ethics

buryblooms

FRANCESCO
ORSI

value
theor
y

Value
Theory

BLOOMSBURY ETHIC S
Series Editors:
Thom Brooks, Reader in Law, Durham Law School, UK
and Simon Kirchin, Philosophy, University of Kent, UK
Bloomsbury Ethics is a series of books on established and new
areas in moral philosophy. Each book is designed both to
introduce upper- level undergraduates and postgraduates to a
key field in ethics, and to develop a particular viewpoint within
that field designed to appeal to researchers. All areas of moral
philosophy are covered, from the theoretical to the practical.
New proposals are always welcome. Please contact the series
editors.

Titles available in the

series:
Intuitionism David Kaspar
Reasons Eric Wiland
Autonomy Andrew Sneddon

Forthcoming:
Ethics Without Intention Ezio Di
Nucci
Trust, Ethics and Human Reason
Olli Lagerspetz
Moral Principles Maike Albertzart

Moral Motivation
Leonard Kahn
Moral Realism
Kevin DeLapp
Virtue Ethics Nafsika
Athanassoulis

Luck Egalitarianism Kasper

Lippert-Rasmussen
Moral Skepticism
Basil Smith

BLOOMSBURY ETHICS

Value
Theory
FRANCESCO ORSI

L O O

B U

LONDON NEW DELHI NEW YORK SYDNEY

R Y

www.bloomsbury.com
Bloomsbury is a registered trade mark of Bloomsbury Publishing
Plc
First published 2015
Francesco Orsi has
the Copyright,
Act, 1988, to
work.
All rights

Bloomsbury Academic
An imprint of Bloomsbury Publishing Plc

5G Bedford Square
London WC1B 3DP UK

Designs and
be identified
reserved. No

Francesco Orsi 2015

asserted his right under

1385 Broadway
New York NY
10018 USA

Patents
as the Author of this
part of this

publication may be reproduced or transmitted in any form or by any means, electronic or

mechanical, including photocopying, recording, or any information storage or retrieval system,
without prior permission in writing from the publishers.
No responsibility for loss caused to any individual or organization acting on or refraining from
action as a result of the material in this publication can be accepted by Bloomsbury or the
author.

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.
ISBN: 978-1-47252-53G-7

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress
Typeset by Fakenham Prepress Solutions, Fakenham, Norfolk NR21 8NN

CONTENTS

Acknowledgements viii Preface ix

1

Value and Normativity 1

1.1 Introduction 1
1.2 Which evaluations? 2
1.3 The idea of value theory 6
1.4 Value and normativity 8
1.5 Overview 16
1.6 Meta-ethical neutrality 17
1.7 Value theory: The questions 22

Meet the Values: Intrinsic, Final & Co. 25

2.1 Introduction 25
2.2 Final and unconditional value: Some philosophical
examples 27
2.3 Intrinsic value and final value 31
2.4 The reduction to facts 35
2.5 Intrinsic and conditional value 39
2.6 Elimination of extrinsic value? 41
2.7 Summary 43

The Challenge against Absolute Value 45

3.1 Introduction 45
3.2 Geach and attributive goodness 46
3.3 Foot and the virtues 50
3.4 Thomson and goodness in a way 53

3.5 Zimmermans ethical goodness 56

3.6 A better reply: Absolute value and fitting attitudes 58
3.7 Summary 61
4

Personal Value 63

4.1 Introduction 63
4.2 Moore on good and good for 63
4.3 Good for and fitting attitudes 66
4.4 Moore strikes back? 69
4.5 Agent-relative value 73
4.6 Impersonal/personal and agent-neutral/agentrelative 77
4.7 Summary 80
5

The Chemistry of Value 81

5.1 Introduction 81
5.2 Supervenience and other relations 81
5.3 Organic Unities 85
5.4 Alternatives to organic unities: Virtual value 90
5.5 Alternatives to organic unities: Conditional value 93
5.6 Holism and particularism 98
5.7 Summary 100
6

Value Relations 101

6.1 Introduction 101

6.2 The trichotomy thesis and incomparability 101
6.3 A fitting attitude argument for incomparability 104
6.4 Against incomparability: Epistemic limitations 107
6.5 Against incomparability: Parity 109
6.6 Parity and choice 110
6.7 Parity and incomparability 113
6.8 Summary 115
7 How Do I Favour Thee? 117
7.1 Introduction 117

7.2 Three dimensions of favouring 117

7.3 Responses to value: Maximizing 126
7.4 Two concepts of intrinsic value? 130
7.5 Summary 134
8

Value and the Wrong Kind of Reasons 135

8.1 Introduction 135

8.2 The fitting attitude account and its rivals 136
8.3 The wrong kind of reasons problem 141
8.4 The structure of the problem and an initial response
143
8.5 Reasons for what? 145
8.6 Characteristic concerns and shared reasons 151
8.7 Circular path: No-priority 156
8.8 Summary 158
Bibliography 159 Index 169

ACKNOWLEDGEM
ENTS

I wish to thank those who have most inspired and encouraged

me through the years: Eugenio Lecaldano, Tito Magri, Jonathan
Dancy, Philip Stratton-Lake. A collective thanks also goes to
fellow graduate students and colleagues in Rome, Reading,
and Tartu, with a special mention for Valerio Salvi, with whom
countless hours were spent discussing these issues. I am
grateful to audiences and students for their feedback on parts
of this book. I thank my Bloomsbury editors for their
enthusiasm for this project. This book was written with the
help of PUT grant 243 (Estonian Research Council). Finally, I
thank my family and Julia for their patience and constant
support.

PREFACE

We value things: relationships, particular people, particular

events, or particular objects. This means that we feel and
behave towards them as if they possessed a certain value,
whether or not we consciously judge them to be valuable.

Sometimes we also evaluate the same and other things as

good, significant, worthy of our interest (or bad, worthless, and
so on). In this case, we might well fail to feel and behave
towards them as we think we should. But both in valuing and
evaluating things not only do we express our attitudes or
convictions; we do so by employing, more or less explicitly, a
rich and sophisticated conceptual repertoire.
This book is an attempt to introduce and examine some of
this conceptual repertoire, as manifested both in ordinary
evaluative practices and in philosophical theories of what is
good: what is it for something to have value for its own sake?
Can disparate sorts of things (say, people, hats, abstract
objects) all have value? In what sense is, say, health good for
me? And can something be good, but good for nobody at all?
Can anything good come out of what isnt good? Are there
incomparable values?
In order to navigate these different questions, and give
unity to an otherwise diverse subject matter, I have followed a
basic insight: in our evaluative practices not only do we
express attitudes of preference or aversion, but also claim or
at least feel that our preferences and aversions are justified by
good reasons or, as some like to say using newly fashionable
terms, are fitting, suitable, appropriate. Whether this insight
provides a reliable and helpful compass is ultimately for the
reader to decide.

CHAPTER ONE

Value and Normativity

1.1 Introduction
We use value talk all the time. Most of the time, we dont need
to mention the word value or similar terms like worth or
merit. More commonly we use evaluative expressions like
good, bad, better, the best, great, fine, excellent,
poor, terrible and so on. To utter such evaluative words of
course is not always to evaluate something positively or
negatively. The exclamation Good heavens! normally doesnt
say anything about the value of the heavens. Also, we can use
good! or thats good simply to express or report our
satisfaction with something, regardless of whether we think it
worthy of our satisfaction, and so good in an evaluative
sense. At other times still, evaluative words are used to
register other peoples evaluations, without joining in
ourselves: if I know nothing about row-crop tractors, except
the ranking that an expert commission gave of models
produced in 2013, I might call one 2013 model better than
another simply to report the commissions ranking. Of course,
given its primary evaluative association, such a judgement
might be misleading, since you might well infer that I do
believe one model to be better than another (and thus,
perhaps, that I am advising you to get that), whereas all I
mean is really that the commission decided so.

It is also important to keep in mind that evaluation need not

be articulated via value terms, or indeed be verbally
articulated at all. Often we can tell a persons values and
commitments from her behaviour and emotional reactions
much more than from what she says (to others or even to
herself). We can understand a given communitys unspoken
and unwritten values or norms, or at least some of them, by
interpreting their social practices, rituals, internal conflicts, and
so on. However, in this book I will explore conceptual questions
about value, rather than about evaluation as a social and
psychological phenomenon. So I will treat the practice of
evaluation or peoples values as a window which reveals, or
at least purports to reveal, putative truths about value.
Therefore evaluations which are verbally expressed or at least
expressible in value terms are the starting point.
In this chapter I introduce the conceptual background of the
book. First, I draw a distinction between different evaluations,
in order to focus on the value concepts that will be discussed.
Then I define and explain the project of value theory, as I
pursue it here. In section 1.4 I lay out the central assumption
or guiding thread: value is normative. I introduce two views
intended to articulate this notion: the fitting attitude account
and the buck-passing account of value. On this basis, I then
give an overview of the contents of the book, showing how the
diverse structural questions about value here investigated will
be addressed through the lenses of the idea that value is
normative. I close the chapter with a remark about how my
approach will be neutral among different meta-ethical views,
and with a graphical sketch of the contents of the book.

1.2 Which evaluations?

Evaluation comes in so many linguistic forms and spans so
many different categories. An exhaustive classification of

evaluative judgements is a task for another occasion. But

there is a conceptual distinction that can help to isolate the
sort of evaluations that will keep us occupied in what follows.
Some evaluations are pure, or purer than others, i.e. they carry
minimal or no descriptive or informative content about the
thing evaluated. Following Bernard Williams (1985), these can
be called thin evaluations. When a certain concept is used in
such an evaluation, it is a thin evaluative concept. As such,
they contrast with thick evaluations (and thick concepts),
which do carry more than minimal information about the thing
evaluated. It is not easy to find this theoretical distinction
reliably mirrored in aspects of grammar or in characteristic
words, but a few remarks may help to grasp the point.
For instance, thin evaluation can be seen in unqualified uses
of words such as good and bad and their comparatives.
These can be applied to concrete objects or states of affairs:
this movie is bad, following her advice was good. In
philosophy, thin general evaluative statements are often made
about abstract objects: pleasure is good, knowledge is
generally better than ignorance. But one can also use abstract
nouns like value (or good as a noun): friendship is/has the
greatest value, and verbs like to matter: friendship matters
more than honour.
Now, these terms can be qualified by adverbs or adverbial
phrases like intrinsically, in itself, instrumentally, or by the
respective adjectives. Depending on the adverb, the resulting
evaluation will be more or less thick, that is, will carry more or
less descriptive or informative content about the thing
evaluated. Sometimes it wont carry any additional descriptive
content, but simply stress the importance of a value: helping
others in need has incalculable value. In other cases it will
carry some minimal content: to say that knowledge is
instrumentally good hints at the effects or consequences of
knowledge, albeit not saying what these are, except that they
are good.

But in other cases, qualifications can carry quite a lot of

informative content: This photograph has personal value (for
me), Wedding rings have sentimental value (for many
people), This song has no artistic value, Your results are
scientifically worthless, etc. These qualifications assign the
thing evaluated to some category or other of value, and so
imply some substantial information about them, or about their
relations to other things or people. Such information need not
be very detailed, of course. To speak of personal or
sentimental value implies something about the historical or
affective connection of an object to people. To speak of artistic
or scientific value turns ones attention to the sort of features
which justify the inclusion of an object in either category. In
these cases, we are making thicker evaluations.
Also, evaluative words can be used predicatively, as in the
sentences used so far, which have the general form:
x is good.
But very commonly they are used attributively, i.e. they
accompany a noun: Sarah is a great dancer, Liquorice soup
is a terrible entre, WWII was the worst war ever. In general:
x is a good F.
Often, these attributive evaluations are quite thick: for Sarah
to be a great dancer, it must be true that she masters the skills
of a dancer to a considerable degree. So attributive uses of
value terms can convey a great deal of information together
with the evaluation. But this need not be the case: it is not at
all obvious what information is carried by the statement that
WWII was the worst war ever, because it is not at all obvious
what a war has to be like, as a war, to be worse than another.
All we definitely know, of course, is that WWII was a war, and
so that there was fighting and the like (for another grey-area
example: Valerie is a very good child). Attributive evaluations

then are not guaranteed, just in virtue of their form, to be

particularly thick ones.
There are two sorts of concepts which are often said to
belong by their own nature to the thicker end of the thin-thick
continuum, which I call: (1) value-maker concepts, (2) -able
concepts.
Examples of (1) are concepts like courageous, honest,
cowardly, corrupt, elegant, tacky, melodious, insightful and so
on. Employing these concepts, at least normally, means both
evaluating and describing the thing or person one way or
another. For a woman to be courageous, she has to meet
certain descriptive conditions, like being able to overcome fear
in certain kinds of situations.1 While it is controversial how to
best understand these terms, this much can be said: they bear
an intimate relation to some relevant thin evaluative concept.
A courageous woman has something good about her, i.e.
merits a positive evaluation of some sort, at least pro tanto: as
far as her courage goes, and ignoring countervailing factors.
Likewise, a cruel action has something bad or evil about it, at
least pro tanto. I call these value-maker concepts, because
not only do they serve to evaluate, but, unlike thin concepts,
and unlike other thick evaluations, they normally also present
a reason for the evaluation: by saying the she is courageous,
we are also saying that she is good (if she is) because she is
courageous. Being courageous makes her good, at least pro
tanto. However, as before, some of these terms need not be
very thick: for instance, virtuous (as in possessing virtue,

1Of course using further evaluative

terms might be the best way to capture
the relevant conditions, e.g. courage
applies where there is something at
stake that is worth the risk. So the
conditions need not be purely
descriptive.

rather than chaste) does convey the idea that the person is
morally good because of certain personal qualities, but nothing
more than that, partly because what makes a person virtuous
is a much more contested issue than what makes her
courageous or cruel.
Examples of -able concepts are: valuable, desirable,
admirable, enviable, contemptible, etc. These concepts wear
on their sleeves, so to speak, the idea that the thing so
evaluated merits or is worth a certain attitude or response:
valuing, desiring, admiring, etc. Obviously the -able suffix
need not appear: the concepts fearsome, trustworthy,
amusing, shameful have exactly this same structure. 2 The
thickness of these concepts lies not in any particular
descriptive condition the thing so evaluated has to meet, but
rather in the fact that they describe a specific response as
merited. By contrast, good, bad and the like by themselves
tell us nothing about merited responses - or at least, nothing
as specific. The idea of merited, fitting, or appropriate
response, as I explain in a moment, plays a central role in what
follows. But note that, also in this case, there are varying levels
of thickness: e.g. valuable is a rather thin sort of evaluation,
comprising itself a number of thicker possibilities (I can value x
by admiring it, respecting it, etc.).
Evaluations on the thinner end of the spectrum provide the
background for most of the subject matter of this book. This is
because most attention will be devoted to general properties
of value as such: how is the value of something related to what
we ought or have reason to do (in a broad sense)? How does

2Nor is the suffix sufficient for

inclusion: e.g. enjoyable normally
doesnt evaluate something as worthy of
being enjoyed. Perhaps response-worthy
is a more accurate label for this
category.

value depend on non-evaluative features? How do we work

out the value of complex states of affairs? Must there be a
common measure to compare different values? For answering
these questions, thicker evaluations may well provide
examples, but they will be interesting for what they can tell us
about value in general rather than for their own sake. Having
said that, there are two chapters in which thicker evaluation
will be of more direct concern: Chapter 3, where
I consider (and reject) arguments to the effect that thin
predicative evaluation (x is good, period) is based on a
mistake, and Chapter 4, where I take some steps towards
understanding what surely is a thicker type of evaluation,
namely that involving what is good for me.

1.3 The idea of value theory

The name value theory doesnt refer univocally. There are at
least three different sorts of philosophical inquiries which go by
that name.3 One is the general study of values, and includes
the whole of moral philosophy, political philosophy, aesthetics,
and possibly epistemology, regarded as a normative discipline.
In this sense, it contrasts with all other theoretical branches of
philosophy, like metaphysics or philosophy of language.
A second, much more restricted use, stands for all
substantive views about what is fundamentally good and bad,
and the debate among them. For instance, hedonism is a value
theory, holding that pleasure is the one fundamental positive
value. As such, it is opposed to other theories, e.g.
perfectionism, holding that the excellent development of
certain abilities is the fundamental positive value, or to some
pluralist theory, holding that there is more than one

3Im drawing here on Schroeder (2012).

fundamental value, for example pleasure, knowledge, and

moral virtue.
In this sense, often synonymous with axiology (from the
Greek axia, value), value theory is usually contrasted with at
least two different kinds of inquiries. One is normative theory,
regarded as the elaboration of a view about what is
fundamentally right or wrong. To the extent that value theory
need not be specially concerned with moral values, their
subject matters are distinct. But of course, normative theory
and value theory often need to work in tandem: whether one is
a utilitarian, a deontologist, or a contractualist, a great part of
morality will have to do with morally appropriate responses
towards what is good and bad. Moreover, the distinction tends
to blur out in the case of elaborating a theory of virtue:
virtuous traits are both plausible candidates for fundamental
values, and have an intimate connection to right and wrong
action. Second, despite the names, normative theory and value
theory are both normative inquiries, in the broad sense that to
adopt a certain normative or value theory is to come to believe
that one ought or at least has reason to behave (act, desire,
feel, judge, etc.) in certain ways for certain fundamental
reasons. (For example if hedonism is true, then in principle one
should take pleasure, ones own or others, as the fundamental
goal.) Together then they stand in contrast with meta-ethics,
conceived
as
the
philosophical
study
of
linguistic,
metaphysical, psychological, and epistemo- logical questions
about ethics, spanning morality and non-moral values alike.
While certain meta-ethical views may ultimately have
normative implications for conduct, the guiding aim of metaethics is not evaluative, prescriptive, or directive, but rather
explanatory, reconstructive, or descriptive.
However, this is a book in value theory in a third and
distinct sense. As already hinted above, I will ask certain
general questions about value. I will not engage in axiology
(value theory in the second sense), since I will not argue for or

presuppose any particular view about what is fundamentally

valuable. Only for the sake of argument will I assume in some
places that a certain axiological view (e.g. hedonism) or a
certain evaluative intuition (e.g. about the value of
punishment) is correct. Nor will I argue for or against any
particular normative theory. To this extent, this is a work in
meta-ethics. But while some of the questions have a decidedly
meta-ethical flavour (How should we define value? How is
value related to non-value? Does the idea of absolute value
make sense?), others may have a more direct normative
import (How should we, in general, relate to value? Are certain
values incommensurable?). And I wont ask traditional metaethical questions about objectivity, realism, motivation, or the
nature of value judgements. Value theory, in this third sense,
purports to explore a host of structural questions about value,
in the hope of understanding at least some salient properties
that things have in virtue of being good, bad, and so on.
Also, it is important to keep in mind a terminological issue.
When we speak loosely of value(s) as a countable term, we
may refer to three different things: (i) the object or state of
affairs that has value, (ii) the features that make an object or
state of affairs valuable, and (iii) the fact that an object or
state of affairs has value (is good, bad, etc.). An axiological
theory (a value theory in the second sense above) is typically
a theory about the features that ultimately make anything
valuable, that is, about (ii). As a result of the value-making
job, we can then speak of values in senses (i) and (iii): there
will be objects and states of affairs that are made valuable, or
are bearers or carriers of value, and there will be facts or
truths about such things being good or bad. In other words, the
idea or fact of value-making is logically prior to the existence
of valuable objects and of evaluative truths. But since here I
am not doing value theory in the sense of providing an
axiology, when I loosely speak about e.g. the value of
friendship or of a wedding ring, I mean to refer to (iii): the fact

or truth that friendship or the wedding ring is valuable, rather

than friendship as a value-making feature, or friendship as a
bearer of value (although it will of course be true that
friendship is a bearer of value).4

1.4 Value and normativity

Evaluative concepts in a very broad sense include not only
good, bad, etc., but also ought, right, wrong, duty, obligation,
practical or epistemic rationality, and good reasons for actions,
attitudes, or beliefs. Certainly we can use such concepts to
evaluate people and situations from diverse standpoints, like
morality, prudence, rationality, and so on. However, in this
book I am concerned with evaluation in a narrower sense: the
sense at issue when we judge things to be good, bad, and the
like, with or without qualifications, or when axiologists
construct their theory of value.
The other concepts mentioned above (ought, etc.) are
sometimes grouped together as normative concepts, with
ought or reasons usually singled out as more basic than
others. Other labels given are deontic (but this seems to

4A fourth sense of values of course

refers to the evaluative properties
themselves: goodness, badness, etc. But
since these properties are always
attached to something (x is good, x is
a good F, etc.), then investigating the
nature of such properties (as in the FA
approach I explain below) is tantamount
to investigating facts or truths about x
being good, x being a good F, etc.

betray an excessive focus on duty, or ought in a moral sense),

and directive or prescriptive (but to say

VALUE AND NORMATIVITY20

you have a reason to prefer a light comedy to a historical

drama, if it makes for a better night out, while no doubt a
normative statement, doesnt have the force of a command or
a prescription). Normative, however, should not be thought to
refer to anything as norm- or rule-based as a quasi-legal
normative system. In this broad sense, commonly used in
philosophy, normative truths can bear the form of a rule as
well as that of particular statements about what you should do
here and now. Nor are actions the only normatively relevant
objects: intentions, desires, emotions, beliefs are all just as
relevant, insofar as it makes sense to hold that there is or there
could be good reasons for and against them. T. M. Scanlon
labels all these responses as judgement-sensitive attitudes, in
the sense that an ideally rational person would come to have
[them] whenever that person judged there to be sufficient
reasons for them and [they] would, in an ideally rational
person, extinguish when that person judged them not to be
supported by reasons of the appropriate kind (1998: 20). 5
Distinct as they may be, still the evaluative and the
normative bear an intimate relation: value is normative. This
can be said to be one of values fundamental properties. The
exact articulation of such property is a matter of debate among
philosophers, and we will be directly concerned with such
debate in the final chapter. But this is a guiding assumption of
this book: truths about value, at least, regularly entail
normative truths of some sort about actions or attitudes. Here
are some examples: that a certain hotel is overall better than
another seems to entail that you should prefer it to the other;
that a certain conduct was morally good seems to entail that
there is some reason to admire it; that a movie is terrible
seems to entail reasons to avoid going to see it, and so on.
What is relevant is that the connection between value and
reasons is entirely general: it applies to values or value claims
and reasons as such. It is not because of what better hotels are
like that they should be preferred, but simply because of what
it is for something to be better than something else. Of course,
the specific attitude or action that is appropriate to take will
depend on the valuable object at stake: for instance, a great

5Ill come back to this in Chapter 7.

VALUE AND NORMATIVITY

21

painting should be appreciated in a certain aesthetic way, and

a poor one aesthetically censured, while morally

10

VALUE THEORY

outstanding conduct is to be praised, and evil conduct

blamed or condemned. But what remains constant across
these diverse cases is that a positive value is to be responded
with a positive attitude, and a negative value with a negative
attitude of some sort.
The most straightforward way to articulate the idea that
value is normative is to say that what it is for something to be
good is nothing else than for there to be reasons, in principle
for anyone, to respond favourably to it, and likewise for what it
is to be bad, better, etc. Goodness, or being good, simply
consists in the existence of such reasons. Evaluative truths are
really normative truths in a shortened form. In the
philosophical literature, there are two existing accounts of this
form. One has come to be known as the fitting attitude account
of value (FA):
x.

FA: x is good = it is fitting to respond favourably to (favour)

FA has a long and distinguished history, starting with the

German philosopher Franz Brentano.6 The crucial term is of
course fitting. This is meant to be a fully normative term,
expressing the idea of a normative or ideal match between the
object x and the favourable response. But other terms have
been used in this connection: appropriate, suitable, correct,
worthy, deserved, merited, required, ought (x is good = x
ought to be favoured). While these terms have different
connotations, they are all supposed to be general normative

6A full list and history of FA advocates

is beyond the present aims. But any list
will include: Brentano (1969 (1889)),
Sidgwick (1981 (1907)), Broad (1942),
Ewing (1948, 1959), Chisholm (1986),
Lemos (1994), Mulligan (1998), Zimmerman
(2001). See Rabinowicz and R0nnowRasmussen (2004) for other names, and
Dancy (2000) for a short but telling
historical reconstruction.

VALUE AND NORMATIVITY

11

terms. In particular, they should not be assumed to have a

moral meaning.
The other account has been called by Scanlon the buckpassing account of value (Scanlon 1998: 95-8):
Buck-passing Account 1: x is good = x has the property of
having other properties that provide reasons to favour x.
This differs from FA in two respects. The first is that it uses the
normative concept of a reason to favour. The second is that it
makes clear a certain structural claim about value: xs being
good (and other value properties on the thinner end of the
spectrum7) is not itself a reason to favour x, but rather is the
fact that x has other features, i.e. other than goodness, which
provide reasons to favour x. To talk of xs goodness thus is not
already to indicate one reason to favour x, but rather to
announce that there are reasons to favour x. In this sense
goodness passes the normative buck, i.e. the ability to
provide reasons for attitudes, down to the features which make

7This is important. FA articulates the

normativity of goodness and other thin
concepts (and obviously of what I called
-able evaluative concepts). But is it
applicable to what I called thick
value-maker concepts? For example,
courage makes a person good or
admirable, but the property of being
courageous may not be the property of
having other properties which give reason
to admire the courageous person. Courage
doesnt have to pass the buck to
courage-making features: it is itself a
perfectly good reason to admire
somebody. The normativity of these thick
evaluative concepts seems to lie in
their immediate reason-providing (or goodmaking) ability, not in their being
analysable in FA terms.

12

VALUE THEORY

x good, rather than keeping the normative buck, i.e. providing

a reason itself. We can represent the buck-passing structure as
follows:
x has certain properties (e.g. is pleasant) ^ (there are
reasons to favour x = x is good).
The arrow ^ signifies that being pleasant provides or grounds
a reason to favour x, and correspondingly grounds xs
goodness. Scanlons account incorporates this aspect of value
into the very definition, but also advocates of FA have been
clear about this. So A. C. Ewing writes:
[T]he reason why it is proper to admire anything must be
constituted by the qualities which make the object of
admiration good ... The ground [for an attitude] lies not in
some other ethical concept, goodness, but in the concrete,
factual characteristics of what we pronounce good. Certain
characteristics are such that the fitting response to what
possesses them is a pro attitude, and that is all there is to it.
We shall not be better off if we interpolate an indefinable
characteristic of goodness besides, for it is no easier to see
that it follows directly from the nature of things that they
are good than it is to see that it follows directly from their
nature that they are fitting objects of a pro-attitude. (1948:
158, 172)
Both FA and the buck-passing account then make two central
claims:
1 Reduction Claim: value properties such as goodness are

reduced to normative relations of fittingness of, or of

there being a reason for, attitudes;
2 Normative Redundancy Claim: value properties such as
goodness do not themselves provide reasons for
attitudes (make attitudes fitting) over and above the
good-making features which already provide reasons
for attitudes (make attitudes fitting).
The two claims are related. 8 If the Reduction Claim is true, then
for x to be good is for x to have properties that provide reasons

8See Stratton-Lake and Hooker (2006).

VALUE AND NORMATIVITY

13

to favour x. Consequently, if xs goodness were a reason to

favour x, such reason would consist in the fact that x has
properties that provide reasons to favour x. But what kind of
reason would this be? First, suppose we ask Why should I
favour x? The answer because x has properties that provide
reasons to favour x is very unsatisfying. What we want to
know is what reasons there are for favouring x, and all we are
told is the fact that there are reasons. In a reason-giving
exchange, at least, mentioning goodness is little or no use.
Second, the complex fact about reasons in which goodness
consists obviously depends on other properties already
providing reasons to favour x: for instance, that x is pleasant.
To hold that goodness provides a reason of its own over and
above these other features, or on the same standing as them,
is to ignore this fundamental dependence. Imagine again a
reason-giving exchange: Why should I favour x? Because x is
pleasant, and x is good. Xs goodness would not add any
normative weight to the case for favouring x. The case for
favouring x has already been made in order for x to be good. In
this sense, if goodness is reduced to a fact about reasons (or
about fitting attitudes), goodness would be redundant as a
provider of reasons. On the FA/buck-passing picture, then,
value is normative because it just is a normative relation, and
not because it is a provider of good reasons or something that
itself makes an attitude fitting.
Is there any significant difference between the concepts of
fittingness and reasons? Should one choose FA or the buckpassing account? Arguments can be made on either side. In
favour of FA, one might say that to have a reason to do
something, one must be able to do it (reason, like ought,
implies can), whereas this doesnt hold for fittingness, since it
expresses a more ideal sort of norma- tivity. And we can clearly
imagine things being good or bad in a world where there are no
agents capable of responding as they have reason to: e.g. the
plight of dinosaurs extinction was arguably something bad, if
non-human pain is bad at all, but nobody could do anything
about it, and therefore nobody had reason to do anything
about it. This doesnt change the fact that the plight of
dinosaurs merited, even back then, a negative response.

14

VALUE THEORY

However, it is still true that, in such situations, if there were

agents capable of responding, they would have reason to do
something about it. So one might simply qualify Scanlons
account accordingly:
Buck-passing 2: x is good = x has the property of having
other
properties that provide reasons to favour x, or would provide
such reasons to suitably situated agents.9
Another possible reason in favour of fittingness comes from
what is known as the wrong kind of reasons problem (WKR). 10
There can be all sorts of reasons for positive attitudes towards
things, people, and situations, which are not good. A commonly
used example asks us to imagine that a demon will torture us
or the whole humankind if we fail to admire him. It seems that,
based on this threat, we have abundant reasons to admire him.
If the buck- passing account is true, then it follows that the
demon is good. But clearly the demon is evil rather than good,
in part precisely because of the threat. His threat is a wrong
kind of reason to admire him, because it doesnt make him
good or admirable. Such a reason to admire him does nothing
to show the demons positive value. So, one reaction to this
example might be that, whatever contingent reasons there
might be to admire the demon, that doesnt change the fact
that it is unfitting to admire him: a being like the demon simply
doesnt deserve admiration. Fittingness truths might seem to
have the sort of fixedness that goes well with truths about
value.
But buck-passers can again modify their formulation, so as
to pick out the right sort of reasons as those that stem from
fittingness:11

9See Dancy (2000), Suikkanen (2004),

Bykvist (2009), Orsi (2013b).
10See Rabinowicz and R0nnow-Rasmussen
(2004), who elaborate on previous work
by Crisp (2000), DArms and Jacobson
(2000a, 2000b).

VALUE AND NORMATIVITY

15

Buck-passing 3: x is good = x has the property of having

other
properties that provide reasons of fittingness to favour x, or
would provide such reasons to suitably situated agents.
On the other hand, advocates of a buck-passing account might
complain that the notion of fittingness really belongs to the
evaluative rather than the normative realm. This might be for
two reasons. First, the idea of an ideal match applies also
outside of the domain of human responses: a key fits a
keyhole, a chord fits a certain melody, a certain trait makes a
species fitter for a certain environment, etc. These statements
express evaluations of a functional, aesthetical, or biological
kind, without directly addressing what anyone should do.
Second, there is a similarity between reasons and fittingness
that is worth bringing out: certain features of x provide reasons
for an attitude towards x or an action, just like an attitude
towards x or an action fits certain features of x. In other words,
both notions relate objects and their properties with responses.
But reasons seem to essentially relate to the agent or the
subject of those responses in a way that fittingness need not.
Reasons are always reasons for someone to do something, or
they are not reasons at all, while a certain attitude might be
called fitting prior to being fitting for someone to take. Of
course, when appropriate, we can always add some for A
qualification, but so can we add like qualifications to claims
about value: e.g. it is (would be) good of you to act generously.
But intuitively we dont think that goodness requires mention
of an agent, even when it is the goodness of an action. So,
since normativity has to do with claims addressed to agents,
and reasons, but not fittingness, essentially apply to agents,
reasons may seem better equipped to express the normativity
of value.
However, these remarks are insufficient to make a case in
favour of reasons over fittingness. First, the fact that the word
fitting and related words have other, non-normative uses
shouldnt give us pause, because the same applies to the word

11In Chapter 8 I will explore in detail

this and other strategies to address the
wrong kind of reasons problem.

16

VALUE THEORY

reason. When we talk about the reason why Im ill or why my

shoes got broken, we are not even in the business of
evaluating anything, but rather explaining a certain event.
Such polysemy is no obstacle to isolating a clear normative
sense for these words. Moreover, the same concept can be
expressed by other words like deserved or merited, whose
meanings are unmistakeably normative.
Regarding the second remark, there can indeed be an
intuitive difference between reason (essentially agential,
belonging to agents) and fittingness. But, for one, it shouldnt
be overstated. The agent argument in a reason-predicate
often has to be left unfulfilled - whose are the counterfactual
reasons to prevent the plight of dinosaurs extinction? Do we
learn anything normatively significant by assigning those
reasons to potential agents? Secondly, the fact, if it is such,
that fittingness, like goodness, is not essentially agential might
cut both ways. If reasons are relations to agents, and value is
not, then a definition of value in terms of reasons looks less
promising than one in terms of fittingness, because it would
metaphysically require more from value than we might have
thought: the presence, albeit possibly only counterfactual, of
agents. Perhaps we can accept that there is a level of
normativity where the specification of an agent or a subject is
an inherent possibility (after all, fitting responses will be
someones responses), but no agent need feature as a
separate term of the normative relationship.
If this is true, and given added reasons of convenience (the
modified buck-passing accounts above are complicated to
keep in mind or state), in what follows I will mostly talk in
terms of the fitting attitude account of value, or more
precisely, I will speak as if FA provides the true account of the
normativity of value. The question whether it is the right
account will be directly addressed in the final chapter, where I
will consider some objections against FA and possible
alternatives to it. But this suspension of judgement will not
undermine the explanations and arguments to follow. I will use
FA as a clear articulation of the fundamental normativity of
value, but it is really the latter idea that, depending on the
context, will play the role of a constraint to be met, a
desideratum, or a helpful explanatory tool.

VALUE AND NORMATIVITY

17

1.5 Overview
So here is an overview of the contents in the light of this
assumption. In Chapter 2, I put FA to work in order to highlight
a number of different value concepts: intrinsic value, final
value, unconditional value, and others. FA, as formulated
above, should be taken as a definitional schema whose details
need to be filled depending on what sort of value we are
concerned with. For example, if final and instrumental value
are different kinds of value, such difference will have to be
captured at a normative level, e.g. as a difference in the
attitudes that are appropriate to each kind of objects. FA will
also have a role as the basis for an objection against the
proposal to collapse the final/instrumental value distinction
into the intrinsic/ extrinsic distinction. In Chapter 3, I use FA to
reply to a sceptical challenge: the idea that only attributive
value claims ultimately make sense, i.e. those expressible in
statements like x is a good F. In Chapter 4, I show how the
concept of personal value, or what is good for someone, makes
good sense once it is understood along the lines of FA. In
Chapter 5, I discuss various accounts of so-called organic
unities: complex states of affairs whose value seems to be
higher or lower than the sum of the values of the parts making
up the state. FA here has a role as a constraint that, for
instance, Moores account seems unable to meet (parts would
have values but no corresponding attitude would be fitting),
while other proposals (a conditional account) have better
resources to meet it. In Chapter 6, I explore value relations: are
being better/ worse than and as good as the only ways in
which we can compare things? Can we always compare things
value-wise? Here FA will serve as a tool to illustrate the
possibility of different value relations and the idea of value
incomparability. In Chapter 7, I directly take up the question of
what we should do with value, so I will delve deeper into the
nature and diversity of fitting attitudes - what is behind the
anodyne notion of favouring something. In Chapter 8, as I
said, I question FA as the true account of value, considering
important objections to it, possible refinements, and sensible
alternatives. But, whether ultimately correct or not as a
definition of value, a fitting attitude approach is here shown to

18

VALUE THEORY

be a framework within which several questions about value

can be asked and answered.
Apart from taking this approach as a necessary and fruitful
assumption, this book advances no general thesis about value.
It is rather meant to be a dispassionate investigation into the
many forms that value can take and to address some hard
questions raised by values. But readers will find an overall
resulting picture which is by no means neutral, for two
reasons. First, arguments in favour and against certain
particular theses will be presented and assessed in each
chapter, and naturally my treatment will incline towards one or
the other thesis: while no conclusive argument will be offered,
and no arguments will be conclusively refuted, the burden of
proof will eventually be laid on some thesis or other. Second, it
is typically arguments for certain reductive theses which will
be criticized: the collapse of final and intrinsic value, the
(connected) idea that only states of affairs are bearers of
value, the elimination of absolute value or of personal value,
the denial of a plurality of value relations, the reduction of
fitting responses to one basic kind of response. So, a certain
methodological pluralism will emerge as a result, usually
justified by an application of the fitting attitude account to the
specific question at hand. This is not to say that FA requires a
pluralist picture of value: some of the reductive theses above
have been advocated by philosophers who fully endorse FA.
But if value is normative, then reducing possibilities for value
means reducing normative possibilities, and - at times - it will
appear that such a reduction needs more justification than it
might otherwise have seemed.

1.6 Meta-ethical neutrality

There is one respect in which this book, and its reliance on the
idea that evaluative truths regularly entail normative truths, is
meant to be completely neutral. The fitting attitude account
(or some other articulation of the normativity of value), while
no doubt a meta-ethical thesis, need not be regarded as a
contender in certain meta-ethical debates about value, namely
about the nature of value judgement, the existence of

VALUE AND NORMATIVITY

19

evaluative truths, or the question whether value is a natural or

a non-natural property. This point might not be immediately
obvious: doesnt FA assume that (1) value judgement is
cognitive rather than non-cognitive - a belief about an
evaluative matter of fact, rather than a desire or emotion; (2)
there are evaluative truths; (3) value properties are nonnatural properties, since fittingness and the like are nonnatural just as goodness and the like? Lets consider these
seeming assumptions in turn.
1. FA does not assume cognitivism, for two reasons. First,
FA can be put forward as a general conceptual thesis
about the content of value judgements: when we judge
that something is good, we also judge (or are
committed to judging) that it is fitting to favour it.
Scanlon himself sometimes speaks in these terms, and
it is notable that Allan Gibbard, a prominent noncognitivist, accepts accounting for notions such as
goodness or wrongness in terms of the rationality (or
their being warranted) of certain feelings and attitudes,
like approval, admiration, remorse, indignation, etc.
(Gibbard 1990). It is then a further question whether
these judgements of rationality (or fittingness) are to
be understood in non-cognitive terms, e.g. as
expressing a non-cognitive state of acceptance of a
system of norms that permit certain actions and
attitudes (as in Gibbards analysis), or as expressing
beliefs about matters of fact concerning fitting
attitudes (as on a cognitivist approach).
The second reason why FA does not assume cognitivism is
that, even if FA were an account of value properties, and thus
in turn implicitly suggested a cognitive account of evaluative
judgements as beliefs purporting to represent those properties,
still non-cognitivism has over the years developed into what
Simon Blackburn calls quasi-realism (Blackburn 1984). Quasirealism is the project of reconstructing (or earning the right
to) such notions as normative truth, properties, belief, and
knowledge, working with the scant materials of a noncognitivist approach and a generally empiricist worldview.
Accepting a fitting attitude account of value properties doesnt
seem to make such project any harder than accepting a

20

VALUE THEORY

different account of value. If value is normative, a quasi-realist

analysis will deploy whatever resources it already has to
reconstruct such a basic feature of value.
2. John Mackies error theory holds that affirmative
propositions about value, when understood as implying
objective and categorical demands or reasons, are all
false. To speak of objective value, or alternatively of
attitudes that are fitting regardless of what anyones
actual attitudes are like, is to commit oneself to a
metaphysically and epistemologically queer view
(Mackie 1977). However, FA doesnt assume the truth
of any evaluative proposition. It only means to
articulate what evaluative properties would have to be
like, if there were any, and so, if any proposition about
them were true.
3. Another remark worth making in connection with
Mackie relates to the third question about the
natural/non-natural distinction. Mackie takes common
sense to be committed to the existence of value or
fittingness conceived as entities constituting nonnatural facts: facts which lie outside the world
described by
the
natural
sciences,
including
psychology, and which therefore require some nonempirical (and, for Mackie, queer) means of accessing
and knowing about them. But in fact FA does not
presuppose any of all this. FA says nothing about
whether value, and therefore the fittingness of
attitudes (or reasons for attitudes, as in the buckpassing account) can be understood in naturalistic
terms, for instance as depending somehow on an
agents actual or potential attitudes. The massive
mistake imputed by Mackie to common-sense
evaluative thinking lies not in FA, but in a certain metaethical (ontological and epistemological) understanding
of the concepts and properties mentioned in the
account.
This point is easy to miss, since proponents of FA have
themselves often hailed it as a welcome alternative to
naturalist and non-cognitive analyses. C. D. Broad, defending
FAs reduction of goodness to fittingness, claims:

VALUE AND NORMATIVITY

21

cannot see the need to have both a non-natural quality

of goodness, grounded on these various natural
characteristics, and a non-natural relation of fittingness
grounded on this non-natural quality. (Broad 1940:
237)12

More recent advocates, such as John McDowell (1998a, 1998b)

and David Wiggins (1987), propose FA (or something close
enough) as a sensibility theory of value with two alleged
benefits: (i) unlike for instance Moores own view, which takes
goodness to be an indefinable sui generis property, FA makes a
connection to human attitudes central to understanding value,
thus clearing somewhat the aura of extra-worldliness about it;
(ii) unlike naturalistic analyses, and unlike otherwise analogous
dispositional accounts of colours and other mind-dependent
entities, the human attitudes FA refers to are not actual or
potential reactions caused by valuable objects, but responses
merited by them - thus preserving the intuitive normativity of
value.13
Further, Scanlon defends his buck-passing account by
claiming it to back up Moores open question argument, which
is aimed to show the failure of naturalistic analyses of value
(Moore 1993: 66-9). For Moore, we can always sensibly ask, for
any naturalistic property C (e.g. being pleasant, or being
approved), whether an object that has C is also good or
12See also Ewings claim to provide the minimum non-naturalist theory of
ethics(1948: 169).

