2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

Ratio (new series) XXI 2 June 2008 00340006

KANTIAN VALUE REALISM

Alison Hills
Abstract
Why should we be interested in Kants ethical theory? One reason
is that we find his views about our moral responsibilities appealing.
Anyone who thinks that we should treat other people with respect,
that we should not use them as a mere means in ways to which they
could not possibly consent, will be attracted by a Kantian style of
ethical theory.
But according to recent supporters of Kant, the most distinctive
and important feature of his ethical theory is not his claims about
the particular ethical duties that we owe to each other, but his views
about the nature of value. They argue that Kant has an account of
the relationship between practical reason and value, known as
Kantian constructivism that is far superior to the traditional
value realist theory, and that it is because of this that we should
accept his theory.1
It is now standard for both supporters and critics to claim that
Kants moral theory stands or falls with Kantian constructivism.2
But this is a mistake. In this paper, I sketch a rival Kantian theory
of value, which I call Kantian value realism. I argue that there is
textual evidence that Kant himself accepted value realism rather
than constructivism. Whilst my aim in this paper is to set out the
theory clearly rather than to defend it, I will try to show that
Kantian value realism is preferable to Kantian constructivism and
that it is worthy of further study.

1. Kantian constructivism
According to value realism, the correctness of our ethical judgements is not dependent on our choices or our attitudes. For
example, it is simply a fact that making promises that you do not
intend to keep is morally wrong.
According to Kantian constructivism, values are not part of the
fabric of the world. Instead, rational agents construct values
1
2

Korsgaard (1996a, 1996b, 2003).

Gaut (1997), Regan (2002).

KANTIAN VALUE REALISM

183

by carrying out certain procedures. Both moral and non-moral

values are constructed, though in different ways.
Moral wrongness is constructed by following the procedure laid
down by the categorical imperative tests. It is morally wrong to act
on a particular maxim if that maxim fails the tests; only if the
maxim passes the test is it morally permissible to act on it. For
example, it is morally wrong to make a lying promise for your own
benefit, because in making a lying promise, you treat the other
person as a mere means, not as an end in herself, and thus your
maxim fails the formula of humanity test (Wood, 1999: 8790;
ONeill, 1989: 1338).
Non-moral values are constructed by the choices of rational
agents: by choosing some goal, a rational agent confers value on
it. For example, I may have a maxim of reading a book. I
confer non-moral value on reading through adopting that maxim
(Korsgaard, 1996a: 122125; Wood 1999: 127).
Korsgaard and Wood have developed an argument that links
the construction of moral and non-moral value, known as the
regress argument (Korsgaard, 1996a: 119124; Wood, 1999:
125132). This argument begins with an explanation of the
source of non-moral values, and terminates with the formula of
humanity.
According to Korsgaard and Wood, when you choose to act,
you set yourself an end to achieve by adopting that end into a
maxim. In their view, there is an important distinction between
setting yourself an end and desiring that it happen. You may
desire that x whilst thinking that x is worthless. But when you
choose to act, you set x as an end and you commit yourself to the
claim that it is good. Any agent is, therefore, committed to believing that her ends are valuable.3 But this commitment is appropriate only if x is in fact valuable. If no ends are valuable, the
commitments that are essential to agency are wholly misguided.
So to defend the presuppositions of agency, we need an explanation of the value of our ends.
One possible explanation is that the source of this value is not
the end itself, but rather the agent herself. By choosing x as an
end, you confer value on it. I call this the conferral model of
3
It is controversial whether an agent must always regard her ends as good, for it seems
that we sometimes might choose to act in ways that we know to be pointless or worthless.
But it plausible that any agent must at least sometimes act for ends that she regards as good,
and she is committed to there being an explanation of the value of those ends.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

