
194
freedom of, 468, 11214, 132
and domain of practical philosophy, 701,
736
and fact of reason, 1046
and moral responsibility, 1938, 203
and ought implies can, 91, 18793, 202
determined by reason, 11317, 1846
rst person, 756, 8891, 205
in nature, 67, 1846
of a being in herself, 92, 198203
postulate, 2036
transcendent faculty of, 11011, 1968
Collins, Arthur, 34
Common Sense theory, 31
consequentialism, 19
and realism, 13
and value of humanity, 213
constructivism, 11, 237, 72, 240
and metaethics, 26
and term idealism, 27

Index
and value, 41
as decision procedure, 245, 39
as nature of agency, 267
compatible with moral realism, 13, 23
in mathematics, 23, 51
in political philosophy, 42
in theoretical philosophy, 23
origin of term, 1011
seen as only alternative to realism, 11, 37
critique. See also method
as division in philosophy, 56, 5968
as project of theoretical reason, 914
Cummiskey, David, 213
DallAgnol, Darlei, 14, 32, 37, 98, 210
Davidovich, Adina, 163
Davidson, Donald, 134
de Almeida, Guido, 98, 235
De Caro, Mario, 29
Dean, Richard, 210
deduction
empirical, 603
metaphysical, 132
of moral law, 247
of the moral law, 21, 65, 96, 1302, 143,
145, 22731, 2345, 238
transcendental, 20, 45, 603, 132
Denis, Lara, 149
determinism, 356, 11420, 18793
and practical freedom, 111
dignity, 223
divine command theory, 16, 177
domain
of ethics as free acts, 736, 205
of philosophy as ultimate end of human
reason, 535, 72
required for a system, 52
separate for theoretical and practical, 701
duty
no direct duties to God, 172
no duty to believe in God, 87
particular duties, 379
pure and empirical, 58
Dworkin, Ronald, 10
Dye, James Wayne, 31
empirical idealism
moral, 202, 240, 2468
and contingent ends, 37, 245
and duties, 379
and fact of reason, 1034
and God, 1779, 245
and moral law, 45
and necessary ends, 3942, 246
and practical reason, 2478

259
and reason, 457, 1456, 247
and value, 224
theoretical, 1920
empirical realism. See also metaphysical
naturalism, nature
moral, 202, 2468
and duties, 379
and free choice, 201
and method in ethics, 648
and moral law, 45
and necessary ends, 3942
and reason, 457, 71, 1435
and value, 21722, 227, 246
theoretical, 1920, 30, 33, 64
and category of causality, 64
ends
end in itself ultimately inexplicable, 2367
objective or necessary, 3942, 209, See also
humanity, value of; value, absolute
of human reason as domain of philosophy,
535
order of. See reason, order of
subjective or contingent, 37, 209, 222
that are also duties, 3940
error theory, 1216, 27, 248
Esteves, Julio, 132
ethics
as discipline in philosophy, 50
transcendental method in, 648, 1302
whether systematizable, 523
evolution
and moral law, 1434
and non humanlike altruism, 62
and transcendental method, 623
of categorical imperative, 247
experience
and fact of reason, 1012
moral, 202, 648, 85, 92, 96, 98100,
1302, 170, 206, See also practical
point of view
expressivism, 1315, 267
fact of reason, 32, 445, 96106, 206
and awareness of value, 219
and freedom, 1046
and metaphysical naturalism, 104
and moral agent, 1034
and realism and idealism, 1034
as act or deed, 1001
as consciousness of moral law, 667
denition of, 100
limits empirical realism, 144
Falkenburg, Brigitte, 34
Formosa, Paul, 12, 14, 25, 37, 240
Frster, Eckart, 59, 1756