13Both McDowell and Wiggins add to these

a further point which I will consider in
Chapter 8: some of these responses (e.g.
admiration or approval) are not intelligible without a prior notion of value.
Their sensibility theory is thus not
meant as a non-circular definition of
value. See also the way Jacobson (2011)
introduces FA as an alternative to
robust realism and dispositionalism
about value.

22

VALUE THEORY

valuable, in a way that we cannot sensibly ask, for instance,

whether a man who is an uncle is also someones brother. This
conceptual openness is taken to be evidence for the
irreducibility of goodness to any naturalistic property. For
Scanlon, the buck-passing account draws a tight connection
between value and reasons which explains the intuition of
openness, and thus underwrites an anti-naturalist conception
of value:
Judgements about what is good ... generally express
practical conclusions about what would, at least under the
right conditions, be reasons for acting or responding in a
certain way. Natural ... facts may provide the grounds for
such practical conclusions ... Judging that these facts obtain
need not involve explicitly drawing these conclusions
however. Questions such as This is C, but is it valuable?
(where C is the term for some natural or metaphysical
property) therefore have an open feel, because they
explicitly ask whether a certain practical conclusion is to be
drawn. (Scanlon 1998: 96-7)14
However, while it is perfectly proper to advance FA as a model
for a non-naturalist theory of value, such a strong meta-ethical
thesis is not to be read straight off the mere formulation of FA.
If value is normative along the lines of FA, this is no blow for
naturalists.
Their job has always been to reconstruct normativity out of
naturalistic (or naturalistically tractable) materials, and
acceptance of FA will not make the naturalist programme any
harder than it already was. 15 For instance, the time-honoured

14See also Stratton-Lake and Hooker

(2006).
15This is not to say that FA is
completely neutral with respect to all
naturalisms. FA, particularly in its
buck-passing version, imposes a certain
structure on value, whereby goodness is
conceptually distinct from the goodmaking properties. This may be a problem

VALUE AND NORMATIVITY

23

ideal observer theory can analyse an attitude being fitting to x

as the attitude that a fully informed, sympathetic, and
imaginatively equipped observer would take towards x. 16 Or,
on the so-called internalist approach to normative reasons,
there being a normative reason for someone to take a certain
attitude towards x would be a matter of them being possibly
motivated to take that attitude, coherently with their other
existing attitudes (Williams 1981). Unlike the ideal observer
approach, internalism does face a difficulty when presented in
combination with a fitting attitude account of value: at least
certain values seem to be such that there is reason for

for those naturalists who identify

goodness with e.g. pleasure, or with a
disjunctive list of whatever natural
properties goodness supervenes on (see
Jackson 1998). On these views being good
is e.g. being pleasant, and being
pleasant, in turn, seems to be what
provides reasons, rather than being the
property of having other reason-providing
properties (i.e. other than pleasure).
So being good is not the property of
having other properties that provide
reasons (the good-making properties),
because one of these properties
(pleasure) is goodness. This naturalist
view sits better with a buck-stopping
account of value, where goodness is
itself a, or the, reason-providing
property, or with a definition of
reasons in terms of value (and so in
turn in terms of pleasure) (see Chapter
8). Alternatively, the superveniencedriven naturalist can in primis accept FA,
and then proceed to identify the
property of being a reason with being

24

VALUE THEORY

everyone, in principle, to respond to them in a certain way, but

inter- nalism makes such universality hostage to facts about
the possible motivation of particular individuals. But it is
noteworthy that internalists are aware of this problem. Some
(like Michael Smith and Christine Korsgaard) accept it, and
argue (or at least hope) for a thesis of convergence: qua
informed, coherent, and rational agents, our motivational sets
would overlap, at least on some issues. In turn, our reasons
would converge. So values can be normative for everyone. 17
Other philosophers (like Bernard Williams) seem to reject the
idea that values can as such be normative for everyone, or at
least that values imply the specific normativity of reasons. In
sum, the stance of naturalist philosophers is to either do their
best to accommodate FA, or (in Williamss case) reject the
assumptions that would make their account conflict with FA.
In this sense, the fitting attitude account is importantly
meta- ethically neutral. FA can be taken on board by

pleasant or with a disjunction of

naturalistic reason- providers. What
matters for this book is that these
naturalists have to rely on some account
of the normativity of value, and this is
all that is needed here.
16The idea goes back at least to Adam
Smith (1759). See Firth (1952), Lewis
(1989).
17See Smith (1994), Korsgaard (1996).
Im taking internalism to be a
naturalist thesis about reasons. This
can be disputed in Smiths and
Korsgaards cases, but the relevant
point is that their views, while aiming
to illuminate the notion of a reason in
further terms (be they naturalistic or
not), rather than taking it as a
primitive, do not thereby conflict with
FA.

VALUE AND NORMATIVITY

25

naturalists, error theorists, and non-cognitivists, not as the

ultimate meta-ethical truth about value, but as a general
truism about value which ones theory will aim to explain or
otherwise come to terms with. To the extent that FA, or
something close enough, is a guiding assumption of this book,
there is no reason for all such philosophers to decline their
interest in the value-theoretic questions explored here.18

1.7 Value theory: The questions

The value-theoretic questions addressed in this book can be
graphically represented as branching out from the basic
assumption. If x is valuable implies that it is fitting to favour x,
then we can ask a number of questions:

18Even the error theorist will be

interested. If their stance is
conservative, and ethics must be
preserved as a useful fiction, then it
will be important to understand the
value-theoretic details of such fiction.
If their stance is revisionary, then it
will be clearer to them how to revise
that part of ethics that deals with
value and fitting attitudes.

26

VALUE THEORY

Figure 1.1

CHAPTER TWO

Meet the Values:

Intrinsic, Final & Co.
2.1 Introduction
To the extent that we justify actions and attitudes in terms of
some value they produce or express, we do not normally need
to reach for fundamental, basic, or ultimate values. If a certain
medicine promises to be effective against my headache, and

VALUE AND NORMATIVITY

27

accordingly I take it (or you advise me to take it), we need not

question whether a headache-free condition would be good in
itself, or because of some other good state related to it - say,
having the capacity to think clearly. If a certain policy is
thought to reduce unemployment or improve working
conditions, we can support it without asking why having a
decent job is a good thing: maybe work, in certain conditions,
is good for its own sake, or maybe it is good because of other
good things to which it connects - financial autonomy, or the
possibility of achievement.
But certainly there are plenty of contexts where our
attention focuses on a deeper dimension of value. For instance,
many of us would take a critical attitude towards a
businessman intent on accumulating money exclusively for the
sake of it, quite regardless of how money impacts on other
aspects of his life. The feeling is that money, however
important it may be, is the sort of thing to be valued not for its
own sake, but for the sake of something else; at any rate, it
cannot be the only thing to be valued for its own sake.
Conversely, someone who appears to treat their own children
exclusively as potential workforce for the family would be the
target of blame: whether or not people can have some
instrumental value, few of us believe that ones own children
can be regarded exclusively as potential workforce. In these
and similar examples, our criticism is based on an intuitive
distinction between what is valuable for its own sake (ones
own children) and what is instrumentally valuable (money).
Similar distinctions are traditionally made by philosophers in
their search for the basic elements of a good life, and for
ultimate standards of action: the attempt is to pick out what, in
some sense, matters fundamentally, as opposed to what
matters for the sake of something else. The aim of this chapter
is to make distinctions between value concepts which are often
not distinguished, either within philosophy or in ordinary value
talk: final value, exclusively final value, unconditional value,
intrinsic value, necessary/essential value. The idea is to alert
the reader to the importance of keeping these concepts
separate. On the argumentative plane, I provide some defence
for the idea that something can be valuable for its own sake
(i.e. finally valuable, the sense in which ones children are

28

VALUE THEORY

valuable and money arguably is not), and yet doesnt have to

be unconditionally valuable, nor must it owe its value to its
intrinsic properties. This idea is worth exploring, since it has
been assumed by most philosophers that what is valuable for
its own sake can only be valuable in itself, that is, in isolation
from other things, and not conditionally on other values. Final
value, according to this tradition, must be the stopping place of
our evaluations, or is not really final. I will argue that this
tradition, which has a distinguished history and has recently
been revived, considerably restricts the ways in which it is
appropriate to value things, and depends on a questionable
assumption on what it means to be final. In section 2.6 I also
explain why we should distinguish merely instrumental
extrinsic value from other forms of extrinsic value.

2.2
Final and unconditional
value: Some philosophical
examples
When philosophers build their substantive axiologies, i.e. their
theories of what is good and bad, they generally seek to
identify what is ultimately good and bad. In stating their views,
they use various phrases. One is good for its own sake. For
instance, after an examination of peoples opinions and values,
Aristotle thus describes the sense in which eudaimonia
(happiness as a state of flourishing) is the ultimate good:
If, then, there is some end of the things we do, which we
desire for its own sake (everything else being desired for the
sake of this), and if we do not choose everything for the
sake of something else (for at that rate the process would
go on to infinity, so that our desire would be empty and
vain), clearly this must be the good and the chief good ...
Now such a thing happiness, above all else, is held to be; for
this we choose always for self and never for the sake of
something else, but honour, pleasure, reason, and every
virtue we choose indeed for themselves (for if nothing
resulted from them we should still choose each of them),
but we choose them also for the sake of happiness, judging
that by means of them we shall be happy. Happiness, on the

VALUE AND NORMATIVITY

29

other hand, no one chooses for the sake of these, nor, in

general, for anything other than itself. (Nicomachean Ethics
I 2, 7)19
Another commonly found phrase is good in itself, equally
used by otherwise very different philosophers such as Kant and
Bentham. Kant applies it to the good will, i.e. willing or
intending to act according to the moral law:
It is impossible to conceive anything at all in the world, or
even out of it, which can be taken as good without
qualification, except a good will ... A good will is not good
because of what it effects or accomplishes - because of its
fitness for attaining some proposed end; it is good through
its willing alone - that is, good in itself. (Kant 1964: 61-2)
Bentham proposes instead a hedonist view of value:
Strictly speaking, nothing can be said to be good or bad, but
either in itself, which is the case only with pain or pleasure;
or on account of its effects, which is the case only with
things that are the causes or preventives of pain and
pleasure. (Bentham 1970: 11)
Yet another phrase is good/desirable as an end, for instance to
be found in Mills statement of hedonism (here misleadingly
identified with utilitarianism, which includes but does not
reduce to a theory of the good):
The utilitarian doctrine is, that happiness is desirable, and
the only thing desirable, as an end; all other things being
only desirable as means to that end. (Mill 1861: IV 2)
How are we sure that Aristotle, Kant, Bentham, and Mill are
talking about the same thing? First, despite different
terminologies, eudaimonia, pleasure, good will, or happiness
are all contrasted with things that are good for the sake of
something else, that is to say, things that are good because

19Aristotle is obviously not the first

philosopher to draw such distinctions.
For example, see Plato, Republic 357.

30

VALUE THEORY

they stand in a certain (typically but not necessarily)

instrumental relation to the former items. For example, from a
hedonist point of view, health is good insofar as it produces,
facilitates, or simply contains experiences of pleasure, or
prevents painful experiences. Devoid of such relationships to
pain and pleasure, health would be neither good nor bad.
Second, they all mean to pick out what can be described as
final values. Since in value theory this term is not always
understood in the same way, here is the idea we are working
with. Something is finally valuable when it is valuable for its
own sake, that is, as suggested in Chapter
1, when we have reason or it is fitting to favour or value it for
its own sake. The idea is fully normative: something that is
finally valuable deserves, as such, a certain kind of treatment.
However, beyond this shared concern for final values in this
sense, other notions are mixed in the quotations above, and it
is instructive to tease them apart. One is the concept of
something that is not only finally valuable, but exclusively
finally valuable: all the value it has or could ever have is final
value. This is what after all for Aristotle sets eudaimonia apart
from things like reason and virtue, which are finally valuable
but not exclusively finally valuable, since they are valuable also
for contributing to other things, such as happiness. Also Kant is
explicit about this, as he claims that the usefulness of the good
will could not add any value to it (1964: 62). And the same
goes for hedonists: of course a particular experience of
pleasure could give us further future pleasures, and to that
extent be instrumentally as well as finally good, but there is no
non-pleasure-related state which might make an experience of
pleasure additionally valuable in this way. Pleasure could not be
worth seeking for reasons other than pleasure: it could only be
good for its own sake. So, alongside contrasting final to nonfinal values, these philosophers meant to pick out exclusively
vs. non-exclusively final values.
It is important to note, however, that the search for
exclusively final value is not inevitable. Those who hold a
pluralist axiology, whereby at least two things are finally
valuable - e.g. pleasure and virtue - would have no problem
admitting that virtue is both finally valuable, and also valuable
when and because it produces pleasure, and vice versa for

VALUE AND NORMATIVITY

31

pleasure. One such example is W. D. Rosss axiology, which

includes innocent pleasure, knowledge, and virtue as final
values (2002 [1930]). Exclusively final value thus is a notion
that only monist axiologies require: if there is no other final
value than V, then there is nothing else for the sake of which V
could be valuable. Of course, if you are a pluralist, you might
still believe that each of your chosen two (or more) final values
is exclusively finally good: e.g. that the value of virtue cannot
be enhanced by pleasure or vice versa. But the monist needs
to believe in a value that is exclusively final.
A second notion implicit in the quotations above is that
those final values are also unconditional values, in the sense
that their value does not presuppose the value of anything
else. Again, if you are monist, e.g. you think that pleasure is
the only final value around, then it follows that the value of
pleasure cannot depend on the value of anything else. For an
example of a conditional value, Kant famously mentions
happiness: someones happiness is valuable only if they are
worthy of it (due to their good will). In this sense, Kant denies
that happiness as such is valuable for its own sake. It is a
problematic question whether Kant thereby counts as a
monist, since in some interpretations it appears that now two
things are finally valuable: good will (unconditionally), and
deserved happiness (conditionally on the value of good will).
The possibility of such an interpretation already suggests
that final value need not be unconditional. For a different
example, consider a valuable relationship such as friendship.
Friends are usually disposed to help each other out, and they
take pleasure or derive happiness from their relationship.
Plausibly, if their happiness and altruistic dispositions had no
value, their friendship would have no value either. In this
sense, the value of friendship is not unconditional. But this
does not mean that we are to value friendship simply for the
sake of these other things (happiness, altruistic action).
Similarly, the value of altruistic action itself can plausibly be
said to presuppose the goodness of happiness and the
badness of suffering: if suffering were not bad, there would be
no merit in alleviating others suffering. But this, again, does
not mean that altruism is to be admired and encouraged
simply for the sake of minimizing suffering. In sum, goods like

32

VALUE THEORY

friendship, altruism, or happiness (in Kants view) can be finally

valuable and yet (unlike Aristotles eudaimonia or Kants good
will) not unconditionally so.
To prevent possible misunderstandings: of course, even
values that are in this sense unconditional (i.e. not conditional
on other values) may depend on the obtaining of certain
enabling conditions; e.g. nobody can have a good will if they
are not sufficiently mentally sophisticated creatures. Being a
mentally sophisticated creature is a necessary condition for
someone to have a good will, and therefore for the value of the
good will to be instantiated. Moreover, plausibly nobody can be
a mentally sophisticated creature in isolation from others (say,
their parents): there being other people around is another
necessary condition for the value of good will. But this is a
dependence on factual, rather than evaluative, conditions:
enabling conditions of this sort do not refer to the value of
being mentally sophisticated or the value of there being
people around. In this sense, an unconditional final value can
and normally does depend on factual conditions.
If we might be tempted to see the presence of evaluative
enabling conditions as undermining the status of a given value
as final (wrongly, as I have briefly suggested), no similar
temptation should arise with regard to factual enabling
conditions. Consider another example. Any experience we may
have depends on us being alive: dead people have no
experiences. Therefore also the value of our experiences (e.g.
our pleasures) depends on us being alive. But it does not
follow that experiences are valuable for the sake of being alive.
Rather, being alive appears valuable, if at all, for the sake of
the experiences it makes possible. The value of being alive (as
an enabling condition) is therefore non-final (see Chapter 5 for
further discussion of enabling conditions).20

20Henry Sidgwick makes it clear that the

factual or at least conceptual
dependence of beauty and knowledge on
minds is compatible with favouring
beauty and knowledge as final ends though he eventually does not take them
to be so (1981: 114).

VALUE AND NORMATIVITY

2.3

33

Intrinsic value and final value

The phrases for its own sake and in itself naturally suggest a
further consideration: what makes these things finally valuable
must somehow be found in the valuable thing, rather than in
something totally or partially outside it. This is why such
values have historically been called intrinsic, and therefore
contrasted with extrinsic ones. G. E. Moore so defines intrinsic
value: To say that a kind of value is intrinsic means merely
that the question whether a thing possesses it, and in what
degree it possesses it, depends solely on the intrinsic nature of
the thing in question (1993 [1922]: 286). And in order to
determine what has intrinsic value, he devised the so-called
isolation test: it is necessary to consider what things are such
that, if they existed by themselves, in absolute isolation, we
should yet judge their existence to be good (1993 [1903]:
236). Now, certain things just cannot exist in absolute isolation:
e.g. a state of pleasure necessarily has a subject and (at least
for some sorts of pleasures) an object (what it is pleasure at).
So the test seems unserviceable as it stands: we cannot
coherently imagine a world containing states of pleasures but
no subject and no object of those pleasures. However, in the
light of Moores definition of intrinsic value, we can propose an
amended isolation test: in order to determine the intrinsic
value of x, we have to focus on xs intrinsic nature alone, and
on that basis see if we judge x to be good - presumably, this is
something we can do with pleasures and other states. 21
For instance, Kants good will is said to keep its value
regardless of any relation to actions it may or may not
originate. For a classical hedonist, it is the intrinsic qualities of
pleasures (such as their sensory quality, intensity, or duration)
which determine their value: if I feel good because I think I
have won the lottery, then my state has the same intrinsic
value whether in fact I have or have not won the lottery. And
intrinsic value can belong, of course, also to goods, e.g. the
beauty of a work of art, whose intrinsic nature is a matter of
certain relations among different parts (say, colours and

21See Lemoss intentional isolationism

(1994: 10-11).

34

VALUE THEORY

shapes being in a certain arrangement): such relations make

up the work of art for what it is. In this sense, they are
internal as opposed to external relations, such as the fact that
the work of art provokes emotional experiences to viewers, or
that it is a unique piece in its genre. Such external relations do
not affect the intrinsic value of the object as a beautiful work
of art. As for Moore himself, the isolation test delivered the
result that the enjoyment of personal relationships and the
appreciation of beauty are, not the only intrinsic values, but
the highest ones.
But is all final value intrinsic value? So did Moore assume.
And until not long ago, such a question might have sounded
otiose. If something is to be valued for its own sake, then it
seemed obvious that what makes it worth valuing in such a
way must be found in the intrinsic nature of the object. And if
what makes something worth valuing is to be found partly
outside the object itself, then it seems that the value of the
object cannot be final. However, precisely these statements
should already make it clear that final/ non-final and
intrinsic/extrinsic
value
are,
at
least,
conceptually
distinguishable pairs: final/non-final refers to the correct or
appropriate way of valuing something, while intrinsic/extrinsic
refers to the metaphysical location of the good-making
properties of something. So a Moorean view which equates
final and intrinsic value need not be obviously true.
In particular, many believe that final values can be extrinsic:
objects which are valuable for their own sake partly thanks
to their relations to other objects. This can happen in many
ways.22

22See Korsgaard (1983), Kagan (1998),

Rabinowicz and Ronnow-Rasmussen (1999),
Langton (2007). Korsgaard, who basically
single-handedly started this line of
thought, went on to argue for an
ultimate dependence of final value on
the extrinsic property of being valued
by a rational agent. But this claim of
hers is best understood as a general

VALUE AND NORMATIVITY

35

1 One is where final value depends on the value of other

things. As suggested above, if friends happiness

were not valuable, and friendship did not contribute
to friends happiness - both facts being conceived of
as externally related to the friendship - probably
friendship would not be valuable for its own sake.
For a more concrete example: a particular fur coat
might be regarded as valuable for its own sake, as
an outstanding piece of handicraft, yet so only
assuming an appropriate evaluative background. If
fur coats were not in general instrumentally valuable
for protection against the cold they provide, this
particular coat could not have any value, let alone
any final value.23

meta-ethical thesis about the nature of

value, rather than as the denial that
there is any real intrinsic value (see
section 1.6). Being valued by a rational
agent, after all, is not to be counted
among the good-making properties on top
of whatever intrinsic or extrinsic
properties already make an object (e.g.
a fur coat) finally valuable. In other
words, intrinsic value need not be
metaphysically mind-independent value
(this is a confusion which arguably
Moore himself caused in his 1922 essay).
23Jonathan Dancy would claim that if the
enabling conditions (factual or
evaluative) for something to be a value
are extrinsic properties, but the ground
of value (the good-making feature) is an
intrinsic property, then the value is
still intrinsic (2004a: 172). In the
example, friendship could still be an
intrinsic value, even if happiness is an
extrinsic enabling condition for its

36

VALUE THEORY
2 Another

case is where final value might be

enhanced by an objects relational properties: a
fine work of art, beautiful and thus already
valuable on account of its intrinsic features, might
have its value increased by its being a unique or
rare piece (a kind of relational property, since it
implies the absence of other things, or many other
things, like it). And such uniqueness or rarity need
not make the
object additionally valuable simply in instrumental
terms, say because it increases its economic value, or
because it provides particular pleasure to viewers or
owners.
3 Yet another case is where the final value of an object does
not just counterfactually depend on the value of
something else, but results from or is exhaustively
grounded on some relation to other objects. Uniqueness
or rarity can give an otherwise unremarkable item (say, a
stamp) a new final value. A certain type of car might
have the ability to race at unusually high speed: this
relational, indeed causal, feature may conceivably make
the car valuable (e.g. worth maintaining) for its own
sake. Were it not for this feature, the car would simply
have the instrumental, non-final value that most other
cars have. Or consider Napoleons hat: a dull, worn-out
accessory, but it is not absurd to believe it worth
preserving for its own sake simply because it belonged to
such an extraordinary person. In each of these cases, of
course, there is also a long story to be told about
counterfactual evaluative dependences. For instance, in
the case of the car, we are dealing with something like
the value of pushing physical boundaries; in the case of
the hat, the value of keeping traces of historically crucial
figures, and in turn the value of history. So in these
cases, final value is extrinsic in two ways: first, it is
grounded in relational properties of the object, and
second, such relational properties make the object finally
valuable only on the condition of something else than the
object itself being valuable.

VALUE AND NORMATIVITY

37

If these examples are persuasive, then final value can be

extrinsic. It also follows that we cannot always apply Moores
isolation test to discover what has final value. Remember that
we had to imagine the object in isolation from any other thing.
Then we would miss the final value of Napoleons hat, since we
would ignore who the hat belonged to, and likewise the final
value of a rare stamp, since we would ignore how many other
stamps like it are around. We would miss the additional value
of a unique artwork, since we wouldnt know that there are no
other exemplars of that. We would miss the final value of the
fur coat, since we would have to ignore the point of producing
and wearing fur coats. And we would possibly mistake the final
value of any given friendship, since we wouldnt know if it is
actually good for the friends - we would ignore an external
condition for its final value.

2.4 The reduction to facts

Given the plethora of examples available, it seems that the
burden of proof lies on those who take final value to be always
intrinsic value, i.e. value which depends only on an objects
intrinsic features. How could one hope to show that? One
strategy starts from the question: what are the real bearers of
value, i.e. what sorts of entities really have value? Most of the
examples assume that individual objects (coats, hats,
stamps ...) are the sort of entities that can have final value. The
strategy invites us to reconsider this assumption. It seems that
we could, in principle, always understand or translate the
putative extrinsic final value of an individual object in terms of
the final value of a fact, or a state of affairs, which includes the
object together with the relevant relational properties. For
example, the final value of Napoleons hat can be seen as
really the final value of the fact that
(F1) there is a hat which belonged to Napoleon.
Alternatively, it could be seen as the final value of the fact that
(F2) this hat belonged to Napoleon.

38

VALUE THEORY

Either F1 or F2 would be both worth valuing for its own sake,

and in virtue of their intrinsic or internal properties: that the hat
belonged to Napoleon is indeed what these facts are all about,
be it F1 or F2. So this translation strategy, or move from
objects to facts, would give the result that a Moorean needs: a
picture where final value is always intrinsic. Similar moves can

VALUE AND NORMATIVITY

39

be made, in principle, with other putative cases of extrinsic

40

VALUE THEORY

final value.24
It is interesting to note that the F2-style translation, which
avoids the existential commitment, is perhaps preferable to F1
for two reasons. First, it is not obvious that to value something
for its own sake always means to value its existence. This
might work for artefacts which are worth preserving (and thus
keeping into existence), but what about wishing a painless
death to our sick dog as a way of valuing him for his own sake?
Second, precisely this point shows that in general we value the
existence of x only if we value x in the first place. F1-style
translation instead sees valuing xs existence as a prior
condition on valuing x, so it seems to get things the wrong way
around.
But why adopt such a reduction or translation manoeuvre in
the first place? Why cant we be happy with final values that
are extrinsic? One immediate reason might be simplicity.
Extrinsic final values give us a less unified picture of how value
comes into existence. And, as we have seen, we wouldnt be
able to use Moores isolation test, which might have given us a
handy procedure to look for the things to care about for their
own sake.
A more theoretical reason, suggested by Michael
Zimmerman (2005: 194), is that if there are long stories to be
told about, e.g. why Napoleons hat is a valuable thing, i.e.
stories that must refer to values other than the value of the
hat itself, then what we have is at most a derivative value. But
final values, in principle, should rather play the role of
endpoint values: once we reach them, no helpful explanation
of them can be given, except saying that they are good as
such - i.e. for their own sake. Thats why extrinsic values of
the sort indicated above cannot be final values for
Zimmerman: explanation (or rather justification) does not stop
at them. Only values that are intrinsic can play such an
endpoint role, because theres no looking beyond them to
explain why they are valuable. And it seems that only facts,
and not individual objects, can be guaranteed to play this
double role of carrying final, i.e. non-derivative, and intrinsic
value.
Moreover, the reduction to facts does correspond to a
natural way of talking about value, so its not an ad hoc or

VALUE AND NORMATIVITY

41

merely technical move. We obviously do not only evaluate

objects or persons, but also facts or situations: we say that it is
good that this and that happened, or that it would be better if
that did not happen, and so on. And it also matches with the
idea that many of our fitting responses to value are what are
generally called propositional attitudes, such as desire or
preference, e.g. desiring that the war end. If, as claimed in
Chapter 1, values must be the sort of things towards which
fitting responses can be directed, then it is easy to see why
values at least could consist in proposition-like entities, such as
facts or states of affairs.
However, the reduction manoeuvre is not easily
accomplished.
I
will mention two challenges. First, since the strategy
has to be carried out for the whole spectrum of final
values, its proponents are committed to the claim that
only propositional attitudes are fitting responses to
final value. If final values are all facts such as F2, then
it seems that attitudes, such as respect, preservation,
or certain forms of love, which are thing- or personoriented, rather than fact-oriented, cannot be
appropriate responses to final values. We respect and
love individual people (or animals), and this is a
different attitude than loving that such people have
this or that characteristic. We want to preserve
Napoleons hat, rather than the fact that it belonged to
Napoleon - indeed, preserving the latter seems to
make almost no sense. According to the reduction
manoeuvre, thing- or person-oriented attitudes might
still be in some sense fitting to their objects, but they
wont match with final value: final value will rather
belong to the sorts of states which we can value by a
relevant propositional attitude. So the fittingness of
preserving Napoleons hat will reduce, it seems, to the
fittingness of valuing or cherishing F2, the fact that the
hat belonged to Napoleon. While this latter example
may give some plausibility to the strategy (after all
preserving the hat is a way of valuing F2), it is not
equally clear that, e.g. respect and love for individual
persons can always be seen as fitting because of the

42

VALUE THEORY

fittingness of a kind of respect and love towards some

fact about these persons. In a sentence, the reduction
of finally valuable things to finally valuable facts
commits one to a reduction of fitting responses (to final
value) to fact-oriented attitudes, and the latter
reduction might be hard to sustain, given the variety in
the kind of responses that we normally identify as
valuing for its own sake. Perhaps then bearers of final
value are just as various as our responses: We value
many different kinds of things, including at least the
following: objects and their properties (such as
beauty), persons, skills and talents, states of character,

VALUE AND NORMATIVITY

actions,

accomplishments,

activities

43

and

pursuits,

44

VALUE THEORY

relationships, and ideals (Scanlon 1998: 95). 25

Second, the theoretical justification suggested by
Zimmerman for regarding all final value as intrinsic is weak.
There is no initial reason to think that final values, understood
as things worth valuing for their own sake, also need to play
the role of non-derivative or endpoint values, in the sense that
no explanation of their value can be given, except pointing to
their own intrinsic nature. Conceptually, we are again dealing
with different notions: final value, which refers to the
appropriate way of responding to something, and nonderivative value, a structural-metaphysical notion. Consider
again friendship. The value of friendship, and thus the respect
a particular friendship deserves from third parties, can be
partly explained by reference to other values: the mutual
desire for happiness that friends typically have is good; the
happiness thus achieved is also good; and, at least on some
views, friends must make each other happy by and large in
morally permissible ways. The value of friendship then is not
an endpoint value, because appreciating the value of any
given friendship is not independent of appreciating the value
of other things (happiness, morality). But valuing, e.g.
respecting, a particular friendship on the condition that it
doesnt involve immoral conduct, obviously does not mean
valuing it for the sake of something else, namely moral
permissibility. One can recognize a value as conditional on
other values, and in this sense derivative, and still fittingly
care about it for its own sake.
What does seem to be true is that, if a final value is
derivative or conditional on other values, then this reflects on
the appropriateness of responses. If I care about Napoleons
hat for its own sake, but show otherwise utter indifference
towards Napoleons sword, or indeed to objects belonging to
comparable historical figures (say, Sitting Bulls war bonnet),
then my valuing is not of the fitting kind, precisely because it is
insensitive to the structure of the final value towards which it
is directed (assuming, of course, that the hat is only valuable
for its historical connection). But such interrelation among
fitting attitudes does not subtract anything from the final value
of the object - it is not as if its value gets thinned out in the
network of attitudes required towards similar objects.

VALUE AND NORMATIVITY

45

In the attempt to reduce final values to intrinsic, nonderivative values we can see again the concern with
unconditional values which occupied Aristotle and Kant.
Moreover, if such attempt is carried out by the reduction
manoeuvre of construing the value of individual entities as the
value of facts including those entities, what we get are
essential values, i.e. values which depend on the essential
properties of the valuable object. It is an essential property of
F2 that it concerns Napoleons hat and its having belonged to
Napoleon: a different hat or a different owner would make up a
different fact. In general, a fact like that x is P has the
essential property of concerning x, or ascribing P to x. If a fact
is valuable because of such intrinsic property, then it is also
essentially valuable. In this sense, final values turn out to be
incorruptible, that is, they remain constant for as long as their
bearer remains what it is. It is a good question, to say the
least, whether it is a defensible consequence of such a view
that only unconditional and essential values are really worth
valuing for their own sake, whereas values that are had

46

VALUE THEORY

conditionally or contingently do not deserve that same kind of

VALUE AND NORMATIVITY

47

response.26

2.5

Intrinsic and conditional value

Having defended the possibility of final extrinsic values, we can

now also clarify the relations between intrinsic/extrinsic and
unconditional/conditional
value.
Lets
consider
first
unconditional value. Unconditionally good things do not
depend on the value of anything else. Therefore whatever it is
that makes them good must be looked for among their intrinsic
properties: in their intrinsic nature. As said, even unconditional
values might depend on factual enabling conditions: the
unconditional value of pleasure depends on the subject being
alive. Now, such enabling conditions can be intrinsic or
extrinsic properties of the valuable object; e.g. there being
other people around might be a factual extrinsic condition for
the value of a good will. But this doesnt mean that therefore
the good will has an unconditional extrinsic value. This is
because Moores definition of intrinsic value and the amended
isolation test for it presuppose that whatever necessary
conditions for x to exist and be intelligible as such, be they
intrinsic or extrinsic properties of x, have already been counted
in. So, for instance, it would make no sense to apply Moores
test to the good will, and conclude that the good will has no
intrinsic value because one instance of the good will in a world
without other agents would strike us as valueless. Rather,
when we apply the test, we should imagine a world where
necessary conditions for a good will to exist and be intelligible
as such are taken for granted (such as the existence of other
people besides the agent). Factual dependence on external
conditions doesnt make a value extrinsic.
What about conditional value? Here a certain object x is
dependent for its value on the value of something else. Now,
this something else can be intrinsically or extrinsically related
to x. Earlier I considered the value of a friendship, and
suggested that it may depend not only on whether it produces
happiness for both friends, but on whether such happiness is
indeed good for them or more generally valuable. Happiness,
and its value, are in this way extrinsically related to any given

48

VALUE THEORY

friendship: when we apply Moores test on friendship, its

producing valuable happiness or not is one of those facts we
should abstract from, because it is not a necessary condition
for any friendship to exist - friendship can produce misery as
well as happiness. So the value of friendship appears to be
conditional and extrinsic (but not for these reasons non-final,
as argued above).
On the other hand, if the something else is intrinsically
related to x, then x has conditional but intrinsic value. A good
example is offered by Thomas Hurkas theory of virtues, in
which virtues are attitudes of this form: loving (desiring, etc.)
the good for its own sake and hating (avoiding, etc.) the bad
for its own sake (Hurka 2001). Virtues are therefore secondorder final values. Their structure makes their value at once
conditional, because it presupposes the first-order value of the
objects of love and hate, but also intrinsic, since it is an
intrinsic feature of a virtue that it involves an attitude towards
the good or the bad.
An instructive contrast between intrinsic and extrinsic conditional value emerges from two different readings of Kants
claims about deserved happiness. If we take the valuable state
of affairs to be S is happy and deserves being happy (i.e. has a
good will), then its value is conditional on the value of the
good will, and yet intrinsic, since Ss having a good will is a
part of the state of affairs: it is intrinsically related to it. On the
other hand, if the valuable state of affairs is simply S is
happy, then its value is conditional on Ss having a good will,
but extrinsic, because Ss having a good will is externally
related to this state of affairs.
The crucial point from this discussion is that intrinsic value
can be conditional value. This is significant, since it is easy to
confuse the two notions: it is natural to wonder, how can
somethings value be intrinsic, belong to the thing itself, if it
depends on the value of something else? But just like we
shouldnt assume final value to be necessarily intrinsic, so

VALUE AND NORMATIVITY

should

we

not

assume

that

intrinsic

value

49

is

always

50

VALUE THEORY

unconditional.27

2.6 Elimination of extrinsic value?

There is a lingering concern with extrinsic values. If x owes its
value to the value of something else y, then the worry is that x
really has no value at all. Some philosophers are tempted to
draw this conclusion when considering extrinsic value of the
instrumental sort. Thus Ross on the value of acts: Whatever
value [an act] has independently of its motive is instrumental
value, i.e. not goodness at all, but the property of producing
something that is good (2002 [1930]: 133, my emphasis).
There is a sense in which the act does not contribute any value
to the world. In computing how much value the world contains,
we are not going to add instrumental value on top of whatever
value the act has caused. A world where the same valuable
states of affairs occur through other means would contain the
same amount of value (other things being equal). Hence the
act is good-causing, but not literally good.
Similarly, we may call a medicine healthy, but the truth is that
it is health-restoring (there is nothing healthy about a pillshaped chemical aggregate). What is healthy is the state you
are in upon taking the medicine.
But could we draw eliminativist conclusions for all extrinsic
value? For instance, some philosophers speak of signatory
value: an X-ray is signatorily good if it indicates something
else that is good (e.g. that the tumour has gone). (X-rays of
course are in general also instrumentally good, as aids to
medical knowledge.) Or there can be contributory value: a
certain motif in a painting is good in this sense if it contributes
to the (aesthetic) value of the painting as a whole. It is good
as a part. It would be tedious to recount all forms of extrinsic
value, because it would require drawing a list of all relevant
relations: causing, being a sign of, being a part of a whole,
being historically connected to (as in Napoleons hat), etc. But
the question now is: if we grant that one sort of relation
(instrumental) to value means that an object (an act, say)
really is valueless, why not generalize and conclude that the
concept of extrinsic value is, at best, a handy way of talking

VALUE AND NORMATIVITY

51

about value, but does not capture a genuine evaluative reality?

And if extrinsic value doesnt capture a genuine evaluative
reality, then a fortiori all final value must be intrinsic.
While not everyone agrees with Ross, it might be worth
ending this chapter by suggesting why, even if he is right
about instrumental value, not all extrinsic value can be
eliminated like that. The argument for elimination assumes
that what has instrumental value does not contribute any
value over and above the value of its causal consequences.
However, this does not seem to apply to the cases of extrinsic
final values considered above.
First, it is true that Napoleons hat has no value apart from
its historical connection. But Napoleons hat contributes value
precisely in being an exemplar of a supposedly valuable
category of things - objects (maybe of a certain kind) that
belonged to important historical figures. Destroy the hat, and
you have directly reduced the amount of value contributed by
this category. On the other hand, once you imagine away acts
and in general things that only have instrumental value - while
keeping constant the amount of valuable states of affairs they
would otherwise produce - you will not have diminished in the
slightest the amount of value in that world. And this is after all
what makes sense of why it is appropriate to value the former
for its own sake, and the latter for the sake of something else.
Second, regarding values such as friendship which are
conditional on external factors, the elimination argument
would work if it were true that the fact of friendship would
contribute no value, over and above the value contributed by
other facts which are conditions for friendships value (e.g.
mutual happiness produced within a friendship). But it is hard
to see how to show this much. While it might be held that a
given friendship loses in positive value if it makes one or both
friends worse off, what loses in value is the complex of
expectations, mutual feelings, shared history, etc., which
define friendship for what it is, and which - prior to
philosophical arguments to the contrary - constitute its final
value and its specific valuable contribution to the amount of
value in the world. Of course some philosophers (e.g.
hedonists) would be ready to make the substantive claim that
friendship is only instrumentally valuable, e.g. insofar as it

52

VALUE THEORY

promotes the general happiness. But this move would not be

acceptable, since here we were looking at possible reasons
why extrinsic values are eliminable (and a fortiori non-final)
qua extrinsic. It seems that no general argument for
elimination can be found.