184

ALISON HILLS

value: the theory that our rational nature, our capacity to set and
pursue ends, is the source of the value of our ends. Korsgaard and
Wood appear to accept the conferral model, and argue that, as
the source of all value, rational nature itself must be valuable,
indeed that it must be more valuable than any of our ends. They
conclude that rational nature is unconditionally valuable, and
that we are required to respect every rational agent as unconditionally valuable. This requirement is expressed by the formula of
humanity and the particular ethical duties that can be derived
from it.
The conferral model is only one possible explanation of nonmoral value. An obvious rival is value realism. Our ends might be
valuable independently of their having been chosen by us. Korsgaard defends the conferral model by arguing that, for a number
of reasons, Kantian constructivism is more appealing than value
realism.
She claims that Kantian constructivism offers an explanation of
why ethical statements are true, namely, in terms of the correctness of the categorical imperative procedure that is used to construct them: making false promises is morally wrong because a
maxim of false promising fails the formula of universal law and
the formula of humanity tests. Realism cannot offer any explanation for why ethical statements are true, for there is no explanation; it is simply a fact that making false promises is wrong. In
response to someone who questions whether it really is wrong to
make a promise you do not intend to keep, or whether she really
has reason not to do so realism can offer only an expression of
confidence, not an adequate answer (Korsgaard 1996b: 39).
In addition, realism is committed to the existence of moral
values: queer intrinsically motivating entities, which we can
know about only through an epistemologically problematic
faculty of intuition (Korsgaard, 1996b: 35, 378). Kantian constructivism is not committed to the existence of these strange
objects; instead, it is merely committed to maxims, formulated
and adopted by rational agents, and to the existence of certain
procedures for testing maxims.
Third, she argues that realism has a problem explaining
why, even if we were to discover these peculiar facts, we should take
them to have any connection to what we have reason to do (Korsgaard 1996b: 44). By contrast, Kantian constructivism explicitly
links what is wrong with what we have reason not to do, because the
procedure for testing maxims at the same time tests whether the
2008 The Author
Journal compilation 2008 Blackwell Publishing Ltd

KANTIAN VALUE REALISM

185

maxims are morally permissible or forbidden, and whether they

are suitable to be used as practical reasons, i.e. whether they can be
offered to any other rational agent as a reason.4
Korsgaard concludes that Kants moral theory should be
accepted, because the Kantian constructivist account of the
nature of value is so compelling, compared with value realism.

2. Problems for Kantian constructivism

Kantian constructivism attempts to give an account of both moral
wrongness and non-moral value. But there are serious doubts as to
whether it can give an adequate account of either.
In the first place, constructivism is in danger of falling foul of
the same problems as realism.5 According to Kantian constructivism, there are correct procedures for determining what is morally
wrong: the categorical imperative test. It is simply a fact that this is
the correct procedure and it is not at all clear that facts about
correct procedures are genuinely less ontologically and epistemologically problematic than facts about what is morally right or
wrong.
The conferral model of non-moral value is clearly intended to
help constructivism avoid this problem. According to the conferral model, the rational will is the source of the value of everything
else in the world. As such, it is unconditionally valuable, and we
are required to respect rational wills as ends in themselves not as
mere means. The conferral model is supposed to explain the
special value of a rational will, which in turn explains the correctness of the formula of humanity.
But appealing to the conferral model to defend Kantian constructivism is helpful only if its explanation of how value is constructed from rational choice is actually coherent. The conferral
model implies that objects do not have value before they are
chosen, and by choosing them, we confer value on them. But how
can we create genuine value out of nothing, simply by deciding to
make something our end? The idea that value can be created an
act of choice is quite mysterious, arguably at least as mysterious as

4
5

O Neill (1989: 18, 81104).

Shafer-Landau (2004: ch. 2).

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

186

ALISON HILLS

the claim that value exists in the world independently of our

choices.
Moreover, even if we were to accept that we really can confer
value on objects by choosing them, it is far from obvious why the
capacity to choose itself should be valuable at all, let alone that
it should be supremely, unconditionally valuable. In general, the
source of F-ness may not be itself F. Nor is the problem merely
that the source of value might not in fact be valuable. Given the
conferral model, it is hard to see how rational nature could possibly be valuable. If nothing has value antecedently to its being
chosen, there can be no reason for choosing one option over
another, since they are all equally worthless. A rational choice
between these is reduced to a mere plumping for one over the
others. Furthermore, it is hard to see that the ability to plump
for one worthless option rather than another is valuable at all,
let alone that it is of supreme value. These difficulties are forcefully pressed by Regan (2002) who concludes that rational nature
cannot be valuable in a Kantian world.6
Since we cannot explain how a rational choice can create value
and how the ability to plump for one worthless thing rather than
another could be supremely valuable, the benefits that Kantian
constructivism seemed to have over value realism are illusory.
Since Kantian ethics seemed to depend on the success of Kantian
constructivism, this is a serious problem for Kantian moral philosophy as a whole.

3. Kantian value realism

Kants ethical theory, as I understand it here, make two distinctive
claims. First, it endorses the formula of humanity: we ought always
to treat rational nature as an end, not as a mere means. And
second, it offers an explanation of the formula of humanity: rational nature is supremely valuable because it is the source of all
other values.
Some have taken the failure of Kantian constructivism as a
reason to reject the claim that rational nature is the source of

6
Regan (2002: 267). See also FitzPatrick (2005) for a very thorough critique of Kantian
constructivism.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