260

Index

freedom. See also reason, freedom of; choice,

freedom of; will
acting under idea of, 1336
and boundary between theoretical and
practical, 231
and fact of reason, 106
and primacy, 86
as idea of reason, 231
as postulate, 47, 83, 1046, 2036
assumed in action, 8891
cannot be proved, 1302
multiple meanings of term, 108
practical, 11011, 1212, 1836, 197, 220,
243
transcendental, 12230, 134, 184
Friedman, Michael, 29
Frierson, Patrick, 186
Fugate, Courtney, 196
geometry, 20, 24
God, 41, 70, 121, 142
and metaphysical naturalism, 14950
and moral law, 172, 175, 177
and moral realism, 179
and realism, 13
as postulate, 424, 83, 14950, 1523, 203
belief in
compared to opinion and knowledge,
15660
function for action, 15960, 1635,
16770
not a duty, 87, 163
subjective not objective, 1589, 1635,
173
concept of, 1523
and nature, 152, 169
and religious practice, 1702
constructed by practical reason, 1667,
173, 1767
equated with pure practical reason, 175
immanent not transcendent, 16770
scope of term, 149
stems from practical reason, 1656
Guyer, Paul, 31, 153, 163, 210, 213, 229, 237
Hare, John, 177
Henrich, Dieter, 98, 133, 136, 229, 234
Herman, Barbara, 21
highest good, 423, 71, 87, 94, 149, 156,
1679, 173, 175, 245
as systematic relation, 73
not a motive to action, 878
possibility of natural ground, 1635
Hill, Thomas, 11, 245, 38
history, 35

Hffe, Otfried, 34
Hudson, Hud, 187
humanity. See also value, absolute
as end in itself, 3942
value of, 402, 20917
and arbitrariness, 2212
assigned by reason, 223
based on natural property, 2202
not an intrinsic property, 21718, 2202
Hume, David, 31
Hussain, Nadeem J. Z., 26
Hutter, Axel, 55, 58
hypothesis, transcendental, 149, 158, 171
compared to belief, 15960
compared to idea, 1536
compared to postulate, 162
not endorsed in use, 1556
idea, 42, 119, 230
and boundary between practical and
theoretical, 231
and domain of philosophy, 535
and transcendental realism, 142
as heuristic device, 154, 219, 238
compared to hypothesis, 1536
of God. See God, concept
of hell not coherent, 171
required for architectonic unity, 512
idealism
empirical. See empirical idealism
moral. See moral idealism
transcendental. See transcendental idealism
immortality, 423, 70, 83, 149, 203
independence
as metaphysical separation, 15, 125
as unchangeable by another, 1267, See also
reason, structure of
intelligible world, 1368, 167, 2247, 229,
2335
as moral ideal, 2245
realized in nature, 2257
intuition, 23, 345, 117, 218
forms of. See also space, time
Johnson, Robert, 21113, 217
judgments, 11920, 1336, See also reason,
freedom of
Kain, Patrick, 25, 37, 39, 43, 102, 13941, 170,
177, 240
Kant, Immanuel
as reective rather than active person, 94
religious beliefs of, 149
Kielkopf, Charles, 101
kingdom of ends, 26, 38, 73, 245

Index
Kinnaman, Ted, 56
Kitcher, Philip, 12
Kleingeld, Pauline, 55, 978, 1001
Korsgaard, Christine, 11, 21, 234, 26, 41, 143,
210, 237
Krasnoff, Larry, 10
Kuehn, Manfred, 149
Larmore, Charles, 38
logic
as division in philosophy, 56, 58
Loparic, Zeljko, 29
Louden, Robert, 38, 169
Macarthur, David, 29
Manchester, Paula, 52
mathematics, 23, 31, 35, 51, 58, 66, 128, 161,
223
Meerbote, Ralph, 134, 187
metaphysical naturalism, 27, 2933, 36, 64,
2414
and duties, 39, 243
and empirical power of choice, 1856
and empirical reason as cause, 11420
and fact of reason, 104, 243
and free choice, 243
and freedom of choice, 457, 201, 203
and God, 14950, 177, 243
and moral law, 243
and necessary ends, 402
and obligation, 43, 243
and priority of action, 96
and reason, 2423
and structure of reason, 1434
and transcendental freedom, 125
and transcendental method, 603
and transcendental method in ethics, 648
and value, 2224, 244
denition of, 29
in theoretical philosophy, 33, 241
limited to ethics, 323
metaphysics
as division of philosophy, 56
method
and logic, 56
and systematic organization of cognition,
523
in ethics
compared to chemistry, 24, 656
transcendental, 9, 21, 32, 5968, 76
transcendental, 29, 107, 174
Mieth, Corinna, 128, 1323
mind, 27
and evolution, 623
nature of, 17