2.7

Summary

In this chapter I have distinguished several pairs of value

concepts:
final vs. non-final value: what is fitting to favour for
its own sake vs. what is fitting to favour for the
sake of something else;
MM

exclusively final vs. non-exclusively final value: what is

fitting to favour only for its own sake vs. what is fitting to
favour for its own sake and for the sake of something
else;

unconditional vs. conditional value: what is fitting to

favour (for its own sake) independently of whether it is
fitting to favour something else vs. what is fitting to
favour (for its own sake or not) not independently of
whether it is fitting to favour something else;
intrinsic vs. extrinsic value: what is fitting to favour
because of its intrinsic properties vs. what is fitting to
favour (for its own sake or not) partly because of its
extrinsic properties;

MM

essential/necessary vs. contingent value: what is fitting

to favour in all possible worlds where it occurs vs.
what is fitting to favour in some but not all possible
worlds where it occurs.
I

have argued that we should keep an open mind as to

how these concepts relate to one another. In particular,
I have suggested that final value need not be
unconditional value, and criticized a strategy to equate
intrinsic and final value based on the idea that only
facts or states of affairs are bearers of final value.
Since the resulting view is that there can be final

VALUE AND NORMATIVITY

53

extrinsic values, then I had to defuse a worry that the

very category of extrinsic value could be dispensed
with.

CHAPTER THREE

The Challenge against

Absolute Value
3.1 Introduction
Things can be good or bad relatively, i.e. in relation to other
things: good for people or animals, good as objects of a certain
kind, good to buy/enjoy/know ... But the philosophical search
for general final values has traditionally been the search for
values which are not in this sense relative, but absolute: what
is (finally) good simpliciter, or plain good. With the possible
exception of Aristotle, for whom the concept of what is plain
good overlaps with the concept of what is good for human
beings, the philosophers mentioned in the previous chapter
(Kant, Mill, Moore, Ross) believe that their chosen final values
(good will, deserved happiness, pleasure, beauty enjoyed, etc.)
are, in this sense, things that are absolutely good. Moreover,
while our everyday evaluative talk no doubt makes extensive
use of good and bad in their relative senses (more or less
explicitly), it is not uncommon to find oneself assessing general
or particular situations and outcomes with no specific
relativization in mind:
military intervention in Syria is a really bad idea,
human extinction would be the worst possible
outcome, critical thinking is better than passive
acquiescence.

54

VALUE THEORY

Notice: we are not necessarily thinking that these things would

be good or bad for or relatively to someone or something, nor
that military intervention is really bad as an idea, or that

VALUE AND NORMATIVITY

55

human extinction is bad as an outcome, or that critical thinking

56

VALUE THEORY

is better as ... (what?).28

Some philosophers have argued that all such talk of absolute
value is meaningless. In this chapter I explain the challenges
brought by P. Geach, P. Foot, and J. J. Thomson. Then I present
one reply by M. J. Zimmerman, notice one problem with it, and
briefly suggest how to remedy it.

3.2 Geach and attributive

goodness
Peter Geach starts from looking at how we ordinarily use terms
such as good and bad. Like most other adjectives, in English
we use good both in predicative ways: the meal was good,
its good that you came early (= that you came early is
good), etc., and in attributive ways: that was a good meal,
this is a good knife, etc. This is a fact about grammar. But
what Geach focuses on is the logical behaviour of our ordinary
terms: what kind of inferences are legitimate, given our
ordinary use of these terms? Only this kind of analysis will show
something philosophically interesting about the concept of
goodness. Geach defines as logically predicative an adjective A
if x is an A y (where y is a noun) splits up logically into x is
A and x is a y; otherwise it is logically attributive (1956: 33).
The intuitive idea is clear enough: a logically predicative
adjective allows certain intuitive patterns of inference that an
attributive one does not. For example:
1

This is a square photograph.

This is a postcard.

3 Therefore, this is a square postcard.

This is an acceptable inference. Therefore square is
predicative. But consider an analogous inference with the
adjective fast:
1 Sue is a fast turtle.
2 Sue is an animal.
3 Therefore, Sue is a fast animal.

VALUE AND NORMATIVITY

57

The conclusion is false. But the point is not just that the
conclusion is false, but rather that it is not warranted, given
the premises. This can be seen by considering the following
inference, which shows how good, like fast, is attributive:
1 Sarah is a good lawyer.
2 Sarah is a chess player.
3 Therefore, Sarah is a good chess player.

58

VALUE THEORY

Now, Sarah is a good chess player (lets suppose). But clearly

the premises are insufficient to establish that. The inference is

VALUE AND NORMATIVITY

59

invalid, even if premises and conclusion happen to be true. 29

According to Geach, what explains the intuitive difference
between the inferences with x is a square y and those with
x is a fast (good) y is that the premises have a different
logical structure: the former is decomposable (it can split), the
latter is not (it cant split). X is a square y has the following
structure:
x is a square y = x is a y and x is square.
This structure makes explicit why the inference above is
acceptable. It is acceptable because it is valid:
1 This is a photograph and this is square.
2 This is a postcard.
3 Therefore, this is a postcard and this is square.

Now, if x is a good y had a similar logical structure, it would

produce equally acceptable inferences. But it doesnt. So x is a
good y must have a different structure - it cannot split into x
is a y and x is good. Therefore, in this sense, Geach claims
that:
1.

The adjective good is logically attributive.

60

VALUE THEORY

But presumably an analysis of the logical behaviour of good

will not just show us something about the word good, but

VALUE AND NORMATIVITY

about the concept of goodness itself:

61

30

2.

3.

If good is logically attributive, then the concept of

goodness is attributive.
Therefore, the concept of goodness is attributive.

If the concept of goodness is attributive in the sense that x is a

good y is not decomposable into x is good and x is a y, this
has obvious epistemological consequences (ibid.: 34). Knowing
that a Volvo is a good car doesnt involve joining two separate
pieces of information: that a Volvo is a car and that it is good.
Its not even clear what it means to know the latter as such.
Likewise, knowing that Sarah is a good lawyer doesnt involve
knowing that Sarah is good. If this is understood as Sarah is a
good person, this information may be quite irrelevant to
knowing whether Sarah is a good lawyer. Compare with this is
a red car. In this case there are two relevant pieces of
information to be learnt separately: that it is a car (suppose we
are colour-blind), and that it is red (if we dont know what a car
looks like).
And if the concept of goodness is attributive, this constrains
the ways in which we can meaningfully think about the
property of goodness:
4.

If the concept of goodness is attributive, then there is

no such thing as being just good or bad, there is only
being a good or bad so-and-so. (ibid.: 34)
That is, if good is attributive, i.e. it is a mistake to think of x
is good as a free-standing proposition detachable from x is a
good y, then it is a mistake to think there can be a freestanding property of goodness that can equally attach to
different kinds of objects. If Sarah being a good lawyer and the
Volvo being a good car were a matter of Sarah being good and
the Volvo being good, then we could think of Sarah and the

62

VALUE THEORY

Volvo as possessing a common property. But the idea that

VALUE AND NORMATIVITY

63

goodness is attributive rules this out.31

Positively, goodness is always relative to some specification
of the thing or person we call good or bad: good as a lawyer,
good as a car, good as food. In general, the only acceptable
form is x is a good K, where K stands for some appropriate
specification of the sort of thing that x is, or that x is good as
(a folded sweater can be good as a pillow without being a
pillow). Goodness is always goodness of its, or of a, kind. But
what to do with predicative uses of good?
5.

If the concept of goodness is attributive, then any

occurrence of predicative uses of good, e.g. in Julien
is good or in helping others in need is good, can be
explained away as an attributive use in disguise, where
an appropriate accompanying substantive can be
found, or else it is meaningless (i.e. doesnt correspond
to a legitimate concept).

Therefore all talk of values, final or intrinsic, either needs to be

understood as elliptical for talk of attributive value, or else is
meaningless. Consider claims traditionally made in axiology:
pleasure is good is either meaningless, or else might be
understood as pleasure is a good state of mind; likewise,
justice is good perhaps can be interpreted as justice is a
good arrangement. However, Geach warns that the
substantive to which good can logically attach must be
contentful enough as to convey either a criterion of identity or
a standard of goodness (ibid.: 41), and unlike functional or
role terms such as knife or hitman, it is not clear that the
notions of a state of mind or an arrangement can convey by
themselves a standard of goodness for things of that kind.

64

VALUE THEORY

Likewise for terms like state of affairs or (possible) world. 32

Geachs argument thus presents a challenge to substantive
value theory as classically understood, because it condemns
as meaningless the application of value concepts to objects or
states of affairs in the absence of appropriate standards of

VALUE AND NORMATIVITY

65

evaluation for those kinds of things. 33 But such is the use of

value concepts in philosophical axiologies which aim to tell us
what is just finally valuable, or provide a list of such just finally
valuable states of affairs, as we have seen in the previous
chapter by considering claims by Kant, Mill, Moore, and Ross.

3.3 Foot and the virtues

Philippa Foot targets the notion of an unqualifiedly good (and
better, best, etc.) state of affairs as used, in particular, by actconse- quentialists in trying to construct their moral view that
an action is right if, and only if, there is no alternative action
which brings about a better state of affairs. Consequentialism
owes its appeal to the compelling thought that it can never be
right to prefer a worse state of affairs to a better (Foot 1985:
198). If talk of preferences is turned directly into talk of actions,

66

VALUE THEORY

we arrive at act-consequentialisms moral view. 34 Foot thinks

we can dispense with consequentialism in all of its forms only
by dispensing with its compelling thought. In turn, we can do
that only by throwing into doubt the very legitimacy of talk of
simply better or worse states of affairs. Her argument is based
partly on (1) some kind of semantic continuity, and partly on
(2) eliminating alternatives.
1 Foot offers the sketch of a meta-ethical view of what it

is to claim that things are good, better, etc., when no

further
specification is given, e.g. thats a good thing. When we
use such unqualified evaluations about teams in a contest,
or about people we care about, it seems clear to her that
our claims have only speaker-relative truth conditions. They
signal, or rather report, that the speaker is interested in
whatever they claim to be good, whether for their own sake
or for the sake of others. So, Foot argues, why suppose that
we use evaluative terms in a completely different sense
(i.e. non-speaker-relative) when we comment on some
event that doesnt involve us personally, e.g. that was very
bad when talking about the disasters of some distant
earthquake (ibid.: 201)? At least, any proposal to interpret
such statements in a non-speaker-relative way - e.g. the
proposal that we mean to ascribe to the event some
property such as being simply bad - needs special
justification.
Foot considers and eliminates two alternative
interpretations. According to the first, what we mean to say
in contexts where we are not personally involved is that
things are good, bad, better, etc., from an impersonal point
of view. Foot quickly dismisses this, by arguing that the
concept of an impersonal point of view can only be
understood by reference to the concept of maximum
welfare. But the claim made by welfaristic consequentialism
that the (impersonally) best state of affairs is the one
which contains maximum welfare is not supposed to be
true by definition.
The second interpretation is: some state of affairs is
good, bad, better, etc., from the moral point of view. Foot
claims that simply adding this qualification will not make

VALUE AND NORMATIVITY

67

the concept intelligible. The qualification must be given

some content. And the most plausible content refers to the
concept of a moral person: there are some things that a
moral person must want and aim at in so far as he is a
moral person and ... he will count it a good thing when
these things happen and a good state of affairs either
when they are happening or when things are disposed in
their favour (ibid.: 204). What this interpretation shows,
however, is that the concept of a good state of affairs
presupposes moral concepts, and so cannot be used as part
of a morally neutral premise to arrive at moral
conclusions. For instance, the idea of happiness as a
good state of affairs from a moral point of view
presupposes the idea of a benevolent agent who must
aim at such a state. Likewise, other states of affairs we
might be inclined to call good will refer back to some
virtue or other. In this sense, the best interpretation we

68

VALUE THEORY

can give of a good state of affairs is parasitic upon

VALUE AND NORMATIVITY

69

moral concepts.35
If Foot is right, then the claims about unqualified value made in
axiology are really speaker-relative, or are simply a by-product
of our moral thinking, or else are meaningless. In this sense,
Foots challenge is somewhat less radical than Geachs: she
accepts the legitimacy of talk of goodness that is not reducible
to talk of attributive goodness, but points out its dependence
on moral concepts.
While less radical, Foots view has two consequences which
significantly reduce the importance of traditional axiology. First,
if the only acceptable claims about good states of affairs are a
by-product of moral thinking, then such evaluative claims
cannot serve as an independent basis for all or even some
moral claims, contrary to what both consequentialists and
some non-conse- quentialists (e.g. Ross) believe. Axiology
would lose its role as a theoretical guide for practical decisions.
Second, Foot must regard as meaningless any evaluative claim
that doesnt obviously correspond to what the moral person
would want. For example, if the advancement of knowledge or
the enjoyment of beauty for its own sake are not intelligible
aims of a moral person as such (what kind of moral virtue
would have that as its aim?), then accordingly the claims that
knowledge or the enjoyment of beauty are valuable for their
own sake are meaningless, or at best only have speakerrelative
meaning. It is easy to see how this undermines the sort of
claims that are ordinary currency in axiology. Like in Geachs
view, it is not that such claims are false, but rather that the
very debate on whether they are true or false is based on a
misconception.

3.4 Thomson and goodness in a

way
Following Geach, Thomson accepts that good is an attributive
adjective. But she rejects the implication that good must
always be accompanied by an appropriate substantive, or else
it cannot be applied. She thinks that Geachs conclusion is both
not sufficiently radical and too narrow. It is not sufficiently

70

VALUE THEORY

radical, since it implies that goodness still is a respectable

property, only a relative one like bigness. If goodness is
relative in this way, one could still in principle construct
piecemeal axiologies for the kinds of things which admit of
standards of evaluation (and notably, Geach thinks one can do
this for a good person and a good action). But Geachs
conclusion is also too narrow, because x is a good K is only
one of the many ways in which one can meaningfully use
good and its cognates: things can be good to eat, drink, read;
good for use in doing something; good for someone or
something; good as a thing of that kind, and so on. Indeed,
Thomson argues that understanding x is a good K depends on
knowing, from the context of utterance, in which of these ways
of being good one means that x is a good K. One, but only one
among the possibilities, is that x is good as a K, which is what
Geach seems to mean by the idea that the accompanying
substantive must be able to convey standards of goodness. In
general, x is good is always elliptical for x is good ..., where
... is replaced by an adjunct, i.e. an answer to in what
way(s) is x good?. Otherwise put, x is good is a structurally
incomplete expression, which needs to be resolved in order to
appear in a meaningful sentence.
In Thomsons view, all goodness is goodness in a way
(1997: 276), or alternatively, somethings being good is
nothing over and above one or more of the different ways of
being good. It follows that goodness is neither an absolute
property, nor a property simply relative to a certain kind of
thing. Metaphysically speaking, Thomsons view can be
interpreted in two rather different ways:
(say, a book) or a state of affairs, a person, etc., and
some appropriate adjunct (an activity: to read, a
person: for you, a comparison class: compared to
other philosophy books, etc.):
x is good [insert appropriate adjunct] = x G adj.a (x G
adj.b, etc.)
In a sense, this would not be much different from
Geachs thesis, except for the larger variety of
adjuncts which Thomson admits. For this reason, this

VALUE AND NORMATIVITY

71

interpretation doesnt seem to be what Thomson has in

mind.
2 Being good is not a relational property which, so to speak,
remains stable across different kinds of relata, but rather
there are as many goodnesses as there are ways of
being good. In this sense, e.g., being good to read, being
good for your soul, and being good as a philosophy book
are three different relational properties: (subscribed
letters: a = activity, f = for, k = relative to the kind).
x is good to read = x Ga read
x is good for your soul = x Gf your soul
x is good as a philosophy book = x Gk philosophy books

This seems closer to Thomsons intentions. If Thomson is right,

then there is no room for unqualified value claims such as
pleasure is good, pain is bad, deserved pain is better than
undeserved pain and so on, nor could one make similarly
unqualified claims such as acting so and so will make things
better, or this and that will make the world a better place.
These formulations seem to intentionally evade the question,

72

VALUE THEORY

good, bad, better. etc., in what way?36 Consequently, we cannot

hope to build normative claims, i.e. claims about what we have
reason (morally or otherwise) to do, at least in part on the basis
of unqualified value claims (this is what some deontologists,
such as Ross, think they can do). Much the less can one hope
to build a full normative theory on such basis (this is what
classic consequentialists, like Mill and Moore, think they can
do).
Thomson does provide her own basis for (moral) normative
claims: the virtues, which in her view constitute second-order
ways of being good. They are second-order since they
depend on one particular way of being good: being good for. A
second-order way of being good/bad, such as being just/unjust,
being kind/cruel, etc., is such only insofar as being just, kind,
etc., is, in more or less direct ways, good for people, i.e. insofar
as whatever else may be true of the people among whom we
live, it is better for us that they have the trait than that they
not have it (ibid.: 284). But here we are concerned not with
Thomsons normative views, but with the idea itself of
goodness in a way, which Thomson presents as an alternative
to Moorean or unqualified goodness. Despite the differences
between ways of being good, Thomson accepts that they share
a core feature which makes them all ways of being good.
Thomson claims: Intuitively, for a thing X to be good in one
of the first-order ways is for X to benefit someone or some
thing Y (which might or might not be X itself) in the
appropriate way, or to be capable of doing so (ibid.: 289).
Examples: something is good to look at, if looking at it pleases
people; someone is good at playing chess, if she is capable of
winning (often enough), thus benefiting herself in this sense; a
frying pan is good for use in making omelettes, if it answers to
what people making omelettes want from it. Finally, some X is
good for Y, if X conduces to Ys doing what it is designed to do
(if Y is an artefact), or if X conduces to Ys reproductive
success (in case of animate objects), or, in case of human
beings, if X conduces to Ys reaching one or more of the goals
Y would have under ideal conditions (that is, were Y fully

VALUE AND NORMATIVITY

73

informed about relevant facts, and their goals were not the

74

VALUE THEORY

result of some inappropriate influence) (ibid.: 296).37

VALUE AND NORMATIVITY

3.5 Zimmerman's ethical

goodness

75

76

VALUE THEORY

Michael Zimmerman is among the few who have directly

VALUE AND NORMATIVITY

77

responded to the Geach/Foot/Thomson challenge. 38 He makes

three main points, taking Thomson as his principal opponent.
The first point is that Thomsons view seems incoherent. On
the one hand, Thomson claims that if goodness is always
goodness in a way, then there is no such property of being just
good. On the other hand, she clearly believes that also being
good for someone or something is always being good for
someone or something in a way, and we have just seen such
ways of being good for. But she doesnt therefore conclude that
there is no such property of being just good for. So, at the very
least, she owes an explanation of why being good is different
from being good for (Zimmerman 2001: 20).
Zimmermans second point is that even if goodness is
always goodness in a way, it simply doesnt follow that
goodness is not a property, but only that goodness is a
determinable property, i.e. a property that needs to be
determined or specified in one way or another. Such is the
case for being shaped or being coloured: nothing can be
shaped without being shaped some way (square, circular, etc.),
but this seems no reason to deny being shaped or being
coloured the status of properties (ibid.: 19-20).
Zimmermans central reply, however, starts from noticing
that the property (or concept) traditionally assumed by Moore
and other axiologists is not some form of pure, unadorned
goodness, but intrinsic goodness. And being intrinsically good
seems certainly one way in which things can be good.
Moreover, intrinsically good seems to be logically predicative
rather than attributive: from Judith being pleased is an
intrinsically good state of mind, and a state of mind is a state
of affairs, we can infer Judith being pleased is an intrinsically
good state of affairs (ibid.: 21). But, as Zimmerman notes,
Thomson could still ask: what kind of way of being good is
that? And two natural answers are not satisfactory:
1 x is intrinsically good when xs goodness depends only
on xs intrinsic properties (see Chapter 2). This
answer, true
or not in its own right, seems to appeal to some concept
of pure, unadorned goodness, which Thomson would
reject.

78

VALUE THEORY

2 x is intrinsically good when x is non-derivatively good.

But xs being derivatively good means that xs
goodness is owed to some extent to the goodness of
something else. And this (despite its adverbial form) is
not really a way of being good, but rather a way of
being related to something else which is good (in some
way). (Recall the doubts in Chapter
2 about whether instrumental value is a form of value at
all.) A fortiori, being non-derivatively good is not a way
of being good either.
Zimmermans answer to Thomson is that being intrinsically
good, in the sense that interests value theorists, coincides with
being ethically good: common candidates for intrinsic
goodness, such as pleasure, knowledge, or beauty are such
that there is a moral requirement to favor them (welcome
them, admire them, take satisfaction in them, and so on) for

VALUE AND NORMATIVITY

79

their own sakes (ibid.: 24, authors emphasis). 39 Being

ethically valuable in this sense seems indeed to be a distinctive
way of being good: what is good for people, or good as an
object of its kind, might or might not be also ethically valuable.
Moreover, the notion of ethical value doesnt make any appeal
to pure, unadorned goodness. If so, it seems we can restore
the idea that when we talk of what is just good, or of good
states of affairs, and so on, we are ascribing a specific
evaluative property: being ethically good.
However, it can be easily seen that Zimmermans proposal,
as it stands, and as he explicitly notes (ibid.: 26, 28), is not
distant from Foots idea that any meaningful concept of good
state of affairs must implicitly refer to what a moral person
would aim at. On the reasonable assumption that a moral
person would favour (or aim at) x for its own sake if and only if
there is a moral requirement to favour x for its own sake, the
two proposals really overlap.
But as we have previously remarked, if Foots thesis is correct,
then axiology would lose its role as a guide for moral claims,
nor would it be possible to debate the merits of things like
beauty or knowledge in the absence of a concept of a moral
virtue which has it as its point to aim at or favour such things
for their own sake.
At the very least, we need to convince ourselves that the
moral person as such aims at or favours such things as
knowledge or beauty for their own sake, or these things wont
have intrinsic (i.e. ethical) value. Perhaps she does so aim, but
the point is that axiology is thus made hostage to such a
question. And this seems to invert the order of explanation in
traditional axiology, in which first certain intrinsic goods are
specified, and then certain attitudes towards them are

80

VALUE THEORY

declared to be virtues and vices. 40 Therefore, pending further

explanations, it is not clear that Zimmermans answer to
Thomson can reinstate the ambitions of substantive value
theory.

VALUE AND NORMATIVITY

3.6
A better reply: Absolute
value and fitting attitudes

81

82

VALUE THEORY

However, Zimmermans insight can be salvaged by removing

VALUE AND NORMATIVITY

83

the reference to a specifically moral requirement.41 In effect,

this amounts to answering Thomsons challenge by offering a
fitting attitude account (FA) of the way in which things like
pleasure, knowledge, etc. are claimed to be simply good: x is
good if and only if it is fitting to favour x. In axiology, we will be
interested in final values, i.e. in what it is fitting to favour for
its own sake. It is indeed peculiar that Thomson never seriously
considers this long-standing idea about value. Perhaps this can
be explained by her focus on Moore: Moore famously rejects

84

VALUE THEORY

any such definition of intrinsic value.42 But he sometimes

remarks that x has intrinsic value and some such phrase
about worthy or fitting attitudes are logically equivalent, and
this is all that matters for present purposes: whether
definitional or not, the connection of final value to fitting

VALUE AND NORMATIVITY

85

attitudes tells us something relevant about the way of being

86

VALUE THEORY

good that being finally good is.43

Recourse to FA as specifying a way of being good would
reinstate the role of axiology also against Foots challenge.
Evaluative claims will be conceptually independent of moral
ones: that it is fitting to favour x doesnt entail that it is
morally fitting to favour x, or that a morally sensitive person
would favour x. Of course, for some xs it is morally fitting to
favour them for their own sakes: peoples deserved happiness,
morally good traits, etc. But this will be a fact pertaining to the
specific nature (i.e. the supervenience base) of the value in
question, and not a truth about final value as such.
Conceptual independence allows one to construct some or
all of ones moral claims on the basis of axiology. For instance,
the compelling moral thought which Foot ascribes to
consequentialism:
it cannot be right to prefer a worse state of affairs to a better
seems to simply derive from a non-moral thought that is
equally compelling (indeed self-evident, given a fitting attitude
account of better):
it cannot be fitting to prefer a worse state of affairs to a
better,

VALUE AND NORMATIVITY

87

plus the plausible assumption that if F-ing is not fitting, then F-

88

VALUE THEORY

ing is not morally right either.44

It is not clear how a defender of Thomson or Foot might
object to this amendment of Zimmermans rejoinder. Perhaps it
will be said that a not-further-specified concept of fittingness
presents the same difficulties as goodness simpliciter. One idea
might be that something must be always fitting to favour one
way or another. But this demand is satisfied by FA: final values
are fitting to be favoured for their own sake. However we
understand this phrase and the associated attitude, it seems to

VALUE AND NORMATIVITY

89

provide a characterization of one way of being fitting to

90

VALUE THEORY

favour.45
Another idea might be that we should specify not only the
mode or direction of being fitting to favour (final or non-final),
but also its normative dimension, like Zimmerman does by
talking of moral fittingness. And if we do this, then again we
wont have an independent notion of final value which we can
use in constructing normative claims. In reply, we can say that,
even if it were possible to always assign a normative
dimension to fittingness claims, e.g.:
x (beauty/knowledge/humour ...) has final value = it is
aesthetically/epistemically/comically fitting to favour x for its own
sake
when final value is concerned, we assume that favouring these
things is also and simply normatively fitting. In terms of
reasons, the question is whether the various aesthetic,
epistemic, comic, and moral reasons stand as genuine reasons
to favour x for its own sake. This is what matters, if final value
is supposed to be a guide for attitudes and action. Mere
assignment to a putatively normative dimension doesnt add
anything significant in this respect.
Yet another objection might be that fittingness, like other
normative notions, doesnt hang in the air, but must attach to
agents. Again, FA can accept this: final values are fitting to be
favoured for their own sake, in principle, by any agent who is
capable of favouring them. If, however, the demand is that FA
provide an answer to why this agent?, then it is a demand too
many. For all that FA says, it might be that there is some kind
of further story to tell about the relation between fittingness
and particular agents (perhaps something along the lines of
Williams internalism about reasons), but providing such a
story involves taking a step into meta-ethical analysis. FA, on
the other hand, is supposed to hold true regardless of whether
evaluative and normative judgements are to be explained in

VALUE AND NORMATIVITY

91

realist (naturalist or non-naturalist) or non-realist terms (see

92

VALUE THEORY

section 1.6).46

3.7 Summary
In philosophy and ordinary life, it is not uncommon to evaluate
things absolutely: from no particular point of view, or relative
to no particular standard. Many philosophers have argued that
all such talk is ultimately meaningless (Geach, Thomson) or at
best derivative upon moral concepts (Foot). By discussing and
refining Zimmermans reply to this critical challenge, I have
tried to show how a fitting attitude account, or at least
elucidation, of what is meant by absolute value can make such

VALUE AND NORMATIVITY

93

a concept both intelligible and useful in the construction of

94

moral theories.47

VALUE THEORY

VALUE AND NORMATIVITY

95

CHAPTER FOUR

Personal Value
4.1 Introduction
In ordinary conversation, it is extremely common to evaluate
how good or bad things are for determinate people, how they
affect their interests, welfare, or well-being, whether they
benefit or harm them. We can assess experiences, events,
relationships, or whole lives as being good in this sense. We
can also regard particular objects as having personal value for
some people: a photograph, a ring, etc. And philosophers have
engaged in theorizing about what makes a life good in this
personal or prudential sense: hedonism, eudaimonism,
perfectionism, objective list theories all try to give answers to
this question, quite independently of whether pleasure,
eudaimonia, etc., are also good in some absolute sense. In this
chapter, I stand aside from such substantive questions, and
investigate the very concept or property of being good for
someone. Starting with G. E. Moores critical view, I will then
explore how a fitting attitude account of value can help here. I
will then discuss the distinct concept of agent-relative value,
and understand its connection with personal and impersonal
value.

4.2 Moore on good and good for

The history of philosophical approaches to the concept of being
good for someone, like for the concept of intrinsic value, can
be made to start with Moore. In conjunction to his enthusiasm
for intrinsic, and indeed absolute, goodness, we find him

96

VALUE THEORY

deeply sceptical about the viability of goodness for. Moores

discussion takes place in the attempt to refute ethical egoism,
i.e. the thesis that everyone ought to pursue their own good, or
what is good for them, as the only end. Moores aim is to show
that good for me or my own good are unserviceable concepts,
at least if one wants to build an ethical theory out of them.
Moores view then is nicely symmetrical to the attempt, by
Geach and others, to discredit absolute good, at least if one
wants to build an ethical theory or certain ethical claims out of
it. Therefore it is worth keeping in mind his critical argument.
Moores central claim is that when I talk of a thing as my
own good all that I can mean is that something which will be
exclusively mine, as my own pleasure is mine ... is also good
absolutely; or rather that my possession of it is good
absolutely. The good of it can in no possible sense be private
or belong to me; any more than a thing can exist privately or
for one person only (1993: 150). In the terms introduced by
Geach (see previous chapter), Moore seems to treat good for
A as a logically predicative adjectival phrase, because it can
be logically split up: if x is good for A, then x is good
(therefore, good absolutely) and x is somehow related to

VALUE AND NORMATIVITY

A.

97

Given this premise about the meaning of good for

me, it can be seen how ethical egoism collapses. If my
own good is something absolutely good, then anyone
in principle ought to pursue it, and the same goes for
anyones own good: any agent in principle ought to
pursue any other agents own good. The distinction
between my own good, or what is good for me, and the
good of others, or what is good for others, no longer
holds any fundamental normative significance.
There are a number of replies one could give to Moores
argument. One is this: the ethical egoist need not make any
claim about goodness. She could simply claim: everyone ought
to pursue their own happiness (or pleasure, or success, etc.) as
the only end. Whatever else might be said against such a view,
it is left untouched by Moores argument. A second reply might
be that all they mean by x is good for me is x benefits me,
and then interpret the latter as a non-ethical claim like x
increases my chances of getting what I (really) want, or some
hypothetical claim about what I would desire if I were in ideal
conditions of information and psychologically well-functioning.
Again, this sort of view would make egoism immune from
Moores criticism.
However, such replies explicitly turn the concept of good for
into a non-normative concept: what is good or bad for me
would not be immediately related to what I or others have
reason to do. Moreover, consider the second reply: x is good for
A = A would desire x if A were in ideal conditions of information
and psychologically well-functioning. Couldnt A desire what is
absolutely good or good for others as well as what is good for
her? Being desired by such an ideal subject might be a
48

98

VALUE THEORY

necessary but not a sufficient condition for something to be

VALUE AND NORMATIVITY

99

good for her.49

So a more interesting reply to Moore would try to
rehabilitate the intrinsic ethical significance of the concept of
good for as distinct from absolute goodness. Henry Sidgwick,
writing a few years before Moore, thought the concept made
perfect sense if we take ultimate good on the whole for me to
mean what I should practically desire if my desires were in
harmony with reason, assuming my own existence alone to be
considered (1981: 112). That is, Sidgwick appears to explain
the concept of good for someone in terms analogous to the
fitting attitude account: good for me is what it is fitting for me

100

VALUE THEORY

to desire (or I have reason to desire), assuming my own

VALUE AND NORMATIVITY

101

existence alone to be considered.50

Now, this latter qualification does not quite sound right: if
we abstract from the existence of other human beings,
probably a lot of what we usually think is good for us would be
left out (e.g. personal relationships). Moreover, the qualification
makes Sidgwicks definition rather close to Moores approach:
what is good for me is what I ought to desire that occur in my
life considered in isolation from others. As in Moore, the for
me in good for me then seems to refer to a particular
location where good things (things one ought to desire) take

102

VALUE THEORY

place, rather than being a way for things to be good. 51

Therefore, there is no guarantee that all I ought to desire that
occur in my life are things that are good for me. Some such
things might be the possession of virtuous character traits
which, no matter how good absolutely, might conceivably fail
to be good for me in the relevant sense of benefiting me.
The difference with Moore is that Sidgwick takes the fact
that, so to speak, I am that location to be normatively
significant: from the point of view of the universe (a favourite
Sidgwickian phrase) my good (e.g. my happiness) is as
important as others, all else being equal, but from the point of
view of my existence as such my happiness acquires an
additional significance. Therefore ethical egoism makes
rational sense. What is more, one cannot be rationally
compelled to take the point of view of the universe as the
dominant one.

VALUE AND NORMATIVITY

4.3

Good for and fitting

attitudes

103

104

VALUE THEORY

Despite issues of formulation, Sidgwicks idea paves the way to

VALUE AND NORMATIVITY

105

a more general reply to Moore. 52 Suppose that value is what it

is fitting to favour. If x is good for A, then this suggests that A
is implicated one way or another in the fact that it is fitting to
favour x. Here are some proposals:
a) It is fitting specially for A to favour x.
b) It is fitting to favour x for As own sake.

It is fitting for A, and/or for those who care for A,

to favour x for As own sake.
d) It is fitting, for those for whom it is fitting to care for A,
to favour x for As own sake.
In each proposal A has a different role. In (a), she is the subject
to whom a certain normative predicate applies. In terms of
reasons, we can say A has special or personal reasons to
favour x. While these reasons might be shared with others, it is
crucial that they need not be so shared. However, just having
special reasons to favour some x doesnt entail that x benefits
the subject of those reasons. I might have a special reason to
favour repaying my debt to the bank (I incurred it!), but we can
c)

106

VALUE THEORY

easily imagine how repaying my debt might fail to be good for

VALUE AND NORMATIVITY

me.53

107

108

VALUE THEORY

In (b), A has the role of a patient rather than that of an

VALUE AND NORMATIVITY

109

agent: someone for whose sake it is fitting to favour x. 54 Here

we get closer to the idea that x somehow benefits A, or makes
A better off (and conversely, if x is bad for me, then x harms
me in some way). Lauras recovery from illness is fitting to
favour for Lauras sake: it is good for her. However, consider a
thief getting away with theft. We can easily imagine how
getting away with the crime might be good for her. But is it
fitting to favour such a state of things for her own sake? Some
things that are good for people might be unfitting to favour, or
fitting to disfavour, and a fortiori not fitting to favour for those
peoples sake.
Or at least, it is not fitting for just anyone to so favour them.
Proposal (c) builds on this idea: only for or by some people is it
fitting to favour certain things for a persons sake. In (c) A is a
term of the relation both as a patient and as the, or an, agent
from whom a certain response is called for. By whom is it
fitting to favour getting away with crime for the criminals
sake? Well, probably only by the criminal, although it is an
interesting question whether the sympathy felt by those who
care for her is to some extent a fitting reaction. For instance,
one might say that a complete stranger has no business caring
for criminals, and to this extent their sympathizing is simply
not fitting or even positively unfitting (think about a stranger

110

showing

VALUE THEORY

their

sympathy

by

praying

for

the

criminals

VALUE AND NORMATIVITY

111

success).55
Therefore a better idea is that only those for whom it is
fitting to care for A will also fittingly favour things for As sake.
This is proposal (d). The proposal has two virtues: first, it
restricts the pool of favouring agents: what is good for A need
not concern just anyone who could want or do something for
As sake; second, it maintains the idea that what is good for A
has to do with what benefits A (something one should want for
As sake). One problem is that (d) makes goodness for hostage
to there actually being somebody whose caring for A is fitting.
And we want to allow that things can be good and bad for A,
even if A herself cannot favour or disfavour them (because she
is an infant, or in a coma, or dead, etc.), and even if there
happens to be nobody else around to care for A. But a simple
modification might help:
e) x is good for A = it is or would be fitting, for those

for whom it is fitting to care for A, to favour x for

As own sake, if they could.
If we take (e) as our favourite understanding of goodness for,
then ethical egoism will be the thesis that everyone ought to
pursue what it is fitting to favour for their own sake (by those
for whom it is fitting to care for them). This is a recognizable
form of egoism, if a little bit complicated in its formulation. And
even if one doesnt follow the egoist in thinking that this is the
only ultimate ethical principle, it seems that (e) can make

112

VALUE THEORY

sense of good for as a valid normative category quite

VALUE AND NORMATIVITY

113

independent of absolute goodness.56

4.4

Moore strikes back?

Moores attack on personal goodness was based, as seen, on

the notion that what is good for me must be good (in some
other, non-personal, sense), and must be related to me, in an
ownership relation of some sort. The fitting attitude account
that has emerged denies both conditions. First, x can be good
for A without being good in some other non-personal sense.
The thief getting away with theft is one such case. Second, x
need not be owned by A, or even be a state of affairs in which
A is essentially involved. For instance, that my favourite
football team wins the league is such a state: I dont need to
exist in order for them to win the league. Lets assume it is
good for me. This is because it is fitting that those, who
fittingly care for me, favour such a state for my own sake. Of
course, the reason why so favouring this state is fitting will be
something like: hes going to be happy for his team. Wishing
x for As sake makes sense only if x is somehow thought to

114

VALUE THEORY

benefit A.57 But the bearer of personal value need not be any

VALUE AND NORMATIVITY

particular state belonging to the person.58

115

116

VALUE THEORY

But it is hard to shake the feeling that nothing can be good

VALUE AND NORMATIVITY

117

for me or contribute to my well-being, if its not already good. 59

For example Scanlon (who in many respects lies at the
opposite of Moore) writes:
It would make sense to say that I work hard at philosophy
because I believe it is worthwhile, or because I enjoy it, or
even because I long for the thrill of success. But these
things in turn are not desirable because they make my life
better. Enjoyments, success in ones main aims, and
substantive goods such as friendship all contribute to wellbeing, but this idea of wellbeing plays little role in explaining
why they are good. This might be put by saying that wellbeing is what is sometimes called an inclusive good - one
that is made up of other things that are good in their own
right, not made good by their contributions to it. (1998: 127)
Scanlon seems to express two ideas here. The first is that the
way we think about what is good for us is not independent of
what we think is simply good or worthwhile. We desire what is
good for ourselves under the guise of the simply good, or at
least not in contrast with it. An account that makes personal
value independent cannot make sense of this fact. However,
this point just is not very convincing. The intellectually honest
thief (and those who care about her) may well recognize that
her getting away with crime is, simply, bad, and yet still
believe it is good for her. She might be wrong as a matter of
fact (maybe being caught would be better for her), but she is
not conceptually confused.
The other idea is that contributing to well-being doesnt
explain why things like friendship or enjoyment are good - if
there is any order of explanation, it goes from goodness to
well-being. This remark, however, is compatible with thesis (e)
above. Defending a non-Moorean account of goodness for
doesnt commit one to taking good for or well-being as
somehow prior to goodness. Indeed, we should not expect that
the fact that x contributes to someones well-being explains
why x is good non-personally, since the point stressed is

118

VALUE THEORY

precisely that the two concepts are independent.60 All such an

account entails is that (e.g.) friendship is (generally, by no
means always) both simply good, and good for those who
participate in it. Indeed, this very case suggests another
example of conceptual independence: particular friendships
might be good for the participants, but simply bad - say,
because they revolve around sadistic activities against
strangers.
The Moorean might insist at this point that if getting away
with crime and immoral friendships are really good for these
people, then they must at least be pro tanto simply good, even
if they are simply bad overall: that is, there is at least some

VALUE AND NORMATIVITY

119

respect under which they are simply good. 61 However, in the

light of the general connection between value and fitting
attitudes, this means or entails that there is some feature
which makes it to some extent fitting for anyone to favour
immoral friendships and getting away with crime, even if
overall it is fitting to disfavour them. It is not clear, to say the
least, that it is fitting even pro tanto for the victims of such
states of affairs to favour them. Moreover, there is a general
worry here: it is difficult to believe that whether something is
good for me should conceptually depend on what it is fitting for
anyone to do, no matter whether or how related to me they
are. It is a conceptual truth that if x is good for me, and x, y,
you, and I are similar in all relevant respects, then y is good for
you. In this sense what is good for me conceptually depends on
what it is fitting for others to do for your sake (and vice versa).

