MD

Metaethics and Its Discontents: A Case Study of Korsgaard
(Version 1.7)
DRAFT: PLEASE DO NOT QUOTE OR ATTRIBUTE WITHOUT PERMISSION
Nadeem J. Z. Hussain
Department of Philosophy
Stanford University
Stanford, CA 94305
Nadeem.Hussain@stanford.edu
(650) 725-9690
Nishi Shah
Department of Philosophy
Amherst College
Amherst, MA 01002
npshah@amherst.edu
(413) 374-1229
Version 1.8, Rev. 3 July 24, 2005

Metaethics and Its Discontents: A Case Study of Korsgaard
Realism and expressivism are both true in their way. But establishing that realism is true
in that sense is not the end of moral philosophy, in either sense of “end”: it is only the
beginning.1
Expressivism, I believe, is like realism also true after all, and also in a way that makes it
boring.2
When we think of the subject this way, we will not be inclined to think that there is a
difference between doing “meta-ethics” and doing “normative” or practical ethics. The
attempt to specify the meaning and reference of an ethical concept will point fairly directly
to practical ramification. This represents another way in which constructivists break with
the platitudes of twentieth century ethics.3
1. Introduction
These striking passages from Christine Korsgaard express one way of giving shape to an
otherwise inchoate discontent with traditional metaethics that has been growing in recent years
among philosophers with otherwise widely differing views. The discontents we have in mind do
not simply champion a competitor to the likes of noncognitivism or realism; they disapprove of
supposed presuppositions of the existing debate: an indefensible distinction between metaethics
and normative ethics, a misplaced priority of the theoretical over the practical, outdated
conceptions of mind and language, or a reductive, scientistic naturalism. Metaethics somehow
misconceives the task of understanding ethics. However, it would be inaccurate to say that our
discontents simply want metaethics to stop. There are acceptable versions of some of
metaethics’ concerns though in altered form. The point is not to generate a new theory within
metaethics, but, in a sense that will need articulation, to go beyond metaethics by replacing it
1
Christine Korsgaard, "Realism and Constructivism in Twentieth-Century Moral
Philosophy," The Journal of Philosophical Research APA Centennial Supplement (2003): 118.
2
Korsgaard, "Realism and Constructivism," 122 n. 49
3
Korsgaard, "Realism and Constructivism," 121 n. 44.
Version 1.8, Rev. 3 1 July 24, 2005

2
with something that is more responsive to the concerns that really animate, or should animate,
the search for the “sources of normativity”. This desire to transform or reform traditional
metaethics is spurred by the recognition that all existing metaethical theories fail to vindicate
some central feature of our moral practices. Rather than conclude that our moral practices need
reform, these discontents draw the opposite conclusion that the way philosophers practice
metaethics must change.
Such discontent is widespread though not always published. It is also hard to engage with
since the differences with traditional metaethics go deep enough that it can feel as though two
different paradigms are talking past each other. Our goal here is to begin this engagement by
focussing on the work of one philosopher, Christine Korsgaard. She is widely taken as someone
who has expressed, and argued for, one type of discontent. It shares much with other types, both
Kantian and non-Kantian, and thus our discussion, focussed though it is on her work—indeed, on
only parts of her work—has broader implications. Our official thesis is perforce restricted to our
case study: we argue that Korsgaard fails to go beyond metaethics.4
In our experience, it immediately strikes some as quite implausible that Korsgaard is trying
to do any such thing. It is worth beginning with a brief consideration of this response. The
response comes in two conflicting forms. First, that Korsgaard is not engaging with
contemporary metaethics at all; she is just doing something else. Second, that Korsgaard is just
trying to present another theory within traditional metaethics. She is not a discontent in our
sense; no doubt she sees something wrong with the theories on offer, but she has no aspirations
to, somehow, go beyond metaethics.
Neither form of the response can be correct though—at least not obviously so. First,
Korsgaard repeatedly and explicitly sets up her own account of normativity in contrast with
“substantive realism” or “dogmatic rationalism”, two labels that she uses more or less
interchangeable. She insists that the “realism” her view is meant to be contrasted with is not the
general normative position that there are correct answers to normative questions about what we
4
This is consistent with her having made important contributions to the existing debate in
metaethics. We think that many of her claims about normativity are ones for which she has
given compelling arguments, that these are claims which might well show that some versions of
traditional positions in metaethics will not work, and that these are claims which any plausible
theory of metaethics will have to accommodate. However, these claims do not comprise a new
position within metaethics, nor a position beyond metaethics.
3
should do––a position that merely marks out a contrast with nihilism.5 Rather, she emphasizes,
the realism she wants to oppose is, as she puts it, “substantive moral realism”. It is “the view
that there are answers to moral questions because there are moral facts or truths, which those
questions ask about”.6 According to substantive realism, moral requirements must be given
“some sort of ontological foundation, by positing the existence of certain normative facts or
entities to which moral requirements somehow refer”.7 Substantive realism thus is clearly a view
identified by specific metaphysical and semantic commitments. She also repeatedly sets up her
own view in contrast with a position, usually associated with Hume, which she calls
“empiricism”.8 It is harder to pin down precisely what she takes this position to involve, but,
perhaps it is fair to say, it seems to involve an empiricist epistemology, a naturalist ontology, and
a methodological tendency towards reductions to this naturalistic ontology.
In any case, it seems clear that Korsgaard identifies her opponents by commitments
traditionally used to identify metaethical theories and not theories in normative ethics. Thus
whatever else she is doing it seems clear that she takes her own position to somehow provide a
superior alternative to such metaethical theories. However, in order to demonstrate the
superiority of her position over traditional metaethical theories she must engage with metaethics
to some extent. These theories were articulated to answer certain questions that traditional
metaethics posed. If her theory is better, then, at the cost of absurdity, it must be better at
answering the questions metaethics has traditionally asked, or, as our language of going beyond
metaethics has meant to suggest, better at answering questions that, in some sense, metaethics
should be asking. Her project thus cannot simply bypass metaethics altogether.