KANTIAN VALUE REALISM

187

value. But we really need to question one particular interpretation of that claim: the conferral model. The view that objects are
valuable because value is conferred on them when they are
chosen is untenable; but we need to distinguish the view that
rational choice is the source of value, from the different claim,
that rational nature is the source of value. Provided we reject the
former, we can still maintain the latter, by accepting an account
that I call the constitutive model of rational nature as the
source of value.
Suppose that all rational wills, that is, all persons, are valuable.
We should therefore respect all persons and refrain from treating
them as mere means. We can call these our moral ends.
But how can we explain the value of our other, non-moral ends?
In order to function as a rational agent, you need to adopt and
pursue some ends. In some respects, it does not matter which
ends, because in setting and pursuing many different ends, you
can be acting rationally. Non-moral ends are valuable not because
they are chosen by the rational will. Instead, their value depends on
their contribution to the functioning of a rational will.
Suppose that you have been told by your doctor to take up some
kind of sport for the sake of your health. As far as you are concerned, the value of sports like tennis and squash depends on the
fact that if you take them up as a hobby, you will exercise and
benefit your health. Of course, there are important differences
between different sports: tennis will contribute more in this
regard than darts, so tennis is more valuable to you than darts, and
you have more reason to take up tennis.
According to the constitutive model, ends that we set acquire
value in the same way that tennis and darts acquire value from
their contribution to your health. Tennis is a way of exercising: it
is a constitutive means of physical exercise. Similarly, setting a
particular object as your end and pursuing it is not an instrumental means to be a rational agent, it is a way of being a rational
agent. For example, you may choose to read the Critique of Pure
Reason as your end. According to the constitutive model, this
project is valuable as a constitutive means of exercising your rational nature. It, or some similar project, is worth setting as an end,
because it is worthwhile exercising your rational nature.
Many objects are suitable to be set as ends. But some would
contribute more than others to your exercising your rational
agency. Some projects are more complex than others, and require
you to set more than one end; it may be difficult to identify a
2008 The Author
Journal compilation 2008 Blackwell Publishing Ltd

188

ALISON HILLS

suitable means to achieve the end, or the means may be very

difficult to carry out, and so on.7
Since the value of the ends that we set for ourselves depends on
their contribution to our exercising our rational nature, then
simply by choosing an object, you cannot make it any more valuable than it was before.8 So the constitutive model is not a version
of constructivism. We do not construct value through our choices;
rational nature is valuable whether or not we recognize that it is
and choose to treat it as valuable. Our non-moral ends are valuable in virtue of their contribution to our functioning as rational
agents: we do not confer value on them through our choices.
The constitutive model of value is better classed as a version of
value realism: a distinctively Kantian value realism.
The point of defending the claim that rational nature is the
source of value is to support the view that rational nature has
dignity and that we should treat it with respect. According to the
constitutive model, the value of all our non-moral ends depends
on their contribution to the exercise of rational nature, so there
could be no reason for any of us to pursue some end at the
expense of rational nature.9 Kantian value realism therefore supports the formula of humanity: it cannot be reasonable to act for
a maxim that expresses a lack of respect for rational nature, so we
are required not to do so.
Kantian value realism has many features in common with
Kantian constructivism. Both claim that the rational will is
supremely valuable, and both claim that value is constituted by
the rational will. Neither is, therefore, a form of what Korsgaard
has called Objective Realism, the view that values would exist in
7
This picture of the relationship between value and the rational will has some similarities to Thomas Hurkas Aristotelian conception of the importance of our rational
nature (Hurka 1993). According to Hurka, we have reason to exercise our rational capacities, and to achieve as much as we can. But this is because our rationality is essential to our
nature, and we have reason to perfect our own nature. The Kantian conception of the
importance of rational nature is similar, but does not rest on a claim about what is essential
to our nature.
8
The value of objects depends on their contribution to rational nature, but once you
have chosen one option over another, you have some reason to pursue that project, since
carrying out the goals you have set yourself is an important aspect of rational action. Others
have reason to respect your choice, rather than trying to force you to do what they would
prefer. In this limited sense, your choice can confer some reason to pursue what you have
set as an end.
9
It as an assumption of Kantian value realism that each of us has reason to respect all
rational wills, not merely our own will. I consider the possibility that we might have reason
to respect only our own rational wills in (Hills 2004, Hills, forthcoming).