120

VALUE THEORY

But whether what is good for me is also simply good seems a

VALUE AND NORMATIVITY

121

further question, not to be decided on a priori grounds alone. 62

Of course this doesnt mean that there are no connections
between good and good for. But these will be rather indirect.
First, it might be plausibly said that what is good for A must at
least be a default good: i.e. it must be the sort of thing that is
in general good (e.g. freedom, friendship), but could in given
circumstances explicably fail to be even pro tanto good, as in

122

VALUE THEORY

the cases cited of freedom from a deserved punishment, and

VALUE AND NORMATIVITY

123

immoral friendships.63
Second, there are structural affinities. The distinction
between final and non-final values holds also for personal
value. We draw distinctions between things that are good for
us for their own sake and those that are good for us for the

124

VALUE THEORY

sake of something else that is good for us. 64 Being rich,

arguably, is good for me to the extent that it allows me to have
the sort of things that directly contribute to my well-being (e.g.
enjoyment, achievement, etc.). Note: just to say that being rich
is good for me because it contributes to my wellbeing by itself
doesnt mean that being rich is instrumentally good for me,
since the same could be said about any x that is good for me.
Indeed, the because here doesnt introduce an explanation,
but rather a reformulation. The question is whether being rich
contributes directly or indirectly to my well-being. In turn, the
question is whether being rich is a condition that I, and those
who fittingly care for me, should favour as such for my own
sake. In general, then:
x is finally good for A = it is or would be fitting, for those for
whom it is fitting to care for A, to favour x for As own sake
for
its own sake, if they could.
One should not be misled by this welter of for the sake of
locutions. The idea is not that x is to be favoured both for As
own sake and for its own sake. For that would make nonpersonal final value a necessary condition for personal final
value. Rather, for As own sake is the logically primary

VALUE AND NORMATIVITY

125

specification of it is fitting to favour x, and for its own sake

126

VALUE THEORY

acts as a modifier of for As own sake:65

it is fitting, for those for whom it is fitting to care for A, to
favour x (for As own sake [for xs own sake]).
This formulation should make it clear that, while final values
can be personal or impersonal, final impersonal value, or what
we identified in the previous chapter as absolute value or value
simpliciter, is distinct in two ways from personal value: (1) it
takes for its own sake as the only necessary minimal
specification of fittingness, i.e. final impersonal values might
be worth favouring for someones sake, but need not, because
the question for whose sake? need not always apply; (2) if
fittingness is a three-place relation between an object (state of
affairs), an attitude, and an agent, then in the case of final
impersonal value it is left unsaid who or what fills in the agent
term of the relation, whereas in the case of personal value, the
agent term is filled by those for whom it is fitting to care for A,
which might or might not include A herself. I turn now to this
second aspect.

4.5 Agent-relative value

Personal value, as we have discussed it so far, must be
distinguished from another concept of which much has been
made recently: agent-relative value. The distinction has in fact
already emerged in discussing proposal (a) above: x is good for
A = it is specially fitting for A to favour x. The objection against
(a) was that it might be specially fitting, for moral reasons, for
A to favour his repaying a debt, while repaying a debt might
not be good for him - it will make him poorer, for instance.
However, the idea that a normative demand attaches
especially to an agent (and not to another), while unable to
capture the concept of good for A, has been thought to
configure a different category of value: x as being good relative
to A.
Agent-relative value, unlike personal value, appears to be a
technical concept, created for specific reasons having to do
with the credentials of consequentialism. In its classical form,
consequen- tialism is roughly the view that we morally ought to

VALUE AND NORMATIVITY

127

maximize final value, where the latter is non-relative, absolute

final value. One implication of this is that, if faced with the
choice between letting two murders happen and murdering
one person ourselves, we morally ought to murder the one
person. This is because we have to choose between two evils
of the same sort (two murders, one murder), and for
consequentialism this is all that morally matters. It would be

128

VALUE THEORY

wrong to choose the greater evil (two murders). Committing

VALUE AND NORMATIVITY

129

murder is the only way to maximize value here.66

Non-consequentialists have complained that it cannot be
right to murder to prevent two murders. From a classical
consequentialist point of view, such complaint appears
groundless: how is it not right to prefer the least bad state of
affairs? However, some conse- quentialists have taken a
different line, one that allows at least in principle for the nonconsequentialist claim. They do not abandon the central idea,
i.e. that we morally ought to maximize value, but relativize the
notion of value to particular agents: each of us morally ought

130

VALUE THEORY

to maximize (final) value relative to them.67 Going back to our

example: if we grant the substantive claim that one murder by
A is worse relative to A than the murders by B and C together,
then A ought not to murder in order to prevent murders by B
and C. When seen as absolute evils, all murders have the same
disvalue (ceteris paribus): therefore the more, the worse,
period. But when their disvalue is relativized to agents, it
seems both possible and plausible to assert that the murder I
commit is worse relative to me than the murders committed by
someone else. Therefore I ought not to murder, since if I
murdered someone I would maximize disvalue relative to me.
This agent-relativized structure can make sense of a range
of situations where we think it morally significant that the
agent is involved. I need to repay my debt to my creditor, but I
learn that I could help B repaying her higher debts to her
creditors. But if I do this, I will be unable to repay my debt to
my creditor. Ceteris paribus, it seems I have an overriding duty
to
repay
my
debt.
According
to
agent-relative
consequentialism, this is because repaying my debt is better
relative to me than repaying others debts. Analogous
examples can be multiplied at will.
But introducing agent-relativity is not just a move to make
consequentialism more plausible. In fact, it affects the way we
conceive of value. If all value is always relative to some agent,
then it is a mistake to think of value that is not relative to
anyone. The most one can say is that certain things have
agent-neutral value: they are good- (or bad-) relative to
anyone, and their value does not vary from agent to agent. The
classical consequentialist then believes that there are only
agent-neutral values. Ceteris paribus, my murder and someone
elses murder have equal value relative to me (and relative to
anyone else, of course). The agent-relative consequentialist
will on the other hand insist that at least certain kinds of goods
and evils cannot be assumed to have agent-neutral value. The
agent-relativist and the agent-neutralist could and perhaps
should agree on a given list of universal final values - As
murder is bad relative to A just as Bs murder is bad relative to
B, and so on. Relativization does not mean private or individual
values, for the reason why As murder is bad relative to A is the
same reason why Bs murder is bad relative to B. But the

VALUE AND NORMATIVITY

131

crucial question is, how good or bad, relatively to each, are

these things? If murder is an agent-relative evil, then Bs
murder might be bad relative to A, but not as bad relative to A
as As own murder, and vice versa as regards B. In other
words, given two instances x and y of the very same value V,
there is no single ranking on which we can say that x is better

132

VALUE THEORY

than, worse than, or as good as y. Rather, there are as many

VALUE AND NORMATIVITY

133

rankings of x vis--vis y as there are agents.68

Authors have noted that the idea that value is essentially
relative to agents had better be something more than just a
technical device, on pain of introducing a purely arbitrary and
ad hoc solution to the problems of consequentialism. It has to
express a concept of value which somehow resonates with our
understanding of value. But some options in this respect are
blocked off: (1) Agent-relative value is not personal value, as
seen above. (2) Agent-relative value is not value relative to
my evaluative point of view, i.e. relative to what I happen to
value. This would make agent-relative consequen- tialism the
subjectivist view that we ought to maximize what we in fact

134

VALUE THEORY

value. But this is not what we are looking for. 69 Also, agentrelative and -neutral value cannot be seen as simply the
evaluative counterparts of agent-relative and -neutral reasons.
Agent-relative reasons, on an intuitive understanding, are
reasons that apply to some agents only, while agent-neutral
reasons apply to everyone.
Now, there is a reason for everyone not to murder. This reason
is agent-neutral. But the act or state of affairs of murdering is

VALUE AND NORMATIVITY

135

held to be an example of an agent-relative value. So, the two

136

VALUE THEORY

ideas are different.70

However, there is a natural home for the notion of agentrelative value. The thesis that value and fitting attitudes are in
broad correspondence seems to allow or perhaps even require
a notion of value relative to. Fitting attitudes will be some
agents attitudes. So, if x is valuable, it always makes sense to
ask for whom it is fitting to favour x. And if it is fitting for me
(but not necessarily for others) to favour x more than y, then
we can say that x is better relative to me than y. This seems to
be what agent-relative consequentialists mean when they talk
about evaluator-relative value (Sen 1983): not what each of
us happens to value, but what each of us should value or,
better, should favour. While FA makes room for agentrelative
value, it must be stressed that it does not follow from FA alone
that there actually are agent-relative values: that is, as far FA
goes, it could be that whatever is valuable is simply valuable
relative to anyone, and exactly to the same degree.
Some may feel that appealing to what we ought to favour
presupposes what consequentialism must explain. But notice:
for agent-relative consequentialism each morally ought to
maximize what they, and possibly only they, ought to favour
the most, i.e. what ranks highest in their rankings. These
concepts are all different: (1) ought (or fitting) to favour
doesnt mean morally ought to favour (see Chapter 3); (2)
ought to favour the most doesnt mean ought to maximize
(see
Chapter
7).
Thus
understood,
agent-relative
consequentialism states an interesting and distinctive
connection between moral concepts and evaluative ones. A
non-consequentialist still has plenty of room to deny (1) that all
we morally ought to do has to do with value (be it relative or
neutral), and (2) that all we morally ought to do with value is to
maximize it.

4.6
Impersonal/persona
l and agentneutral/agent-relative
Impersonal value poses no restriction on the class of favouring
agents. If FA is broadly correct, impersonal value is relational in
its structure just like personal value, but as noted above, it is
left unsaid for whom it is fitting to favour stuff for its own sake.

VALUE AND NORMATIVITY

137

Importantly, it is misleading to think that the class of favouring

agents is everyone. While it may be true that impersonal
value makes claims on everyone, it is still not the same as
value relative to everyone. Impersonal value is meant to
generate agent-invariant rankings, rather than being the result
of uniformly converging agent-variable rankings. The class of
favouring agents is perhaps better expressed by anyone, that
is, no one in particular. Thus impersonal value is properly
agent-neutral value.
Also, if agent-neutrally better is understood as better
relative to everyone, we will have to rule out that things can
have conflicting agent-neutral and agent-relative values. For it
cannot be the case both that x is better relative to me than y
(because x is better relative to everyone than y, and that
includes me) and that y is better relative to me than x
(because y has a particular agentrelative value relative to me).
That is simply inconsistent. On the other hand the idea of
anyone behaves differently, for it refers, so to speak, to an
anonymous agent: as such, I should prefer x to y; while as
this particular involved agent, I should prefer y to x. There is
conflict between agent-neutral and agent-relative value but no
contradiction. As an agent among others, I should prefer my
one murder to two murders; as the agent who would commit

138

VALUE THEORY

the murder, I should prefer the two murders to my one

VALUE AND NORMATIVITY

139

murder.71 The agent-relative consequentialist solves such

140

VALUE THEORY

conflicts by giving priority to the agent-relative value.72

What about the relation between personal and agentrelative values? Well, some agent-relative values neatly
overlap with personal values. What is good for A, in the
definition given above, addresses a restricted class of agents
whose favouring for As sake is fitting: those who fittingly care
for A. So what is good for A will rank higher for these agents
than what is good for B (that is, provided that it is not fitting at
all for them to care for B, or that it is fitting for them to care for
A more than for B). And vice versa for those who fittingly care
for B. Personal values thus seem to generate agent-variable
rankings.
But not all putative agent-relative values are personal
values. Indeed, typical examples of agent-relative value are not
well understood as personal values. Paying my debts is better
relative to me than paying others debts even if the latter is
better for me (say, because it will get me a lot of friends).
Murder by my own hands is supposed to be evil relative to me
even though it might in given circumstances be good for me
(say, its my only way to safely escape imprisonment). Agentrelative values need not be states of affairs that are good for
the agent or for other people involved. As such, then, they are
not reducible to either impersonal or personal values, though
they have something in common with both. Like impersonal
value, agent-relative value need not presuppose fitting
attitudes of care or for-someones-sake attitudes towards
anybody. Like personal value, agent-relative value keeps track
of who is the agent and how she is related to the relevant state
of affairs.
One might complain that if agent-relative value is neither
personal nor impersonal, then it is not a species of value. But
such a complaint can only come from someone who rejects the
whole fitting attitude approach to value. If we think of value in
FAs terms, then there will be as many putative species of
value as the results that we can generate by (sensibly)
toggling the controls within the FA schema. The agentneutral/agent-relative value distinction is the result of toggling
the agent control one way or another - indeed, we could
compare agent-neutral value to the muting of such control.

VALUE AND NORMATIVITY

141

Having distinguished these four forms of value, I do not

mean to deny possible overlaps. Precisely how these values
overlap is a
happen to rank similarly within her ranking. If I can choose between 100
murders and 1 murder by others hands, I ought to choose the 1 murder,
because relative to me 100 murders by others hands are simply worse than
1.

matter of substantive dispute, but the following propositions

should help to give an idea of their possible relations. I have
suggested that (1) if x is agent-neutrally good, then it is
impersonally good, and vice versa, and (2) if x is personally
good, then it is agent-relatively good (where the agent might
or might not be the beneficiary of x), but the converse doesnt
hold. Beyond this, it can be argued that (3) some xs are both
impersonally and personally good (and thus agent-neutrally
and agent-relatively good); (4) some xs are only agentneutrally/impersonally good; (5) some xs are only agentrelatively good. An example of (3) might be my own health: it
matters as both something absolutely good and good for me,
and thus something that has both a position in an agentinvariant ranking of states of health, and one in a ranking
relative to me (and certain others). Examples of (4) might be
certain aesthetic or logical properties (harmony, coherence),
where there is simply nobody to care about for their own sake.
Examples of (5) might be either values that can only be
personal (e.g. a killer getting away with crime, as discussed
above), or agent-relative values that are not personal, e.g. the
honour of my family, which might conceivably fail to
correspond to my or anyones well-being. The following
diagram should help to illustrate these possibilities, while
keeping in mind that one might on substantive grounds reject
some or all of them:

142

Figure 4.1

VALUE THEORY

4.7 Summary

This chapter explored the common-sense notion of something

being good for someone. Pace Moores worries, this concept is
not easily reducible to impersonal goodness, and a fitting
attitude approach can make sense of its distinctive normative
import. I have then examined the relations between personal
value and the distinct idea of agent-relative value.

CHAPTER FIVE

The Chemistry of Value

5.1 Introduction
In this chapter, I examine what we might call the chemistry of
value: how the value of a valuable object gets formed from its
other features, and from other values possibly present in the
object. First

VALUE AND NORMATIVITY

143

clarify the notion of supervenience, and introduce a

distinction between the value-making features of an
object and other conditions that enable such features
to be value-making. Then I discuss the so-called
principle of organic unities: the idea that the value of a
complex object need not be just the sum of the values
of its parts. Noting some problems with it, I contrast it
with
two
alternative
views,
virtualism
and
conditionalism. Finally, I briefly present the ideas of
invariant and default value, with reference to the
recent debate on particularism in moral theory.

5.2 Supervenience and other

relations
Value essentially depends on something else. If a certain
object is valuable, then something other than its being
valuable must make it so. This is reflected in our evaluative
practices: while it might be obvious in many cases why a
certain decision or a certain state of affairs is a good or a bad
one, one is always in principle entitled to an answer as to why
it is good or bad. Nothing can be just good, just bad, or just
better than something else. The concept of supervenience is
an attempt, in part, to articulate rigorously this intuitive idea.
To say that value properties supervene on non-evaluative
properties is to say that value properties depend on nonevaluative ones in a certain way: if two objects (states of
affairs, experiences, actions, or what have you) have exactly
the same non-evaluative properties, then they must have the
same value properties. For example, if my toothache is just as
painful as your toothache, then (other things being equal) they
must be equally bad, i.e. bad to the same degree. But this
positive way of formulating supervenience, while perfectly
accurate, might lead astray into thinking that if two objects
have different non-evalu- ative properties, then they cannot or
do not have the same value properties. But the latter claim is
no part of supervenience (if p, then q does not entail if not p,
then not q!). Moreover, it is obviously false. My toothache
might be caused by poor tooth- brushing, while yours might be

144

VALUE THEORY

caused by a wisdom tooth, but this non-evaluative difference

might not make any evaluative difference: both experiences
can still be bad to the same degree.
In order to avoid such misunderstandings, it is common
(though not more accurate) to describe supervenience in
negative terms: there cannot be a difference in evaluative
properties without a difference in non-evaluative ones. If my
toothache turns out to be worse than yours, then this must be
because of some non-evalu- ative difference between our
aches: maybe mine lasts longer, or is more intense than yours.
It is instructive to point out a few features of supervenience:
1 It is a principle about what necessarily is the case: notice

the cannot. This means that any difference in value

properties requires a difference in other properties. Its
just not possible for two things to be evaluatively
different (say, one good and the other bad) and be
exactly similar in other respects.
2 Supervenience does not tell us which things or states are
good and bad. It is a logical constraint or requirement
on things that are or could be already good or bad
(note that its logical form does not even entail that
there are good or bad things). It is rather about the
relation of properties of a certain kind - whatever they
are, actually instantiated or not - to properties of a
different kind. Nor
does supervenience tells us how to go about deciding
what is good and bad, since it doesnt tell us for any
specific value property (say, the badness of my
toothache) which are the non-evaluative respects on
which it supervenes. In part, this is because it is a
metaphysical relation, not an epistemological one: it is
about how things could (not) be. But even the
epistemological derivates of supervenience (for instance:
you cannot assert that there is an evaluative difference
between x and y if x and y are exactly similar in other
respects) do not tell us what these other respects are except that they are not evaluative ones. In this regard,
supervenience of value is less informative than the
supervenience claims commonly made about other sorts
of properties; e.g. it is generally accepted that mental

VALUE AND NORMATIVITY

145

properties (at least for humans) supervene not just on

any non-mental ones, but specifically on physical
properties of brains (and not, say, on geological or
astronomical properties!).
3 Supervenience does not tell us much about the sort
of relation between value and non-value properties,
except of course that they stand in a modal
relation: any change in the former necessarily
entails a change in the latter.
But we are not told whether value properties reduce to,
consist in, are grounded on, are caused by ... etc. nonvalue properties.
4 The supervenience claim leaves it open whether
some evaluative properties also supervene on
other evaluative or anyway normative properties.
For instance, if we are to believe the fitting
attitude account of value presented in Chapter 1,
then evaluative properties like being good and
being bad supervene on normative properties
such as an attitude being fitting. Likewise, if
intra-evaluative analyses are correct, then e.g.
the property of being good will supervene on
(roughly) the property of being better than
nothing. Also, very often the only manageable
way of stating parts of the supervenience base of
particular evaluative facts will be in terms of
other evaluative, or anyway not simply
descriptive or naturalistic, facts: the action was
admirable because it displayed the proper
amount of courage; a policy is better than another
because it allows people to get what they deserve; a
pleasure has positive value if it is innocent, and so on.
However, the supervenience claim does imply that all
these further evaluative facts in turn supervene on nonevaluative ones: nothing can be just proper, or just
deserved, or just innocent, even if there may be no finite
way to spell out in non-evaluative terms the full
supervenience base of these properties.
The idea of supervenience is intuitively understood in spatial
terms: evaluative properties stand on top of (super-vene on)

146

VALUE THEORY

non-evaluative ones (which therefore sub-vene under them).

This is why it is common to talk of the supervenience base of
evaluative features. The principle of supervenience does not
tell us what is included in any supervenience base; all one can
say is that any given super- venience base includes all the nonevaluative properties which are able to make an evaluative
difference. In other terms, it includes the properties without
which the object or state in question would be evaluatively
different: say, not as good as it is, or not good at all. Therefore
the base of any given value property may be a very diverse
and possibly not entirely discernible set of properties or facts.
By way of illustration, just consider all the negative facts which
are able to make an evaluative difference: e.g. a certain
pleasant experience is good, as long as it does not lead to
painful experiences, it is not a response to your children being
tortured, and so on.
For this reason, some philosophers have thought it useful to
distinguish, within any supervenience base, at least three
different sorts of features or properties: (1) those from which a
value results, or which ground value, or make something good
(also known as the resultance base); (2) those which enable
the former properties to make something good, and whose
absence might disable the good-making role of the former
properties: the enabling conditions; (3) those which intensify or
diminish the value of something (Dancy 2004a: 170-2, 178-81).
As an example, consider an experience of pleasure, say,
obtained from a warm bath. Lets assume that it is good to a
certain degree. On what does its goodness supervene? (1) On
what its goodness results from, what makes it good: arguably,
the very sensation of pleasure. (2) On a number of enabling
conditions, such as the negative facts recalled above: that the
pleasant sensation does not lead to subsequent painful
sensations, it does not distract me from more important tasks,
it is an innocent pleasure, and so on. The idea here is that if
any of these things were not the case, the pleasantness of the
sensation would not count in favour of the experience, that is,
would not make the experience good. But of course, enabling
conditions can also be positive facts: recall the discussion of
evaluative and factual enabling conditions in section 2.2. (3)
Possibly, on a set of intensifying factors: e.g. that the pleasures

VALUE AND NORMATIVITY

147

of the bath come after a long day of work makes the

experience better, or have a higher degree of goodness than it
would have, say, after a lazy Sunday, though that would still be
quite good.
Besides purposes of conceptual clarification, these
distinctions are particularly suited to express certain views
about the computation of value - i.e. about how, given a
certain value assigned to a complex state of affairs, we should
understand the contribution made by the different elements
composing that state. Originally devised by Jonathan Dancy to
support his particularism (see section 5.6), we will see how the
distinctions can also be usefully employed by less radical views
(section 5.5).

5.3 Organic Unities

In Chapter 2 we discussed Moores idea that final value
depends only on intrinsic properties of an object or state, that
is, that final value is intrinsic value. We had examined some
difficulties for this thesis, which now we can put in the
following terms: the supervenience base of some final values
seems to include extrinsic properties, for instance as enabling
conditions, or as part of the resultance base, or again as
intensifiers, so that final value cannot always only depend on
intrinsic properties. But Moores view seems to face another
difficulty. If final value depends on intrinsic properties only,
then extrinsic facts or properties wont be able to affect final
value. Freedom of movement, arguably, has final value, and
the loss of it has final disvalue: we should regret our own or
others loss of freedom of movement, for its own sake. But is
there anything to regret about a jailed serial killer not being
free to move around?
Well, if the thesis that final value is intrinsic value is correct,
then the killers lack of freedom is still finally bad, and just as
bad as an innocent person lacking freedom. The fact that he
lacks freedom as a result of his crimes, being an extrinsic fact,
does not matter for its final disvalue. So, if we consider this
state of affairs: killing people and being jailed for that, we will
have two intrinsically bad things: people being killed, and

148

VALUE THEORY

someone being jailed. We could then evaluate it by just

summing these intrinsic disvalues: say -10 (people being killed)
and -2 (someone being jailed) = -12. But then this state will
turn out to be worse than the state in which people are being
killed and the killer gets away with the crime: -10, for
simplicity. For many, including Moore, this is unacceptable: that
a serial killer is jailed is not worse, but better than if he were on
the loose, and finally so - not just because he is then unable to
harm others, or the victims relatives feel satisfied. Using
Moores isolation test (section 2.3), it seems we would prefer a
world where he is jailed to a world where he runs free, even
ignoring all other consequences. But the thesis that final value
is intrinsic value cannot underwrite this intuition. Is there a way
around it?
The answer, given by Moore, is that the final value of a state
of affairs composed of parts (a whole) need not be equal to the
sum of the final values of the parts - i.e. the values they have
intrinsically, or on their own (Moore 1993: 79). Some wholes
are organic unities in this sense. Another example might be the
following. If pleasure as such has final intrinsic value, then
being pleased as such has some positive value. Now, cats
being tortured is rather bad. If cats being tortured = -10, and
being pleased by something = +2, then we might expect that
the value of being pleased by cats being tortured equals -8,
and as such is at least a bit better than cats being tortured. But
thats clearly unacceptable (at least for many, Moore included):
the whole being pleased by cats being tortured is definitely
not better, and likely worse, than cats being tortured. So its
value cannot result from a sum of the intrinsic values of the
parts. Or take artistic examples. The final value of the Mona
Lisa must result from its intrinsic properties. However, arguably
the painting would have little to no aesthetic value if we simply
summed the final values of each brushstroke. Conversely, if we
join together two pieces of music, each beautiful on its own, we
might get something quite cacophonic. As we sometimes in
fact say, it is the whole that counts.
But how can positive (negative) value result from a whole
composed only of bad (good) parts, or even parts which are
neither good nor bad? Stating the principle of organic unities is
one thing, but accounting for how value gets formed in an

VALUE AND NORMATIVITY

149

organic unity is quite another. One idea is that, in such cases,

the relevant state of affairs S is composed by two sorts of
entities: the parts of the whole, and the whole as a distinct and
additional entity, e.g.:
S: (a) people being killed, (b) killer being jailed, (c) (a&b).
It is important here to understand that (c) is not quite the same
thing as the state of affairs we are evaluating: although
obviously overlapping with it, for the purposes of computing
value (c) is to be understood as a constituent of the state of
affairs under consideration, so that its value contributes to, but

150

VALUE THEORY

in principle does not exhaust, the value of the state of affairs. 73

This approach is suggested in the distinction that Moore draws
between the value of a whole as a whole and its value on the
whole:
the value which a thing possesses on the whole may be said
to be equivalent to the sum of the value which it possesses
as a whole, together with the intrinsic values which may
belong to any of its parts. (Moore 1993: 263)
The overall value of a state of affairs like a serial killer being
jailed, that is, its value on the whole, results from the sum of
three factors: the values of the two parts (a) and (b), plus the
value of the overlapping whole as a whole ((c) above), which is
determined organically - i.e. doesnt need to correspond to the
sum of the values of (a) and (b). Moore thus makes it clear that
he does not deny the additivity of value, i.e. the idea that value
must ultimately be computed by performing the operation of

VALUE AND NORMATIVITY

151

summing.74 Only, when we are dealing with complex states of

affairs, we should take into account as a further addend the
value of the organic unity (should there be one), on top of the
values of the parts. So the value of the state of affairs will be
something like this:
(i) Ss value on the whole = (a) (-10) + (b) (-2) + (c) (4) = -8.
By giving (c) an ad hoc value of +4, we get the result that a
serial killer being jailed is at least better than people just being
killed (-10), though it is still a bad state of affairs. If one instead
wants to claim that retributive punishment has positive value,
then the value of (c) will have to be increased accordingly, say
+14:

152

VALUE THEORY

(ii) Ss value on the whole = (a) (-10) + (b) (-2) + (c) (14) = 2. 75
The doctrine of organic unities, so understood, seems rather
problematic for two reasons. The first issue is what I call
attitude mismatch. The introduction of the value of the whole
(c), as an additional element to be computed in calculating the
total value of S, has strange implications. Remember that (c) is
in effect overlapping with S. But how can the value of (c)
(value as a whole) diverge so much from the value of S
(value on the whole)? The problem becomes more evident
when we understand final values as the objects of fitting
attitudes. Consider example (i). Here it is fitting to have a
negative attitude towards S, but also fitting to have a positive
attitude towards (c). Or consider example (ii). Here it is fitting
to have a positive attitude towards S, but a much more intense
positive attitude towards (c). Such mismatches of fitting
attitudes are, at best, very surprising, since S and (c) come
down to the same metaphysical thing: a serial killer having
killed people and being jailed. S and (c) overlap, since (c)
contains whatever is included in S. There seems to be no nonevaluative difference between them which would warrant

VALUE AND NORMATIVITY

153

having attitudes differing in kind or even in degree towards the

154

VALUE THEORY

two states.76
When we consider the relation of values to fitting attitudes,
a second problem emerges, which I call normatively idle
values. Recall that Moore is drawn to the doctrine of organic
unities due to the thesis that final value is intrinsic value. This
means that, e.g., if pleasures of a certain intensity have some
positive final value, then the pleasure felt by the sadist in
seeing cats being tortured carries the same positive final value
as an equally strong but innocent pleasure. Occurring in
different contexts or wholes does not affect their intrinsic
value. As Moore explicitly says: The part of a valuable whole
retains exactly the same [final] value when it is, as when it is
not, a part of that whole (1993: 81). However, it is not clear
whether, while admitting that it is overall fitting to condemn
someone being pleased by torture, anyone would be prepared
to say that it is also fitting, at least to some extent, to respond
favourably to the fact that they are pleased, e.g. by viewing
the pleasure as such as improving things somewhat. In this
instance, it seems, favouring pleasure is just not a fitting
response at all. However, if so much is in fact granted by the
organicist, then we have a normatively idle value: a putative
positive value unrelated to positive fitting attitudes. This is an
implication that even someone, like Moore, who rejects a fitting
attitude analysis of value should view with suspicion. As we
saw in Chapter 1, nobody can really deny the existence of a
regular connection between value and fitting attitudes.
Of course, the organicist might bite the bullet, and concede
that even the sadists pleasure has some positive value and
merits some kind of positive response, albeit one that is
outweighed by the overall negative response to the overall
state of affairs A takes pleasure in torture. But this is not an
attractive view. Or the organicist might reject this particular
example, i.e. reject the suggestion that being pleased by
something has some final positive value regardless of what
pleases one. This can mean that being pleased, as such, has
neither positive nor negative value, but is neutral (equals 0).
This answer will not do. If something is evaluatively neutral,
then intuitively it is fitting to be indifferent to it - neither favour
nor disfavour it. But is there any room for indifference in a case
like this? Indifference seems rather unfitting here. And if the

VALUE AND NORMATIVITY

155

organicist agrees on this, then we have again a kind of value

156

VALUE THEORY

(admittedly, a neutral one) which doesnt tie up with fitting

VALUE AND NORMATIVITY

157

attitudes.77

5.4 Alternatives to organic

unities: Virtual value
These might not be knock-down objections to the doctrine of
organic unities, yet they do motivate the search for alternative

158

VALUE THEORY

approaches.78 One such approach starts exactly from examples

such as sadistic pleasure, and claims that states like being
pleased, as such, are evaluatively inadequate: not good, bad,
or even neutral. It follows that being pleased never features as
a part whose final value has to be computed to arrive at a
wholes value, since it simply has no value. It also follows that
no question of fitting attitudes arises. To be more precise:
faced with an evaluatively inadequate state, the fitting
response to it is to withhold favour, disfavour, or indifference,
just like a movie one hasnt seen yet. What we have, then, is
not a bad whole with some good or neutral and some bad
parts, but an undivided and uniformly bad state of affairs:
being pleased by seeing torture. There is no room for favouring
the state of being pleased, since this state as such has no
actual final value. If it has no (actual) value, then a fortiori it is
not a normatively idle value. Nor is there any necessity to
postulate the value of a state of affairs as a whole as opposed
to its value on the whole. So, no problem of unwarranted
attitude mismatches can arise.
But how can we explain the resultance of value from a state
which essentially contains evaluatively inadequate elements?
People being tortured is pretty bad, but that plus other people
taking pleasure in that is even worse. How does this approach
explain the difference clearly made by the sadists pleasure?
Zimmerman (1999, 2001: Chapter 5) claims that being pleased
has virtual value: if it is fitting to e.g. disfavour a state of affairs
whose existence entails being pleased (e.g. the state being
pleased by torture), then being pleased is virtually bad. By the
same token, if the state being pleased by others happiness is
actually good, then being pleased (there) is virtually good.
Virtual value is a derivative and therefore extrinsic form of

VALUE AND NORMATIVITY

159

value, since it depends on the actual value of the state of

160

VALUE THEORY

affairs the object takes part in.79

One problem, however, is that if pleasure only has virtual
value, we cannot say that pleasure is what makes the
evaluative difference between torture and sadistic torture. Any
evaluative weight that pleasure carries is logically derived
from, or owed to, the value of the larger state of affairs. In the
order of explanation, actual value is prior to virtual value. So
this approach has to take the difference between torture and
sadistic torture somehow as a given. Sure enough, the
virtualist can say that sadistic torture is worse because of the
presence of pleasure (supervenience: evaluative differences
require non-evaluative ones). However, the virtualist cannot
say that pleasure acts as an additional bad-making feature.
Pleasure somehow makes things worse without adding any
negative weight of its own. In general, since virtual value
presupposes the actual value of a state of affairs, virtual value
cannot contribute to explaining actual value.
The virtualist approach is designed to avoid recourse to
organic unities, while endorsing the main thrust of Moores
philosophy: actual final value is intrinsic value, and as such
doesnt vary in kind or degree from context to context. But can
the approach be generalized across all putative cases of
organic unities? Here a second problem emerges. Where the
value of intentional states is in question, the idea may seem
plausible enough: sadistic pleasure is an intentional attitude
towards something, and like for other attitudes (belief, desire,
fear ...), we cannot really pass an evaluative or epistemic
judgement on them in abstraction from the object of the
attitude. But now consider other alleged cases. For example,
the claim that evaluative progress (for a life, a career, a
relationship ...) is better than evaluative decline, even if the
relevant good and bad stages have precisely the same
degree of value:
Progress: Going from B (bad) to G
(good) is better than
Decline: Going from G to B.
The doctrine of organic unities, as we have seen, will compute
the values of G and B for each state, and then place an extra
value in the very wholes, so that Progress comes out better

VALUE AND NORMATIVITY

161

than Decline. While that move has its problems, the virtualist,
on the other hand, seems forced to ascribe no actual value to

162

VALUE THEORY

the parts of either Progress or Decline. 80 And we should

withhold responses towards the parts, since these only have
virtual value. But this amounts to rejecting the terms of the
example. In Progress and Decline there must be things (states
of affairs) with actual positive and negative values - otherwise
how could Progress be actually better than Decline? Progress
and Decline stand in an actual second-order evaluative relation
(it is better that good follows bad than that bad follows good),
but as such they must be able to relate to actual first-order
values.
Moreover, suppose we define virtual value as follows:
Necessarily, S is virtually intrinsically good [bad] to a certain
degree if and only if, for some state S whose existence
entails that of S, the contemplation of S and S alone
requires that one favour [disfavour] S to a corresponding
degree.
Even conceding for the sake of argument that G and B have
merely virtual values, it turns out that, in Progress, B (by
assumption something bad) is virtually intrinsically good, since
presumably we are required to favour Progress. On the other
hand, in Decline, G (by assumption something good) is
virtually intrinsically bad, since presumably we are required to
disfavour Decline. This is surely a surprising implication:
whether actual or virtual, the values of G and B cannot switch
polarity from one case to the other without changing the very
terms of the example. In sum, when we move to computing
the value of complex non-intentional states, the virtualist
approach loses some of its appeal.

5.5 Alternatives to organic

unities: Conditional value
We have seen that organic unities are problematic for a
number of reasons. We have also seen that organic unities are
a natural consequence of the following theses:

VALUE AND NORMATIVITY

163

A) Actual final value is always and only dependent

on (supervenes on) the intrinsic nature of an

object.
B) (following from A) An object retains its actual final
value across contexts and wholes in which it may
appear.
If virtualism does not work as an alternative, then it seems
that the easy solution is to give up A, and B in turn. Indeed, in
Chapter 2 we have seen some reason to doubt A. The
alternative - lets call it conditionalism - claims that:
C) Actual final value can depend on (supervene on) an

objects intrinsic and extrinsic properties alike.

D) (therefore) An object need not retain its actual final

value across contexts and wholes.

How does conditionalism explain the value of the two states of
affairs we have been using as examples?
S: A serial killer being jailed for his crimes.
T: Taking pleasure in others pain.
In both cases, one need not suppose that the parts or
components which would or might be good (or bad) on their
own, namely freedom and pleasure, carry over their value to
these contexts (this is what D tells us). Therefore there is no
need for an additional entity, overlapping with the states of
affairs, whose value as a whole is able to make the overall
values of S and T (on the whole) turn out different from the
sum of the values of their parts.
Regarding S, remember that we want to be able to say that
punished crime is, as such, better than unpunished crime. The
conditionalist can arrive at this claim by using the distinctions
sketched in section 2. The fact of desert acts here as a
condition: it changes the value-making contributions of
punishment and freedom. For instance, we can say that in the
serial killers case desert not only disables the intrinsic evil of
being punished and the intrinsic positive value of freedom, but
also enables punishment to switch from bad to good, or
freedom from good to bad. So punished crime is better than
unpunished crime because the former contains one bad feature
(crime) and one feature that is neutral or even good (being

164

VALUE THEORY

punished), while the latter contains the same bad feature

(crime) and one feature that is bad or neutral (freedom):
Crime Punished>Crime Unpunished:
Bad+Neutral>Bad+Bad, or
Bad+Good>Bad+Neutral.
In these explanations, we can see that the kind (polarity) or the
degree of the values of punishment and freedom are
conditional on an extrinsic factor, namely desert.
Regarding T, we can say that pleasure, when enjoyed by
sadists, is simply and finally bad. We can explain this because
sadism disables pleasure to be good-making, and enables it to

VALUE AND NORMATIVITY

165

be a bad-making feature in this context.81 Note that this is not

the same as saying that pleasure as such is an evaluatively
inadequate state, but rather that the degree and kind of value
it does actually take may be conditional on its object, and thus
cannot be assumed to be good or bad or neutral across all
contexts.
Because of this different analysis, the problems of attitude
mismatch and normatively idle values do not arise. We can
simply be happy, or at least satisfied, for S on the whole
without any competing (in kind or degree) fitting attitudes
towards the whole as a whole. And we can agree that pleasure,
in the case of the sadist, has no idle positive value, but indeed
has a non-idle negative value, and accordingly we can have an
unmixed fitting attitude of condemnation towards it.
However, the conditionalist analysis invites two objections.
The first we can call the arbitrariness objection. Consider
retributive punishment again. On what grounds do we decide
that desert makes punishment a good thing (or neutral, or less
bad), rather than making a bit less bad the evil caused by the
punished criminal, and worse the evil caused by the
unpunished criminal?
In this case, we probably rely on an intuitive sense that the
degree of evil caused by the crime just cannot be improved or
made worse by anything external to it, so we are led to focus
on the other component (the punishment). But in other cases,
the choice seems more arbitrary. Take sadistic pleasure. Why
does sadism affect the value of pleasure, rather than making
worse the fact of people being tortured? Put in another way,
why does sadism disable the good of pleasure, rather than
intensifying the evil of people being tortured? Or take a case of
disproportionate attitudes. An extremely selfish person is
disproportionately concerned for herself more than for others.
But which of the two concerns does the lack of proportion
affect, turning it bad from good or at least from neutral? Is
great concern for oneself a bad thing, when accompanied by
little concern for others, or is it the other way round? The vice
of selfishness, it seems, lies precisely in the combination, or
the whole, rather than in the degree of intensity of either

166

VALUE THEORY

concern (Hurka 1998).82 While the conditionalist might find

more or less satisfying answers to each case, adopting the
doctrine of organic unities has an apparent advantage in that it
simply does not require an answer to such location issues.
Any value to be explained can be placed in the whole rather
than in one of its parts. In this sense, organic unities treat their
parts symmetrically, whereas condition- alism needs to justify
its asymmetric treatment of parts.
A second problem with conditionalism can be called the
adjusted attitude objection. Consider yet again retributive
punishment. If desert makes the deprivations suffered by the
criminal a (finally) good thing, then the appropriate response
to them should be pure happiness (for its own sake). But, as
Hurka points out, presumably even for fans of retributivism
retribution involves or should involve a distinctively sombre or
subdued emotional tone, one suffused with regret (1998:
310). Organic unities go some way towards justifying such
reaction, since the evil of punishment has not been cancelled
or reversed by desert. While the overall fitting response is one
of satisfaction, there is room for the sombre tone to which
Hurka alludes, given that the evil of punishment is still one of
the addends. But on the conditionalist analysis, there seems to
be no justification for adjusting ones feelings based on what
the criminal is suffering. Analogously, consider the good of
compassion, understood as being pained at someone elses
(undeserved) pain. The organicist makes sense of why, while
admiring compassion, we may think twice before getting
people to be compassionate: after all, while the whole has
positive value, pain felt by the compassionate preserves its
intrinsic negative value as pain here as elsewhere. But the
conditionalist will locate only positive value in such pain (Olson
2004a: 40) - the final value of compassion must after all result
from something good in it.
Lets consider briefly two possible replies to this second
objection. One is that the conditionalist need not take
retribution to be good, in order to explain why it is better as
such than absence of retribution. This easily explains the
sombre tone to be adopted: even on a conditionalist analysis,
the state of affairs is a bad one. And if one were to insist that
retribution is good (and not just better as such than its

VALUE AND NORMATIVITY

167

absence), then it is less clear why we should adjust our

feelings based on what the criminal suffers - of course, since
the state of affairs implies the evils of crime, one should not be
too cheerful about it.
This reply, however, may not work for the case of
compassion, where one wants to say that being compassionate
is a good character trait. Why think twice before encouraging
something good? Naturally there may be all sorts of contingent
reasons: e.g. compassion may make some people very weak or
vulnerable. But the objection is that there is something
intrinsic to compassion which justifies such hesitation, namely
the pain in which compassion consists.
Here the conditionalist can allow that the pain felt makes
compassion somewhat bad, only, not finally bad (bad for its
own sake) but personally bad: bad, to some extent, for the
subject who feels it. The overall appropriate attitude to a
compassionate person, particularly for those who are supposed
to care about what is good and bad for her, will take this into
account, by moderating admiration (based on the positive final
value of compassion) with a cautionary attitude (based on its
partly negative personal value) (see Chapter 4). A similar
explanation might also be applied to retribution. Even when
retribution is held to be finally good, the criminals loss of
freedom can be personally bad, i.e. bad for him. This implies
that the criminal, and those who are specially related to him,
have some reason to disfavour (e.g. be saddened by) the
punishment. A sombre emotional tone by a bystander would in
turn be justified not by the final disvalue of the criminals loss

168

of

freedom,

VALUE THEORY

but

as

way

of

respecting

the

sadness

VALUE AND NORMATIVITY

appropriately felt by friends and relatives of the criminal. 83

169

170

VALUE THEORY

Such explanations have a theoretical burden, in that they

VALUE AND NORMATIVITY

171

require introducing a notion of personal value. 84 And we have

172

VALUE THEORY

not found a general solution to the arbitrariness problem. 85 The

virtualist analysis also has its limitations. It remains to be seen
whether organicists can do any better, saddled as they are
with the issues of attitude mismatch and normatively idle
value, plus general doubts about the Moorean view of final
value, which is what motivated the doctrine of organic unities
in the first place.