Second, passages such as the ones we started the paper with make it clear that Korsgaard
does not want to simply present another theory within metaethics. In addition to those passages,
there are other passages in which she claims that Kantian constructivism represents an alternative
to the view of normative concepts that is shared by contemporary cognitivists and
noncognitivists alike.9 According to Korsgaard, once we accept the correct account of normative
5
Christine Korsgaard, The Sources of Normativity (Cambridge: Cambridge University Press,
1996) 35.
6
Korsgaard, Sources 35. Emphases in original.
7
Christine Korsgaard, "The Normativity of Instrumental Reason," in Ethics and Practical
Reason, ed. Garrett Cullity and Berys Gaut (Oxford: Oxford University Press, 1997), 218.
8
Korsgaard, "Normativity of Instrumental Reason," 218-19.
9
Korsgaard, "Realism and Constructivism," 106.
4
concepts we will see how the existing positions have some truth to them. We do not claim that
the full interpretation of these passages, and their contexts, is obvious. Indeed much of the
burden of this paper is an attempt to articulate how Korsgaard understands metaethics, and the
positions within metaethics, in order to provide an interpretation of such statements.
Nonetheless, these statements do constitute prima facie evidence for, and provide some content
to, our claim that Korsgaard intends, both here and elsewhere, to go beyond metaethics. Within
contemporary metaethics the distinction between cognitivism and noncognitivism is indeed
usually taken to provide a complete partition of the space of theories.10 Thus a rejection of both
types of theories, and a rejection of some purported shared assumption, is just the kind of
wholesale rejection of contemporary metaethics that we want to mark by our talk of going
“beyond” metaethics. Our use of “beyond” is further justified by the kind of charge apparently
laid against realism and expressivism in the quotes above. The charge is not simply of
falsehood—the kind of charge that a normal, straightforward competitor would make. Indeed,
Korsgaard grants that the traditional theories are, in some sense, true. Rather, her charge is that
the full range of metaethical theories, and the assumptions about normative concepts shared by
them, presumably, somehow miss the point of moral philosophy. As the quote about “the
platitudes of twentieth century ethics” makes clear, the very distinction between metaethics and
normative ethics is part of the approach to moral philosophy that Korsgaard thinks we should
reject. Thus when Korsgaard argues against, say, realism, and presents her Kantian view as an
alternative to realism, the kind of rejection of realism involved is not just the kind of rejection
that, for example, a traditional non-cognitivist would argue for. It is not simply the argument
that realism is false and that the presented alternative does a better job of answering the same
questions as realism was intended to. Something clearly more radical is intended here. The
Kantian alternative is not merely another theory within metaethics; rather, the very
presuppositions of metaethics are to be rejected—indeed the very idea, apparently, that there is
some identifiable, distinctive domain of metaethics is to be rejected. This, then, still roughly, is
what we mean by the aspiration to go “beyond” metaethics.
10
Fictionalist accounts are perhaps sometimes taken as an exception. We do not think that
this possibility shows that Korsgaard’s target is not all of metaethics. In any case, fictionalism
does not really provide an alternative to the traditional cognitivist and noncognitivist theories
either (see Nadeem J. Z. Hussain, "The Return of Moral Fictionalism," Philosophical
Perspectives 18, no. 1 (2004)).
5
We are not the only ones to have noted this aspiration of constructivism and it is not
restricted to Korsgaard’s version. Darwall, Gibbard and Railton in their survey of “fin de siècle”
ethics, interpret at least some strains in constructivism as aspiring, as they put it, to “have
sidestepped traditional metaethics” or to “render traditional metaethics obsolete”.11 In their very
brief survey, it seems that in the end they despair of explaining how constructivism could really
do any such thing, but it is just this apparent strain to sidestep or render obsolete that we are
pointing to with the language of “going beyond”. Finally, we think, but will not be able to argue
here, that this desire to go beyond metaethics is a shared by the wider group of discontents we
alluded to at the beginning.
The central task of this paper, then, is that of interpreting Korsgaard’s texts to see how her
own position is supposed to contrast with both realism and expressivism, both cognitivism and
noncognitivism, both dogmatic rationalism and empiricism, and thus go beyond metaethics. We
defend our thesis that she fails to go beyond metaethics by arguing that, in fact, she fails to
distinguish her position from either realism or expressivism and fails to provide an account of
normative concepts that shows what is wrong with such theories. If successful, we will have
provided a challenge to those who would like to leave behind metaethics to tell us exactly what
the mistake is supposed to be.
2. Metaethics and Normative Ethics
In order to go beyond metaethics, Korsgaard has to engage with traditional metaethical

theories at least to the following extent: she needs to show us what is wrong with both realism
and expressivism. Even if realism and expressivism turn out to be in some sense true, Korsgaard
thinks that they are also failures in some sense. Furthermore, it cannot just be that they fail for
some extrinsic purpose, so to speak—as a screwdriver may fail when used to pound nails in to
wood—rather they fail for the task for which they were, or should have been, intended. Or
rather, perhaps, they fail at this task once this task is properly articulated. Her own position thus
11
Darwall, Gibbard and Railton also note Kantian constructivism’s apparent aspiration to
“have sidestepped traditional metaethics” or to “render traditional metaethics obsolete”
also suggest that constructivism may be trying to sidestep traditional metaethics though they
seem in the end to despair of explaining how it could really do that (Stephen Darwall, Allan
Gibbard, and Peter Railton, eds., Moral Discourse and Practice: Some Philosophical
Approaches (New York and Oxford: Oxford University Press, 1997) 14-15.
6
has to involve a deep criticism of realism and expressivism and, a fortiori, must differ from
either realism or expressivism.
Our general strategy will be to argue that what are supposed to be claims that conflict with
realism or expressivism in fact fail to do so. We will rarely attack the arguments for these claims.
What we will attack instead is the argument against realism and expressivism—and thus,
presumably, metaethics—based on these claims. These claims (and the arguments for them of
course) fail to undermine realism or expressivism because Korsgaard fails to show that they
actually conflict with realism or expressivism in the first place. They fail to conflict because
though they may appear to be metaethical claims they in fact are not obviously so and indeed are
most charitably interpreted as either claims within normative ethics or normative psychological
claims in the philosophy of action, claims compatible with several different metaethical accounts
of those same claims.