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

KANTIAN VALUE REALISM

189

a world devoid of creatures who see and respond to them. As far

as both Kantian value realism and Kantian constructivism are
concerned, it is true that the rational will brought value into the
world. But they have very different accounts of the way in which
it did so. Kantian constructivism links value to rational choice.
This is the basis of all its problems in explaining how we can create
value, why the capacity to create value is itself valuable and why a
capacity to choose between worthless things is valuable. Kantian
value realism does not have these problems, because it does not
claim that the value of objects is dependent on their being chosen
by the rational will.10
4. Textual support for Kantian value realism
Textual evidence in favour of Kantian value realism is not hard to
find. The first chapter of the Groundwork begins with what certainly appears to be a straightforward statement of value realism:
It is impossible to conceive of anything at all in the world, or
even out of it, which can be taken as good without qualification,
except a good will (G. 4: 393).
And it continues in the same vein:
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.
Considered in itself it is to be esteemed beyond comparison as
far higher than anything it could ever bring about merely in
order to favour some inclination, or, if you like, the sum total of
inclinations (G. 4: 394)
Here, Kant says that there is something that is good without
qualification. This thing is the good will (which is later explained
to be a rational will, that is, a will that adopts and acts upon the
categorical imperative to respect humanity).
10
Other commentators on Kant have emphasised the importance of the value of
rational nature (Herman 1993: 20240), or of the value of the freedom to set and pursue
ends (Guyer 1996: 4203; Guyer 1998: 335; Guyer 2000: 12931,14855). But they do not
go on to explain how rational nature is the source of the value of all other ends.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

190

ALISON HILLS

Anyone who rejects Kantian value realism must claim that Kant
chose an exceptionally misleading way of stating his views in these
striking passages. They might try to defend this claim by appealing
to the structure of the Groundwork. Kant begins with ordinary
rational knowledge of morality and in the course of each chapter,
moves to a more philosophically sophisticated theory. Perhaps the
first paragraphs set out a version of value realism that Kant will
later repudiate as philosophically nave. But Kant never suggests
that common sense is substantially mistaken about morality. In
fact, he says that ordinary common sense moral views are substantially correct, and simply need philosophical elucidation.11 In the
rest of Groundwork, he sets out to explain what it means for the
rational will to be supremely valuable, not to criticize that view.
It is obvious that Kant did not reject Kantian value realism later
in the Groundwork in any case, since a particularly clear statement of
the theory can be found in the section on the formula of humanity:
Suppose, however, there were something whose existence has
in itself an absolute value, something which as an end in itself
could be a ground of determinate laws; then in it, and in it
alone, would there be the ground of a possible categorical
imperative- that is, of a practical law.
Now I say that man, and in general every rational being, exists
as an end in himself, not merely as a means for arbitrary use by
this or that will.
(G. 4: 428)
Kant could hardly make his views any clearer: the existence of
rational beings has in itself an absolute value. There are many
other passages in which he makes similar remarks.12
There is good textual evidence that Kant accepted Kantian
value realism. But there are also a number of objections to this
interpretation. First, Kant frequently emphasizes the importance
of principles in ethics rather than values. For example, in the
Critique of Practical Reason, he claims that we cannot understand
what is valuable until we know the correct principles for action.13

11
Human reason can, in matters of morality, be easily brought to a high degree of
accuracy and precision even in the most ordinary intelligence (G. 4: 391).
12
See also G. 4: 396, 4: 397, 4: 429, 4: 462.
13
C.Pr.R. 5: 624.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

KANTIAN VALUE REALISM

191

In the Groundwork he explicitly considers a theory he calls perfectionism that seems very like value realism, and rejects it.
But in both cases the evidence is misleading, for Kant does not
reject value realism as such, but only particular versions of it. In
the Critique of Practical Reason, it is not value that he rejects in
favour of principle, but conceptions of the good that are based on
happiness or desire.14 In the Groundwork he makes it quite clear
that a formal principle can be based on an objectively valuable
end such as rational nature.15 He rejects perfectionism because
he rejects theories based on the effect of willing, rather than the
nature of willing itself.16 In neither case does he question value
realism as such.
The second reason for doubting whether Kant endorsed
Kantian value realism concerns epistemology. Kant says that it is
possible for all of us to have moral knowledge, for example, he
thinks that we all know that humanity is of supreme value. If we
construct value, then it is fairly easy to see how we can know what
is valuable. But if we do not, if value realism is correct, then our
epistemic access to value is much more puzzling and in need of
explanation. Kant does not seem to find moral knowledge puzzling, and this is some reason to favour a constructivist interpretation of his theory. But nevertheless it is compatible with realism
that we do have epistemic access to truths about value. Kant was
clearly more interested in setting out his moral theory and in
questions about freedom than in discussing moral epistemology,
but this may be simply because he was confident that we actually
have access to values, not because he denied that values are real.
So this is not a decisive reason against interpreting Kant as a value
realist.
The final concern is that the conception of rational nature
required by Kantian value realism is incompatible with that set out
by Kant in the Groundwork. According to Kant, when an agent
decides to act she is free to choose between her desires, to set the
objects of any of them as an end, or indeed to set as an end some
object that she does not desire. Setting an end includes incorporating that end into maxim, which, provided that that maxim
could be a universal law, she may then pursue.

14
15
16

C.Pr.R. 5: 212, 5: 27.