5.6 Holism and particularism

Remember the claim made by the conditionalist:
D) An object need not retain its actual final value

across contexts and wholes.

This claim allows for contextual determination of both the
degree and the kind or polarity of a certain value. As such, it
stands against the claim (B) that an object retains its kind and
degree of final value across all contexts or wholes, which has
been variously labelled as the thesis of universality (Lemos
1994) or invariability (Zimmerman 2001). However, (D) as such
doesnt make value a radically local (i.e. contextual) or
variable affair. A conditionalist might say that while certain
things change their final value from one context to another,
other things never do, and these things might well be wholes.
For instance, sadistic torture might be always finally evil,
regardless of context. Or the virtues, even if built from objects
of conditional or variable value (pain, fear, or other attitudes),
might always be finally valuable. The conditionalist may go on
to produce a coherent list of such invariable values, and thus
offer a pluralist axiology. Or they may even claim that only one
sort of thing has final value, although it doesnt always have it,
because its value can be disabled in certain conditions, or
doesnt always have it to the same degree, because of
intensifying or diminishing conditions. This is what a
conditionalist hedonist might say: only pleasures have, i.e.
ground, final value (and pains disvalue), but not all pleasures
are valuable or valuable to the same degree. For example,
malicious pleasures are not good, and some social or
intellectual pleasures are better than some bodily pleasures (a

VALUE AND NORMATIVITY

173

claim that could turn out crucial for J. S. Mills qualitative

hedonism, see Olson 2004a: 48).
Conditionalism appears tantamount to what Dancy defines
as holism in the theory of value: a feature or part may have
one value in one context and a different or opposite value in
another (Dancy 2005: 333). And Moores thesis (B) is
equivalent to what Dancy calls atomism: what is a (final) value

174

VALUE THEORY

in one context is the same (final) value in all contexts. 86

However, Dancy does think of value as essentially a variable
affair, and in analogy with his theory of reasons, argues from
holism to particularism in the theory of value, which would
sound like this: the possibility of evaluative judgement does
not depend on a suitable provision of evaluative principles
(Dancy 2004a: 7). Value particularism stands opposed to value
generalism, the view that evaluative judgement cannot do
without a suitable provision of principles, i.e. invariant
evaluative truths, as exemplified for instance by pluralism and
conditional hedonism above. More precisely, particularism
doesnt deny that there might be invariant values, only their
existence is not required for evaluative thought, and so we
shouldnt expect to find any. It is indeed a chief question in the
debate over particularism whether, once Dancys holism is

VALUE AND NORMATIVITY

175

accepted, there is any room or motivation left for a principled

176

VALUE THEORY

picture of a pluralist or monist sort.87

Dancys particularism grants the existence of what he calls
default values alongside purely contextual values:
some features [e.g. causing needless pain] come switched
on already, as it were, though they can be switched off by
other features; others do not come switched on, but they
can be switched on by a suitable context. (2004a: 185)
As a matter of fact, some default values might be invariant
across contexts. But, as values, they always susceptible to
variation, though any such variation in value would require
explanation. Default values thus cannot form a fixed list of final
values, as in the pluralist view sketched above. What about
conditional hedonism? It might seem that pleasure behaves
precisely as a default value there. However, important
differences remain between a conditionalist axiology and
particularism. First, if the idea is to provide suitable guidance
for evaluative judgement, we would expect conditional
hedonism to keep variability under control, namely by
specifying in broad terms the sorts of conditions that can
switch off (or increase and decrease) the value of pleasure.
For particularists, contextual variability cannot, and anyway
doesnt have to, be controlled in such ways. Second, even if
conditional hedonism adopted an open-ended conception of
such conditions, it could not accept the idea of purely
contextual values: simply, context cannot give rise to other
final values besides states of pleasure. A monistic (or for that
matter pluralistic) structure, while compatible with Dancys
holism, must stop short of embracing the full range of
possibilities opened up by holism.

5.7 Summary
Understanding the chemistry of value requires understanding
the many ways in which value is explained by and related to its
super- venience base. Moreover, when it comes to complex
objects of value, there are competing accounts of how their
value depends on the values of the components or parts. I
have presented three approaches (organicism, virtualism,

VALUE AND NORMATIVITY

177

conditionalism), and noted how they all suffer from difficulties,

notably (but not exclusively) having to do with how it is fitting
to respond to such complex states of affairs. Still,
conditionalism might seem a better approach, to the extent
that it stems from a less rigid picture of intrinsic and final value
than the other approaches assume.

CHAPTER SIX

Value Relations
6.1 Introduction
In the previous chapter we explored different views about how
the value of a complex state of affairs results from the values
of its parts. Among other things we saw some reasons to prefer
an account which doesnt assume the value of objects to be a
fixed, permanent, or invariant matter, even when it comes to
their final or intrinsic value. In this chapter, I examine another
phenomenon which appears to be at odds with a tidy picture of
value: the idea that some values are incomparable or
incommensurable with others. I will present arguments from
Raz and Anderson meant to establish that value comparisons
are not always possible, and then discuss arguments from
Regan and Chang meant to defend value comparability. This

178

VALUE THEORY

will hopefully clarify at least one of the ways in which values

VALUE AND NORMATIVITY

179

are possibly related to each other.88

6.2

The trichotomy
thesis and
incomparability

It is natural to assume that, given two objects or states of

affairs X and Y, if we can compare them at all, then we can
always (in principle, if not in practice) rank them against each
other in one of these three ways: X is better than, worse than,
or as good as Y. For example, other things being equal, an hour
of innocent pleasure is better than an hour of undeserved pain;
unpunished crime is worse than punished crime; one printed
copy of Van Goghs Sunflowers is as good as another exactly
identical printed copy. This has been called the trichotomy
thesis (Chang 1997b: 4).
But sometimes we seem to be less confident making such
comparisons. Joseph Raz gives as an example the choice
between a career as a lawyer and a career as a clarinet player
(1986: 342). Neither career seems better than the other, all
considered. Are they therefore equally good? If they were, then
a slight improvement in either career would make choosing
that career better than the other. So, suppose you will earn
$1,000 more as a lawyer. Raz suggests that this slight
improvement is not sufficient to make the lawyers career
better than the clarinet players. But this is how a tie between
equally good alternatives can normally be broken: e.g. the fact
that a dress at shop A costs even a little bit less than the same
style of dress at shop B makes it better to choose shop A, other
things being equal. So, he concludes, the two careers are not

180

VALUE THEORY

even equally good.89 If the trichotomy thesis is true, then they

are incomparable in value: no positive ranking or comparison
between the two is possible.
Note the structure of the problem: the two items are incomparable under a certain aspect, i.e. their value, or more
precisely their value as careers. The two careers can of course
be compared under other aspects: maybe the clarinettists
career is longer than the lawyers career, possibly the lawyers
career allows more leisure time than the clarinettists, etc. So
they need not be incomparable under every aspect (are any
two things as a matter of fact totally incomparable?). But, as
we know by now, value supervenes on non-value. So a full
survey of the features which make either career valuable
should also bring out the reasons why the two are incomparable in value.
There are two possibilities here. One is that, in fact, the two
careers are comparable under every non-evaluative aspect,
but there is no way of settling which of these comparisons
matter more, nor whether they equal each other out. The idea
is that the same values, or rather the same value-making
features are present in both cases, albeit to different degrees
in each. As a clarinet player, you will earn enough money to
live on, but not as much as you would as a lawyer. As a lawyer,
you will cultivate your intellectual skills, but not as well as you
would as a musician. And so on. In this case, each valuemaking feature is shared, but there is no overarching common
measure or method to balance their contributions against each
other. We can call this overall incommensurability.
The other possibility, perhaps more realistic for this sort of
case, is that each career contains some value-making feature
that the other simply lacks. Perhaps the kind of aesthetic
sensibility you develop as a clarinettist finds no match in a
lawyers career, and the same can be said mutatis mutandis
for the dialectical skills required for a lawyer. Further, it is not
clear that the one value-making feature contributes more value
than the other. Nor is it plausible to say they contribute value
in equal measure: aesthetic sensibility and dialectical skills are
just very different abilities. So, in this case, the items are
incomparable because they are partly made valuable by
distinct features whose values are, at least as far as they go,

VALUE AND NORMATIVITY

incomparable.90

This

we

can

call

181

pro

tanto

182

VALUE THEORY

incommensurability.91
So incomparability between two objects (choices, states of
affairs, etc.) can happen either because there is no way to
overall measure the importance of their non-evaluative
features, even if shared to some extent, or because their nonevaluative features have not enough in common for their
evaluative contributions to be compared with each other.
What if there were only one fundamental value, e.g.
pleasure? The lawyers career and the clarinet players career
would have a common value-making feature: the pleasant
experiences they contain (ignoring for arguments sake the
pleasure of other people involved). This would rule out what I
call pro tanto incommensurability, since both careers would
contain the one value-making feature. But it is interesting to
note that, by itself, such a monistic view wouldnt rule out
overall incommensurability. For suppose the clarinet players
career contains less intense pleasures, but longer lasting ones
than the lawyers. How do we balance intensity and duration of
pleasure against each other? There may be no way to do this
for each case, nor is there any pressure to conclude that such
careers must therefore be equally good. The availability of a
common measure, in other words, doesnt guarantee
comparability between any two items.
And on the other hand, a pluralist view, whereby there is
more than one fundamental value (e.g. pleasure and virtue),
makes pro tanto and overall incommensurability conceptually
possible, but not inevitable. Spending money on a morally
good cause might be better than spending it for ones
entertainment, even if these choices concern completely
different values, with nothing in common. In other words,

VALUE AND NORMATIVITY

absence

of

common

measure

doesnt

183

make

value

184

VALUE THEORY

comparisons impossible.92

6.3

A fitting attitude
argument for
incomparability

Apart from an appeal to intuition in particular cases, can a

general argument from the nature of value be made for the
possibility of incomparability? So argues Elizabeth Anderson.
She starts out from a premise that fitting attitude minded
philosophers should agree on: there is nothing else for a value
to be but something that guides our deliberations and attitudes
in practice (1997: 91). To speak of options whose values are
comparable or incomparable is thus to speak of the attitudes
one should take towards them. If two options are comparable
in value, then it makes sense (or it is fitting) to express
attitudes of a comparative sort towards them: desire (admire,
etc.) the one more than the other, or favour them equally. And
conversely, if it makes no sense to express such attitudes
towards any two given options, then the goods are
incomparable. Anderson calls this the no good reasons
principle for incomparability (ibid.: 99).
She argues that the principle generates many plausible
cases of incomparability, ranging from aesthetic cases
involving crosscategory items (e.g. novels and sculptures) to
tragic dilemmas involving choices between two loved people.
In all these examples, it seems, it is positively unfitting to hold
comparative attitudes. The project of universal aesthetic crosscategory comparison can only be for philistines, snobs, and
prigs (ibid.: 99), that is, for people whose sense of aesthetic
appreciation is distorted by various factors. The proper
development of an aesthetic sensibility is incoherent with the
notion of a universal art ranking, and the attitudes that would
go with it. In the case of a value comparison between loved
ones, it would need to be true both (i) that there is one single
common attitude to be expressed e.g. both towards ones
mother and towards ones friend, and (ii) that there is a
possible answer to the question how intense such an attitude

VALUE AND NORMATIVITY

185

does each deserve?. But these assumptions are not always

justified: (i) different people deserve different sorts of attitudes
(e.g. filial love and friendship), (ii) even when two people
deserve the same sort of attitude, e.g. respect, the notion of
deserving respect of intensity n may not (always) be
applicable (ibid.: 103). A decision between incomparable
options might still be possible, but it will have to be grounded
on something else than their comparative values. Moreover
(although Anderson doesnt mention this possibility), when
decision is not at stake, it might be fitting (for me as well as for
others) to refuse to compare ones actual career as a musician
to a merely potential, alternative career as a lawyer.

186

Andersons

VALUE THEORY

argument

has

certain

force,

when

VALUE AND NORMATIVITY

187

reconstructed this way.93 Note, however, that it cannot simply

depend on the no good reasons principle. That principle
should be acceptable also to a philosopher who denies
incomparability, because it expresses a general truism about
the relation between value concepts (including comparability)
and fitting or appropriate attitudes. What her argument
depends on is a certain substantive interpretation of this
truism. In particular, she denies that there is always some
single kind of attitude fittingly directed to different valuable
objects or people. This is the crucial point: there is good reason
to compare two items only if there is good reason to hold the
same sort of attitude towards each, with a different or identical
intensity or polarity as the case may be. If x and y are
comparable, then it must be fitting to have e.g. a desire-like
attitude towards each - and then, towards either, a stronger,
weaker, positive, negative, or equal desire-like attitude, as the
case may be.
Now, defenders of FA might well agree that different kinds of
responses are in general fitting towards different objects of
value (see Chapter 7), but in this connection Anderson needs a
stronger claim, namely this: similar kinds of responses are
unfitting towards different objects of value; at least in certain
cases, that is, similar responses are ruled out. For instance,
Anderson implies that it is unfitting to desire ones mothers
staying alive more than to desire to save a friendship, and vice
versa, or to desire these things equally. Even if it might turn
out that we ought to save our mother, this would not be
because saving our mother is what we ought to desire the
most. A different FA theorist might instead believe that, even if
different objects do merit different responses (filial love,
friendly loyalty, etc.), comparison might at least in principle
always have a point, to the extent that even very different
objects also merit similar kinds of responses, such as desire or

188

VALUE THEORY

preference.94 This is not meant per se as an objection to

Andersons view, but it points out the sort of argumentative
burden she has to face.

6.4

Against incomparability:
Epistemic limitations

Donald Regan defends both the trichotomy thesis, and what he

calls the complete comparability view. In one single
formulation: given two items sufficiently specified as far as
their value is concerned, it will be true that either is better than
or as good as the other (1997: 129). On this view, then,
putative instances of incomparable values are really cases
where one of the three relations holds, but we do not know
which one.
Regan, like Anderson, starts out from a general view about
the relation of value and attitudes. Regans central claim, at
least for one of his arguments, is that [v]alues, properly
apprehended, motivate (ibid.: 141), and motivate to the
appropriate degree. So, consider the choice between the
clarinettists and the lawyers career: since both are valuable to
some degree, an ideal agent, who properly apprehends values,
would be motivated by each to some degree. But then the
ideal agents choice according to the stronger motivation will
thereby reveal which one is better, i.e. how they compare (if
there is a motivational tie, i.e. the ideal agent has equally
strong motivations, then they are equally good). In other
words, when we consider any putative case of incomparability,
we should not assume that we know that neither option is
better than the other: as fallible and less than ideal agents, we
might have failed to properly apprehend the relevant values.
There are two weak points in Regans argument. The first is
that the limitation attributed to actual agents cannot be simply
epistemic or propositional, i.e. such that we do not know
whether a career is better than the other. Our predicament, as
should be clear from the explanation in terms of pro tanto and
overall incommensurability, is rather such that we dont know
how to balance or weigh the pros and cons of the two careers
against one another. This is a practical limitation. But if this is
so, then we cannot rule out, as a candidate explanation of our
predicament, that there is no way to balance them: there

VALUE AND NORMATIVITY

189

might be no way to operate the weighing mechanism in this

case, because no such mechanism is available. To use another
metaphor: if I dont know how to ride a certain bike, it might
simply be that the bike is not fit for riding.
Regan might at this point retort that we dont know how to
compare, not because no comparison is possible, but because
we are not ideal agents. This brings us to the second worry.
Regan assumes that the attitudes and dispositions of an ideal
agent are always comparative in kind. Faced with any valuable
state of affairs, an ideal agent is essentially willing (and of
course able) to assign it a position in a ranking of options. But
this is a substantive assumption, which doesnt simply follow
from the idea that values, properly apprehended, motivate. To
be sure, Regan is in no worse dialectical position than
Anderson on this score. However, his assumption does follow
from a further claim he makes early in the article: the proper
ultimate aim of practical reason is to produce the best state of
affairs possible (ibid.: 131). If the ideal agent is the (ideal)
personification of practical reason, then her ultimate aim is to
produce the best state of affairs, and this aim requires taking
an essentially comparative attitude towards options.
However, now that the source of Regans assumption has
been exposed, it is clear that it needs further defence on two
counts: (1) the ultimate aim might be producing not the best,

190

VALUE THEORY

but the sufficiently good;95 (2) even if the best is a legitimate

ultimate aim, it might not be a mandatory aim - Regan seems
to think that values always impose what have been called
peremptory reasons, such that we ought to act on them or

VALUE AND NORMATIVITY

191

respond in a certain way to them.96 But many have suggested

alternative roles or sorts of reasons: some reasons are enticing
rather than peremptory (Dancy 2004b), they make options
eligible rather than required (Raz 1999), or they justify rather
than require action (Gert 2004a). It remains an open question
whether an ideal sensitivity to reasons means that an agent
ought to aim at producing the best state of affairs, and in turn
ought to hold the essentially comparative attitude required for
that aim. Regan might be right, but more argument is needed
to establish complete comparability.

6.5

Against incomparability:
Parity

Even if Regan is wrong, still one should not therefore jump to

the conclusion that the legal career and the musical career are
incomparable. Ruth Chang has forcefully argued for the
existence of a fourth comparative relation besides being better
than, worse than, and equally good: being on a par with. Chang
agrees with the Razian argument that, if a small improvement
on the legal career doesnt make it better than the musical
career, then the two careers are not equally good (nor is the
one better than the other). But she argues that a further
possibility needs to be considered before declaring them
incomparable. So she defends comparability precisely by
denying the trichotomy thesis.
Her chaining argument starts from the idea that some
comparisons can be made even between very diverse items: in
her example, Talentlessi, a very bad sculptor, is clearly worse
(with respect to a relevant value such as creativity) than
Mozart, a great composer. Then she appeals to the principle
that small unidimensional differences cannot change two
comparable items into incomparable ones. So a slightly better
sculptor than Talentlessi is still comparable to Mozart. Apply
this in turn to ever more creative sculptors, and you will be
forced to hold that Michelangelo is comparable to Mozart,
despite their diversity. If neither is better than the other, nor
are they equally good, then we need a fourth comparative term
to describe their relation (Chang 2002: 674-5).
There are two worries about Changs position. First, the
structure of Changs chaining argument looks suspiciously like

192

VALUE THEORY

the famous sorites paradox. Consider John, who has a head full
of hair, and Jack, who has only one hair. John is hairier than
Jack: this is the relation they are in. Now, add one more hair to
each successive Jack. If the principle of small unidimensional
differences is true, then the relation between the two items
would be preserved through any small unidimensional change
in one of the two. It follows that John must be hairier than even
a very hairy Jack - who has the same number of hairs as John.
But this is absurd.
The sorites paradox is supposed to show that certain
predicates (being hairy, being bald) have no precise or
determinate boundaries: they are vague predicates. By the
same token, one could say that the predicate being
comparable to is vague as well. Talentlessi is worse than
Mozart, but it might still be indeterminate whether and how
Michelangelo is comparable to Mozart. There might be no fact
of the matter about that: it is neither true nor false that one is
better than the other or as good as the other. The chaining
argument thus seems to motivate the view that comparative

VALUE AND NORMATIVITY

193

terms are vague, rather than showing the need for a fourth

194

VALUE THEORY

comparative relation.97
Chang replies to this worry by offering a general reason why
comparative terms such as better than cannot (always) be
treated as vague predicates (ibid: 682-6). The comparison
between Mozart and Michelangelo is a hard one to make. If
comparative terms were vague terms, this comparison would
be analogous to the decision over a borderline case of, e.g.,
being bald. Chang points out that we proceed in very different
ways in these two cases. In cases of borderline application of
vague terms, we can resolve a dispute as to whether Jack is
bald essentially by flipping a coin. This is in effect tantamount
to stipulating a new non-vague predicate (being bald*) and
applying it to Jack. But, Chang observes, we cannot do this in
the case of value comparisons between Mozart and
Michelangelo: there seems to be a substantive disagreement
which cannot be resolved by flipping a coin, nor (equivalently)
by stipulating a new term of comparison (being better than*)
and applying it to either. This would be to change the topic of
our disagreement.

6.6

Parity and choice

So, despite having a sorites-argument form, Changs chaining

argument does not require that comparable be a vague
predicate, and can support the idea that we need a fourth
comparative relation. A second worry about her argument,
however, is that she has not done enough to support the
existence of a new fourth comparative relation as basic as
being better, worse, and equally good.
Joshua Gert (2004b) has articulated this worry. He starts
from an assumption which readers of this book will find
congenial (and which Chang has no reason to reject either):
talk of values is or is closely tied to talk about justified (fitting,
appropriate, etc.) attitudes. In particular, for Gert, talk of value
comparisons is closely tied to talk of permissible choice: A is
worse than B when choice of A over B is rationally
impermissible. So, given what Chang has told us, parity can in
principle be defined as follows: A and B are on a par if and only
if it is not a mistake, or irrational, to choose A over B, or B over

VALUE AND NORMATIVITY

195

A, and ... this may continue to be true even if one of the items
is slightly improved (2004b: 506). Given the value/rational
choice link, this means that A and B are on a par when neither
is better than the other, nor are they equally good (remember
that if they were equally good, any slight improvement on
either would make it better than the other).
However, Gerts point is not just that parity is definable in
terms of other value relations. Rather, the category identified
by As and Bs that are neither better than one another nor
equally good, and yet comparable, is not well described by the
relation of parity. The cases at issue can be represented as
involving overlapping ranges of strength in rationally
permissible preferences (or choices), e.g.:
1 A, B: 8-12, 8-12 (= we can have a preference for A to

any degree between 8 and 12, no less and no more,

and the same for B).
2

C, D: 9-10, 9-10.

E, F: 10-12, 10-20.

G, H: 5-10, 9-15.

196

VALUE THEORY

In (1), the ranges are perfectly overlapping, and this for Gert

VALUE AND NORMATIVITY

197

we can call a case of parity. 98 In (2) the ranges are narrower

and approximate one unique value (a one-point range). Since a
unique value would mean equality, Gert calls this case rough

198

VALUE THEORY

equality.99 In (3) and (4) the permissible ranges are quite

markedly different: F has a much higher top bound than E; G
and H are almost disjointed. So, even if still partially
overlapping, the relative items cannot be said to be on a par.
The moral is that Changs parity covers too many different
cases in which it is true that it is not a mistake to choose A
over B, or B over A, and this may continue to be true even if
one of the items is slightly improved. Parity is, at most, just
one possibility in a class of cases that can be described using
the usual comparative relations.
Chang (2005: 344-7) has replied to Gerts objection by
defending parity as involving distinctive constraints on the
sense in which it is permissible to prefer A to B and vice versa.
Also equally good items are such that it is OK to prefer either
(at least in the choice-related sense of preference: otherwise,
equally good items are such that one should be equanimous
towards them, i.e. not prefer one to the other). What makes
parity different, however, is that over a series a choices among
items on a par we might end up with less value than we could
have had. So its not always permissible to choose either of two
items on a par. This is due to the immunity to slight
improvement which is characteristic of parity. Suppose A and B
are on a par: A is a decent cup of espresso, B is a good cup of
tea. A+ is by definition an improvement on A (a slightly better
cup of espresso), but still is on a par with B. The first choice is
between A+ and B: they are on a par, so we can choose either.
Suppose we choose B. The second choice is between B and A:
even if the items are again on a par, now it would be a mistake
to choose A over
B, because we would end up with less value than we
could have had, had we chosen A+ at first. We would
end up with a worse cup of espresso than we could
have had. In this sense, the first choice constrains the
second, even if both choices are between items on a
par. No such constraints apply to equality, since serial
choices between equally good items are guaranteed to
result in identical amounts of value.

VALUE AND NORMATIVITY

6.7

199

Parity and incomparability

Changs defence of parity stands against the trichotomy

thesis: if items are comparable, then there will be four and not
only three possible ways of comparing them. But it is also
meant to remove ground for the thesis that incomparability is
possible and indeed inevitable. The idea is that, before
declaring two items incomparable, we now need to make sure
that they are not on a par. And this might prove rather difficult,
since Razs argument from slight improvements was shown to
be compatible with the possibility of parity.
Moreover, the choice among incomparable items seems to
be subject to the same constraints as the choice among items
on a par that we have seen just seen. If incomparable items
are such that it is always OK to prefer either, then over a
series of choices we would get the same undesirable results as
with items on a par. A pattern of unconstrained successive
choices between incomparable items, distributed across a
lifetime, might land us with a life barely worth living, that is, a
life from which value has been pumped away by each
subsequent choice, and justifiably so, because at each choice
situation we were permitted to choose either item (Chang
1997b: 11).
However, by way of defending the possibility of
incomparability against the comparativist alternative offered
by parity, two things should be kept in mind. First, if parity
holds, then it is not clear what it means to compare the value
of items on a par. On top of ascertaining the values of each
item, comparing involves an attitude (or an act) of relating
their values to each other in such a way as would yield some
novel fact about them. When we compare the weights of two
objects, we get to know which one weighs more. When we
compare the ages of two people, we get to know who is older.
This is not only true of numerical comparisons: when we
compare the taste of two chocolate bars, we get to know
which one is sweeter. And so on. However, when we compare
items that are on a par, we dont get to know any such novel
fact, except what we already knew, e.g. that Mozart is great
and Michelangelo is great. Of course, we do get to know
something negative about them: that neither is better than or

200

VALUE THEORY

as good as the other. But it is not clear why we should

interpret this negative finding as a successful comparison
rather than as a failure of comparison.
Second, even if we can make some sense of the notion of
comparing items on a par, there is still plenty of room left for
incomparability. The remark by Chang above, about series of
permissible choices resulting in value being pumped away, is
indeed evidence that incomparable items, like those on a par,
are not such that it is always OK to choose either. But in itself it
doesnt show that the only way around value pumps is to
impose
constraints
on
permissible
choices
among
incomparables, as it is for items on a par. Another reaction
might be to simply deny that the idea of incomparability is tied
at all to that of permissible choice. Andersons cases above
pointed to a rather different understanding: two items are
incomparable when it is unfitting to compare them
when there is sufficient reason against comparing
them. In a similar vein, Rabinowicz suggests that
incomparable items are such that the absence of a

VALUE AND NORMATIVITY

201

disposition to choose either is required (2008: 25-7).100

Parity, on the other hand, even when subject to the
necessary constraints to avoid value pumps, is
essentially related to the idea that it is OK to prefer
either of two items on a par, and thus that it is not
unfitting to hold a comparative attitude towards them.
These definitions leave it open, of course, that other noncomparative attitudes or dispositions might still be fitting
towards either item, and thus that we might be able to make a
justified decision among incomparables, particularly if and
where deciding is indeed better than not deciding at all. This is
clearest in the career choice. Even if neither career is better as
a career, we can still legitimately embrace either on the basis
of considerations that do not pertain to their comparative
values, that is, that do not show the two careers to be
positively related: for example, that one career is not worse
than the other could be such a justifying consideration (see
Hsieh 2007). In some range of cases, we can even flip a coin or
let non-rational motives decide, provided that our decision
does not reflect an evaluative comparison or an intention to
choose one option over another. Or as Rabinowicz says: We do
make a choice if we have to, but this choice is made without
the conflict of reasons being resolved (2008: 26).

6.8 Summary

In this chapter, I have focused on one value relation, besides

the usual ones of being better/worse than, and as good as,
namely the negative one of two values or value instances
being incomparable.
I
have considered arguments by Raz and Anderson in
favour of the possibility of value incomparability, and
two different rejoinders from the comparativist camp,
one defending the traditional trichotomy (Regan), the
other pointing to a fourth possibility of comparison
(Chang). In all these cases, the relation of value to
fitting attitudes helped to shed light on the issues, or
even provide considerations in favour of either side.

202

VALUE THEORY

CHAPTER SEVEN

How Do I Favour Thee?

7.1 Introduction
The fitting attitude approach to value appeals to the idea of
favouring valuable things for their own sake or for someones
sake. But what exactly does this mean? Here I try to articulate
somewhat this concept along different dimensions (kind,
telos, and degree of favouring), while bearing in mind that a
full account of fitting attitudes can be given only
simultaneously with a substantive value theory or axiology,
which is not our present aim. But I will critically discuss a
substantive monistic proposal, namely the idea that promotion
or maximization must be the fundamental fitting response to
value. In the end I consider a thesis which, if true, would
undermine the value-theoretic debate about the nature of
fitting attitudes, as well as in other areas, as a case of
philosophers talking past each other.

7.2 Three dimensions of favouring

If value and fitting attitudes are in broad correspondence with
each other, as we have been assuming so far, then the full
story about particular values or valuable states of affairs will
require a specification of the relevant fitting attitudes. So far
we have been working with the notion of favouring (and
disfavouring) an object or a state of affairs. But this is a generic
notion, meaning nothing more than an attitude of being in
favour of (or against) something, a pro-attitude (or a con-/anti-

VALUE AND NORMATIVITY

203

attitude) towards something.101 Thin as it may sound, the basic

idea is not completely empty: favouring must be, at least, a
positive intentional state, a favourable response towards a
certain object of thought, or towards a fact. This rules out two
sorts of mental states from figuring in the relevant class:
(i) non-intentional mental states such as: feeling hungry
(itchy, sleepy, etc.), being in a certain mood (bored,
euphoric, etc., but not about anything), sensing
pleasure (as opposed to taking pleasure in something),
and so on; (ii) intentional but not valenced (positive,
negative, neutral) mental states: mere believing,
contemplating, imagining, supposing, remembering,
and so on (believing that something is good should
count as a possible pro-attitude here).
It seems to me that favouring can be determined along
several dimensions:
1 What kind of attitude it is: a desire, a practical stance,
2

an emotion?
Its direction or telos: for the sake of what or whom?
3 Its intensity: how strong, and how stronger than
attitudes towards other things?

These three dimensions, while perhaps not exhaustive, are

typically present, though often implicitly, when we make
ordinary value judgements. Here is a rough illustration.
1. The first dimension depends on a number of features of
the valuable object. For example, we can consider its
existential, modal, or probabilistic status. Something is
intrinsically good, Ewing suggests, when, on its own
account, we ought to welcome, rejoice in it if it exists,
seek to produce it if it does not exist. We ought to
approve its attainment, count its loss a deprivation,
hope for and not dread its coming if this is likely, avoid
what hinders its production, etc. (1948: 149, my
emphasis). Or again, it can matter which ontological
category a value belongs to. For Ewing, an intrinsically
good experience (e.g. an innocent pleasure) is fitting to
be pursued, but an intrinsically good action is as such
admirable or worthy of approval, though not
necessarily fitting to be pursued or chosen for its own

204

VALUE THEORY

sake (ibid.: 153-5). Furthermore, as hinted in section

2.4, there seem to be attitudes that are fittingly
directed towards individual objects and people, and
only indirectly towards states of affairs or propositionlike entities containing those individuals: preserving,
protecting, cherishing, respecting, trusting, loving and
hating people. And conversely there are attitudes
primarily directed at states of affairs rather than
individuals, like desiring, taking pleasure in, preferring
(at least in one sense), promoting. According to some
philosophers, there is one kind of practical attitude
(promotion or maximization) that enjoys some kind of

VALUE AND NORMATIVITY

205

normative priority among others. I will come back to

206

VALUE THEORY

this in section 3.102

2. I used the for the sake of locution to distinguish final
from non-final value (Chapter 2), and personal from
impersonal value (Chapter 4). Final value is to be
favoured (in some of the ways above) for its own sake.
Personal value is to be favoured for someones sake.
Lets examine final value first. In Chapter 2, I discussed
and tentatively argued for the claim that final value
can be intrinsic or extrinsic, i.e. can supervene on
internal or relational properties of the valuable thing,
where these also include relations to other valuable
things.
In the case of extrinsic final value, relational properties can
play different roles within the supervenience base (see section
5.2). Consider our familiar Napoleons hat. Its final value
results from its having belonged to a historically important
man. And the significance of this fact is conditional on, or
enabled by, history having some sort of final value (at least for
the sake of argument). If history didnt matter, Napoleons hat
would be just another worthless accessory. For more abstract
sorts of final value, e.g. the value of a certain friendship,
likewise some enabling evaluative conditions need to hold,
such as its not being immoral, or its being good for the
participants.
While it seems clear that intrinsic and unconditional values
are to be favoured for their own sake, it might not be
immediately obvious how we can favour extrinsic and
conditional values for their own sake. For example,
Zimmerman claims that we should favour final value for its
being what it is ... and not for the sake of its relation to some
other valuable state (Zimmerman 2001: 91). Is Zimmermans
conception of favouring for its own sake also applicable to
extrinsic and conditional values? I think so. Two things should
be clear from the discussion in Chapters 2 and 5: (i) extrinsic
value is not necessarily instrumental value; (ii) an enabling
condition for value is not a ground of value.
Lets start with (i). An object like Napoleons hat is to be
favoured (e.g. preserved) for the sake of its relation to
Napoleon, but this is not an instrumental relation to Napoleon
or to any valuable state. The object is fitting to preserve

VALUE AND NORMATIVITY

207

because of the historical role of Napoleon, and ultimately,

because of the value of history. Now, in a sense, this puts the
hat in a relation to some other valuable state: Napoleons
deeds, and ultimately world history. But note that it makes no
sense to say that the hat itself produces or is a means to
history, as if history were a separable outcome caused by the
hat. Rather, the connection to Napoleon forges a unique link
between the hat and history, analogous to the link between a
wedding ring and a relationship, such that the loss of the hat
(or of the ring) constitutes a loss in value even in the presence
of other hats (or rings) similarly historically connected.
Therefore, to say that the hat has to be preserved for the sake
of history, which is surely correct, implies (rather than

208

VALUE THEORY

contradicts) that it has to be preserved for what it is, as

VALUE AND NORMATIVITY

209

required by Zimmermans definition. 103

Now to (ii). If the value of friendship requires the value of
other things connected to friendship, this doesnt mean that it
is for the sake of its relation to these other things that it is
fitting to favour (engage in, respect, etc.) a given instance of
friendship. It only means that our favouring friendship should
not occur in isolation from our favouring those other things.
This is because the conditions for the value of friendship affect
but do not exhaust its value. Friendship is valuable as the kind
of relationship it is, whose value doesnt simply result from
being morally OK or good for the participants. So we can, and
should, value friendship for what it is, when those conditions
obtain.
The cases above can be contrasted with extrinsic relations
which do exhaust a things value, and therefore make it fitting
to favour it for the sake of a relation to some other valuable
state. The obvious example is instrumental value. The radiator
in my room, or the fact that it works, should certainly be
favoured (during winter), but our evaluative attention or focus
is or should be confined to its benefits to me and others. This is
the value for the sake of which we should favour the radiators
working.
I havent delved as deep into for someones sake
attitudes. We shouldnt assume that these have
something more substantial in common with for its
own sake attitudes than their being positive responses
towards something. Their being essentially personoriented might make them rather different in kind.
First, there is a notable structural difference. In
impersonal final value, the valuable object is that for
the sake of which an attitude is fitting. In personal
value, it isnt: we favour an object for someones sake,
because we are interested in its being good for a
person, and ultimately in the person herself. Second, it
seems clear that desiring for your own sake that you
live a happy life is not the same as desiring, for its own
sake, that you live a happy life. The level and kind of
concern displayed for you is different in the second
case, even if you are within a state I desire finally. The
difference shows up, at least in part, in the emotional

210

VALUE THEORY

reaction to the frustration of my desire: sadness in the

VALUE AND NORMATIVITY

211

first case, but only disappointment in the second. 104

Third, final and personal value can also be combined,
as explained in section 4.4: when we correctly favour
something for someones sake, we can still make a
mistake as to how we favour it. For example, it might
be a mistake for me to favour, for your sake, your
getting as rich as possible for its own sake. So it seems
that we are dealing with related but different attitudes.
An analysis of for someones sake attitudes is beyond the
scope of this book. The basic idea is that we favour x with an
eye to someone (else) (Rannow-Rasmussen 2011: 56), and in
particular with an eye to how certain properties of x relate to
certain properties of a person. If we think that pleasure is
finally good for someone, then this value will supervene on
some relation between pleasure and the person: intuitively,
pleasure directly contributing to her well-being. The problem
with this is that, in order to explain which properties of a
persons are relevant, we have referred back to the concept of
well-being or goodness for. Perhaps a non-circular account of
good for is unavailable. Or perhaps we could appeal to some
psychological description of the attitude of caring for someone.
In any case, this point serves again to underscore the
difference between for its own sake and for someones sake
attitudes: in the former case, we are not necessarily concerned
about anybodys well-being. Perhaps we should conclude that
the word sake is semantically ambiguous (ibid.: 60).
3. Value comes in degrees, and so should fitting attitudes.
Some finally valuable things might be much better or
much worse than others, and our fitting responses
should be sensitive to these differences. Likewise, a
minor evil is fittingly responded to with mild disfavour,
while in response to a major evil the same mild
disfavour is not a fitting response, or anyway less of a
fitting response than strong disfavour.
One could suggest that talk of milder or stronger favour and
disfavour is really reducible to talk of preferences. That the
object has a given degree of value entails that the object has a
certain position in a fitting ranking of things, but not that it
should be favoured or disfavoured with a corresponding
strength. And preference, in turn, may not be reducible to

212

VALUE THEORY

greater favour: e.g. to prefer the lesser evil to the greater is

VALUE AND NORMATIVITY

213

not to favour the former more than the latter, because that

214

VALUE THEORY

would be to take a positive attitude towards evil.105

One problem with this reductive approach is that talking
about degrees of value is simply not tantamount to making
value comparisons, and thus to preferring something: we are
not always interested in betterness when e.g. we judge a
certain state of affairs to be very good (e.g. we need not imply
that it is better than most states of affairs of that kind). A
second problem is that, even if we accept preference as a
distinctive attitude, there is still an issue about by how much
one should prefer one thing to another. For a schematic
example: ceteris paribus, preferring one minute of pain to two
hours with the same intensity as one prefers one hour of pain
to two hours seems intuitively wrong. Both one minute and one
hour of pain are better than two hours, but one minute is much
better, and our preferences should reflect that. For a less
schematic example: we should prefer a good pianist to a bad
one more than we should prefer a mediocre to the same bad
one. Betterness comes in degrees, and so should the relative
intensity of preferences.
It is worth mentioning another challenge to the idea that
fitting attitudes come in degrees. Degrees of value might be
said to correspond to degrees of fittingness, rather than
degrees of attitudes. A good (10 units of pleasure), it could be
said, is better than another (5 units of pleasure) when it is
more fitting to favour the former than it is to favour the latter,
rather than when it is fitting to favour the former more (i.e.
more intensely) than the latter. What is gradable here is the
property of being fitting, not the intensity of attitudes. Even

VALUE AND NORMATIVITY

215

admitting this conceptual possibility, 106 the proposal has two

significant difficulties.
First, lets think about the notion of more fitting than. In
one case, we have a good x of degree 10, a generic attitude of
favour towards it, and (lets suppose) a property or relation of
fittingness between the two to a degree of 10. In another case,
we have a good y of degree 5, the same generic attitude of
favour towards it, and a property or relation of fittingness
between the two to a degree of 5:
(1) xG10 & xFavour F10]
(2) yG5 & yFavour [^ F5]
If this is what is meant by (1) being more fitting than (2), i.e.
something like five times more fitting, then it borders
nonsense. This way of putting things might suggest that (1)
contains more fittingness than (2). Or alternatively, that (1) is
a better specimen of fittingness than (2). But such claims
(whatever exactly they mean) are clearly unjustified. In the
two cases we have a good of a certain degree, and the same
kind of response to each good. Provided that the response is of
a fitting kind (and it is), (1) and (2) can only be equally fitting.
Maybe more fitting is more charitably interpreted as: the
reason to favour x is stronger or weightier than the reason to
favour y. Degrees of fittingness should be understood as
degrees of normative strength or weight, a notion familiar from
talk about reasons and deontic language. But this proposal is
also untenable, as it stands. If reasons here are simply another
way to talk about fittingness, the idea then must be that
reasons of fittingness to favour x are stronger than reasons
of fittingness to favour y. But as we have just seen, as far as
fittingness goes, (1) and (2) are on an equal footing. So favour
as such is no more fitting or appropriate to x than to y. So
there cannot be stronger reasons of fittingness to favour x.
Now, the case for x is stronger than the case for y, but this
normative truth should be expressed differently. While it is not
more fitting to favour x than to favour y, it is fitting to choose x
over y, should it come to that. And this is because, given how
good x and y are compared to each other, it is fitting to favour
x more than y, and in turn it is fitting to choose what we should
favour the most.