This strategy may appear to run the risk of ignoring Korsgaard’s intention to undermine the
distinction between normative ethics and metaethics. However, our strategy is compatible with
taking this intention seriously. Korsgaard cannot undermine the distinction simply by assuming it
away.12 The distinction is to be undermined as a consequence of her arguments against realism
and expressivism. An argument for the claim that her view is different from, and conflicts with,
either realism or expressivism cannot presuppose that the distinction has been undermined.
Indeed, we will argue that part of the explanation for why she fails to go beyond metaethics
is that she, like many other discontents, fails to appreciate all the ramifications of the traditional
distinction between normative ethics and metaethics. Blame should be shared here by the
practitioners of traditional metaethics; the distinction is usually briefly mentioned and often
accompanied by a note of warning that it may not in the end be defensible. This belies what we
think is the philosophical reality, namely, that there is a family of important distinctions behind
the use of the traditional distinction between metaethics and normative ethics—distinctions
which we ignore at our philosophical peril. We begin therefore with an attempt to rearticulate
the distinction between metaethics and normative ethics.
12
This claim should be qualified. One could posit the rejection of a distinction and then show
how the rejection of the distinction in fact leads to a better theory. This would give reason to
reject the distinction. The sense however in which the resulting theory is a better theory, and a
fortiori what makes it a different theory, cannot then in turn presuppose the distinction.
7
We take normative ethics to be continuous with our everyday human practices of making
normative judgments. As Shelly Kagan puts it, normative ethics “involves substantive proposals
concerning how to act, how to live, or what kind of person to be”. If anything distinguishes it
from our everyday practice, it is that it “is concerned with stating and defending the most basic
moral principles” that, given further input about specific circumstances, tell us how to act, how
to live, or what kind of person to be in those specific circumstances. These basic moral
principles thus “systematize” and, in one sense, ground our more specific moral judgments.
They ground these specific moral judgments by showing that they can be derived from more
general moral judgments, the general moral judgments that just are these basic moral
principles.13 The radical claim that no systematization is possible, expressed perhaps by some
forms of particularism, is the limiting case of theories in normative ethics.
Metaethics is the second-order investigation of the practice of making moral judgments—the
everyday practice and the more philosophical practice of normative ethics. Metaethical
investigations do not attempt to figure out which moral judgments we should make, but rather try
to understand what is going on when we engage in the practice of making moral judgments. The
point of metaethics is to give an account of what it is to think, and express, a normative thought,
not to tell us which normative thoughts to think.
This description of the metaethical task allows us to see the hidden unity behind the lists of
“second-order” questions that are usually used to identify the domain of metaethics.14 We need
to know what kinds of mental states are being had when we think moral thoughts and if these
states have content, then we need to know what the content is—questions of moral psychology.
We express these mental states in language and so an account of what the content is will have to
fit with an account of “semantic function of moral discourse”—questions of meaning.15 If the
function is to state facts, then a full metaethical account will state what kind of fact these are and
how we can know them—questions of metaphysics and epistemology.
Non-reductive realism and non-cognitivism are examples of positions that give competing
answers to these questions. A non-reductive realist might claim, for example, that the moral
judgment that x is right expresses a belief that x has a normative property. He might accept a
13
Shelly Kagan, Normative Ethics (Boulder: Westview Press, 1998) 2.
14
Consider, for example, Alexander Miller, An Introduction to Contemporary Metaethics
(Cambridge: Polity Press, 2003) 2 and Kagan, Normative Ethics 4.
15
Miller, Contemporary Metaethics 2.
8
referential semantics according to which moral language is about moral facts constituted by non-
natural normative properties. And he might hypothesize that we access these facts through some
faculty of non-empirical, rational intuition. Non-cognitivists, on the other hand, usually reject a
referential semantics for moral terms. They claim that moral judgments do not express truth-
evaluable beliefs in normative facts, but express non-depictive motivational states such as
desires, preferences, or emotions. Non-cognitivists are therefore free to accept an ontology
restricted to natural properties. They do need to provide some account of what we are doing
when we say, for example, that we know that x is wrong; however, they will deny that the
relevant epistemological account will involve a story of how we are tracking a sui generis realm
of normative facts.
Now, as we have already noted, the usually much briefer statements of the distinction
between normative ethics and metaethics are also usually followed by a general note of warning
that the distinction may not in the end be defensible. Take Shelly Kagan again. He follows the
statements above about normative ethics with the usual identification of metaethics by a list of
questions. He then goes on to argue that in fact there is a “continuum rather than a sharp line”
between normative ethics and metaethics.16 We are willing to grant in principle that there is a
continuum of some kind, but we think that it is crucially important to be very clear about the
kind of continuum present. It is precisely this clarity that we will try to provide next. The
fundamental thrust of our discussion, and the conclusion that we will need for the rest of the
paper, is that one can have a recognizable theory in normative ethics—one can have made
significant progress in normative ethics—without having covered much ground at all at the
metaethical end of the continuum. This is important for us precisely because we want to claim
that Korsgaard’s views, though recognizable views in normative ethics, do not commit
themselves sufficiently on metaethical matters to conflict with, and thus be alternatives to,
realism or expressivism.
Let us begin then with Kagan’s arguments for a continuum. Kagan focusses on the
justification of moral claims including the very general ones of concern to normative ethics:
When doing normative ethics we try to defend and justify substantive normative claims.
But to do this, obviously enough, we have to have views about what it is you need to do to
provide a moral claim with a good defense. That is, in doing normative ethics we will be
16
Kagan, Normative Ethics 5-6.
9
presupposing some kind of account—either a developed one or at least a working

understanding of one—of justification in ethics. But the topic of justification in ethics is
itself one that actually belongs to metaethics. In short, doing normative ethics requires
having views about metaethical issues. (5)
Put this abstractly, Kagan invites potential confusion. Consider the kind of justifications we
often appeal to when engaged in normative ethics. We often appeal to intuition: we ask what
judgment a proposed moral principle would require for a particular situation and then ask
ourselves whether the judgment intuitively seems the right one to make. If it does not, then this
is often taken as a defeasible reason to reject the moral principle. Our intuitions, or our
willingness to override intuitions, though seem sensitive to evidence that a moral principle does a
good job of systematizing our other particular judgments or leads us to think that we now have
clear intuitions about a case that puzzled us before. Now it is certainly right to say that in
engaging in such a process of “reflective equilibrium” we are presupposing that this process
yields justification. We may thus be presupposing that the correct metaethical account contains
an epistemology that vindicates the claim that this process yields justification.