261
moral agent
and belief contrasted with opinion and
knowledge, 15660
and belief in postulates, 85
and causality of reason, 1203
and fact of reason, 1013
and rst-person freedom, 756
and God, 1779
and objectivity, 19, 648
and priority of action over belief, 8891,
936
and reason, 457, 756
and truth, 13
as being in itself, 28, 33, 67, 136, 198203
as subject rather than object, 1516, 47
empirical, 1618, 202, 39, 648, 968,
1034, 11420, 12932, 15660, 179,
1846, 189, 201, 222, 239
in constructivism, 247
in denition of moral realism, 14
obligation tied to, 434
possibility of non humanlike, 18, 145
praise and blame, 1938
purely rational, 98100, 129
transcendental, 1618, 202, 648, 968,
1034, 1302, 179, 211, 2224, 245
moral anti-realism, 11, 211
problems with term, 1112
moral idealism, 68, 1456, 2448, See also
empirical idealism, transcendental
idealism
and freedom of choice, 47
and obligation, 43
and postulates, 423
and priority of action, 94
denition of, 14, 244
separates reason and reality, 240
use of term, 12, 27
moral law, 445, See also categorical
imperative
consciousness of, 667
different from categorical imperative,
98100
possibility of deduction of, 1302
relation to concept of God, 15770
reveals freedom, 92, 11213, 1359, 188,
194, 205
stems from structure of reason, 125
ultimately inexplicable, 2356
why a priori, 95
moral nonrealism, 11, 13, 16, 27, 38
denition of, 14
moral principle, property, or object
contrasted with one required for morality,
15

262

Index

moral realism, 9, 16, 68, 2448, See also

empirical realism, transcendental
realism
and contingent ends, 37
and freedom of choice, 47
and obligation, 43
and postulates, 423
and priority of action, 94
and structure of reason, 13946
at empirical and transcendental levels, 212
contemporary, 9
Cornell, 14
denition in relation to moral agent, 1418,
244
denition in relation to truth, 1114
need to assess in parts, 22, 37
procedural, 23
relation to naturalism, 9
requires reason to reveal reality, 240
moral world. See intelligible world
morality, 71
common moral practice, 93
relevance of transcendental/empirical
distinction for, 20
ultimate inexplicability of, 2348

ONeill, Onora, 11, 245, 46

objectivity, 19, 1445, 224, 227, 240
obligation, 434, 98100, 135, 145, 211
ontology, 323, 36, 76, 1423
and transcendental method in ethics, 65
and value, 21820
no ontological claims in practical
philosophy, 946
ought implies can, 91, 18793, 202

empirical, 54, 57
practical, 44, 568, See also reason, practical
and action, 72, 946
as distinct set objects, 701
as pragmatic use of knowledge, 74
contrasted with theoretical, 6873
domain as free acts, 6876
domain not empirical, 75
inner boundary of, 22934
outermost boundary of, 2289, 2348
structure of, 558, 91
theoretical, 44, 568, 228, See also reason,
theoretical
and knowledge, 72
as distinct set of objects, 701
contrasted with practical, 6873
transcendental, 5960, 174
physics, 30, 345, 61, 75, 167
Pierce, Drew, 169
Pihlstrm, Sami, 60
Plato, 29, 201, 224
Pollok, Konstantin, 31, 34
postulate, 83, See also freedom, God,
immortality
and idealism, 1779, 245
and primacy of practical reason as belief,
848, 93
as positing existence to theoretical concepts,
1623
constructed by practical reason, 1667
denition of, 152, 1612
immanent not transcendent, 15770, 204
theoretical use of term, 1523, 162
practical point of view, 25, 28, 756, 8891,
946, 1013, 1336, 1456, 239,
2478
primacy of practical reason, 8194, 162
and integration model, 83
dened as priority of action, 82
dened as priority of belief, 81
priority of action, 82, 8897, 102, 105,
2056, See also reason, practical
priority of belief, 8190, 97, 103, 105,
149, 206, 248, See also reason,
practical
Proops, Ian, 98
psychology
empirical, 31, 33, 357, 44, 62, 74, 956,
104, 11618, 1846, 242
moral, 26, 111
rational, 122
questions of reason, 578, 157

philosophy
denition of, 51, 534, 723

Railton, Peter, 11
Rawls, John, 10, 21, 24, 98

naturalism, 2736
and moral law, 44
and postulates, 423
contemporary, 2930
epistemological, 29
Kant's use of term, 27
metaphysical. See metaphysical naturalism
denition of, 241
methodological, 2831, 956
denition of, 29, 241
relation to realism, 9
semantic, 29
nature, 30, 336, 71, 84, 168
empirically real, 613, 11420
order of, 11819, 152, 18993, 2334
reason in, 11618, 1203