216

VALUE THEORY

The second difficulty with more fitting than is that it cannot

hope to replace the relevant distinctions. Quite simply, we do
evaluate attitudes, feelings, and emotions on the basis of their
intensity, and other relevant features such as duration,
resistance to certain challenges, and so on. With regard to
different objects or states of affairs, some degrees and levels
of concern, desire, satisfaction, fear, frustration, etc., are more
or less proportionate than others. This often has to do with our
particular position as situated agents (e.g. normally we should
care more for the near and dear than for strangers) or as
epistemic subjects (e.g. normally we should fear the likely
more than the unlikely). But just as often it has to do with
differences in the personal and impersonal values we are
responding to. It would be a mystery why we should respond to
value by having certain attitudes, while a central feature of

VALUE AND NORMATIVITY

217

attitudes, namely their strength, in a broad sense, is not to be

218

VALUE THEORY

involved in this process.107

Once we admit degrees of attitudes, there arise intricate
questions regarding how symmetrical relations between
degrees of value and degree of fitting attitudes should be. For
example, suppose I am enthusiastic about a minor
achievement (say, ironing a shirt). My favouring response is of
a fitting kind, but not maximally fitting: I should be satisfied
rather than enthusiastic. Is my response thereby unfitting or
only less than maximally fitting? On the other hand, notice that
when a response is of the unfitting kind, e.g. if I am happy
about an earthquake in Japan, there seems to be no limit to
how unfitting my response can be - there is no maximal unfittingness. Again, another question is whether favouring the
neutral (what is neither good nor bad) is positively unfitting, or

VALUE AND NORMATIVITY

219

simply not fitting, and whether disfavouring the neutral is

220

VALUE THEORY

similarly unfitting or simply not fitting.108

There seems to be no reason to expect a perfect symmetry,
based on conceptual considerations alone. These are
substantive issues, which should not be decided in advance by
a fitting attitude account of value. If we want to leave these
questions open to substantive disagreement, though, we had
better formulate FA accordingly. For instance, we cannot define
finally good to degree n simply as fitting to be favoured for its
own sake to some degree, because this would allow any
degree of favouring towards the finally good to be fitting,
regardless of the degree of goodness. But we have seen that
not only different degrees of favouring can be more or less
fitting, but excessive favouring of some good could be even
unfitting (and the same might be true for defective favouring,
and mutatis mutandis for excessive/defective disfavouring of
the bad). On the other hand, we dont want to define good to
degree n as fitting to be favoured precisely to degree n,
because we want to allow at least some degrees of favouring
<n and >n to be fitting rather than not fitting, albeit less
fitting than favouring to degree n. It seems then that all we can
say is that if x is finally good to degree n, then x is fitting to be
favoured for its own sake to a degree sufficiently close or
proportional to n, where room is left for determining this
sufficiency from a very inclusive range (favouring x to degree
n is perfectly fitting, but any favouring will still be fitting) to a
very exclusive one (only favouring x to degree n is fitting).
Moreover it is important, in addressing these questions, to
separate the (un)fittingness of a response from its value. Being
enthusiastic about a minor achievement might be unfitting, but
still good, or at any rate better than being indifferent towards
it. Likewise, favouring what is neither good nor bad might be
unfitting, but again still good, or at least not bad. Questions
about the value of fitting and unfitting attitudes will be decided
by ones substantive theory of value. There does not seem to
be a conceptual requirement that good responses, i.e.

VALUE AND NORMATIVITY

221

responses it is fitting to favour, need to be themselves fitting

222

VALUE THEORY

responses.109

7.3

Responses to value:
Maximizing

Lets go back to the first dimension: what kind of (dis)favouring

is appropriate to what? On the one hand, we can be as liberal
and pluralistic as we like. There will be appropriate responses
to some things and not others, in virtue of their different
properties: appreciating an artwork, praising a persons moral
qualities, preserving a forest, and so on. And some responses
might be common to many things: e.g. admiration, care,

VALUE AND NORMATIVITY

223

respect seem to apply to people as well as to valuable objects

224

VALUE THEORY

and states of affairs.110

However, for some philosophers all such responses are
appropriate only to the extent that they express some single
master fitting attitude to final value. In the broadly
consequentialist tradition, final value is, first and foremost,
something to be maximized: we should make it the case that

VALUE AND NORMATIVITY

225

the world contains as much value and as little disvalue as

226

VALUE THEORY

possible.111
How we do this is largely an empirical question: admiring,
respecting, preserving value are usually all fine ways to
maximize it, nor do we need to be constantly and actively
engaged in producing value, like workers in some sort of value
factory. But sometimes maximizing makes it necessary to
sacrifice one instance of a value for a bigger amount of the
same value, or for another weightier value. We have seen this
in Chapter 4 while discussing agent-relative value: it might be
necessary to kill one in order to save two from being killed. It
might be necessary to wage war in order to protect peace in
the long run, or to prevent some major injustice. It might be
necessary to destroy Napoleons hat in order to raise historical
awareness.
Notice: the philosophical issue here is not (just) whether all
things considered this is what one should do. Rather, it is about
the best way to construe such value conflicts. If maximization
is the only appropriate response to value, then killing, waging
war, or destroying Napoleons hat are prima facie wrong or
unfitting only because they are actions which prima facie fail to
maximize the relevant value (human life, peace, history), and
not because they express some other unfitting attitude
towards those values (say, disregard or disrespect). The choice
then is not between respecting human life (peace, history) and
maximizing human life (peace, history), but rather between
two potentially maximizing strategies: maximizing a certain
value by respecting it or maximizing it by not respecting it. And
once the conflict is seen from this angle, it becomes obvious
that one should simply choose the option which does (or is
expected to) maximize the relevant value, and in this sort of
case, respecting the value simply is not a way to maximize it.
But is this the right way to view such conflicts? The answer
depends on whether there is any reason to assume
maximization as the fundamental way of favouring value.
Philip Pettit suggests that universalizability provides such an
argument (1997: 142-3). Suppose I claim that respecting,
though not necessarily maximizing, peace is the fitting attitude
for me (or more realistically for a country) in a given
circumstance: it is fitting to abstain from waging or engaging in
war. Now, fittingness claims, like other normative claims, are

VALUE AND NORMATIVITY

227

universalizable: in making this claim, I commit myself to the

claim that it is fitting for anyone else in the same relevant
circumstances as mine to respect peace. If this is so, Pettit
argues, I am committed to subscribing to one single kind of
consideration that would call for all agents (similarly situated)
to respect peace. So, universalizing an attitude of respect
towards peace is tantamount to being committed to or desiring
a world where everyone respects peace. But a world where
everyone respects peace is a world where peace is maximized.
This is because Pettit defines respecting or rather honouring
or instantiating a value as follows: to instantiate a value is to
behave in the way that would promote [maximize] the value in
a world, roughly, where others were equally compliant (ibid.:
127). So respect really commits one to maximization, rather
than being a distinct and possibly incompatible attitude. Those
who think they can respect value but eschew a commitment to
maximization are simply confused.
There seem to be two problems with Pettits argument. One
is this. Even if it is true that a world where peace (or some
other value) is universally respected is a world where peace is
maximized, it doesnt follow that I, as a respecter of peace,
am thereby committed to maximizing it, period. All I am
committed to, it seems, is desiring a world where peace is
maximized by being universally respected. In turn, this
commitment seems compatible with a refusal to accept
maximization by other means, e.g. engaging in war to protect
long-term peace, which may involve attitudes and behaviour
contrary to respect for peace.
A second problem with Pettits argument is that it simply
assumes that the values to which respect is an appropriate
response are always, also, values which it is appropriate to
maximize, whether or not maximization dominates respect. But
arguably not all such values are maximizables, such that it
always makes sense to desire, and in turn bring about, more of

228

VALUE THEORY

them rather than less.112 Scanlon hints at this point regarding

friendship: it seems overblown to say that what is important
about friendship is that it increases the value of the state of
the universe in which it occurs (Scanlon 1998: 88). Anderson
sums this up in general terms: Since there are no generally
valid practical maximizing principles for intrinsic value, we say
that intrinsic value cannot be increased by increasing the

VALUE AND NORMATIVITY

229

number of its bearers (Anderson 1997: 97). 113 Honouring

friendship for its own sake has nothing to do with preferring a
world where there are 10 friendships to one where there are 9.
Respecting humanity has nothing to do with wishing that more
human beings are created. Note: the point is independent from
the question, addressed in Chapter 2, of whether individual
objects or proposition-like entities are the primary bearers of
final value. Even if we agree that only facts carry final value
(e.g. that A and B are friends; that she is a human being, etc.),
it does not follow that preferring or producing more rather than
less of them is always appropriate.
The advocate of maximization might be tempted to reply
that if increases in the number of value bearers do not increase
value in general (as Anderson says), or the value of the state of
the universe (as Scanlon says), then maximizing a certain
value simply does not require increasing the number of value
bearers of that kind, but rather it requires holding the whole
panoply of attitudes Scanlon and Anderson recommend.
This answer, however, seems to trivialize the notion of
maximization as a distinctive kind of response to value. If
maximizing a given value V amounts to making sure that
things go as well as possible with respect to V, or that the
world is as good as can be as far as V is concerned, then we
are back to a generic sort of concern or favouring towards V. To
be sure, the notion wont be completely empty of significance:
after all, it is a concern that things go as well as possible, as

230

VALUE THEORY

opposed to well enough.114 But what has disappeared is the

distinctive, and therefore more controversial claim that the
more final value (and the less final disvalue), the better, and

VALUE AND NORMATIVITY

231

with it the corresponding idea that our attitudes should be

232

VALUE THEORY

properly sensitive to this principle.115

7.4

Two concepts of intrinsic value?

Throughout the book, we have explored a number of

contrasting views:
1 final value is always intrinsic (and unconditional) value

vs. final value can be extrinsic (and conditional);

2 final value bearers are always states of affairs (or

proposition-like entities) vs. final value bearers can be

individual objects or persons;
value is always impersonal vs. value can be personal;
4 if something has final value, it has it (and to the same

degree) no matter where it occurs vs. something can

alter its final value in different contexts;
5 final value is fundamentally comparable vs. there can be
incomparability;
and in this chapter:
6 there is one fundamental kind of fitting response to

final value (maximization) vs. there can be

different kinds of fitting responses.
Ben Bradley (2006) has argued that these disputes are to a
large extent merely apparent, since there is no common
concept of final or intrinsic value shared by the disputing
parties prior to taking a stand on these issues. Rather, there
are two distinct concepts of value, which Bradley calls Moorean
and Kantian: Moorean value is by and large defined by the lefthand claims in 1-6, while

VALUE AND NORMATIVITY

233

Kantian value is defined by the right-hand ones. 116 If the theses

and antitheses in 1-6 define different concepts of value, then
the disputing parties are really talking past each other. By way
of an analogy, the dialectical situation here might be similar to
a disagreement on Gods properties. Theologian A argues that
God is all-benevolent, while theologian B denies that. They
seem to be disagreeing over a common subject matter. But it
turns out that theologian B also denies that God has a personal
nature - for B, God is rather a force of nature, energy, or
something of that sort. Now it seems that A and B are really
talking past each other: if they dont agree on whether God has
personal nature, they must be having two different concepts of
God to begin with, and the dispute over God being allbenevolent dissolves. Likewise, if the Moorean and the Kantian
dont agree on something so basic as, say, the nature of final
value bearers, then any further disagreement about value
must be treated as merely apparent.
Now, among the claims defining Moorean value Bradley
includes a normative claim about practical reasons (or fitting
attitudes and actions):
PRO: When something is intrinsically good, someone has a
good
reason to try to promote it, or preserve it, or make it true, or
bring it into existence. (Bradley 2006: 120)
If Bradley were right, then the foregoing disagreement
between the maximizer (or promoter) and the non-maximizer
would be based on a mistaken assumption. There simply isnt
one common value concept or property on which we can
disagree regarding its normative implications for conduct.
Rather, there are two concepts of value, each defined, inter
alia, by what responses are appropriate to the value bearer:
there is Moorean, i.e. maximizable value, to which PRO applies,
and there is Kantian, i.e. non-maximizable (or at least, not
necessarily maximizable) value.
In this book I have proceeded contrary to Bradleys claim. In
Chapter 1 I presented the general framework under which
disputes such as those 1-6 should be embedded: to be
valuable is to be the object of fitting favourable attitudes. And
in Chapter 2 I defended the assumption that different

234

VALUE THEORY

philosophers such as Aristotle, Kant, Mill, and Moore are

discussing about a common subject matter, by pointing to the
fact that they all accept, in one form or another, a notion of
what is valuable for its own sake, that is, final value, as
opposed to what is valuable for the sake of something else
that is valuable. On the basis of this assumption, disputes
involving 1-6 have been treated as genuine disagreements
regarding the articulation and specification of further features
of final value. For instance, dispute no. 6 clearly flows from the
general fitting attitude characterization: which favourable
attitudes, if any, are fundamentally fitting towards final value?
Bradley doubts that FA provides sufficient common ground
among the disputing parties. He writes:
What would be problematic for my thesis would be if there
were a unifying value concept V such that the disputes
between Mooreans and Kantians discussed in this paper
could plausibly be seen as disputes over the nature of V. But
FA does not provide, and is not intended to provide, any
such unifying concept. (ibid.: 127)
I have three replies to give here. First, it is not clear that FA
does not provide a unifying value concept. Certainly FA is
intended to articulate a central, non-negotiable fact about
value to which both Mooreans and Kantians are responsive. At
each juncture of the debate over 1-6, FA can reasonably be
expected to stand as a common constraint which determines
the benefits and the costs of a given thesis. In particular, we

VALUE AND NORMATIVITY

235

have seen how FA determines some costs for the Moorean

236

VALUE THEORY

theses 1-4 (Chapters 2, 4, 5).117

Second, it is true that FA is an abstract schema, and as such
leaves open many fundamental questions about the nature of
final value. And Mooreans and Kantians answer these questions
- fill in the schema - in their own distinctive ways, as claims 1-6
show. In doing so, each develops a view about the logic and
structure of final value. It is also true and interesting that
claims 1-6 form rather unified clusters: if you accept the view
that final value is always intrinsic, then you will look for the
sort of value bearers which deliver that result, and in turn will
accept only fitting responses which could logically be directed
at these value bearers; or instead you could start with a view
about fitting responses to value, and work your way up from
there through the metaphysics of value. However, its essential
that each of the claims in 1-6 needs to or at least can be
argued for, and this can only be done (and is typically done)
by, inter alia, arguing against the seemingly opposing view.
Bradleys point would imply that such argumentative attempts
are misguided. But this would take the ground away from both
Mooreans and Kantians. How else can Mooreans build their
picture of value, if not by, at each step, showing the difficulties
of the contradictory thesis?
Third, note a certain asymmetry between Mooreans and
Kantians. The latter typically (though not always) hold views
which recognize part of the truth as lying in the corresponding
Moorean thesis. For example, Kantians do not need to deny
that some value bearers are states of affairs, or that the value
of some things remains constant across contexts, or that
maximization is an appropriate response to certain goods.
What they deny is that the Moorean claims hold necessarily. By
Bradleys lights then, these Kantians are guilty of
misunderstanding their own views. They would be mistaken in
thinking that what they are doing is proposing a more generous
metaphysical and normative picture of the same thing (i.e.
final value), about which the Moorean gives a more rigid or
unified account. While it can be argued that a Kantian view
may suffer from such a pluralistic stance (how can the same
thing, i.e. value, come in so many different shapes and sizes?),
it is less plausible to think that Kantians are wrong to
understand themselves as pluralists in value theory.

VALUE AND NORMATIVITY

237

7.5 Summary
In this chapter I have tried to clarify in general terms the nature
of fitting attitudes. I have defended to some extent a pluralistic
view about the ways in which value can be favoured. Finally, I
have claimed that there is a genuine debate here, as elsewhere
in value theory, while agreeing with Bradley that answers to
certain questions (e.g. about value bearers) may constrain
answers to other questions (e.g. about fitting attitudes). The FA
approach, even when not taken as a definition of value,
provides sufficient common ground among discussants.

CHAPTER EIGHT

Value and the Wrong

Kind of Reasons
8.1 Introduction
In Chapter 1, I introduced the idea the value is normative as a
general and essential property of value. I used the fitting
attitude account as a historically and theoretically important
example of how the normativity of value can be articulated:
FA: x is good = it is fitting to respond favourably to (favour) x.
I have relied on FA as the working assumption throughout the
book. In this chapter, I reconsider the merits of FA as a
definition of value, as an answer to the question what is
value?. The main problem for FA is that not any fitting
favouring attitude towards x, or not any reason to favour x,
implies that x is good.
I
will explore in some detail different strategies FA
theorists have devised to solve this problem. The

238

VALUE THEORY

conclusion will be that further work is needed, and

possibly we should consider the merits of alternative
accounts.

8.2
The fitting attitude
account and its rivals

FA is not the only account that aims to make sense of how

value is connected to normative concepts such as reasons,
ought, or fittingness. Remember FAs two central tenets:
1 Reduction Claim: value properties such as goodness are

reduced to normative relations of fittingness of, or of

there being a reason for, attitudes.
2 Normative Redundancy Claim: value properties such as
goodness do not themselves provide reasons for
attitudes (make attitudes fitting) over and above the
good-making features which already provide reasons
for attitudes (make attitudes fitting).
Any genuine alternative to FA must deny either or both central
claims. So here is a list of alternatives:
Early Moorean Account: denies (1), because it turns it
around. Normative relations are reduced to evaluative
properties. In Principia Ethica, Moore holds there that while
intrinsic goodness is indefinable, deontic concepts such as
rightness or duty can be defined in terms of value:
To ask what kind of actions we ought to perform, or what
kind of conduct is right, is to ask what kind of effects such
action and conduct will produce ... Every judgement in
practical Ethics may be reduced to the form: This is a cause
of a good thing ... [R]ight does and can mean nothing but
cause of a good result. (1993:196)
It is rarely noticed, but very relevant in this connection, that
Moore also analysed non-moral concepts like fittingness in
terms of intrinsic value:

VALUE AND NORMATIVITY

239

[B]y saying that a certain relation between two things is

fitting or appropriate, we can only mean that the existence

240

VALUE THEORY

of that relation is absolutely good in itself. (ibid.: 152) 118

VALUE AND NORMATIVITY

241

How to precisely and satisfactorily spell out an Early Moorean

242

VALUE THEORY

Account may require some effort, 119 but the general idea
should be clear enough: evaluative concepts are basic, and
normative concepts are to be explained in evaluative terms. On
this approach, value is normative because normativity is
already evaluative: you cannot understand what a reason or an
ought is, without an understanding of what is good, bad, and
the like. It should be noticed that (2), the Normative
Redundancy Claim, follows from the Early Moorean Account
just as it does from FA: if being good serves to define the
notion of a reason, then it cannot also be a provider of reasons.
The existence of a reason to favour x would mean, inter alia,
that x is good, intrinsically or otherwise. For x being good to
provide a reason to favour x would be for x being good to make
it the case that x is good. But that is an empty claim.
Therefore, also on this account it is not goodness, but goodmaking properties which provide reasons.
Necessary Equivalence Account (or Later Moorean Account):
Moore came to reject his own definitions of deontic terms in
evaluative terms, settling for the view that there is a necessary
equivalence between evaluative and normative facts, but not

VALUE AND NORMATIVITY

243

any relation of identity or reduction. 120 Whenever a certain

normative claim is true, so is a relevant evaluative claim, and
vice versa. In the same way, we could say that a triangle
having three sides and a triangle having three angles are
necessarily equivalent but not identical properties: they
necessarily occur together (no possible triangle has one but
not the other property), but they still are different properties of
triangles. So (1) is rejected because normative and evaluative

244

VALUE THEORY

truths are tightly related, but not in any reductive way. What

VALUE AND NORMATIVITY

245

about (2)? Here there are two possibilities:121

Necessary Equivalence with Redundancy: this view accepts
(2). Fittingness and value are necessarily related, but not
because value makes attitudes fitting, or provides reasons for
attitudes. Of course, such a view has its challenges to face.
First, it needs to establish
(2) on grounds other than the identity or reduction of one
sort of property to the other. Second, it needs to
provide an alternative explanation of why the
necessary equivalence holds: it cannot be a mere
coincidence that value is always normative. Proponents
of this view may say that the equivalence is explained
by goodness and fittingness sharing their resultance
bases (see Chapter 5): whatever makes an attitude
fitting also makes a state of affairs good, e.g. a certain
experience being pleasant both makes it fitting to
favour it and makes it good. And if this is true, then
goodness cannot be a reason-provider or fitting-maker:

246

VALUE THEORY

if it were, then it would also be a good-maker, but that

VALUE AND NORMATIVITY

247

is again an empty claim.122

Necessary Equivalence without Redundancy: according to
this view it is fitting to favour x whenever x is good, and vice
versa, because xs goodness makes it fitting to favour x, or
provides reason to favour it. The necessary equivalence is
explained by goodness and other like evaluative properties
being themselves reason-providers. Value is normative in the

248

VALUE THEORY

sense that it is normatively significant:123 it provides reasons.

So this view denies (2). In Scanlons terms, goodness or being
valuable doesnt pass the buck: When something has the right
natural properties it has the further property of being valuable,
and that property gives us reason to behave or react in certain
ways with regard to it (Scanlon 1998: 97).
To some extent, common sense supports this view. We often
mean to justify our choices, preferences, or attitudes by simply
appealing to the goodness or betterness of an action, or the
badness of the alternatives, and so on. And some philosophers
also have resisted the pressure to pass the buck. For instance
Ross argues against FA on the following basis. If good were
defined as fitting to be admired, then good would be defined as
fitting to be thought of as good. This would already involve
moving in a circle. But, further, the circle would be a vicious
one, because if we ask on what ground a thing is worthy of
being thought to be good, only one answer is possible, namely
that it is good (Ross 1939: 278-9). So goodness has a role to

VALUE AND NORMATIVITY

249

play as an ineliminable fitting-maker. 124 Also, when it comes to

other normative claims, e.g. moral ones, Ross seems to argue
that the prima facie duty to promote intrinsic value is grounded

250

VALUE THEORY

on the fact that something has intrinsic value (Ross 2002:

VALUE AND NORMATIVITY

251

24).125
The structure of this view then is:
x being pleasant (e.g.) ^ x being good ^ it is fitting to
favour x.
According to this view, there is a certain division of explanatory
or grounding labour: certain properties make things good, and
goodness in turn grounds normative truths.
Once the structure of the view is clarified, it cannot be
argued against it that goodness does not add any normative
weight to an already given list of reasons to favour x: until we

252

VALUE THEORY

mention goodness, simply there are no items on that list. 126

Further, in a reason-giving exchange because it is good
seems to be an acceptable answer to the question why should
I favour x? Of course, it wont be a very informative answer,
but this is because it needs to be supplemented by a
specification of the good-making features, rather than of
further reasons to favour x. So it seems that any good
objection to this view must be directed not so much at its
structure, however baroque it may seem, but rather at its
implications: e.g. that there is only one kind of ultimate reason
in favour of actions and attitudes, namely goodness and its
cognates. A pluralistic account of reasons might perhaps be
preferable.
It is indeed this kind of objection that Scanlon mentions as
one of his arguments against such a view:
[M]any different things can be said to be good or to be
valuable, and the grounds for these judgements vary widely.
There does not seem to be a single, reason-giving property
that is common to all these cases. (1998: 97-8)
Note that proponents of Necessary Equivalence without
Redundancy can accept the first part of the quotation: many
different things can make things ultimately good. The objection
lies rather in the second part: it is difficult to believe that,
given pluralism about the good (about ultimate good-makers),
there is only one and always the same reason to favour these
different things. What is odd, according to one way of reading
Scanlon, is the idea that once we have learnt that innocent
pleasures are good, and on this basis it is fitting to favour
them, we have thereby also learnt why it is fitting to favour
other disparate good things like virtuous characters or
friendship - as if understanding virtue or friendship were not
necessary to appreciating the reasons to favour them, and how
we should respond to them, because understanding goodness
suffices for that.
In sum, we can have different models of the relation
between fittingness (or reasons) and value. Supposing that

VALUE AND NORMATIVITY

253

(a) x is good if and only if (b) it is fitting to favour x,127

there are competing explanations of why it is so. Schematically:
1
2

Fitting Attitude Account: (b) defines (a);

Early Moorean Account: (a) defines (b);
3 Necessary Equivalence with Redundancy: (a) and (b) both
result from x having certain other properties;
4

Necessary Equivalence without Redundancy: (b) results

from (a).

8.3 The wrong kind of reasons

problem
These alternatives to FA are well worth mentioning, because
there is one major stumbling block to FAs success, namely the
wrong kind of reasons problem (WKR), which I briefly
introduced in Chapter 1. There I gave the example of the evil
demon who will torture us (or the whole humankind) if we fail
to admire him. It seems that, based on this threat, we have
abundant reasons to admire him. If the buck-passing account is
true, then it follows that the demon is good. But clearly the
demon is evil rather than good, in part precisely because of his
threat. His threat is a wrong kind of reason to admire him,
because it doesnt make him good or admirable. Such a reason
to admire him does nothing to show the demons positive
value.
In Chapter 1, I suggested that buck-passers might approach
a solution to the problem by adverting to fittingness: whatever
reasons there might be to admire him, it seems that admiration
just doesnt fit a being like the demon. So the buck-passing
account could be formulated as follows:
Buck-passing 3: x is good = x has the property of having
other properties that provide reasons of fittingness to favour
x, or would provide such reasons to suitably situated agents.
Or alternatively, and more simply, one could adopt FA. But,
arguably, this is more like the promise of a solution than a
genuine solution to WKR. Until we obtain an account of
fittingness, or of reasons of fittingness, we are not saying much
more than this: x is good when there are reasons of the right

254

VALUE THEORY

kind to favour x. The crucial and thus far unanswered question

is: what makes reasons of fittingness reasons of the right kind?
On the other hand, alternatives to FA do not seem to face
WKR with the same urgency. Of course, WKR presents itself as
a counter-example to the claim of necessary equivalence: for
some xs, there are reasons to favour x, but it is not the case
that x is good. However, on the Early Moorean Account, value
is conceptually primary: we start out with value judgements,
such as that the demon and his threats are evil, and in turn
there are reasons to despise the demon and against admiring
him. If there are also reasons to admire him, then these must
stem from some other value; plausibly here, the value of
minimizing suffering. On the thesis of Necessary Equivalence
without Redundancy, likewise, value judgements are prior. If
there are reasons to admire the demon, it doesnt yet follow
that the demon is valuable, because those reasons might be
grounded in the goodness of something else (again,
minimizing suffering). Also the thesis of Necessary Equivalence
with Redundancy seems to avoid WKR: true, a ground for
reasons to favour x is, in principle, also a ground for xs value
(a value-maker of x), but we can consult our intuitions about
value to determine that the demon is not valuable, and thus
whatever reasons to favour him there might be are not also
considerations that make him valuable. In other words, on
these accounts we are free to appeal to the whole range of
evaluative truths in order to explain away apparent counterexamples to necessary equivalence, because we can have an
independent purchase on the notion of value. Not so with FA:
an appeal to values here would be circular, since value is what
FA is trying to understand.

8.4
The structure of the
problem and an initial
response
Given the logical possibility of scenarios like the evil demon,
WKR is generated by a set of assumptions about such a
scenario:

VALUE AND NORMATIVITY

255

Value: some thing x has a certain value (good, bad,

better than y, etc.).
Attitude: there are reasons of some kind to hold an
attitude that is unfitting to xs value.
Non-circularity: these reasons cannot be defined as
being unrelated to xs value.
Several philosophers have pointed out one obvious feature of
the evil demon case: the reason to admire the demon is not
provided by a feature that belongs to the demon. It is
provided, rather, by some state of affairs which will come
about if we admire the demon: the minimization of suffering.
The fact that this state of affairs is in this case, so to speak,
under the demons control, does not mean that it provides
reasons only in connection with the demons threat
we already have reasons, independent of anyones
threat, to seek minimization of suffering. In one word,
the reason to admire the demon is not object-given:
not provided by the demon or any of his properties.
Generally object-given reasons are contrasted with
state- or attitude-given reasons: reasons provided by
features of the very attitude for which there is a
reason. In this case, one can say that the reason to
admire the demon is provided by the instrumental
properties of admiration: the fact that admiration will
spare us from suffering. But, as I just implied, a better
grounding for this reason seems to be given by the

256

VALUE THEORY

absence of suffering, which in this unusual case we

VALUE AND NORMATIVITY

257

would achieve by admiring the demon.128

The next claim then is that only object-given reasons are
reasons of fittingness. This seems a plausible enough claim:
fittingness between attitudes and objects requires a match
between the attitude and some relevant properties of its
object, rather than properties of something else. If the reason
to admire the demon is not to be searched in the demon, then
admiration is not a fitting response to the demon. Therefore it
is not the case that the demon is good or admirable.
This initial reply, however, does not work for the whole
range of scenarios which present WKR problems. I have three
sorts of cases in mind. First, consider a very cruel joke being

258

VALUE THEORY

made about somebody.129 On this basis, lets suppose that

there is overall reason not to be amused by the joke. On the
basis of FA, it would seem to follow that the joke is a bad one.
But, barring a strong form of moralism about the comic, this
need not be so: the joke can still be very funny, albeit cruel,
and so a good joke, though one that we ought not to be
amused by. The fact that it is cruel is a wrong kind of reason to
make it bad as a joke. But note that the fact that it is cruel is
an object-given rather than attitude-given reason against being
amused by it. It is not a matter of what would happen if
amusement at the joke were shown in the face of the targeted
person. Cruelty is a property possessed by the joke, i.e. by the
object of amusement. In other words, some object-given
reasons for or against amusement need not correspond to the
value of the joke.
Second, there are possible (indeed, quite realistic) scenarios
where object-given reasons to prefer an object to another do
not necessarily coincide with its being better than another. A
father might have reason to be happier about his own
daughters recovery from illness, than about someone elses

VALUE AND NORMATIVITY

259

recovery from an even worse illness. 130 The fathers reason for
this stronger attitude is object-given rather than attitude-given,
because it is provided by his daughters recovery and by her
being in a closer relationship to him, rather than by some
property that his having a stronger attitude would possess. But
such a reason is of the wrong kind to make his daughters
recovery better than the strangers. In this and the previous
example, it seems that morality or personal relationships can
offer lots of object-given reasons for or against attitudes which
do not necessarily bear on the value of the object. There is an
element of fittingness in such responses, but not the sort of
fittingness which goes along with the value of their objects. It
seems that we need finer distinctions than the objectgiven/attitude-given one to capture the relevant reasons of
fittingness.
The third case is one where the distinction simply doesnt
apply. Incentives like the demons threat need not give reasons
for attitudes at all, but simply reasons for actions, like cheating
in a game of chess: if you cheat, the demon will spare us from

260

VALUE THEORY

torture.131 Intuitively this is a reason of the wrong kind to make

cheating an admissible move. Cheating, like admiring the
demon, remains unfitting, despite there being an overall
reason in favour of it. But what rules out such incentives as
reasons of fittingness cannot be anything about a mental state
as opposed to its object, since we are not being asked to have
any mental state in the first place. Again, we need a different
criterion if we want to explain the general difference between
reasons of fittingness and other reasons.

8.5 Reasons for what?

Some philosophers address WKR by denying or qualifying one
of the assumptions of the evil demon scenario, namely that in
these cases there really are reasons to admire the demon, or
that there are reasons to primarily admire him. These solutions
focus on the nature of the relevant response rather than on the
reason-giving features. Lets consider them in turn.
Practical Reasons. It may be argued that in the evil demon
case we do not really have a reason to admire the demon, but
rather a reason to want to admire him, or to try to admire, or
to bring it about that one admires him. That is, the threat of
being tortured if we dont admire the demon provides a reason
to have a second- order attitude such as wanting to admire
him, but not a reason to admire him. After all, it may be said,
this is perfectly coherent with FA: having the attitude of
admiration is an (instrumentally) good thing, for it saves us
from torture, so we have reason to have a pro-attitude towards
having this attitude. On the other hand, there being no reason
to admire the demon for his own sake (to have a first-order
pro-attitude towards him), the conclusion is avoided that the
demon is intrinsically good.
The claim that in the evil demon case we have reason to
want to admire him is correct. The proponent of this solution
however has the burden of explaining why we dont also have
a reason to admire the demon. Both John Skorupski (2007) and
Ingmar Persson (2007) introduce a parallel with reasons to
believe. When it is useful to believe that p (e.g. that God
exists) - as it is useful to admire the demon - does this give

VALUE AND NORMATIVITY

261

both a reason to want to and make yourself believe that p and

a reason to believe that p? Both authors answer in the
negative, although for different reasons.
Skorupski offers two considerations. First, he illustrates a
general distinction between what he calls epistemic reasons
(reasons to believe), evaluative reasons (reasons to feel or
have certain emotions), and practical reasons (reasons to do
something). Then he points out that evil demon cases offer in
the first place reasons to do something, namely bring it about
that one admires the demon. Likewise, if the evil demon
threatens to torture us if we dont believe that p, we have
reason to bring it about that we believe that p. But these are
all practical reasons, not reasons to admire or to believe.
Second, if the usefulness of admiring x or believing that p were
also a reason to admire x or believe that p, then we would
need to defend a pragmatist theory of reasons to admire (or to
believe), which Skorupski rules out as unusual (2007: 12).
Skorupskis two points are connected. There are different
kinds of reasons, and the considerations that bear on one kind
of reason (e.g. the usefulness of an attitude, which bears on
practical reasons) cannot bear on another kind of reason
(epistemic or evaluative). The usefulness of admiring x cannot
be a reason to admire x but only a reason to bring it about that
one admires x.
However, apart from ruling out a pragmatist theory of
reasons to admire, Skorupski does not tell us exactly why we
should not admire the demon although we should bring it
about that we admire him. This is quite a paradoxical situation:
why should we bring it about that we do x if x is not what we
should do? After all, we will be safe from the demons threat if
we admire him, not just if we make ourselves admire him. Note
that there is an important difference here between actions
such as bringing ones admiration about or making oneself
admire and, for instance, trying or intending to admire. It is
true that, in principle, one could have a certain prize attached
to merely the trying or the intention, without this implying that
one should do what one should try or intend to do, as in the
famous Toxin Puzzle, where one gets rewarded for intending to

262

VALUE THEORY

drink a toxin but has no reason to drink it. 132 If we could be

saved from the demon just by trying to admire him, then it
would not follow that we have any reason to admire him. But,
as Skorupski recognizes, we are safe only if we bring it about
that we admire him. And bringing about is, arguably, a
success-term: we bring x about only if x occurs. In other words,
the case is set up so that we have reason to succeed; and
since succeeding here means admiring the demon for his own
sake, it seems to follow that, if we have reason to bring it
about that we so admire him, then we also have reason to so
admire him.
Secondary Reasons. Persson doesnt deny that we also have
reasons to admire the demon, but these are secondary
reasons. Consider Pascals wager. The probability of being
amply rewarded after death if we believe that God exists, plus
the negligible costs of believing in God, overall constitute a
reason to want to believe that God exists. Is this also a reason
to believe that God exists? It is a pragmatic reason to so
believe. But what is characteristic of these reasons is that we
cannot directly respond to them by forming a belief, because
we cannot see these reasons as justifying the relevant belief by
supporting its truth. The fact that, if we believe, we might be
rewarded after death does not make it true or more probable
that God exists. So pragmatic reasons are only secondary
reasons to believe (Persson 2007: 4-5). They are, primarily,
reasons to want to believe, as the direct response to the fact
that believing that God exists will reward us is to want to have
that belief. Analogously, the prospect of torture gives us only
secondary reasons to admire the demon. This is because we
cannot directly respond to the belief that the demon will
torture us if we dont admire him by admiring him, for we
cannot see that belief as justifying admiration towards the
demon (ibid.: 7). Our direct response will rather be a desire to
admire him (a second-order attitude). Consequently, reasons of
fittingness should be identified with primary reasons only, that
is, those considerations the direct response to which is a

VALUE AND NORMATIVITY

263

certain pro-attitude, and that primarily justify such a pro-

264

VALUE THEORY

attitude.133
Perssons central claim is that we cannot see admiration for
the demon as justified by the belief that he will punish us if we
fail to admire him. Persson thinks that this is because in all
probability, there is no practical reasoning attributable to you
which has this [admiration] as its conclusion and the belief as
its premise (ibid.: 7-8). Such a reasoning would have the form:
(i) I believe that the demon will punish me if I dont

admire him;
(ii) I desire not to be punished;
(iii) Therefore I admire the demon.

Persson thinks that only (iv) is the proper and direct conclusion:
(iv) Therefore I desire to admire the demon.