It is unhelpful though, we think, to describe the acceptance of such presuppositions as
“having views about metaethical issues” (5). To see this, it helps to note that everyday moral
claims involve this presupposition as well. If I sincerely and literally assert that killing John was
wrong, then I am presupposing, it might plausibly be argued, that the correct metaethical account
will not be an error theory. There are three things to notice about this presupposition. First, for all
that has been said, this presupposition is compatible with the full range of non-error-theoretic
metaethical theories both cognitivist and non-cognitivist. Second, obviously, the presupposition
is compatible with having no specific positive metaethical account and no arguments for any
metaethical view. Third, the presupposition is merely that. It is defeasible and may turn out to
be false once we actually look at the balance of metaethical arguments. Thus the sense in which
such a presupposition may involve “having views about metaethical issues” is an extremely thin
one. To put the point starkly, imagine the look of puzzlement, or worse, on the astronomer’s
face were one of us philosophers to say, “Yes, I too have views about astronomical issues: I
think there are stars”.
Returning to the more complicated case of reflective equilibrium justifications of moral
principles does not change anything significant. Again we may well be presupposing the falsity
of error theory and we are perhaps presupposing systematicity in the normative realm and a

10
tendency of our intuitive judgments to lead to the correct moral judgments. However, again,
there are versions of the full range of non-error-theoretic metaethical theories both cognitivist
and non-cognitivist that are compatible with these presupposition. These presuppositions are
again compatible with having no positive metaethical account and are again defeasible.
Rather than thinking of these presuppositions as constituting a metaethical view, it makes
more sense to think of them as features of the practice of making moral judgments that the
correct metaethical account needs to explain. Our practice of making moral judgments—
including normative ethics—presupposes, arguably, that these judgments are true and that they
are justified. That is a deep feature of our practice and one that metaethics has to explain
whether or not the metaethical account vindicates them.
Kagan though seems to think that he has another, and better, argument for his continuum
claim. When “we attempt to provide a basis or foundation for the substantive moral claims of
normative ethics” we will
inevitably ... appeal to larger metaethical conceptions of morality’s purpose and point.
That is, in the course of defending a given theory about the foundations of normative
ethics, when we try to explain why it is that the various features of that theory should seem
attractive and plausible, inevitably the claims we make will themselves simply be
metaethical claims about the nature of morality. At a deep enough level, normative ethics
does not merely draw upon metaethics—it simply becomes metaethics. (6)
Kagan makes two errors here. First, not every claim about the nature of morality is a metaethical
claim in any interesting sense. Consider the claim that the point of morality is fundamentally the
to promote the welfare, well-being or happiness of human beings. This could, plausibly, be a
premise in an argument against the claim that trees have intrinsic moral rights and thus could be
part of an argument for a particular set of basic moral principles. Furthermore, such a claim
defeasibly constrains metaethics—in accepting it we would, in the sense discussed above,
presuppose a metaethical account that was compatible with it. However, we are certainly not
very far along the path of explaining what it is to think a normative thought. The fundamental
reason for this is that the claim is itself a normative claim. No doubt it is a normative claim
about a range of normative claims (e.g. moral claims should advocate the promotion of human
welfare), but it does not tell us whether, for example, in making a moral claim I am expressing
the mental state of belief.

11
Second, for all that has been argued so far, there are metaethical theories that explicitly do
not think that, in one important sense, ultimate justification requires drawing upon metaethics, let
alone becoming metaethics. Such metaethical theories will be particularly important for the
purposes of our argument with Korsgaard in this paper. It helps, though, to start at the other end
of the spectrum with the kind of metaethical theory that comes closest to the view that at a deep
enough level, in some sense, normative ethics “becomes metaethics”. Take a reductive,
naturalist moral realism, such as the view that the semantic function of the concept of good is to
refer to the property of pleasure. On such a view moral claims turn out to be, in the end, claims
about the instantiation of the naturalistic property of pleasure, a property that is usually taken to
be accessible by empirical means. Given such a view, a final defense of a moral claim might
conceivably point to the naturalistic reduction provided by the metaethical theory and then
appeal to the relevant empirical data to defend the moral claim. Or, perhaps more plausibly, the
metaethical theory might be appealed to in some way as a “deep” defense of the method of
reflective equilibrium; that is, the reduction would be combined with the claim that reflective
equilibrium is, despite surface appearances, an essentially empirical method of discovering these
facts.
What is important to note for our purposes though is that the justification of a normative
theory does not have to proceed in this way. Non-cognitivism, for example, entails that
appealing to a non-cognitivist account does not support any particular set of moral principles
over any other.17 Any defense of moral principles will always require explicitly making, or
presupposing, further normative claims. The non-cognitivist account tells us what we are doing
when we make these further normative claims but does not itself tell us which ones to make. For
the non-cognitivist there is no going from “is” to “ought”—there is no stepping outside the circle
of normative claims. Therefore, given that the non-cognitivist metaethical theory is a purely
descriptive theory, it does not help us defend any particular normative claim over any other.
Now, again, there is still a weak continuum point that, we suppose, remains. If one’s
opponent appeals to, say, reductive realism on behalf of some set of normative claims, then one
might appeal to non-cognitivism—one might move to metaethical ground—to defeat this appeal.
As always with justification much will turn here on how hard one is being pressed and by whom
17
Whether as a matter of psychological fact it tends to undermine our commitment to
morality is another matter. The sense of “support” here is the sense in which a premise supports
a conclusion in an argument.
12
and in what context. In many contexts, normal “reflective equilibrium” arguments will be
treated by participants in moral discussion, whether everyday discussion or the discussions
constituted by articles in philosophy journals, as sufficient justification for moral principles. The
fundamental point is that one can do vast stretches of normative ethics without settling
metaethical issues. And the fact that one may have reached a plausible position in normative
ethics may leave open important metaethical issues. We must not allow ourselves to forget this
in our rush to accept some plausibly weak version of the continuum point.