G. 4: 427.
G. 4: 4414, G. 4: 394.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

192

ALISON HILLS

Kant regards any action that is free in this way as supremely

valuable.17 This initially appears to support Kantian value realism.
In order to act freely, we set ourselves some end and incorporate
it into a maxim. Since free action is valuable, these ends may be
valuable on the basis of their contribution to our acting freely.
But it is an important part of Kantian value realism that it can
explain the different values of different ends, on the basis of their
differing contribution to rational activity. Just as tennis was more
valuable to me than darts when I was looking for a way of getting
exercise, so different ends may be more valuable than others
when I am looking to exercise my rational nature: some qualify as
achievements, others do not.
But it is far from clear that this key idea of the value of more
extensively exercising ones rational agency is compatible with
Kants view of free action. In order to act freely, any end is sufficient, however straightforward and simple. There seems to be no
reason to choose studying Shakespeare over counting blades of
grass; they are of equal value as potential contributors to our
exercising our rational agency in the Kantian sense.
The problem in reconciling this aspect of Kantian value realism
with Kants conception of the value of free action would be very
serious if the two accounts were incompatible. If that were so, in
order to respect Kants account of free action, we might have to
accept that all the potential ends that we could choose were
equally valuable. This is not very plausible and it would certainly
reduce the appeal of Kantian value realism
Kant emphasized the value of free action generally in the
Groundwork and elsewhere because free action is essential for the
possibility of acting morally, and obeying the moral law is acting
freely. In this context, differences between different ways of exercising ones rational nature are simply not relevant.
But Kant does address the question of which non-moral
projects we should pursue. He claims that we have a duty to
exercise our rational nature more extensively for he thinks that we
have a duty to develop our talents (G. 4: 430). He might therefore
accept that it is better to pursue more complex ends rather than
less, to exercise and develop greater skills rather than allow them
to atrophy, and so on. It is perfectly compatible with Kants theory
that there are two aspects to the value of rational nature: the value
17

Guyer (1996: 4203; 1998: 335; 2000: 12931, 14855).

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

KANTIAN VALUE REALISM

193

of exercising ones rational nature, by acting freely in obeying the

moral law, and the value of exercising ones rational nature more
extensively, thereby choosing more valuable non-moral ends. And
since that the two aspects of the value of rational nature are
compatible, the constitutive model can be accepted as an addition
to Kants account of the value of free action, in order to give a
Kantian account of non-moral value.
There are strong textual grounds for attributing Kantian value
realism to Kant, rather than Kantian constructivism. At several
places in the Groundwork, he quite explicitly states that rational
nature is supremely valuable. There are no decisive reasons for
rejecting this reading, and since Kantian value realism is a coherent theory whereas Kantian constructivism is not, we ought to
interpret Kant as a realist rather than as a constructivist.
5. Ends are valuable for their contribution to rational nature
Since Kant is such an important moral philosopher, it is worth
knowing whether he accepted value realism rather than constructivism. But if this question is to be of more than historic interest,
Kantian value realism must be an appealing theory in its own
right. I will not attempt to give a full defence of the theory
(ultimately I will suggest some problems for it), but I will try to
show that it fits well with many of our common sense judgements
about value, and that it is worth taking seriously as a moral theory.
According to Kantian value realism, humanity, that is rational
agents like ourselves, are supremely valuable. We must treat each
person as valuable for her own sake, not as mere means. It is
morally wrong to sacrifice someones life for the sake of satisfying
our desires, or even to maximize happiness.
Many people are likely to find this view of the value of humanity
attractive, especially those who object to viewing people, as simple
versions of utilitarianism do, as mere receptacles for utility. Kants
claims about how we should respect other adult humans are very
appealing. But serious questions remain about the overall theory.
A critic of Kantian value realism might question whether its
central idea even makes sense. Kantian value realism is coherent
only if our rational nature can have value independently of the
value of the objects we pursue. It is a fair question whether this is
even possible.
When we decide which of the non-moral ends we might choose
are worth pursuing, one of the ways in which we commonly
2008 The Author
Journal compilation 2008 Blackwell Publishing Ltd