This may be true. However, it remains to be explained why our

practical reasoning should be as Persson says it is. That is to
say, why is it that admiration is at best only an indirect and
secondary response to (i)? In the epistemic case, believing that
God exists is only an indirect response to the fact that we will
be rewarded after death if we so believe, because we cannot
see the fact of reward as supporting the truth of the belief that
God exists. This is in line with the function of belief, which is to
fit the facts. In the practical and emotional case, our proattitude, say, admiration for the demon, is an indirect response
to the fact that if we dont admire him he will punish us, but
why? What we need here is something analogous to the
specification in the epistemic case: p epistemically justifies our
belief that q if p supports qs truth or makes it more probable.
With the justification of pro-attitudes, the criterion may be: p
practically justifies a pro-attitude for q if p makes q valuable.
Since the demons determination to torture us if we dont
admire him clearly does not make the demon valuable, neither
can it justify admiration for the demon.
But this criterion is not available to somebody upholding FA.
Remember that here we are trying to explicate the notion of a
primary reason for a pro-attitude, which in turn is needed to
analyse the notion of x being valuable. If we adopt the
following as a criterion: p practically justifies (thus is a primary

VALUE AND NORMATIVITY

265

reason for) a pro-attitude for q if p makes q valuable, then the

analysis would be patently circular. Those who propose to solve
the wrong kind of reasons problem by distinguishing between
primary and secondary reasons for pro-attitudes, as
Rabinowicz and R0nnow-Rasmussen remark:
need to clarify, without taking the notion of value for
granted, what makes something a reason for wanting (or
trying) to have a certain attitude towards an object rather
than a direct [i.e. primary] reason for having the attitude in
question. (2004: 413-4)
Persson tries to give such a clarification in terms of what
attitudes are justified by what, but the analogy with epistemic
reasons leaves him without a similarly deep explanation of
such an idea, apart from a circular one. Epistemic justification

266

VALUE THEORY

can appeal to truth, but practical justification cannot appeal to

VALUE AND NORMATIVITY

267

value.134
There is also a more general problem for these approaches:
they dont seem to work for the other WKR scenarios
mentioned above, namely the moral reasons against being
amused by a cruel joke, the partial reasons to prefer ones
daughters recovery, and the rewarded cheating scenario. In all
these cases of wrong kinds of reasons, it is simply implausible
that the relevant considerations do not provide reasons (or
primary reasons) for the relevant attitudes (rejecting the cruel
joke, preferring ones daughters recovery) and actions
(cheating under the demons threat).
A Brentanian Solution? Danielsson and Olson (2007) argue
for a different distinction. There are reasons to hold an attitude,
which they call holding-reasons, and reasons for the
correctness of that attitude, which they call content-reasons.
The crucial notion here is correctness. There is a difference
between a belief being correct, i.e. true, and the fact that we
have reason to have the belief. Pascals wager provides such
an example: the cost-benefit analysis of belief in God gives us
a reason to believe that God exists (a holding- reason), but
doesnt make it correct to believe that God exists (its not a
content-reason). The same point applies to pro-attitudes: pace
Skorupski and Persson, the demons threat does give us a
perfectly good reason to admire the demon, but still doesnt
make it correct to admire the demon. In turn, FA should
understand xs value in terms of there being content-reasons
to favour x, i.e. considerations that make it correct to favour x.
Despite the attitude-oriented labels, this approach can also be
extended to actions: the demons threat gives us a holdingreason to cheat in the game, but not a content-reason - it
doesnt make the cheating moves correct.
However, this much we knew already from Chapter 1: we
knew that to solve WKR we had to focus on reasons of
fittingness, or of correctness, as in Brentanos terminology.
Danielsson and Olsons contribution points to a solution

268

VALUE THEORY

without providing one.135 Moreover, it doesnt seem to apply to

the other WKR scenarios. There are perfectly good contentreasons for rejecting the cruel joke (or for not being amused by
it): it is, from a moral point of view, a correct or fitting response
to it. There are content-reasons to prefer ones daughters
recovery to a strangers: it is a correct response, from a
parents point of view. But these content-reasons dont show

VALUE AND NORMATIVITY

269

that the joke is a bad or unfunny one, or that ones daughters

270

VALUE THEORY

recovery is better than a strangers.136

8.6
Characteristic
concerns and shared
reasons
The next two solutions I consider aim to say something more
about reasons of fittingness, without relying on an objectgiven/state- given distinction, and without ruling out wrong
kinds of reasons as merely practical reasons or secondary
reasons for attitudes. The first is by Justin DArms and Daniel
Jacobson. They propose to isolate reasons of the right kind for
an attitude (more precisely, an emotion) as reasons given by
the features which are the emotions characteristic focus:
our proposal is that to think an emotion a fitting response to
some object is to think there is (pro tanto) reason, of a
distinctive sort, for feeling the emotion toward it. Such
judgements differ from other sorts of appraisals of an
emotion, in virtue of the kind of reasons they concern.
Reasons of fit are those reasons that speak directly to what
one takes the emotion to be concerned with, as opposed to
reasons that speak to the advisability or propriety of having
that emotion. So reasons of fit for fear are roughly those that
speak to whether or not something is a threat. (2006: 107-8)
Analogously, reasons of fit for admiration are those that speak
to what admiration is concerned with, as opposed to reasons
that speak to the advisability or propriety of admiring. For
example, the prudential or moral reasons we have for
admiration in the evil demon case are of the wrong kind,
because the threat to torture people is not a characteristic
concern of admiration. Likewise, the cruelty and more generally
the immorality of a joke may give decisive reasons against
being amused by it, but still are not a characteristic concern of
amusement: the joke is not thereby shown to be unfunny.
In order to understand the relevant notion of fittingness of a
response, we need some story as to what is the distinctive
concern of that response. Fear is concerned with threats, so it
is fitting when it is a response to genuine threats; shame, as
DArms and Jacobson also suggest, is concerned with what we

VALUE AND NORMATIVITY

271

recognize as a social disability, so it is fitting to be ashamed of

ones social disabilities, and so on. The notion of a responses
characteristic concern seems to advance our understanding of
reasons of fittingness. In order for it to work as an analytical
proposal, we should be able to grasp a responses
characteristic concern in a sufficiently independent manner
from the value to be analysed: we should be able to
understand something to be a threat or a social disability quite
independently from its being fearsome or shameful.
Even if this desideratum is satisfied for certain responses,
however, there are two outstanding issues that require
attention. First, for other responses it is not clear whether we
can identify their distinctive concern without employing the
very value we are supposed to analyse, or a generic value
term. Take admiration. Its distinctive concern (the kind of
things it is typically directed to) may be said to be, roughly,
human strengths and achievements. But not just any
achievement will prompt and justify admiration: rather, those
achievements that manifest a sufficiently high degree of value,
because of their being excellent, uncommon, or challenging for
the person to realize. The problem is not so much that the
notion of an achievement is evaluative - FA need not be
committed to reducing thick as well as thin values - but rather
that there might no way to pick out significant from minor

272

VALUE THEORY

achievements other than by saying they are better or higher in

VALUE AND NORMATIVITY

273

value.137 There is a threat of vicious circularity.

Second, some responses are so generic that they do not
even seem to have a distinctive concern. At least prima facie,
attitudes such as desire or preference, unlike specific
emotional states, may be prompted by far too many diverse
circumstances for us to be able to pick out a set of features as
their distinctive concern - unless by employing generic value
concepts such as the good, the better, and the best. DArms
and Jacobson are aware of this difficulty, but they argue that
an FA analysis must primarily be concerned with specific value
concepts and properties (the fearsome, the shameful, the
funny, etc.) and thus with specific responses (2000b: 747).
However, this reply assumes that WKR cases framed in terms
of generic value concepts and generic responses must be
intelligible in more specific terms, or else are spurious. But
then consider again the father and daughter case. DArms and
Jacobson might say that the preference for a daughters
recovery is intelligible as a distinctive emotion, such as
parental love. However, the facts that justify the fathers
preferential response are within the distinctive concern of that
emotion, since they include his daughters wellbeing. If so, the
fathers response is of the fitting kind according to DArms and
Jacobsons criterion. But still, his daughters recovery is not
better than a strangers recovery from an equal or worse
illness. Moreover, as seen above, a solution to WKR should
throw light on a notion of fittingness that applies to actions as
well as attitudes, whereas DArms and Jacobsons view seems
limited to emotions.
Mark Schroeder has recently proposed a different criterion,
based on the idea that activities (and not just attitudes) are
often governed by standards of correctness: games have rules
specifying which moves are correct, etiquette specifies correct
ways of setting the table, and so on. This notion of correctness,
in turn, is understood in terms of right kind of reasons, which
for Schroeder are shared reasons in the following sense:
The right kind of reasons with respect to any activity, A, are
all and only those reasons which are shared by necessarily
every able person engaging in A, because they are engaged
in A, together with all reasons which are derivative from
such reasons. (2010: 39)

274

VALUE THEORY

If the evil demon threatens torturing unless you put out fish
knives for a meat-based meal, this is a wrong kind of reason
with respect to the activity of setting the table, in the sense
that it wouldnt be shared by every table-setter because they
are engaged in table- setting. Even if every possible tablesetter were threatened by the evil demon, and so there was a
necessarily shared reason to put out fish knives, such a reason
wouldnt stem from the fact that they are engaged in tablesetting. Therefore putting out fish knives is not a correct way to
set the table for a meat-based meal, no matter how strong
(wrong kind of) reasons there might be.
How does Schroeders view apply to the other WKR
scenarios? (i) The reason to admire the demon for his own sake
might be likewise shared by all admirers, but still not because
they are engaged in admiration. In this sense, admiring the
demon is not a correct way of admiring demons or anyone. So
the demon is not admirable. (ii) The reason not to be amused
at the cruel joke might well be actually shared by everyone,
but again not because we are engaged in the activity of
appreciating jokes, or something like that. Engagement in this
activity does not require lack of amusement towards cruel

VALUE AND NORMATIVITY

275

jokes. From this point of view, it is not a correct response, and

276

VALUE THEORY

therefore doesnt show the joke to be unfunny.138

(iii) The father and daughter case is however trickier,
because it is not clear what is the relevant activity.
Suppose it is showing love for Laura (this particular
child). Since not everybody who is engaged in this
activity would necessarily prefer Lauras recovery to
the strangers (suppose you are a friend of Lauras, but
the strangers son), we could say that it is not correct
to prefer it, and hence whatever reasons her father
has, they cannot show that Lauras recovery is better
than the strangers. This would solve WKR, but it
seems based on a very narrow reading of the relevant
activity. Mention of individuals in the specification of an
activity might be appropriate in the analysis of what is
good for someone, but we cannot assume that there is
only personal value at stake here (indeed, her fathers

VALUE AND NORMATIVITY

277

preference doesnt have to be good for Laura, who

278

VALUE THEORY

might not know about it).139

Suppose then the relevant activity is showing love for ones
child or being a decent parent. Now, this activity does seem
to generate shared reasons, for those engaged in it, to prefer
ones childs recovery over a strangers, in any situation like
the father and daughter case. And these are not entirely
general reasons (like moral reasons not to laugh at a cruel
joke), because they only apply to those who are parents, and
insofar as they are parents. But still, despite the preference
being in this way correct, the daughters recovery is not better.
Schroeder might pick preferring as such as the relevant
activity. Clearly, not anyone engaged in the bare activity of
preferring has a reason to prefer Lauras recovery presumably the stranger and his associates dont. The fathers
reason is not one necessarily shared, hence it is not of the
right kind. The problem with this reply is that it is not clear that
preferring constitutes an activity with its own standards of
correctness. True, transitivity may provide one such standard:
if we are engaged in preferring A to B, then we have reason to
prefer A to C, if we also prefer B to C. But is there anything
more substantial one can say about preference? Clearly there
is a notion of a correct preference which includes more than
honouring transitivity. But what this more consists in cannot
be gleaned just by reflecting on the nature of preference except of course the truth, unhelpful here, that in general we
have reason to prefer the better to the worse. Preference is too
thin an activity to deliver (significant) standards of
correctness, and in turn shared reasons.
Of course, Schroeder could reply that if there are no (significant) standards of correctness for preference, and a fortiori no
reasons of the right or the wrong kind to prefer, then this case
simply doesnt pose a WKR problem, if formulated in terms of
preference. But this move, like DArms and Jacobsons rejection
of generic responses, would greatly reduce the appeal of FA as
a general account of the evaluative in terms of the normative.
All we would be left with are piecemeal analyses of those
values which pair up with specific emotions or with activities
governed by standards of correctness. Moreover, not only
would this approach be limited in scope, but it would also be
far less interesting: it only takes a little bit of semantic

VALUE AND NORMATIVITY

279

competence, and not a philosophical discovery, to know that

e.g. the admirable must be what it is correct to admire. The
philosophical work then would be merely one of trying out
different criteria to isolate the relevant notion of correctness:
reconstructing the path that led us where we already are. This
is why Schroeder claims we should be confident that a solution
to WKR can be found (ibid.: 26-8). But such confidence can be
had only based on either of two assumptions: (1) a reduction
of value in general is not our aim, (2) value in general wont
present additional difficulties once the piecemeal analysis is
carried out successfully. The first assumption is, arguably, too
modest; the second, as the discussion above is meant to show,
is unjustified - whatever solution may work for admirable, it
may not suit concepts like preferable and in turn good or
better.

8.7

Circular path: No-priority

If the WKR problem should prove resistant to general solutions,

one should indeed reconsider the merits of those alternatives
to FA, outlined above, which do not suffer as greatly from WKR,
because they dont reduce value to reasons or fittingness. But
there is one methodological move still available: reject or
modify the assumption of non-circularity. The idea is suggested
- though not as a reaction to WKR - by both McDowell and
Wiggins. For instance, Wiggins writes:
[T]here will often be no saying exactly what reaction a thing
with the associated property will provoke without direct or
indirect allusion to the property itself. Amusement for
instance is a reaction we have to characterize by reference
to its proper object, via something perceived as funny (or
incongruous or comical or whatever). There is no objectindependent
and
property-independent,
purely
phenomenological or purely introspective account of
amusement. (Wiggins 1987: 195)
McDowell puts it hypothetically: If there is no comprehending
the right sentiments independently of the concepts of the
relevant [evaluative] features, a no-priority view is surely

280

VALUE THEORY

indicated (1998b: 160). One way to elaborate on this is to say

that the relevant responses, and in turn the notion of them
being fitting, are not prior to and independent of the concept of
value, because they include as part of their content a relevant
value judgement. And vice versa: the notion of value is not
prior to the notion of fitting attitude, since it can only be
understood in these terms. Moreover, it is not the case that
truths about fittingness (or reasons) need to be in general
definable as truths about value, as in the Early Moorean
Account.
Nor is this just a Necessary Equivalence Account, since the
claim made is one of identity rather than necessary cooccurrence:
No-Priority FA: x is good = x has properties that make it
fitting,
inter alia, to judge that x is good.
If we think of value in this way, then what we need is simply an
account of the right kinds of reasons to judge that p, which will
deliver the result that the demons threat is not a right kind of
reason to judge that the demon is good: for instance, it doesnt
make the proposition the demon is good probable or provide
evidence for it. Above, I noted that a solution such as Perssons
relied on a problematic analogy between beliefs (or in general
truth-aimed states) and pro-attitudes. But on a no-priority view
one would only need to adopt an account of reasons to believe

VALUE AND NORMATIVITY

281

(or to judge), without worrying whether a similar story can be

282

VALUE THEORY

told about admiration and the like.140

VALUE AND NORMATIVITY

283

However, even leaving aside the thorny question of whether

284

VALUE THEORY

the kind of circularity of this view is acceptable or not, 141 a nopriority view seems to misconceive the normativity of value as
theoretical or epistemic normativity: value essentially requires
judgements of value, rather than pro-attitudes or actions. Of
course, for any given valuable object, there will be a fitting
emotion or attitude, of which the value judgement is a
component. But the way in which a no-priority view helps in a
solution to WKR makes the judgement the crucial element
common to all relevant fitting attitudes. And conversely, any
sort of attitude which doesnt include such a judgement (and a
fortiori, actions) will not be a fitting response to value. This is a
substantial constraint on the range of fitting attitudes. So a no-

VALUE AND NORMATIVITY

285

priority view of this kind cannot provide a satisfactory solution

286

VALUE THEORY

to WKR.142

8.8

Summary

This book has, at many junctures, proceeded on the

assumption that a fitting attitude account of value, or some
other account of the normativity of value, serves as a
constraint to bear in mind while pursuing value theory. In this
chapter I have questioned whether FA is actually a superior
account, by discussing the wrong kind of reasons problem and
alleged solutions to it: object-given vs. attitude-given reasons;
evaluative/primary
vs.
practical/secondary
reasons;
fittingness/correctness accounts. While we are on the search
for a completely satisfactory solution, it might be worth
considering the merits of alternative accounts of value.

BIBLIOGRAPHY

Anderson, E. (1993). Value in Ethics and Economics.

Cambridge, MA: Harvard University Press.
(1997). Practical Reason and Incommensurable Goods, in
Chang 1997a.
Aristotle (350 BC ). Nicomachean Ethics (trans. W. D. Ross).
http://classics.mit.edu/Aristotle/nicomachaen.1.i.html/
Arneson, R. (2010). Good, period. Analysis, 70(4): 731-44.
Arrhenius, G. (2005). Superiority in value. Philosophical
Studies, 123: 97-114.
Arrhenius, G. and Rabinowicz, W. (2005). Millian superiorities.
Utilitas, 17: 127-46.
Baron, M. (1997). Kantian Ethics, in Baron et al. 1997.
Baron, M. et al. (1997). Three Methods of Ethics. Oxford:
Blackwell.
Beardsley, M. C. (1965). Intrinsic value Philosophy and
Phenomenological Research, 26: 1-17.
Bentham, J. (1970 [1789]). An Introduction to the
Principles of Morals and Legislation. London: Athlone
Press.

VALUE AND NORMATIVITY

287

Blackburn, S. (1984). Spreading the Word. Oxford: Clarendon

Press.
Blanshard, B. (1961). Reason and Goodness. London:
George Allen & Unwin.
Bradley, B. (2002). Is intrinsic value conditional?.
Philosophical Studies, 107: 23-44.
(2006). Two concepts of intrinsic value. Ethical Theory
and Moral Practice, 9: 111-30.
Brannmark, J. (2008). Excellence and means: on the
limits of buck-passing. The Journal of Value Inquiry, 42:
301-15.
Brentano, F. (1969 (1889)). The Origin of Our Knowledge
of Right and Wrong (ed. R. M. Chisholm). London:
Routledge & Kegan Paul.
Broad, C. D. (1930). Five Types of Ethical Theory. London:
Routledge & Kegan Paul.
(1940). Review of Foundations of Ethics (by W. D.
Ross). Mind, 49: 228-39.
(1942). Certain Features in Moores Ethical Doctrines,
in Schilpp 1942, 41-68.
Broome, J. (1997). Is Incommensurability Vagueness?, in
Chang 1997a.
Bykvist, K. (2009). No good fit: why the fitting attitude
analysis of value fails. Mind, 118: 1-30.
Byron, M. (ed.) (2004). Satisficing and Maximizing: Moral
Theorists on Practical Reason, Cambridge: Cambridge
University Press.
Carlson, E. (2001). Organic unities, non-trade-off, and the
additivity of intrinsic value. The Journal of Ethics, 5: 33660.
Chang, R. (ed.) (1997a). Incommensurability,
Incomparability, and Practical Reason. Cambridge, MA:
Harvard University Press.
(1997b). Introduction, in Chang 1997a.
(2001). Against constitutive incommensurability or
buying and selling friends. Philosophical Issues, 11: 3360.
(2002). The possibility of parity. Ethics, 112: 659-88.
(2005). Parity, interval value, and choice. Ethics, 115: 33150.
Chisholm, R. M. (2005 [1972]). Objectives and Intrinsic
Value, in T. R0nnow-Rasmussen and M. J. Zimmerman
(eds) (pp. 171-9), Recent Work on Intrinsic Value.
Dordrecht: Springer.
(1981). Defining intrinsic value. Analysis, 41: 99-100.

288

VALUE THEORY

(1986). Brentano and Intrinsic Value. Cambridge:

Cambridge University Press.
Conee, E. (1982). Instrumental value without intrinsic
value?. Philosophia, 11: 345-59.
Cook, P. (2008). An augmented buck-passing account of
reasons and value: Scanlon and Crisp on what stops the
buck. Utilitas, 20: 490-507.
Crisp, R. (2000). Review of value ... and what follows (by J.
Kupperman). Philosophy, 75: 458-62.
(2005). Value, reasons and the structure of justification:
how to avoid passing the buck. Analysis 65: 80-5.
(2008). Goodness and reasons: accentuating the
negative. Mind, 117: 257-65.
Dancy, J. (2000). Should we Pass the Buck?, in A. O. Hear
(ed.),
The Good, the True and the Beautiful (pp. 159-73).
Cambridge: Cambridge University Press.
(2004a). Ethics without Principles. Oxford: Clarendon Press.
(2004b). Enticing Reasons, in R. Jay Wallace et al. (eds),
Reason and Value: Themes from the Moral Philosophy of
Joseph Raz (pp.
91- 118). Oxford: Oxford University Press.
(2005). The Particularists Progress, in T. R 0nnowRasmussen and M. J. Zimmerman (eds), Recent Work on
Intrinsic Value (pp. 325-47). Dordrecht: Springer.
(2007). Was Moore Right about Punishment?, in S.
Nuccetelli
and G. Seay (eds), Themes from G. E. Moore: New Essays
in Epistemology and Ethics (pp. 325-42). Oxford: Oxford
University Press.
Danielsson, S. and Olson, J. (2007). Brentano and the
buck-passers. Mind, 116: 511-22.
DArms, J. and Jacobson, D. (2000a). The moralistic
fallacy: on the appropriateness of emotions.
Philosophy and Phenomenological Research, 61: 65-90.
(2000b). Sentiment and value. Ethics, 110: 722-48.
(2006). Anthropocentric Constraints on Human Value, in R.
Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 1.
Oxford: Clarendon Press.
Darwall, S. (2002). Welfare and Rational Care. Princeton:
Princeton University Press.
Davidson, D. (1963). Actions, reasons, and causes.
Journal of Philosophy, 60(23): 685-700.
Ewing, A. C. (1948). The Definition of Good. New York:
Macmillan.

VALUE AND NORMATIVITY

289

(1959). Second Thoughts in Moral Philosophy. London:

Routledge & Kegan Paul.
Feldman, F. (1998). Hyperventilating about intrinsic
value. Journal of Ethics, 2: 339-54.
Firth, R. (1952). Ethical absolutism and the ideal
observer. Philosophy and Phenomenological Research,
12: 317-45.
Fletcher, G. (2012a). Resisting buck-passing accounts of
prudential value. Philosophical Studies, 157(1): 77-91.
(2012b). The locative analysis of good for formulated
and defended. Journal of Ethics and Social Philosophy,
6(1): 1-27.
Foot, P. (1985). Utilitarianism and the virtues. Mind, 94: 196209.
Geach, P. (1956). Good and evil. Analysis, 17: 33-42.
Gert, J. (2004a). Brute Rationality: Normativity and
Human Action. Cambridge: Cambridge University Press.
(2004b). Value and parity. Ethics, 114: 492-510.
Gibbard, A. (1990). Wise Choices, Apt Feelings.
Cambridge, MA: Harvard University Press.
Gregory, A. (2014). A very good reason to reject the
buck-passing account. Australasian Journal of
Philosophy, 92: 287-303.
Griffin, J. (1986). Well-Being: Its Meaning, Measurement
and Importance. Oxford: Clarendon Press.
Heathwood, C. (2008). Fitting Attitudes and Welfare, in
R. Shafer- Landau (ed.), Oxford Studies in Metaethics,
vol. 3 (pp. 47-73). Oxford: Oxford University Press.
Hieronymi, P. (2005). The wrong kind of reason. The
Journal of Philosophy, 102: 437-57.
Hooker, B. (2000). Ideal Code, Real World: A RuleConsequentialist Theory of Morality. Oxford: Clarendon
Press.
Hsieh, N. (2007). Is incomparability a problem for anyone?.
Economics and Philosophy, 23: 65-80.
(2008). Incommensurable Values, in E. N. Zalta (ed.), The
Stanford Encyclopedia of Philosophy (Fall 2008 edn).
http://plato.stanford.edu/ archives/fall2008/entries/valueincommensurable/
Hurka, T. (1987). Good and good for. Mind, 96: 71-3.
(1998). Two kinds of organic unity. Journal of Ethics, 2(4):
299-320.
(2001). Virtue, Vice and Value. New York: Oxford University
Press.
(2003). Moore in the middle. Ethics, 113: 599-628.

290

VALUE THEORY

Jackson, F. (1998). From Metaphysics to Ethics.

Oxford: Oxford University Press.
Jacobson, D. (2011). Fitting Attitude Theories of Value, in E.
N.
Zalta (ed.), The Stanford Encyclopedia of Philosophy
(Spring
2011
edn).
http://plato.stanford.edu/archives/spr2011/entries/
fitting-attitude-theories/
Kagan, S. (1998). Rethinking intrinsic value. Journal of
Ethics, 2(4): 277-97.
Kant, I. (1964 [1785]). Groundwork of the Metaphysic of
Morals. New York: Harper and Row.
Kavka, G. (1983). The toxin puzzle. Analysis, 43: 33-6.
Kelly, C. (2008). The impossibility of
incommensurable values. Philosophical Studies,
137: 369-382.
Klocksiem, J. (2011). Moorean pluralism as a solution to the
incommensurability problem. Philosophical Studies, 153:
335-49.
Korsgaard, C. M. (1983). Two distinctions in goodness.
Philosophical Review, 92: 169-95.
(1996). The Sources of Normativity. Cambridge:
Cambridge University Press.
Kraut, R. (2007). What is Good and Why: The Ethics of WellBeing. Cambridge, MA: Harvard University Press.
(2011). Against Absolute Goodness. Oxford: Oxford
University Press.
Lang, G. (2008). The right kind of solution to the wrong kind
of reason problem. Utilitas, 20 (4), 472-89.
Langton, R. (2007). Objective and unconditioned value.
Philosophical Review, 116: 157-85.
Lemos, N. M. (1994). Intrinsic Value. Cambridge: Cambridge
University Press.
(1998). Organic unities. Journal of Ethics, 2: 321-37.
Lewis, D. (1989). Dispositional theories of value.
Proceedings of the Aristotelian Society (suppl.), 63: 11337.
Louise, J. (2004). Relativity of value and the consequentialist
umbrella. Philosophical Quarterly, 54: 518-36.
McAleer, S. (2008). Pettits non-iteration constraint. Utilitas,
20:
59-64.
McDowell, J. (1998a). Values and Secondary Qualities, in
Mind, Value, and Reality (pp. 131-50). Cambridge, MA:
Harvard University Press.

VALUE AND NORMATIVITY

291

(1998b). Projection and Truth in Ethics, in Mind, Value,

and Reality (pp. 151-66). Cambridge, MA: Harvard University
Press.
McKeever, S. and Ridge, M. (2006). Principled Ethics:
Generalism as a Regulative Ideal. Oxford: Oxford University
Press.
Mackie, J. (1977). Ethics: Inventing Right and Wrong. London:
Penguin.
Mason, E. (2011), Value Pluralism, in E. N. Zalta (ed.), The
Stanford Encyclopedia of Philosophy (Fall 2011 edn).
http://plato.stanford.edu/
archives/fall2011/entries/valuepluralism/
Mill,
J.
S.
(1861).
Utilitarianism.
http://www.utilitarianism.com/mill1.htm/
Moore, G. E. (1993 [1903]). Principia Ethica (ed. T.
Baldwin). Cambridge: Cambridge University Press.
(1966 [1912]). Ethics (2nd edn). London: Oxford University
Press.
(1993 [1922]). The Conception of Intrinsic Value, in Moore
1993 (1903).
(1942). A Reply to My Critics, in Schilpp 1942: 535-677.
(1959). Is Goodness a Quality?, in Philosophical Papers (pp.
89-101). London: George Allen & Unwin.
Mulligan, K. (1998). From appropriate emotions to values.
The Monist, 81: 161-88.
Nozick, R. (1981). Philosophical Explanations. Cambridge, MA:
Harvard University Press.
Olson, J. (2004a). Intrinsicalism and conditionalism about
final value. Ethical Theory and Moral Practice, 7: 31-52.
(2004b)Buck-passing and the wrong kind of reasons.
Philosophical Quarterly, 54 (215): 295-300.
(2006). G. E. Moore on goodness and reasons.
Australasian Journal of Philosophy, 84: 525-34.
(2009a). Fitting attitude analyses of value and the
partiality challenge. Ethical Theory and Moral Practice, 12:
365-78.
(2009b). The wrong kind of solution to the wrong kind
of reason problem. Utilitas, 21: 225-32.
Orsi, F. (2013a). Whats wrong with Moorean buckpassing? Philosophical Studies, 164: 727-46.
(2013b). Fitting attitudes and solitary goods. Mind, 122:
687-98.
Parfit, D. (1984). Reasons and Persons. Oxford: Oxford
University Press.

292

VALUE THEORY

(1997). Reasons and motivation. Proceedings of the

Aristotelian Society (suppl.), 71: 99-130.
(2001). Rationality and Reasons, in D. Egonsson et al.
(eds), Exploring Practical Philosophy: From Action to
Values. Aldershot: Ashgate.
Persson, I. (2007). Primary and Secondary Reasons, in D.
Egonsson, J. Josefsson, B. Petersson and T. R 0nnowRasmussen (eds), Hommage a Wlodek. Philosophical
Papers
Dedicated
to
Wlodek
Rabinowicz.
www.fil.lu.se/hommageawlodek
Pettit, P. (1997). The Consequentialist Perspective, in Baron
et al. 1997.
Piller, C. (2006). Content-related and attitude-related
reasons for preferences. Philosophy, 81: 155-82.
Portmore, D. (2005). Combining teleological ethics with
evaluator relativism: a promising result. Pacific
Philosophical Quarterly, 86:
95-113.
Rabinowicz, W. (2008). Value relations. Theoria, 74: 18-49.
Rabinowicz, W. and R0nnow-Rasmussen, T. (1999). A
distinction in value: intrinsic and for its own sake.
Proceedings of the Aristotelian Society, 100: 33-51.
(2003). Tropic of value. Philosophy and
Phenomenological Research, 66(2): 389-403.
(2004). The strike of the demon: on fitting proattitudes and value. Ethics, 114: 391-423.
(2006). Buck-passing and the right kind of reasons.
Philosophical Quarterly, 56: 114-120.
Railton, P. (1986). Facts and values. Philosophical Topics,
14(2): 5-31.
Raz, J. (1986). The Morality of Freedom. Oxford: Clarendon
Press.
(1999). Engaging Reason: On the Theory of Value and
Action.
Oxford: Oxford University Press.
(2001). Value, Respect, and Attachment. Cambridge:
Cambridge University Press.
(2009). Reasons: Practical and Adaptive, in D. Sobel
and S. Wall (eds), Reasons for Actions. Cambridge:
Cambridge University Press.
Regan, D. (1997). Value, Comparability, and Choice, in
Chang 1997a.
(2004). Why Am I My Brothers Keeper?, in R. Jay
Wallace et al. (eds), Reason and Value: Themes from the
Philosophy of Joseph Raz. Oxford: Oxford University Press.

VALUE AND NORMATIVITY

293

Reisner, A. (2009). Abandoning the buck passing analysis

of final value. Ethical Theory and Moral Practice, 12: 37995.
Richardson, H. (1994). Practical Reasoning about Final
Ends. Cambridge: Cambridge University Press.
Ridge, M. (2011). Reasons for Action: Agent-Neutral vs.
AgentRelative, in E. N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy (Winter 2011 edn).
http://plato.stanford.edu/archives/
win2011/entries/reasons-agent/
Rind, M. and Tillinghast, L. (2008). What is an attributive
adjective? Philosophy, 83 (1): 77-88.
R0nnow-Rasmussen, T. (2011). Personal Value. Oxford:
Oxford University Press.
Rosati, C. (2008). Objectivism and relational good.
Social Philosophy and Policy, 25(1): 314-49.
Ross, W. D. (2002 [1930]). The Right and the Good (ed. P.
Stratton- Lake). Oxford: Clarendon Press.
(1939). Foundations of Ethics. Oxford: Clarendon Press.
Russell, B. (1910). The Elements of Ethics, in Philosophical
Essays (pp.
1- 58). London: Longmans.
Scanlon, T. M. (1998). What We Owe to Each Other.
Cambridge, MA: Harvard University Press.
Schilpp, P. A. (ed.) (1942). The Philosophy of G. E. Moore.
Evanston and Chicago: Northwestern University.
Schroeder, M. (2006). Not so promising after all:
evaluator-relative teleology and common-sense morality.
Pacific Philosophical Quarterly 87(3): 348-56.
(2007). Teleology, agent-relative value, and the
good. Ethics 117: 265-95.
(2010). Value and the Right Kind of Reasons, in R.
Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 5
(pp. 25-55). Oxford: Oxford University Press.
(2011). Buck-passers negative thesis. Philosophical
Explorations, 12: 341-47.
(2012). Value Theory, in E. N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy (Summer 2012 edn).
http://plato.stanford.
edu/archives/sum2012/entries/value-theory/
Sen, A. (1983). Evaluator relativity and consequential
evaluation. Philosophy and Public Affairs, 12: 113-32.
Sidgwick, H. (1981 [1907]). The Methods of Ethics (7th
edn). Indianapolis: Hackett.

294

VALUE THEORY

Skorupski, J. (2007). Buck-passing about goodness, in D.

Egonsson, J. Josefsson, B. Petersson and T. R 0nnowRasmussen (eds), Hommage a Wlodek. Philosophical
Papers
Dedicated
to
Wlodek
Rabinowicz.
www.fil.lu.se/hommageawlodek/
Slote, M. (1989). Beyond Optimizing. Cambridge, MA:
Harvard University Press.
Smith, A. (1759). The Theory of Moral Sentiments.
Indianapolis: Liberty Fund.
Smith, M. (1994). The Moral Problem. Oxford: Blackwell.
(2003). Neutral and relative value after Moore. Ethics, 113:
576-98.
(2010). On normativity. Analysis, 70(4): 715-731.
Stratton-Lake, P. (2002). Pleasure and Reflection in Rosss
Intuitionism, in P. Stratton-Lake (ed.), Ethical Intuitionism:
Re-evaluations (pp. 113-36). Oxford: Oxford University
Press.
(2005). How to deal with evil demons: comment on
Rabinowicz and R0nnow-Rasmussen. Ethics, 115: 788-98.
(2009). Roger Crisp on goodness and reasons. Mind, 118:
1081-94.
Stratton-Lake, P. and Hooker, B. (2006). Scanlon versus Moore
on Goodness, in T. Horgan and M. Timmons (eds),
Metaethics after Moore (pp. 149-68). Oxford: Oxford
University Press.
Sturgeon, N. (2010). Normativity. Analysis, 70(4): 744-53.
Suikkanen, J. (2004). Reasons and valuein defence of the
buck-passing account. Ethical Theory and Moral Practice, 7:
513-35.
(2009). Buck-passing accounts of value. Philosophy
Compass, 4: 768-79.
(2009). Consequentialism, constraints and the goodrelative-to: a reply to Mark Schroeder. Journal of Ethics &
Social Philosophy, March 2009: 1-9.
Swanton, C. (1995). Profiles of the virtues. Pacific
Philosophical Quarterly, 76: 47-72.
Szabo, Z. (2001). Adjectives in Context, in R. Harnish and I.
Kenesei (eds), Perspectives on Semantics, Pragmatics, and
Discourse (pp. 119-46). Amsterdam: John Benjamins
Publishing Co.
Tannenbaum J. (2010). Categorizing Goods, in R. ShaferLandau (ed.), Oxford Studies in Metaethics, Vol. 5. New York:
Oxford University Press.
Thomson, J. J. (1997). The right and the good. Journal of
Philosophy, 94: 273-98.

VALUE AND NORMATIVITY

295

(2001). Goodness and Advice. With commentary by Philip

Fisher et al. Princeton: Princeton University Press.
(2006). The Legacy of Principia, in T. Horgan and M.
Timmons (eds), Metaethics after Moore (pp. 233-55).
Oxford: Oxford University Press.
(2008). Normativity. Chicago: Open Court.
(2010). Reply to critics. Analysis, 70(4): 753-764.
Vayrynen, P. (2006). Resisting the Buck-Passing Account
of Value, in R. Shafer-Landau (ed.), Oxford Studies in
Metaethics, Vol. 1 (pp. 295-324). Oxford: Oxford
University Press.
Wiggins, D. (1987). A Sensible Subjectivism?, in Needs,
Values, Truth: Essays in the Philosophy of Value. Oxford:
Blackwell.
Williams, B. (1981). Internal and External Reasons, in Moral
Luck. Cambridge: Cambridge University Press.
(1985). Ethics and the Limits of Philosophy. Cambridge, MA:
Harvard University Press.
Wright, G. H. von (1958). The Varieties of Goodness.
http://www. giffordlectures.org/
Zimmerman, M. J. (1999). The moral aspect of nonmoral
goods and evils. Utilitas, 11(1): 1-15.
(2001). The Nature of Intrinsic Value. Lanham: Rowman
and Littlefield.
(2005). Intrinsic Value and Individual Worth, in T. R 0nnowRasmussen and M. J. Zimmerman (eds), Recent Work on
Intrinsic Value (pp. 191-205). Dordrecht: Springer.
(2007). The good and the right. Utilitas, 19(3): 326-353.
(2009). Understanding whats good for us. Ethical Theory
and Moral Practice, 12: 429-439.
(2010). Intrinsic vs. Extrinsic Value, in E. N. Zalta (ed.), The
Stanford Encyclopedia of Philosophy (Winter 2010 edn).
http://plato. stanford.edu/archives/win2010/entries/valueintrinsic-extrinsic/
(2011). Partiality and intrinsic value. Mind, 120: 447-83.