Finally, let us look at a line of thought that Kagan does not explicitly bring up but that, we
suspect, is also often thought of as undermining the distinction between normative ethics and
metaethics. It too has to do with the matter of justification though here the point is not simply
the matter of whether we can justify some particular moral claim; justify, for example, the claim
that killing S is indeed wrong. Rather it has to do with placing morality within practical reason,
explaining whether we have reason to do what morality demands, and if so, whether these
reasons are derived from another branch of practical reason. There are various ways of
proceeding here. One could point to specific substantive reasons for being moral, say, reasons of
prudence. One might also argue for the necessity of certain claims: one cannot be an agent, say,
unless one accepts the principles of morality.18 The amoralist is inconceivable, someone might
argue.19 It might be thought that since these arguments, if correct, answer our most fundamental
questions about morality (e.g., do we have reason to be moral?) without appealing to traditional
metaethical theories, there must be something wrong with the traditional distinction between
normative ethics and metaethics.
There are two ways of taking such claims of necessity, either of which leave plenty of room
for metaethics. First of all, we might think that these necessity claims are merely descriptive.
The claim that an agent is someone who accepts that he should act morally is on all fours with,
say, the claim that humans are mortal. Understood this way these are not conclusions of
normative ethics though they might be deployed in arguments for normative conclusions. I
might try to convince you to accept the principles of morality by either singing the praises of
being an agent or reminding you that you already think that you are an agent. Here the role the
18
We grant that the word “accept” here requires much explication. Some explication occurs
immediately below; however, we return in more detail to such questions in our discussions in
later sections of the details of Korsgaard’s view.

13
descriptive claim plays in the arguments of normative ethics is essentially no different from the
role many other run-of-the-mill descriptive claims play, (e.g., actions of type x tend to cause
pain).
Since these descriptive claims about agency are claims of necessity, we will feel the pressure
to explain the necessity. Is the necessity, for example, the result of conceptual relations? For
example, it might be argued that it is a conceptual truth about agents as such that an agent
accepts that he should act morally. Explaining the necessity in this way is indeed a second-order
task which, since the content that it is claimed that agents necessarily accept is normative, might
well be a task for metaethics. As far as normative ethics is concerned though these necessary
descriptive claims can be treated just like any other necessary descriptive claim. That is, it is of
no particular relevance to the questions of normative ethics how this necessity is to be explained.
A clear and motivated division of labour between normative ethics and metaethics can still be
maintained.
Secondly, we might though think that these necessary claims are necessary normative claims.
The claim that one cannot be an agent unless one accepts the principles of morality should be
understood as the claim that an agent ought to accept the principles of morality and this “ought”
is inescapable—there is nothing an agent can do to have this “ought” not apply to it.20 The
necessity here is what Fine calls “normative necessity”. This necessity is shared by many other
claims made within normative ethics: maximizing utility is right, killing innocents is wrong, and
so on. Understood this way the claim about agents is a normative claim straightforwardly part of
normative ethics.
The division of labour though is still clear. Metaethics takes up the task of providing an
account of what we are doing when we make such a normative claim and, other things equal,
will attempt to provide an interpretation of the claim that squares with the (apparent) necessity.
Is the necessity the result of the semantics of normative claims? Is it, for example, to be
explained by conceptual relations? Is it to be explained by reductions of the normative to the
non-normative? The apparent necessity, of whatever kind, possessed by the claims is part of the
data the metaethicist must account for or, at theoretical cost no doubt, explain away.
19
Our thanks to an anonymous referee for a different paper for highlighting the importance
in a similar context of the example of the amoralist.

14
Again, just to emphasize the point, if we are convinced of the necessity of certain normative
claims in normative ethics, then we will feel the defeasible pressure to adopt a metaethical
interpretation of these normative claims that can vindicate this necessity. On the other hand, we
can imagine the admittedly unlikely scenario where the best metaethical theory developed to
account for what initially appeared to be contingent normative claims entails that some of these
claims are necessary. Even with these possibilities born in mind, the distinction between
normative ethics and metaethics remains clear. One can make such claims of necessity without
having done much at all towards providing a metaethical account.
Finally, we do count views that argue that there is no way of getting outside of normative
thought to explain it, and therefore that no answers to metaethical questions are possible, as
doing metaethics. However, this type of quietism, which claims that no metaethical theories are
possible is not equivalent to merely failing to state a metaethical position. One might pursue
normative ethical tasks while ignoring metaethical ones, leaving such questions for others to
answer. This acceptance of a division of philosophical labor certainly would not commit one to
the quietist claim that metaethics is impossible. Quietism is a bold position in need of
justification, whereas the decision to pursue normative ethical questions instead of metaethical
ones needs no philosophical defense. The point of metaethics is to give an account of what it is
to think a normative thought or to show that such an account is impossible, not to tell us which
normative thoughts to think or to point out which normative thoughts we cannot help but think.
3. Non-cognitivism
Let us return then to Korsgaard. We have argued elsewhere that Korsgaard fails to
distinguish her position from non-reductive moral realism, what she calls substantive realism.
We argue there that the fundamental explanation for this is a confusion between the tasks of
normative ethics and those of metaethics. A main thread of her argument against the realist is
that the realist fails to adequately place morality with respect to practical reason in the sense
discussed in the previous section. Even if these arguments were successful, and we argue that
they are not, she would not have a produced what she claims to produce, namely, a competitor to
20
Cf. Fine Kit Fine, "The Varieties of Necessity," in Conceivability and possibility, ed.
Tamar Gendler and John O'Leary-Hawthorne (Oxford: Clarendon Press, 2002) on “normative
necessity”.
15
substantive realism. She would instead have presented us with a normative theory that is
compatible with non-reductive realism. Or so we have argued elsewhere.21
It would be a mistake to conclude from those arguments that Korsgaard and the non-
reductive realist have the same metaethical position. The fundamental problem is that her claims
are compatible with different metaethical positions or interpretations. To drive this point home
we present in this section a non-cognitivist reading of the position Korsgaard expresses in the
“The Normativity of Instrumental Reason”. We think that the lessons learnt here can then be
applied to her proposals elsewhere; however, we do not attempt that there. We argue that a
noncognitivist interpretation is compatible with the details of her claims, fits quite naturally with
broader themes in her work such as the importance of the "first-personal", and potentially
provides responses to legitimate metaethical worries about non-reductive realism that may be
lurking behind her own ill-formed objections to that view.