194

ALISON HILLS

evaluate them is in terms of their contribution to the exercise of

rational nature. Other things being equal, an end is more valuable
if satisfying that end would be an achievement. When we assess
whether satisfying some potential end would be an achievement,
we do so on the basis of the effort that is needed to satisfy that end
and the level of skill that is required. For example, it is a greater
achievement to prove Fermats Last Theorem than that there is no
highest prime number, because it takes an immensely greater
amount time, concentration and skill for someone to prove the
former than the latter.18
At least one of our everyday ways of assessing the value of our
ends is in terms of their contribution to the exercise of our
rational nature, and in this regard Kantian value realism fits with
our ordinary practise. But in other ways it does not. Even if we do
assess possible ends in terms of achievement, we do not normally
think of them as merely instrumentally valuable, valuable in terms
of their contribution to rational nature.
But according to Kantian value realism, our potential ends are
in a very special way. Most instrumental means are separate from
the ends to which they contribute. For example, having money is
an instrumental means to buying nice things, so having money is
worthwhile as a mere means, for the sake of an entirely separate
further end. A constitutive means is not separable in the same
way from the end to which it contributes.19 For example, playing
tennis is not an instrumental means to exercising: it is itself a form
of exercise. The two types of activity are not identical, for you can
exercise without playing tennis, but the token activities are identical: when you play tennis, you are exercising. So if exercise is
valuable, your playing tennis is not valuable for the sake of some
further end that is separate from it. There is a sense, therefore, in
which it is true (though perhaps somewhat misleading) to say that
you are playing tennis for its own sake, since your playing tennis is
valuable for the sake of something that is token-identical with it.
Similarly, pursuing our ends is not an instrumental means to
the end of exercising rational nature. For example, your reading

18
Does this mean that we should all be trying to prove Fermats Last Theorem? Obviously that would be absurd, since most of us could make no progress at all towards doing
so. We would have no idea what to do, and would most likely end up doing nothing.
Instead, we should set and pursue ends that we can at least make some progress towards
achieving, so that we do actually exercise our rational nature.
19
For a discussion of constitutive and instrumental means, see Marras (2003).

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

KANTIAN VALUE REALISM

195

a poem may be valuable for the sake of exercising your rational

nature, but when you read a poem you are exercising your rational nature: the two activities are token-identical (though of course
not type identical, for one can exercise rational nature without
reading a poem). So reading a poem is valuable, not for the sake
of some further end that is separate from it, but for the sake of
something token-identical with it. In this sense, it is true to say that
reading a poem is worth doing for its own sake.
Constitutive means are importantly different from other kinds
of instrumental means. Whereas it would be a considerable confusion though no doubt one that many people make to think
that a mere instrumental means, like having money, was worthwhile for its own sake, a constitutive means is in fact valuable for
its own sake, at least in the sense that it is valuable for the sake of
something token-identical with it. So it is understandable that we
talk about our non-moral ends and think of many of them as
valuable for their own sake. This need not conflict with the claim
of Kantian value realism that they are valuable as ways of exercising our rational nature, when that claim is properly understood.
This is not the only criticism that one might make of Kantian
value realism, however. Even if we do evaluate our ends in terms
of the achievement that satisfying them would be, we assess our
ends for many other reasons too, most obviously, in terms of the
enjoyment we expect from satisfying them. Insofar as satisfying an
end contributes to our happiness, we take it to be worth choosing.
This kind of assessment has no necessary connection to the exercise of rational nature. If happiness is valuable for its own sake,
then Kantian value realism is false.
6. The value of happiness
Kants own view of the value of happiness is complicated. There
are passages in which he seems to suggest that happiness is valuable for its own sake, but his settled view in the Groundwork seems
to be that it is not.20 Our reason to pursue happiness is a hypo20
For example, G. 4: 396, suggests that happiness is valuable, though its value is
conditional on its being enjoyed by an agent who has a good will. But in many other places,
Kant claims that it is not valuable and that there is no categorical imperative to pursue
happiness, e.g. G. 4: 399, G. 4: 4189, G. 4: 442. Kant makes similar arguments in other
works, for example, C.Pr.R. 5: 223, 5: 26, 5: 61.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

196

ALISON HILLS

thetical imperative (of a special sort).21 Happiness is not valuable

for its own sake, it is valuable because we choose to pursue it. But
it is an unusual end, because it is one we all do in fact choose.22
Kantian value realism can acknowledge the point of assessing
ends in terms of enjoyment. Since we have set happiness as our
end, it is rational for us to choose a project that we would enjoy as
a means to that end. We are completely right that a project that
will make us happy is worth pursuing. According to Kantian value
realism, there is an important difference between the status of
enjoyment and the status of rational nature, however, because the
value of projects that contribute to enjoyment depends on our
having set that as our end, whereas the value of projects that
contribute to our exercising our rational nature does not depend
on our setting that as our end.
Suppose that you are wondering what to do. You could count
some blades of grass for no particular reason, which would neither
be a substantive exercise of your rational nature, nor particularly
enjoyable. You could lie on a beach all day, which would be more
enjoyable, but would not be a substantial exercise of your rational
nature. Or you could set yourself some demanding task, like
reading the Critique of Pure Reason, the satisfaction of which would
be an achievement and which you would enjoy, or learning some
mathematics, which would also be an achievement, but which you
would hate.
According to Kantian value realism, the demanding tasks are
valuable as potential contributors to the exercise of your rational
nature and consequently it would be reasonable for you to choose
either. Assuming that you have set happiness as one of your ends,
however, you have a reason to choose the Critique over the mathematics, as the former but not the latter will contribute to making
you happy.
There is obviously no reason at all for you to count blades of
grass, a worthless, boring activity. Choosing to waste your life on
such a project could not make it any more valuable. What about
lying on the beach? This would be pleasant, but is no achievement, and so, according to Kantian value realism, a life devoted
to such a pastime would not be worthwhile.23 On the other hand,
Kantian value realism does not require you to maximize the exer21
22
23