Index
realism
moral. See moral realism
reason, 36, 82
active nature of, 712, 100, 137
and system of cognitions, 512
as transcending nature, 122
blamed for wrong action, 1947
causality of, 67, 10930, 138, 1948, 224,
237, 243
and judgments, 1336
in nature, 10910, 11920
transcendentally free, 10910, 12030
ultimately inexplicable, 237
critical use of distinct from practical or
theoretical, 91
denition of, 109, 1203
empirical, 36, 107, 11420, 1846
freedom of, 47, 11423, 1306, 183,
203
human, 535, 92, 151, 164, 228, 237,
240
intelligible character of, 113, 116, 1223,
191, 1956, 237
interest of, 8194, 232
one faculty with different uses, 69
order of, 42, 11819, 152, 2227
practical, 43, 457, 157, See also
philosophy, practical
and action, 946
and concept of God, 1667, 1701, 1757
and empirical idealism, 1456
and is/ought distinction, 6970, 1456,
2324
and value, 218
creates it own objects, 712
meaning of Wille, 1834
primacy of. See primacy of practical
reason
pure, 28, 57, 116, 12230
regulative use of, 34, 36, 1534, 157, 168,
174, See also ideas, postulates
requires an unconditioned, 2358
speculative. See reason, theoretical
structure of, 108, 12230, 13946, 194, 196,
239, 242
theoretical, 946, See also philosophy,
theoretical
and boundary of practical, 2313
and concept of God, 1656
and critical works, 914
and is/ought distinction, 6970, 1456,
2324
and value, 218
higher priority of regarding belief, 848
must conform to practical belief, 814

263
objects given to, 712
same as speculative, 68
Reath, Andrews, 25, 389, 42, 211
Reinhold, Karl Leonard, 196
relativism, 43, 146
Ritchie, Jack, 29
Rohden, Valerio, 24
Rosenthal, Jacob, 128, 1323
Rotenstreich, Nathan, 85
Sayre-McCord, Geoffrey, 1214, 255
Schmidt, Claudia, 36
Schneewind, J. B., 38, 211
Schnecker, Dieter, 2289
science, 28, 302, 57, 61, See also biology,
chemistry, physics
social, 35, See also anthropology, history,
psychology
Sensen, Oliver, 211, 213, 21819, 223
Shah, Nishi, 26
Skorupski, John, 30
Smith, Joel, 29
space, 9, 17, 1920, 33, 35, 238
Studlin, Carl Friedrich, 57
Stern, Robert, 21, 37, 43, 143, 188, 210,
218
Stevenson, Leslie, 158
Strawson, P.F., 63
Street, Sharon, 1213, 16, 24
Stroud, Barry, 60, 634
Sullivan, Peter, 29
Sussman, David, 101
system
nature of systematic organization, 523,
124
structure of ethics, 523, 12930
teleology, 35, 54, 57, 167, 169
theology, 28, 114, 177
time, 9, 17, 1920, 33, 36, 113, 203
timelessness, 12230, 146, 181, 1867, 191,
198203
Timmermann, Jens, 132, 232
transcendental argument. See deduction,
transcendental
transcendental idealism
moral, 202, 457, 2468
and deduction, 247
and duties, 379
and fact of reason, 1034
and God, 1779
and method in ethics, 648
and necessary ends, 3942, 246
and reason, 1435
and value, 21117

264

Index

transcendental idealism (cont.)

theoretical, 1920, 30, 70, 155, 174
and category of causality, 64
relation to naturalism, 301
transcendental realism
moral, 202, 457, 107
and God, 1779
and practical reason, 1423
and reason, 246
and value, 41, 21117
theoretical, 1920
and rationalist ontology, 1423
Vaihinger, Hans, 170
value, 33
absolute
and autonomy, 21014
and idealism, 224
as nonnatural property, 21722
ultimately inexplicable, 2367
as ordering of ends by reason, 2224

of humanity. See humanity, value of

of persons, 47
of subjective or contingent ends, 37, 209
Ward, Keith, 112
Watkins, Eric, 113, 193
will, 36, 84, See also reason, practical
nite rational, 43
freedom of. See choice, reason
identied as reason, 109, 1834
Wille/Willkr distinction, 47, 107, 109,
1823
Willaschek, Marcus, 84, 100, 132, 161, 163,
169
Willkr. See choice, will
Wolff, Robert Paul, 25
Wood, Allen, 14, 26, 323, 38, 87, 139, 145,
149, 166, 187, 189, 210, 212, 21416,
228, 240
Zammito, John, 28, 301