296

VALUE THEORY

INDEX

additivity 87
agent-relative/agent-neutral
value 73-9, 127 doubts
about the distinction 75-6,
78 Anderson, E. 37n. 7, 1046, 114, 129
appropriateness see
fittingness and unfittingness
Aristotle 27, 29, 38, 45
axiology 6-8, 27, 29, 49, 50,
52, 53, 56, 58, 59, 98, 99,
117
bearers of value 8, 35-8, 69,
129, 130-3
pluralism about 37, 133
beauty 31n. 2, 32, 52, 86
Bentham, J. 28 Bradley, B.
87n. 1, 130-3 Brentano, F.
10, 150 Broad, C. D. 19
buck-passing account of
value 10-15, 19, 20, 21n.
15,
1389, 141
arguments against
13-15, 139,
141-3 arguments for
14, 20-1,
13940
formulations of 10,
13, 14,
141
Chang, R. 102, 109-13
comparability, complete 107

conditionalism about value

93-100 adjusted attitude
objection to
96-7
arbitrariness objection to
95-6 consequentialism 6, 502, 55, 59, 73-6, 127 agentrelative 74 contextual value
93, 98-100 correctness 10,
32, 58n. 15, 121, 126n. 10,
150, 153-5
DArms, J. and Jacobson, D.
144n. 12, 151-3 Dancy,
J. 33n. 5, 84-5, 87n. 2, 97n.
13, 98-100, 108, 138n. 5
default value 71, 99
deontology 6, 52, 54, 59n.
17, 76, 139n. 8 duty 8, 74,
136, 137n. 3, 139
Early Moorean Account 1367, 141, 142 enabling
conditions 30-1, 33n. 5, 39,
84-5, 94, 98, 119-20 error
theory 18-19, 22n. 18
eudaimonia 27, 29 evaluative
inadequacy 90-1 Ewing, A. C.
11, 19n. 12, 118
favouring attitudes 9, 60n.
18, 76, 90, 117-18, 145-9
ability to favour 13, 68
comparative attitudes 105-6,
108, 113-14 degree/strength

VALUE AND NORMATIVITY

of 122-6 direction of 60, 11922 kinds of 118-19, 126

fitting attitude account of
value (FA) 10-15, 58, 83,
117, 125-6, 135, 157 and
absolute value 57-61, 66n. 5
and agent-relative/agentneutral value 73, 76-8 and
cognitivism/
non-cognitivism 17-8
and conditional value
120-1 and conditionalism
9 and error theory 18-19
and final value 16, 25-39,
88, 119-22 and
incomparability 104-6,
114
and naturalism/nonnaturalism 19-22
and organic unities 88-90
and personal value 65-8,
72-3,
119-22 and thin/thick
values 11n. 7, 152-3, 156
and virtual value 90-1
arguments for and against
13-15, 139, 141-3 as
leading idea of the book
16, 132, 135 rival accounts
to 136-41, 142,
156
fittingness and unfittingness
5,
10-11, 13-17, 19, 21,
28, 36-8, 58n. 15, 601, 83, 88-9, 119-22,
127, 133, 150-3, 156-7
compared to
normative reasons 1315
degrees of fittingness 122-6
fittingness and agents 13-15,
19, 22, 52, 60-1, 67-8,
72- 3, 73-8, 107-8, 124
and morality 8-9, 10,

297

57-60, 76, 139n. 8,

144, 150, 151 and
value of an attitude
66n.
5, 126
and universalizability 75,
127-8 Foot, P. 50-2, 57-9 for
its own sake 27-9, 31-3, 36,
38, 40, 43, 58, 60, 72,
120-2
for someones sake 66-9n.
10, 72, 121-2, 143n. 11
friendship 30, 33-5, 38, 40,
41n. 9, 43, 70-1, 103n. 3,
105-6, 119-21, 129
Geach, P. 46-50, 53, 64 Gert,
J. 108, 110-12 good as 45,
49, 53-4, 61n. 20 good as an
end see value, final good for
someone see value, personal
good will 27-30, 32, 40-1
good-making properties 4, 8,
11n. 7, 12, 21n. 15, 33, 84,
94, 103-4, 137-8, 140, 142
happiness 27, 28, 29-30, 52
being worthy of 29-30, 40-1
hedonism 6-7, 28-9, 32, 98100 holism vs. atomism 98100 Hurka, T. 40, 41n. 9, 956, 126n. 9
incommensurability, overall
102-3, 104
incommensurability, pro
tanto 103, 104
incomparability/
incommensurability
101-15
arguments against 107-8,
109 arguments for 102, 1046 indifference (as a fitting
attitude) 38, 90, 126

298

VALUE THEORY

13-15 holding-reasons
vs. content- reasons 150
internalism about 21-2,
61 monism and pluralism
about
140
object-given vs. statelove 37, 105-6, 153-4
given 143-4
peremptory vs. enticing
maximization of value 73-6,
108 practical, evaluative,
127-30, 131-2 argument for
epistemic
128-9 McDowell, J. 19-20,
1457, 157
156-7 meta-ethics 7, 17-22,
primary vs. secondary
33n. 5, 50-1, 61 Mill, J. S.
147-50 reasons and ability
28, 55, 98 monism and
13 reasons and agents 14pluralism about value 6, 29,
15 reasons to believe, to
30, 98, 99,
judge
100, 103-4, 117, 126-7,
14650,
157
134, 140 Moore, G. E 20, 55
shared 153-5
on good for 63-6, 69 on
special or personal
intrinsic value 31-2, 34 on
67
organic unities 85-9 on value
see also reasons of
and fitting attitudes 58-9,
fittingness; value and
132n. 17, 136-7 Moorean
fitting attitudes/
conception of intrinsic value
normative reasons:
32, 35, 85-8, 130-3 moral
general connection
theory see normative theory
between, wrong kind
of reasons Normative
Napoleons hat 34-9, 42,
Redundancy Claim 12,
119-20 naturalism 19-22,
136, 137-41 normative
58n. 15, 61, 83-4
theory 6-7, 52, 54-5,
Necessary Equivalence
58-9
Accounts 137
normativity 6-7, 60-1, 83,
with Redundancy 138, 141,
136-8, 157
142 without Redundancy
value and 8-15, 20, 136-41,
138-41,
157
142
Kant, I. 27-30, 32, 38, 40,
131n. 16
Kantian conception of
intrinsic value 130-3
Korsgaard, C. 19-20, 33n. 4
Kraut, R. 61n. 19

no good reasons principle

104-5 non-cognitivism 1718 non-naturalism 19-22,
61 no-priority view 21,
156-7 normative reasons
8-9, 13-15,
60, 124, 137-40
agent-relative 75-6
compared to fittingness

Olson, J. 96, 98, 150 open

question argument 20-1
organic unities 16, 85-90
attitude mismatch
objection to
88-9

VALUE AND NORMATIVITY

normatively idle value

objection to 89-90 ought 8,
10, 13, 76, 108, 136,
144
parity 109-14 and
vagueness 109-10 argument
for 109 objections to 110-12
particularism vs. generalism
99-100 Persson, I. 147-9
Pettit, P. 127-30 Plato 27n. 1
pleasure 6-7, 28-9, 31-2, 845,
89- 92, 98-100, 103-4
sadistic 86, 89-92,
94-6 pluralism about
value see monism
and pluralism about
value pluralism,
methodological 17,
37, 133
Progress and Decline 92-3,
95n. 10
punishment 85-9, 94-7
Rabinowicz, W. 114, 143n.
11, 149
Raz, J. 102-3, 108, 126n. 10,
realism 18-20
reasons of fittingness 14,
124, 141-5, 148, 150-2
Reduction Claim 12, 19, 136,
136-41, 156 Regan, D. 107-8
respect 37, 38, 69n. 10,
105, 126, 127-9
resultance 84, 91, 119, 121,
138,
141
rightness 8, 50, 59, 66n. 5,
136, 137n. 3
R0nnow-Rasmussen, T. 67n.
7, 69n. 11, 121-2,
143n. 11, 149
Ross, W. D. 29, 35n. 6, 41,
59n. 16, 139, 141

299

Scanlon, T. M. 9, 37
on buck-passing 10-11,
20-1, 138-40 on
maximization 129 on wellbeing 69-70 Schroeder, M.
76n. 23, 153-5 sensibility
theory see no-priority view
Sidgwick, H. 31n. 2, 65-6
Skorupski, J. 146-7 Smith,
M. 21, 58n. 15, 77n. 24
Stratton-Lake, P. 12, 143n.
11 supervenience 21n. 15,
81-5, 91, 102, 119, 122
thin/thick values and
evaluations
2- 6, 11n. 7, 152
Thomson, J. J. 538, 60n. 18
trichotomy thesis
101-2, 109
utilitarianism see
consequentalism
value
absolute 45-61, 64-6, 72,
79, 136
arguments against 4655, 61n. 19 defence of
56-61 attributive and
predicative
3- 4, 46-50, 56, 60n.
18, 64
essential 39, 44 extrinsic
31-2, 39-43, 44, 91,
119-21 argument for
elimination of 41-3
final 25-44, 45, 50, 57n.
12, 58-61, 71-2, 75,
85-6,
90- 4, 97-100, 119-22,
127, 129-33
arguments against final
extrinsic value 35-9
exclusively final value 29,

300

VALUE THEORY

41n. 9, 43 and
extrinsic value 32-5,
11921,
131n. 16
and intrinsic value 1617, 26, 31-2, 35-9, 42,
44, 57n.
12, 71n. 17, 85-6, 89,
92- 3, 119, 130, 133 and
personal value 6972,
97, 119-22 impersonal
51, 72-3, 77-9, 121
instrumental 3, 26, 28, 33,
41-3, 120-1 intrinsic 26,
31-2, 33nn. 4,
5, 35-7, 44, 56-7, 119,
130-3
and conditional value
39-41 isolation test for
31-2, 34,
36, 86 and organic unities
85-8 personal 3, 6, 45, 55,
56, 61n. 19, 63-79, 97, 1212, 154 and agent-relative
value
73-9
definitions of good for
65-8
in relation to
impersonal value 6972 Moores argument
against 63-4
unconditional and
conditional 29-31, 33-4, 3841, 43,
120- 1
value and fitting attitudes/
normative reasons: general

connection between
8-10, 16, 71, 76, 89,
106, 107, 110-11,
117, 131, 140-1,
157
value for its own sake see
for its own sake; value, final
value judgements 1-6, 8,
18, 20, 99,156-157 value
pumps 113-14 value theory
debates in 130-3 questions
of 5, 7, 16, 22-3 three senses
of 6-8 virtual value 90-3
virtue and vice 5, 6-7, 29, 40,
41n. 9, 52, 55, 58, 98, 126n.
9
ways of being good 53-6
welfare see well-being wellbeing 69-70, 72, 79, 122 see
also value, personal Wiggins,
D. 20, 156-7 Williams, B. 2,
21-2 wrong kind of reasons
(WKR)
13, 14n. 11, 105n. 6,
135, 141-57 less of
a problem for rivals
to FA 142 WKR
scenarios cruel joke
144 cheating in a
game of chess
145
evil demon 13, 141,
143 father and
daughter 144-5
Zimmerman, M. J. 36-8, 568, 67n. 6, 91-2, 11920

Being good is a relational property which relates an

object
1
5

See Stratton-Lake (2002), Olson (2004a), Dancy (2007) for versions of this
objection.

value. While I accept the distinction

between enabling conditions and grounds,
and hence the possibility that intrinsic
(and final) value is variable (see
Chapter 5), I prefer for illustrative
purposes to use intrinsic value in the
way defined by Moore, whose notion of
dependence admittedly doesnt
discriminate between enabling conditions
and grounds of value.
24See Zimmerman (2001: chapter 3), who
explicitly advocates this move. But the
general idea goes back to Ross (2002
(1930): 112), and has been endorsed by
many value theorists since (Chisholm
(2005 [1972]), Lemos (1994: 20-31). See
Rabinowicz and R0nnow-Rasmussen (1999,
2003) for discussion and criticism.
25See Anderson (1993) for a view at the
other extreme: states of affairs are
never fundamental bearers of final value.
Only individuals (in particular, people)
are.
26Bradley 2006 suggests that Mooreans
(for whom final value is intrinsic and
necessary) and their opponents (whom he
calls Kantians) might be talking past

each other, and simply have in mind two

different concepts of value. I discuss
this in Chapter 7.
27To complete the picture: exclusively
final values can be conditional values.
For instance, Hurkas virtues are
attitudes whose final value is
conditional on the value of their
objects. But Hurka could argue that
virtues are only valuable for their own
sake. Can exclusively final values be
extrinsic? Consider some of our
examples. Friendships value depends on
the value of happiness, but this by
itself doesnt mean that necessarily it
is fitting to favour friendship (also)
for the sake of happiness. Rather it
means that one cannot value friendship
without also valuing happiness. Of
course, plausibly friendship can be
favoured for its beneficial
consequences, so probably it is not an
exclusively final value. Presumably
deserved happiness, for Kant, is a value
that is both extrinsic (given the second
reading in the main text) and
exclusively final (it could only be

worth favouring for its own sake).

28Ishould add that claims of absolute
value are also often expressed by terms
such as matters (more than) or is
(more) important (than), where the
speaker rejects any suggestion that the
thing in question matters or is
important only to them or to anyone else
in particular.
29Geach uses this sort of case to
contrast red and big (small), but a
rather different one with bad: we
cannot infer e.g. that because food
supports life bad food supports life
(1956: 33-4). See Rind and Tillinghast
(2008) for a recent analysis of Geachs
actual examples. See Szabo (2001) for a
discussion of Geachs argument (and its
consequences) in the context of semantic
theory.
30Note that Geach assumes that the
logical behaviour of an adjective A in
its grammatically attributive uses (x is
an A y) decides the question whether
the adjective is as such logically
attributive or predicative. This
assumption requires defence.

31Note that the fact that good is

attributive doesnt mean that good cars
and good lawyers have nothing in common
from an evaluative point of view. They
certainly do have this in common: both
good cars and good lawyers serve well the
purpose or function associated with
being what they are (see e.g. von Wright
1958).
32As Thomson argues, these are not
goodness-fixing kinds (2008: 25).
33Meaningless rather than just false,
since the argument targets the very
intelligibility of claims of absolute
goodness. Compare: x is big, in the
absence of standards of size for xs, is
similarly unintelligible, since we dont
know what would make it true or false.
34It should be mentioned that indirect
forms of consequentialism, e.g. ruleconse- quentialism, refrain from
directly translating talk of preferences
into talk of actions, but they too owe
part of their appeal to the compelling
thought.
35One important consequence is that the
sorts of outcomes one can meaningfully

call better will be limited by an

assessment of what the moral person
would aim at. Foot suggests that killing
one in order to prevent the killing of
two cannot be called better than
letting the two be killed, since the
benevolent person would not aim at the
former: the scope of benevolence is
limited by other virtues, e.g. justice
(ibid.: 206). And it seems that, in the
absence of an independent concept of
good state of affairs, one could not
argue in favour of specifically
consequentialist virtues.
36So they are meaningless, if taken
literally. Thomsons challenge is
clearly also more radical than Foots,
for whom such unqualified claims can
still have meaning as referring to what
a moral person would want for its own
sake.
37On this aspect, though, Thomson (2001:
52-6) seems to reject her former 1997
views.
38See also Smith (2010), Arneson (2010),
Sturgeon (2010).

39Zimmerman doesnt conflate the concepts

of intrinsic value and ethical/final
value. As seen in section 2.3, he
recognizes a distinction between the two
notions, while at the same time arguing
on metaphysical grounds that all
ethical/final value is intrinsic value.
His answer to Thomson therefore shifts
the attention from intrinsic value to
ethical value, but given his views being
ethically valuable is a necessary
feature of what is intrinsically good.
40For instance, Zimmermans own account
of virtue presupposes that there are
intrinsically (ethically) good and bad
states (ibid.: chapter 6.5), whereas in
Foots view it is the theory of virtue
that has priority over intrinsic value
(though virtues for Foot, like for
Thomson, are based on another kind of
value, i.e. goodness for).
41See Ewing (1948: 150-1). See also
Zimmerman (2007: 329-31, 1999).
42See Thomson (2006: 253). As Michael
Smith nicely brings out in Smith (2010),
Thomsons most recent views seem to
acknowledge the concept of correct (or

fitting) attitudes as a meaningful

evaluative concept (e.g. Thomson 2008: 119).
Her only remaining opposition would then
be against understanding goodness simpliciter
in such terms. (She would also oppose
her naturalistic story to an
intuitionist account of fitting
attitudes, but thats a matter for
substantive meta-ethics which is
orthogonal to FAs truth. See section
1.6.)
43For example, Moore (1993: 150, 1959:
94, 1942: 600). See also Ross (1939:
275-6), who rejects an analysis of one
sense of intrinsically good, while
still holding it equivalent to worthy
of admiration. More on this in Chapter
8.
44Given the FA framework, the task for
non-consequentialists is to neither deny
the compelling moral thought, nor take
Foots route, but rather find a way to
avoid saying that e.g. letting two
killings happen is worse than killing
one person; plus, of course, they can
restrict the sorts of things they are
willing to admit as intrinsic goods and

evils (that is, as goods or evils prior to

moral evaluation).
45Thomson briefly suggests that
prefer and favour are arguably
themselves attributive (2010: 756). But
by analogy with goodness the main point
should be that these attitudes are
always in a way or need specification
(favour x how?), and being attributive is
only one such mode of specification
(favour x as a K).
46Richard Krauts recent attack against
what he calls absolute goodness
explicitly acknowledges the cogency of a
fitting attitude sort of reply to
Thomsons challenge (Kraut 2011: 7-8).
His target is rather the thesis
(attributed to Moore and Ross) that
absolute goodness itself plays the role
of a (or the) reason-giving consideration
(on which see Chapter 8). His arguments
are both structural and based on moral
intuition. His positive view assigns the
only relevant practical role to what is
good/ bad for someone (see Kraut 2007). So, as I
see it, Kraut doesnt deny the
intelligibility of a concept of final or
intrinsic value, but rather locates such
value in the property of being good for
someone: there are things which it is
fitting to value for their own sake, and
these things are facts such as that x is
good for someone. His ethical
objection against absolute goodness
(2011: 81ff.) is directed at those value
theorists who take certain things (e.g.

knowledge) as finally good even if not

good for anybody. But the latter are
substantive views about value which, correct
or not, are not required by the concept
of goodness simpliciter as such. Otherwise
said: it is perfectly coherent to
believe both that all that matters is
whether things are good for people - or
sentient beings - and that this matters
for its own sake, i.e. as a final value.
47Questions for further study: What is
the best way to understand goodness of
its kind? What is its normative status?
(see Schroeder 2010, Skorupski 2007, for
a FA-style approach, Brannmark 2008 for
criticism, and von Wright 1958, Mackie
1977, and Thomson 1997 for what I take
to be a different approach).
48Moore also acknowledges that x is good
for A sometimes means A thinks that x
is good (compare p is true for A).
But this presupposes again an absolute
notion of goodness.
49See Zimmerman (2009: 433-4). In
general, writers who favour a fullinfor- mation or ideal
observer/advisor account of value make
it clear that they are not addressing
goodness for in the sense of personal
welfare (Railton 1986: 30, fn.10, Smith
2003). But not all do (e.g. Thomson
1997: 296).
50See Darwall (2002: 31-8) for a
penetrating discussion of Sidgwick.
51See Hurka (2003: 611).
52Moore does consider a fitting attitude
account of x is good for him: it is

peculiarly appropriate that a thing

which will belong exclusively to him
should also by him be approved or aimed
at, but dismisses it since by saying
that a certain relation between two
things is fitting or appropriate, we can
only mean that the existence of that
relation is absolutely good in itself
(1993: 152). Appropriateness, like
deontic predicates (rightness), is
analysed by Moore in terms of intrinsic
goodness. See Chapter 8.
53Zimmerman suggests a view similar to
(a): x is good for P =df. P has a
prudential reason to favor x (2009:
436). While this might get around the
problem of special reasons unconnected
to the agents welfare, the notion of a
prudential reason seems too close to
that of good for in order for the
definition to be informative. Also, this
definition ignores reasons that others
might have to favour x insofar as x is
good for P. Maybe the idea is that,
whatever such reasons others may have,
they are asymmetrically dependent upon
Ps (actual or possible) prudential
reasons. For example, parents have
reason to favour whats good for their
infant child, only because the child
would have reason to favour whats good
for her, if she could (and not vice

versa). Compare with proposal (e) below,

which doesnt give such priority to As
reasons (or to what it is fitting for A
to do). Compare also with Schroeder
(2010: 46-8), whereby the right kind of
reasons need not be As reasons, but
rather those which would be shared by
anyone engaged in the activity of
watching out for A.
54Compare R0nnow-Rasmussen: An object
xs value for a person a (i.e. xs
personal value), consists in the
existence of normative reasons for
favouring/disfavouring x for as sake
(2011: 47).
55Proposal (c) is close to Darwalls
thesis: [W]hat it is for something to
be good for someone just is for it to be
something one should desire for him for
his sake, that is, insofar as one cares
for him (2002: 8).
56A worry about (e) might be that the
notion of caring can only be understood
in terms of good for, so that (e) would
be circular (Fletcher 2012a: 88-9).
57As a friend of yours, it could be
fitting for me to respect your desire to
engage in something that is bad for you
but morally or otherwise good. And it
seems that I would do this for your sake,
though your desire is something
admittedly bad for you. But this is no
objection to the account, if we think
that while your desire is bad for you,
my respecting it is to some extent good
for you. In this case, respect has the
role both of a fitting for someones

sake attitude (me respecting you) and

of a state of affairs that is good for
you (you being respected by me) .
58I note agreement with R0nnow-Rasmussen
(2011: 20), although he thinks these
(and the case of respect in the previous
footnote) are cases where goodness for
doesnt necessarily mean welfare in a
mental state-sense (see 2011: 86-8).
59See Regan (2004).
60In fairness to Scanlon, he is here
arguing against taking well-being as the
master value, like utilitarians
typically do.
61See Fletcher (2012b: 20-1).
62Rosati argues that good for a person
depends on the (non-personal, intrinsic)
value of the person: [with] X is good
for P, we indicate that there is a
reason to promote X with P as the
beneficiary of the action in light of or
out of regard for the value of P (2008:
344). I have no space to discuss this
interesting proposal.
63Or at least it must be a default good
under some relevant specification. A
team winning the league is certainly not
a default good, but as argued it might
be good for someone. See section 5.6 for
a brief discussion of default value.
64Can final personal value be intrinsic,
i.e. based on xs intrinsic properties

alone? That depends: if x is good for A,

and x is not already a state of As,
then such value will be extrinsic (i.e.
result from some relation between x and
A).
65Heathwood (2008: 55) seems to
disagree. See also R0nnow-Rasmussen
(2011: Chapter 5).
66Here I am talking about classical, or
act-consequentialism. A ruleconsequentialist will instead say that
though committing the murder is the only
way to maximize value, what matters is
the value maximized by the
internalization of and conformity to
certain rules, rather than directly by a
single act. Since one such rule might
well be dont kill an innocent human
being, rule-consequentialism might not
require committing the murder. See
Hooker (2000).
67Again, this does not mean: each of us
morally ought to maximize what is good for
them. This is egoism, not agent-relative
consequentialism. See Schroeder (2007:
272-3).
68I n other words, agent-neutral value
means agent-invariant ranking, and
agentrelative value means agent-variable
ranking. A radical agent-relativist can
of course hold that all value is agentrelative in this sense. But while in

principle variable, rankings belonging

to equally uninvolved agents can
converge: relative to C and relative to
D, where C and D are total strangers to
A and B, As and Bs murders should rank
equally. Likewise for equally involved
agents: ceteris paribus, the wellbeing of a son
vis--vis the well-being of a daughter
should rank equally in their parents
respective rankings.
69Also see Schroeder (2007, 2006) for
why agent-relative value cannot stem
from a speaker-relative semantic view of
good.
70Schroeder shows how it is Thomas
Nagels own technical understanding of
reasons for action as reasons to promote
a certain state of affairs that leads
him to assume a tight fit between the
two distinctions (2007: 276-7). For the
debate and literature on agent-relative
and agent-neutral reasons, see Ridge
(2011).
71Suikkanen (2009) similarly speaks of a
conflict between the agent as an
impartial spectator and the agent as a
situated agent. Smith (2003) arguably
suffers from the contradiction problem,
once we formulate his view in terms of
betterness rather than goodness. See
Schroeder (2007: 292^) on this issue.

72The radical agent-relativist will

simply deny that there is any agentneutral value, i.e. any agent-invariant
ranking, and so will solve the conflict
by rejecting its very possibility. This
doesnt mean that according to such a
view when the agent is not involved as an
agent there is nothing she ought to do:
its just that similar things will
73The difference between the state of
affairs evaluated and the whole can also
be appreciated by thinking of the state
of affairs as a complete world where
only the whole and its parts exist, or
at least where only the whole and its
parts have value (see Bradley 2002: 35).
74See Carlson (2001) and Bradley (2002:
35-6). For this reason I have avoided,
in expounding organic unities, talk of
contrast between intrinsic and contributive/ ory
value, value as a part. For example, Dancy
(2004a: 176ff., 2007) points out that,
in an organic unity, parts seem to
contribute a value which, intrinsically,
they
75lack. For example, in deserved pain, pain seems to contribute a value
(makes the whole better) which intrinsically it lacks, if we assume that pain
is intrinsically bad. However, if we understand organic unities as Moore
himself did, we see that, like for all complex objects, in computing the value
of an organic unity we are not done until we compute its value on the
whole: and what makes things better on the whole is not any contributory

value of pain, but rather the intrinsic value of (c) as a whole. Of course, the
value of (c) still supervenes on pain and its being deserved, but this doesnt
mean that pain itself contributes a positive value to (c). The positive value
of (c) simply results from the non-additive combination of (a) and (b): what
value each part contributes is, in a sense, a wrong question to ask about
(c). For the same reason, it cannot be said that in (c) pain doesnt
contribute its intrinsic disvalue, i.e. that, paradoxically, such disvalue is at
the same time still in place but not contributing to the value of (c)
(Zimmerman 2001: 145).3This captures the implausible view that punished
crime (+2) is better than absence of crime (0). The example is merely
illustrative.

76Note that we might agree that the

appropriate attitude towards S outweighs
the attitude towards (c). But the
question is how they can be different.
Also, things dont change if we think of
S as a complete world only composed of
(a), (b), and (c). While
77and (c) are in this sense notionally
different, still whatever differences
there might be they do not warrant being
adverse to world S (since S has negative
value) but being happy for (c), since
(c) contains exactly the same things
included in world S.
78In response to the attitude mismatch
problem, the organicist might try to say
that S is different from the whole (c)
to the extent that the value of (c)
results, non-additively, from a relation
of appropriateness between (a) and (b),
but the value of S doesnt. So (c)
merits a positive response even if S
doesnt. But not all organic unities

need involve appropriateness between

parts (see the value of Progress below).
79Zimmerman speaks of pleasure as having
intrinsic virtual value: of course it is
intrinsic to the extent that such value
supervenes on the nature of pleasure.
But since virtual value also depends on
the actual value of the state of affairs
in which it figures, on Moores
definition of intrinsic value it should
count as extrinsic.
80Zimmerman (2001: 148) indeed makes it
an explicit clause that what has virtual
value doesnt have actual value.
81This is better than: sadism is a
condition that disables pleasure from
providing positive value. In our
example, we want sadist pleasure to be
evil and not just not good.
82The difference between Progress and
Decline, likewise, is subject to the
same arbitrariness problem. Is a good
somehow better than an otherwise equal
good, because it follows a past evil? Is
an evil somehow worse than an otherwise
equal evil, because it follows a good?
83Even if the criminal has no-one left
to care for him, adopting a sombre tone
can be an appropriate reaction to an
imagined group of such people.

84But see Olson (2004a) for a different

conditionalist analysis of such
examples.
85Dancy (2007: 341) suggests that it is
both true that punishment is less bad,
in the context of crime, and that crime
is less bad, in the context of
punishment. Both aspects have their
disvalues decreased, so that punished
crime is better than unpunished crime
(and of course better than punishment
for no crime).
86Of course this is a stipulated
meaning. Elsewhere holism is
associated with the idea that the nature
of an entity or phenomenon cannot be
reduced to its individual parts - in this
sense Moores organic unities are an
holistic approach in the theory of
value.
87See Dancy (2004a), McKeever and Ridge
(2006).
88In a longer work, I would also address
salient properties of the betterness
relation, such as transitivity (if x is
better than y, and y is better than z,
then x is better than z), and the
concept of lexical superiority (x is
lexically superior to y when any amount
of x is better than any amount of y).
89See also Chang (2002).

90This, in turn, might be a feature of

the specific case, or the result of a
constitutive incomparability between these values.
The latter case is sometimes illustrated
by saying that e.g. friendship cannot be
compared with any amount of money, or it
ceases to have the value it has. If any
trade-off between them takes place, it
wont be justified by their relative
values (see Raz 1986, and Chang 2001 for
criticism). Aesthetic sensibility and
dialectical skills are different values,
but presumably not constitutively
incomparable: ceteris paribus, to possess great
dialectical skills is better than to
possess poor aesthetic sensibility, and
vice versa.
91Of course the items must have enough
in common to be at least candidates for
comparability. For example, a legal
career and the number five are not
incomparable, but rather non-comparable.
92Pluralism might be even consistent
with what Richardson calls weak
commensu- rability (1994: 105): the
thesis that in any given conflict of
values, there is a true ranking of the
realization of one value against the
realization of the other value in terms
of some value, which may be one of the
values in question.

93Im interpreting her claims

charitably, since part of her argument
is also that it would be pointless
(rather than unfitting) to e.g. make
cross-category aesthetic comparisons.
But the fact that it would be pointless
to compare them is arguably a wrong kind of
reason to show that the options are
incomparable (see Chapter 8).
94Interestingly, if FA adopted desire as
a master fitting attitude (as presumably
Sidgwick did, and Oddie (2005) has
recently done), then FA would not just
allow but in fact imply value
comparability, at least in principle:
any two valuable states of affairs would
be comparable, to the extent that each
deserves to be desired with a desire of
a given intensity. See Kelly (2008), and
Klocksiem (2011) for criticism.
95See e.g. the debate on satisficing
as an alternative to maximizing: Slote
(1989), Byron (2004).
96For example, talking about reasons for
wanting, he says: wants that are to be
rendered intelligible by their
dependence on reasons are not optional.
An agent ought to be moved by the
perception of a reason (1997: 147). It
is perhaps noteworthy that he stresses

this point while discussing a choice

between pears and bananas.
97See Broome (1997) for such a view.
Space prevents me from discussing it
properly.
98A and B are not equally good for Gert,
presumably because he thinks a slight
improvement in A will increase As range,
say up to 9-13, without thereby making A
and B disjoint, i.e. without thereby
making A better than B (e.g. that would
happen only when As lower range point
goes up to 13). But Chang rejects Gerts
interval model (2005: 340-1). First, she
thinks (1) and (2) are cases of
equality, not parity. Second, in order
for A to be even slightly improved (to
become A+), we should increase the range
uniformly so that A+s range is disjoint
from As range (since thats what it is
to be better, according to Gert himself,
and A+ is supposed to be better than A).
But then any slight but real improvement
on A will make A+ better than B. So
Gerts model (and indeed, for Chang, any
interval model) cannot really capture
the idea of two items that are not
equally good. See Rabinowicz (2008) for
a different modelling based on
permissible preference orderings.
99For this notion, see Griffin (1986),
Parfit (1984: 431).

100According to Chang herself, when it

comes to incomparable items, the choice
function is simply undefined (Chang
2005: 346-7).
101The term pro-attitude was
introduced by Ross in (1939: 237), used
in this connection by Ewing (1948: 14850), and later made famous by Davidson
(1963).
102Another question under this heading
regards the intentional content of the
fitting attitude: does favouring x
require thinking that x is good or
fitting to be favoured? I discuss this
in section 8.7.
103Of course if Zimmerman understands
for what it is as for its intrinsic
properties, then Zimmermans criterion
cannot be applied to extrinsic value.
But this is precisely what should be
left open, or there would not be common
ground between those who think that
final value is always intrinsic and
those who deny it. Also, in the case of
desire, some philosophers use the label
intrinsic desire for the attitude
fitting to final value (Lewis 1989). But

they dont necessarily mean desiring

something for its intrinsic properties.
The idea is just to contrast desiring as
an end vs. desiring something as a means
(to the satisfaction of another desire).
104See Darwall (2002: 1-2, 47, 68-9).
105See Zimmerman (2001: 104), and his
references to Chisholm, Lemos, and
Brentano.
106Fittingness may not always admit of
degrees. Compare: a key either fits or
doesnt fit a keyhole. If a key fits a
hole better than another hole, then it
fits the one and not the other.
107Likewise, we should respond to
evidence not only by acquiring or
discarding relevant beliefs, but by
adjusting our degrees of credence
appropriately.
108See Zimmerman (2001: 103-13).
109In Hurkas view (2001) virtues, qua
intrinsically good responses, are
fitting responses to the good and the
bad. But this is a substantive claim.
110Also, even on a pluralist approach
there can be a broad structure to
appropriate responses. For instance Raz
(2001: 161-9) describes three stages of
correctly relating to value:
recognition, preservation (both of which
fall under respect for Raz), and
personal engagement. The first two
stages are not value- specific, and have
the role of protecting the possibility
of the third stage, in which values are
realized by people engaging with them
in the right ways (preserving a Picasso

painting protects the possibility of

people enjoying it, admiring it, being
moved by it, etc.). Pluralist
philosophers in this sense also include,
among others: Anderson (1993), Swanton
(1995), Baron (1997), Scanlon (1998).
See Nozick (1981: 429-30) for a list of
57 (!) appropriate responses.
111Normally this is meant as a moral
should, since consequentialism is a
moral theory. I treat it here instead as
a generic normative should of
fittingness.
112The problem is not that maximizing
might not be the appropriate response to
e.g. certain forms of artistic
excellence, which depend for their value
on their relative rarity, so that
producing them as much as possible is
nonsense. Pettit might say that if
rarity is a condition for artistic
excellence, then maximizing the value
contributed by excellence will simply
involve attitudes and actions which do
not undermine the rarity of the relevant
skills.
113Anderson defends the more radical
view that intrinsic value cannot ever be
maximized. Here Im discussing the claim
that it cannot always be maximized.
114See Slote (1989) and Byron (2004) on
the debate about satisficing as an

alternative to maximizing.
115Pettit also argues for a constraint
on appropriate responses which only
maximization would satisfy: noniteration. This is an interesting idea,
but it hasnt been sufficiently
elaborated. See Pettit (1997: 261-2),
and McAleer (2008) for a critique.
116Among Mooreans: Moore, Ross,
Chisholm, Lemos, Zimmerman. Among
Kantians: maybe Kant, Scanlon, Anderson,
Raz, and the philosophers who argue in
favour of extrinsic final value, at
least to the extent they do so argue:
Korsgaard, Kagan, R0nnow-Rasmussen and
Rabinowicz. Note that for Bradley the
two concepts, being different, can be
combined within the same normative view:
Kant himself arguably applies Kantian
value to the good will and to persons,
but Moorean value to the state of being
deservedly happy. Kantian value can be
an essential element in a Mooreanly
valuable state of affairs. The label
Kantian is not ideal, though
understandably inspired by Kants claims
about the value and worth of people and
the good will. Nonetheless I cannot
avoid using it in discussing Bradley.

117Moore did not endorse FA as an

analytical thesis about intrinsic value.
But he did accept it as a synthetic
necessary truth. See Chapter 8, and
Olson (2006). And the other Mooreans
mentioned in the previous footnote do
seem to accept some version of FA as a
definition of value.
118'See also Moore (1993: 239).
119Ihave proposed a formulation in (Orsi
2013a).
120In Ethics he makes it clear that the
relation between what is our duty and
what produces maximum intrinsic value is
not analytical (1966 [1912]: 89). In his
Re-ply to My Critics he explicitly acknowledges
his previous view as a mistake (1942:
558-9). Moore was influenced by Bertrand
Russells The Elements of Ethics (1910), where
Russell both takes good to be
indefinable, even in normative terms
(1910: 5-6), and rejects xs
consequences will probably be the best
possible as an analysis of x is right,
though it is a mark which in fact
attaches to all objectively right
actions and to no others (ibid.: 26).
121To my knowledge Moore himself doesnt
explicitly endorse or reject either. See
Olson (2006).
122Roger Crisp (2005, 2008) holds a view
close to Necessary Equivalence with

Redundancy: (at least some) reasons are

provided by evaluative properties,
though not by goodness. Still, the
property of goodness serves to separate
the evaluative from the non-evaluative,
and thus good things will be necessarily
reason-providing. See also Dancy (2000)
for a view close to this one. What both
Crisp and Dancy leave open is whether all
reasons to favour pair up with goodness,
since not all reasons to favour need be
grounded on evaluative properties. So
they hold a weaker thesis: necessarily,
if x is good, then there are reasons to
favour x. (Indeed, Dancy would even drop
the necessity claim and opt for a
regularity thesis.) See Stratton-Lake
(2009) for criticism of Crisps view.
123See Parfit (1997) for the distinction
between normative facts and normatively
significant facts.
124 See also Blanshard: If saintliness and generosity are such as to
merit favouring, it must be because there is something in them
that goes beyond their factual characteristics and equally goes
beyond a mere blank cheque on our favour. What is this?think we
must answer, a goodness that they have already (1961: 286).

125 Rosss overall view is more

complicated, since he also acknowledges
a distinct sense of goodness that is
reducible in FA terms. And his
distinctive deontological position
crucially seems to depend on negating the
necessary equivalence claim: some basic

normative claims, i.e. most of the prima

facie duties, involve relations of
fittingness, but are not about responses
to the good.
126For instance Scanlon: [N]atural
properties provide a complete
explanation of the [practical] reasons
we have for reacting in [certain] ways
to things that are good and valuable. It
is not clear what further work could be
done by special reason- providing
properties of goodness and value, and
even less clear how these properties
could provide reasons for action (1998:
97, my emphasis). For doubts about the
strength and significance of Scanlons
redundancy thesis, see Crisp (2005),
Vayrynen (2006), Schroeder (2011).
127This is not an innocent supposition
(and of course it needs to be qualified,
but here I am proceeding schematically).
As I mentioned in previous footnotes,
one can deny necessary equivalence
altogether (e.g. Dancy 2000), even with
the proper qualifications. Or one can
deny that truths like (b) always require
truths like (a), e.g. when truths like
(b) are about moral normativity, even if
(a)s always require (b)s (e.g. Ross).

128In presenting this reply I am in fact combining two

separate lines of response to WKR. First comes (1) the idea
that reasons to admire the demon are derivative on other reasons
(to minimize suffering), and so they only have a contingent
relation to the object of admiration (i.e. the demon). If so,
then (2) such reasons cannot be said to be object-given.
Putting things in this sequence avoids the complications
mentioned by Rabinowicz and R0nnow-Rasmussen (2004: 407)
about neatly distinguishing object-given and state-given
reasons. See Stratton-Lake (2005) for (1), and Parfit
(2001), Olson (2004b), Piller (2006), Lang (2008) for the
object-given/ state-given distinction and solutions inspired
by that. See Rabinowicz and R0nnow- Rasmussen (2004, 2006),
Olson (2009b) for criticism. For a different solution based
on the dual role of reasons as grounds for attitudes and
subjects motives for those attitudes, see Rabinowicz and
R0nnow-Rasmussen (2004) and Cook (2008) (the idea being that
the fact that admiring the demon will save us from torture
cannot be a subjects motive for admiring the demon for his own
sake, and so is a wrong kind of reason to show the demons
value).
129See DArms and Jacobson (2000a, 2000b). Another example
they use is this: there can be moral or prudential reasons
against envying somebodys fortunes, but this doesnt
suffice to show that her fortunes are not enviable. But in
this case it seems to me that such moral or prudential
reasons would be at least partly attitude-given rather than
object-given, whereas whats wrong with being amused at the
joke is that the joke itself is cruel (or morally
problematic in other ways).
130See Olson (2009a), Bykvist (2009), Zimmerman (2011), Orsi
(2013b).
131See Schroeder (2010).
132See Kavka (1983).
133See also Raz (2009).
134A proposal in this similar spirit is also Hieronymi
(2005). She argues that right kinds of reasons for a given

attitude A are those which bear on a question, the settling

of which amounts to having or forming A. But can the
relevant questions for certain attitudes be specified in
non-evaluative terms? Coming to admire the demon on the
basis of his threat is nothing like having settled the
question of whether he is admirable, but this we knew
already. Hieronymis other notion is that the right kinds of
reasons for A must be reasons that support the commitments
constitutive of A, but she doesnt specify what these commitments
are in the case of favouring and the like.
135See for instance Reisner (2009).
136Olson (2009a) argues that FA can avoid such WKR cases,
since they are based on moral reasons, and these are
different from reasons of fittingness. However, it seems
hard to deny that some reasons of fittingness are moral
reasons (isnt keeping a promise fitting to having made
one?). What Olson needs, again, is a specification of the
relevant reasons of fittingness, not just a bare appeal to
fittingness (or correctness).
137Rabinowicz and R0nnow-Rasmussen (2004: 421).
138Engagement with other activities, such as moral censorship,
might require lack of amusement as a correct response from
everybody, but this would show that the joke has a
different, moral, kind of disvalue. Still, Schroeders
solution to (i) and (ii) is not without problems. (i) He
explicitly appeals to evaluations in defining standards of
correctness for admiration (e.g. admiration for x is correct
if it would not be a bad idea to emulate x (ibid.: 42)).
(ii) Here the solution seems smoother. But its crucial to
its success that Schroeder can describe the relevant
activity, engagement in which would make a moralistic
reaction at best optional and at worst incorrect, in terms
other than appreciating the comic value of jokes, since this
would presuppose the evaluative notion at issue.
139See Chapter 4 for the analysis of personal value.
140Though see Gregory (2013) for the further doubt whether
the notion of a reason (to believe or whatever) is itself

evaluative.
141One reason why it is acceptable might be that, in fact,
in order for a no-priority view to help in a solution to
WKR, the value judgement might as well be understood itself
as a judgement that a given attitude is fitting to x, thus
eliminating a direct reference to good. Values would be
understood in terms of the fittingness of attitudes
including fittingness judgements.
142In fairness to McDowell and Wiggins, they do not claim
that all fitting responses involve value judgements, nor are
they concerned with WKR in the first place.