G. 4: 4145.
G. 4: 415.
Compare Kants discussion of the South Sea Islanders, G. 4: 423.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

KANTIAN VALUE REALISM

197

cise of your rational nature, but merely to respect your rational

nature. It might be compatible with respect for your rational
nature that you could spend an afternoon on the beach, provided
that your other projects involved a more substantial exercise of
your rational nature. As Kant says, we have only a wide, imperfect
duty to develop our talents and skills: we do not have to do so on
every occasion to the maximum degree.24
So we do have reason to promote our own happiness, though
only because we have chosen to do so. What of the happiness of
others? Do we have any reason to care that? Since we have to
respect others as rational agents, and they have all set happiness as
their end, we do have reason to promote the happiness of others,
according to Kantian value realism. If we have a choice between
two projects, one of which will make people happier, other things
being equal, we ought to choose the project that promotes happiness. But we are not to do so if we would have to treat rational
agents as mere means, for that is always morally wrong.
Kantian value realism requires us to be able to recognize more
or less substantial exercises of rational nature independently of
the value of the ends that were being pursued. This fits extremely
well with a common sense assessment of our projects in terms of
whether their success would qualify as an achievement, and how
great an achievement it would be. According to Kantian value
realism, this is the key measure of non-moral value. Projects whose
successful completion would count as an achievement are valuable because the exercise of rational nature is valuable, and the
more substantial the exercise of rational nature, the better. But
Kantian value realism is compatible with value pluralism in the
following sense: we are entitled to use other modes of assessment
of our projects too. We can choose a project on the basis of the
enjoyment it would give to ourselves or others; or on the grounds
that we will be able to carry it out with our friends; or because it is
exciting or frivolous or fun. We are not on each occasion required
to pursue achievement over enjoyment or spending time with our
friends. But we have reason to choose frivolous and fun projects
only insofar as we have chosen to set fun and frivolity as our ends.
The distinctive claim of Kantian value realism is that assessment
in terms of achievement is on a different basis. The value of our

24

G. 4: 430.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

198

ALISON HILLS

projects depends on their contribution to our rational nature

whether or not we have chosen to make achievement one of our
ends; for rational nature alone is unconditionally valuable.

7. Why Kantian value realism?

According to Kantian value realism, Kants ethical theory does
not include an idiosyncratic account of the nature of value, that
is, of what it is to be valuable. It is a mistake to associate his theory
with the rejection of value realism. It is better to interpret it as
a particular version of value realism, as committed to facts about
what is valuable that are not constitutively dependent on the
choices of rational agents.
Rather, the distinctive claims of Kants ethical theory concern
what is valuable. It is distinguished by the fundamental role played
by the value of rational nature. All non-moral ends are valuable
insofar as pursuing those ends contributes to the exercise of our
rational nature.
Korsgaard defend Kants ethical theory by arguing that Kantian
constructivism was a better theory of the nature of value than
value realism. By rejecting Kantian constructivism in favour of
Kantian value realism, have we given up the most powerful reason
to accept Kants ethical theory?
The benefits of Kantian constructivism were for the most part
an illusion. It was questionable whether the commitments of
Kantian constructivism to the correctness of the categorical imperative procedure for testing maxims were less metaphysically or
epistemologically problematic than the commitments of realism
to the existence of moral facts. And the conferral model, which
claimed that value could be created from rational choice, was
deeply flawed and ultimately incoherent. Kantian value realism is
more appealing than Kantian constructivism, since it at least gives
a coherent account of non-moral value.25

25
It is worth noting that value realism in the sense in which I use it here is exclusively
concerned with the relationship between value and practical reason, namely, whether x is
good is constitutively dependent on whether x has been chosen by a rational agent. This
may be compatible with a number of different metaethical theories about the meaning of
statements like X is good or We have reason to promote X. The commitments of value
realism in this sense need not be epistemologically or metaphysically outlandish.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