Korsgaard however officially rejects noncognitivism in favour of her own "constructivist" or
"procedural realist" account of normative concepts. As we will see in the next section, her
rejection of noncognitivism seems to rest on misunderstandings related once again to the failure
to distinguish between different questions about the normative. We then finally turn to an
assessment of whether "constructivism" and its account of normative concepts succeeds, as it is
apparently supposed to, in undermining the distinctions between cognitivism and noncognitivism
and normative ethics and metaethics. We conclude that in its currently undeveloped state, it does
not and that we are thus in the end left with no distinctive Korsgaardian position that can go
beyond metaethics.
3.1 The Non-Cognitivist Interpretation of “The Normativity of Instrumental Reason”
Korsgaard takes as her topic the “normative foundation” of the instrumental principle. She
suggests that she is searching for the an “explanation” of its “normative force”, an account of the
normativity of the instrumental principle, or an account of “what gives the instrumental principle
its normativity”. The interpretive problem we face immediately is that these expressions are
susceptible to two different readings. We could read them as suggesting that she intends to
provide us with either a metaethical account of the normativity of instrumental reason or with an
21
Nadeem J. Z. Hussain and Nishi Shah, "Misunderstanding Metaethics: Korsgaard's
Rejection of Realism," Oxford Studies in Metaethics (2005).
16
account within normative ethics—as we will argue in a moment, indeed two different kinds of
accounts within normative ethics.
It helps here to start with the claim that “no particular thing can be barely valuable or
right”.22 We can draw a distinction between what makes an action wrong, on the one hand, and
what constitutes the normativity or what the property of being normative itself is, on the other.
Thus the fact that brushing my teeth regularly will reduce plaque may make brushing my teeth
good (for me); however, we do not want to claim, presumably, that the property of goodness
itself just is the property of reducing plaque. Or to use a more theoretical example from
normative ethics, I could, perhaps like G. E. Moore, think that what makes states of affairs good
is that they are pleasurable. The good-making property thus might be naturalistic or
psychological. I might deny though that the property of goodness itself is the property of being
pleasurable—it is a sui generis non-naturalistic normative property.23 The two properties are
related but not by identity. One brings about the instantiation of the other, but does not
constitute it. Such right- or good-making properties are examples of what Kagan calls
“normative factors” and as he emphasizes much of the investigation in normative ethics about
normative factors can be, and is, carried out without asking more foundational questions.24
Now Korsgaard could argue for a reductive view that denied this distinction; however, till
she has done that the distinction remains and with it an ambiguity in the expressions she uses to
describe her project. There are perfectly understandable senses of these expressions according to
which one might well say that one is explaining why one ought to brush one’s teeth—and in this
sense uncovering the source of the normativity of teeth brushing––by pointing out that the
brushing of teeth reduces plaque. But these claims are best understood as first-order normative
judgments about what makes brushing one’s teeth good, not as providing a metaethical
interpretation of what it means to say that reducing plaque is good, what metaphysical
commitments such a judgment involves, or how we come to know that brushing one’s teeth is
good.
We should emphasize here that the fact that Korsgaard is talking about the principle of
instrumental reason as opposed to claims about what is good or what is right does not
22
Stephen Darwall, Philosophical Ethics (Boulder: Westview Press, 1998) 8.
23
We do not mean to suggest at all that right-making properties cannot be normative.
24
Kagan, Normative Ethics 17. See the previous section for a discussion of the question
whether at the foundational level there is a distinction between normative ethics and metaethics.
17
immediately help us disambiguate. The distinction we have been emphasizing is plausibly a

general feature of normative claims. For any normative predicate N, if x is N, then x is N in
virtue of or because some other predicate, F, such that x is F. Thus F, is N-making. Take the
predicate “is a reason”. Some consideration, that the grass is brown, can have the property of
being a reason, say being a reason to water the grass. However, intuitively the consideration will
have the normative property of being a reason in virtue of some other property it has, say, the
property of being what I desire not to be the case. The same holds true it seems of many
normative considerations that are reasons. That it would be good for Jane to have company this
weekend might be a reason for Dick to go visit Jane because this consideration has the complex
property of being about the good of Dick’s mother, namely, Jane.
Again a reductive view—here a reduction of the normative to the normative—might deny, in
some way, this distinction: thus one could insist that when the consideration is that x is morally
good or that φ-ing is morally right, then there is nothing further needed for such a consideration
to be a reason to produce x or to φ. Nonetheless the possibility of non-reductive views here too
means that talk of “explaining” or giving the “source” of normativity is potentially ambiguous.
Similar points hold about rules, norms or laws. The range of locutions for norms and for
claims about the correctness, validity, normativity, authority, and so on, of these norms both in
ordinary language and in philosophical discussion is varied enough to add another layer of
ambiguity. However, we can consider a simple case. Talk just of rules and take the rule, R, to
just be a command: φ! For example: Eat your desert after the main course! Now we can make
various claims about such a rule: R is the correct rule, we ought to R, R is the authoritative rule,
and so on. These claims are normative claims and again there is a possible distinction between
the property R has of being the rule we ought to follow and the property that R has that makes it
be the rule we ought to follow. Thus one could have the theistic normative ethical theory that
what makes R the rule we ought to follow is that God commands it without committing oneself
to the meta-ethical theory that being the rule we ought to follow just is being the rule
commanded by God.25
Let us turn then to the instrumental principle. We can now see the ambiguity in Korsgaard’s
description of her task as searching for the an “explanation” of its “normative force”, an account
of the normativity of the instrumental principle, or an account of “what gives the instrumental

18
principle its normativity”. Using our formulation above, we can take the relevant rule or
command, R, to be as follows: Take the means to your end! The relevant normative claim then is
something of the following form:
(1) For all S, S ought to R.