KANTIAN VALUE REALISM

199

When we evaluate Kants ethical theory against its rivals, we

should not primarily be concerned with issues about the nature of
value and its relationship with our procedures for deciding ethical
questions. Instead we will have to tackle substantive problems in
normative ethics. It is beyond the scope of this paper to address
these large and difficult questions here. But I will very briefly
suggest some of the strengths and weaknesses of Kantian value
realism.
Kants ethical theory will appeal to those who hold in high
regard our rational nature: our ability to act for reasons that we
can use to justify what we do to others. It focuses on the appropriate treatment of rational agents, which we are to treat as having
dignity, not mere price. We are required to respect them, to reject
paternalism and to forbid coercion and deception even if good
consequences (in terms of happiness) would follow. We are also
required to respect ourselves as rational agents.
Kantian value realism has problems, however, in accounting
for the significance of pleasure and of suffering, particularly the
suffering of creatures that are not rational (or not fully rational),
such as animals.26 It is questionable whether the theory can give
an adequate account of the value of the environment, other than
as something that sustains the lives of rational agents.27 And
finally, it is not at all clear whether ultimately the value of art, of
human emotion and human relationships can be explained in
terms of the value of rational nature or the choices of rational
agents.
I did not intend to offer a defence of Kants ethical theory here.
Instead, I have attempted to show that Kantian value realism is a
better interpretation of Kants theory than constructivism, that
the theory is coherent, that there is good textual evidence that
Kant accepted it, and that it is worthy of further study. Kant offers
us an account of the ideal moral community, a society of rational
agents who treat each other with respect, which many people find
extremely appealing. He offers one of the most compelling alternatives to consequentialist and welfarist moral theories. And he

26
For an attempt to defend the moral status of animals in Kantian ethics, see Korsgaard
(forthcoming).
27
Of course some people do think the environment is valuable only in this way. See
ONeill (1998) and Wood (1998) for a Kant-inspired discussion of the value of the
environment.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd

200

ALISON HILLS

may be right that we should all strive to bring about the kingdom
of ends.28
St Johns College
Oxford OX1 3JP
alison.hills@sjc.ox.ac.uk
References
FitzPatrick, W. J. (2005). The Practical Turn in Ethical Theory: Korsgaards Constructivism, Realism, and the Nature of Normativity Ethics 115, pp. 651691.
Gaut, B. (1997). The Structure of Practical Reason in eds. B. Gaut and G. Cullity Ethics
and Practical Reason. Oxford: OUP.
Guyer, P. (1996). The Value of Agency Ethics 106, pp. 404423.
(1998). The Value of Reason and the Value of Freedom Ethics 109, pp. 2235.
(2000). Kants Morality of Law and Morality of Freedom in his Kant on Freedom, Law
and Happiness. Cambridge: CUP.
Herman, B. (1993). The Practice of Moral Judgment. Cambridge, Mass: Harvard UP.
Hills, A. (2004). Is Ethics Rationally Required? Inquiry 47.1, pp. 119.
(forthcoming). The Beloved Self: Morality and the challenge from Egoism. Oxford: OUP.
Hurka, T. (1993). Perfectionism. Oxford: OUP.
Kant, Groundwork of a Metaphysic of Morals (G.) trans. H. J. Paton (London: Routledge,
1991). Page references cite the volume and page number of Kants gesammelte Schriften
(published by the Preussische Akademie der Wissenschaften, Berlin: W. de Gruyter, 1902).
Kant, Critique of Practical Reason (C.Pr.R.), trans M.J. Gregor, Cambridge: CUP, 1997; page
references cite the volume and page number of Kants gesammelte Schriften (published by
the Preussische Akademie der Wissenschaften, Berlin: W. de Gruyter, 1902),
Korsgaard, C. M. (1996a). Creating the Kingdom of Ends. Cambridge: CUP.
(1996b). The Sources of Normativity. Cambridge: CUP.
(2003). Realism and Constructivism in Twentieth-Century Moral Philosophy. The
Journal of Philosophical Research APA Centenary Supplement: 99122.
Korsgaard, C. M. (forthcoming). Fellow Creature: Kantian Ethics and our duties to
animals in The Tanner Lectures on Human Values, ed. by Grethe B. Peterson. Salt Lake
City: Utah University Press, Volume 25/26.
Marras, A. (2003). Audi on Substantive vs Instrumental Rationality Philosophy and Phenomenological Research 67, pp. 194201.
ONeill, O. (1989). Constructions of Reason. Cambridge: CUP.
(1998). Kant on duties regarding nonrational nature Proceedings of the Aristotelian
Society Supplementary Volume 72, pp. 211228.
Regan, D. H. (2002). The Value of Rational Nature Ethics 112, pp. 267291.
Shafer-Landau, (2004). Moral Realism. Oxford: OUP.
Wood. A. W. (1998). Kant on duties regarding nonrational nature Proceedings of the
Aristotelian Society Supplementary Volume 72, pp. 189210,
(1999). Kants Ethical Thought. Cambridge: CUP.

28
I would like to thank the AHRC for an award of research leave, during which time this
paper was written.

2008 The Author

Journal compilation 2008 Blackwell Publishing Ltd