Now, a metaethical account would investigate what it is to think (1). It would be concerned with
the usual questions. What mental state does such a claim express? Does the word “ought” refer
to some sui generis normative property? And so on. However, there are many questions in
normative ethics that do not involve answering these metaethical questions. We could be
concerned, for example, with what makes it the case that an agent ought to R. We could also be
concerned with the task of, as we called it in the previous section, placing this claim within
practical reason; explaining whether we have reason to do what we ought, according to (1), to
do. We might wonder how this “ought” claims relates to other “ought” claims, say, the claims of
morality.
It is due to Korsgaard’s failure to clearly distinguish the two different versions of the task of
explaining the normativity of instrumental reason that we are able to give a non-cognitivist
reading of her claims about instrumental reason. We will use Allan Gibbard’s norm-expressivist
account as our basic model for noncognitivism. We do this for two reasons: first, Gibbard’s
account is one of the most developed versions of noncognitivism; second, Gibbard’s account has
resources to respond to some of the traditional dissatisfactions with non-reductive realism that
may be lurking behind Korsgaard’s rejections of realism. For reasons of space, we will of course
have to assume a familiarity with Gibbard’s account though we will discuss some of the parts
that would be controversial in the current context.
For the norm-expressivist, when one makes a normative judgment one expresses an
acceptance of certain norms. Crucially, the state of norm-acceptance is not to be analyzed in the
first instance as a belief in a normative proposition that is capable of being the object of
knowledge, but as a non-cognitive motivational state. Gibbard builds out from an account of
what it is to judge that an action is rational or irrational: to judge that an act is irrational is to
express the acceptance of a norm not to do it. If an agent does not accept norms that tell him not
to do an action, then the norms that the agent accepts permit the action. To judge that an act is
wrong is to judge that it would be rational to feel guilty if I have committed it and resentful of
25
See the useful discussion of these two different versions of the theist view in Francis

19
someone else if he has committed it, and this in turn is to express an acceptance of norms that
permit feelings of guilt if I have committed the act and resentment towards someone else who
has committed the act.
Let us turn now to Korsgaard’s account of instrumental reason. As we noted in the
introduction, in “The Normativity of Instrumental Reason”, as elsewhere, Korsgaard certainly
comes across as engaging with metaethical questions. She introduces “the Kantian conception of
practical rationality” as a “third and distinct alternative” to be distinguished from, and preferred
to, "empiricist" accounts, on the one hand, and "realist" or "dogmatic rationalist" positions on the
other (219). Such metaphysical and epistemic sounding labels are once again hard not to read as
labels for positions identified by metaethical commitments. As we have seen already, the realist
or rationalist position is explicitly identified by its semantic and ontological commitments.
In trying to reconstruct a positive metaethical position from “The Normativity of
Instrumental Reason”, we will end up considering several possibilities for such a reconstruction;
however, we begin with what naturally come across as a family of potentially distinctive
metaethical claims, namely, the claims that the will or action are supposedly constituted by
certain principles or norms.
Her central positive claim seems to be the following:
To will an end just is to will to cause or realize the end, hence to will to take the means to
the end. This is the sense in which the principle is analytic. The instrumental principle is
constitutive of an act of the will. If you do not follow it, you are not willing the end at all.
(244)
The problem, as Korsgaard realizes, is that this does not seem to allow for the possibility of
instrumental irrationality. If it is logically impossible to will an end without taking the means to
the end, then it is impossible to be instrumentally irrational—to will an end and fail to take what
one recognizes to be the means to that end.
To prevent this she makes one negative claim about willing:
So willing the end is neither the same as being actually disposed to take the means nor as
being a particular mental state or performing a mental act which is distinct from willing the
means. (245)
And a positive claim about willing:
Snare, The Nature of Moral Thinking (London and New York: Routledge, 1992) 14-20.
20
[W]illing an end just is committing yourself to realizing the end. Willing an end, in other
words, is an essentially first-personal and normative act. To will an end is to give oneself a
law, hence, to govern oneself. That law is not the instrumental principle; it is some law of
the form: Realize this end. That of course is equivalent to ‘Take the means to this end’.
So willing an end is equivalent to committing yourself, first-personally, to taking the
means to that end. (245)
The most straightforward way to construct a non-cognitivist interpretation of Korsgaard's
claims is to take all talk of willing to express a state of norm-acceptance. This is the obvious
interpretation because thoughts of willing seem to be normative thoughts according to Korsgaard
and taking normative thoughts as expressions of states of norm-acceptance is the heart of the
Gibbardian norm-expressivist strategy. An obvious suggestion for the norm being accepted is
"Realize this end!" To judge
(2) I will this end
is then to express my acceptance of the norm, "Realize this end".
There are at this point two possibilities here for how we could proceed in our interpretation.
Korsgaard, recall, plausibly insists that “willing the end” must be distinct from “actually
pursuing or trying to pursue the means to the end” (245) because otherwise, as she puts it, “there
would be, so to speak, not enough distance between willing the end and willing the means for the
one to require the other” (244). The non-cognitivist though can solve this problem, at least put
this way, while only having one state of norm-acceptance. We are actually not quite sure how
equivalence relations work in the case of laws and commands; however, we will here simply
accept, for the moment, Korsgaard’s claim that “Realize this end” is equivalent to “Take the
means to this end” (245). Given this, our non-cognitivist can now say that in thinking (2) I am
also expressing my acceptance of the norm "Take the means to this end". Thus if I think (2),
then, just as Korsgaard requires, there is a certain question I cannot sensibly ask: I cannot now
wonder whether I should will the means (244). After all, in thinking or saying (2) I have
expressed my acceptance of the norm, "Take the means to this end" and expressing my
acceptance of that norm just is to will the means to this end. However expressing my acceptance
of a norm to take the means to my end is not the same as “actually pursuing or trying to pursue
the means to the end” (245). Neither for that matter is the acceptance of the norm. This just
follows from a general feature of Gibbardian norm-acceptance; expressing one’s acceptance of a

21
norm, or the acceptance of the norm itself, are not the same as actually doing, or trying to do,
everything it requires.
That said, Korsgaard may still not like this one-state solution since, as we saw in one of the
initial quotes, she also wants to reject any view according to which willing the end is the same as
“being actually disposed to take the means” (245). It is plausible that the state of accepting the
norm, “Take the means to this end”, involves a disposition to take the means. However, first, it
would still not be the case that willing the end, the normative judgement on our interpretation,
that I will this end, would be the same as this disposition. Rather it expresses the disposition.
What is really important though, surely, is whether the instrumental principle can be violated.
And this can still happen since having this disposition is not the same as actually taking the
means or trying to take the means.
However, for all Korsgaard actually succeeds in arguing, a two-state solution would also
work. Her main reason for insisting that it is not the case that willing the end is “being in a
particular mental state or performing a mental act which is distinct from willing the means” is
that “no mental state or act can logically necessitate you to be in another mental state or perform
another mental act” (244-45). She thinks that it is the dogmatic rationalist who mistakenly
“conceives willing an end as being in a peculiar mental state or performing a mental act which
somehow logically necessitates you to be in another mental state or perform another mental act,
namely, willing the means” (244). Now, as we have pointed out elsewhere, there is no need for
the dogmatic rationalist to think any such thing. The dogmatic rationalist or non-reductive realist
could just insist that being in one state rationally necessitates being in the other state or
performing the other mental act. They would make a normative claim that being one state
rationally requires being in the other. For example, we can think that because I promised to
return the money, I should; however, this does not require thinking that my promise logically
necessitates that I will return the money. Similarly, we can think that because I willed an end I
am rationally required to will the means without thinking that my willing the end logically
necessitates that I will the means. Korsgaard has given us no reason to think that this kind of
two-state view is unavailable to the dogmatic rationalist, let alone anyone else.
The non-cognitivist version of a two-state solution will, of course, look quite different from
the non-reductive realist’s. The two mental states referred to in the non-cognitivist version
would not be used to analyze the property of willing an end, but would be used to analyze the
judgment that one wills an end. According to this proposal, the judgment that one wills an end is

22
to be understood as expressing the non-cognitive state of accepting the norm “realize this end.”
But the analysis must also refer to the rational connection between this state and a second mental
state. The second mental state would involve accepting the norm "Take the means to this end"
and would be expressed by saying or thinking,
(3) I will the means to this end
This reminds us that our story of the state of acceptance expressed by (2) cannot be complete as
stated no matter which of these two routes we take. What I express has to account for all the
various inferences I could draw from, and to, a claim of the form (2). The full statement of the
content of such a thought would make use of Gibbard’s formalism of sets of factual-normative
worlds.26 This would allow us supposedly to account for inferences from (2) to, for example, the
claim that I am acting or have acted, that I exist, etc. But, if we go with a two mental-state
solution, then one of the norms whose acceptance is expressed by (2) would presumably be a
norm that required the transition to the norm-acceptance state expressed by (3). To put the point
more formally in terms of factual-normative worlds, what we would need is that the set of
factual-normative worlds that represents (2) be a subset of the set that represents (3). (3), we
could thus claim, logically follows from (2)—though notice how this does not require that the
mental states expressed logically necessitate each other. This then would be the sense in which
the two-state interpretation would try to capture the thought that “the principle is analytic” (244).
Again, as Korsgaard requires, there is a certain question I cannot sensibly ask: once I have willed
the end, I cannot now wonder whether I should will the means (244). To think (2) just is to think
that I should will the means, that is, to express my acceptance of a norm that says to transition to
the state of acceptance expressed by (3).27
As we suggested at the beginning of this section, a non-cognitivist interpretation also puts us
in a good position to capture a plausible version of Korsgaard’s point that willing is essentially
first-personal and normative. To start with we must insist that there is a certain interpretation of
26
Allan Gibbard, Wise Choices, Apt Feelings: A Theory of Normative Judgement
(Cambridge: Harvard University Press, 1990) 94-97.
27
We should note that the two-state solution requires that the equivalence relation Korsgaard
appeals to when she claims that “Realize this end ... is equivalent to ‘Take the means to this
end’” (245) allows for there to be two separate thoughts here. We think that it would help the
discussion of Korsgaard’s view if this equivalence relation was carefully spelled out. No doubt
some ways of spelling it out will require refining and adjusting our non-cognitivist interpretation;
however, we do not see any reason to think so far that plausible ways of spelling it out will rule
out a non-cognitivist interpretation of her view.
23
the first-personal claim that we think Korsgaard surely does not intend. Korsgaard surely does
not want to deny that the following inference is perfectly legitimate:
(4) John willed this end
therefore
(5) John is committed to this end
therefore
(6) John is committed to taking the means to this end.
Presumably Korsgaard also does not want to deny that we, and not just John, can claim that John
is irrational if he either does not take the means to the end he has willed or give up his end.28
Here is at least one way a norm expressivist might go about saying what it is to think
thoughts of this kind. Such claims have both a normative and non-normative component. In
saying (4) and (5) I am claiming that John is in a certain state of norm-acceptance. That though
is not a normative claim rather it is a claim describing John's state of mind. But I am also
expressing my acceptance of norms that require that when in this state, John form a transition to
another state of norm-acceptance, (6).29 This second state is one in which John accepts the norm:
Take the means to this end. Whether John actually accepts this second norm neither confirms or
disconfirms my claim, since this aspect of my claim is meant as an expression of my own norms
about which norms John should accept, not as a description of the norms that John actually does
accept. Of course more of the details of Gibbard’s account would be needed to give a full
account of such normative statements about others.30
As we have said already, what is at the heart of the claim that willing is “essentially first-
personal” cannot be the view that claims of the form (4)-(6) are not allowed.31 Rather what is
crucial is that willing is not something that we can talk about if we are merely in the business of
28
This does not of course mean that we have to think that there is a detachable ought here
(cf. John Broome, "Normative Requirements," in Normativity, ed. Jonathan Dancy (Oxford:
Blackwell Publishers, 2000)).
29
For an analogous norm-expressivist treatment of belief ascriptions, see Shah and Velleman
[forthcoming]. They argue that when I claim that John believes that p I am making the
descriptive claim that John is in a certain state of mind whose content is that p and I am
expressing my acceptance of a norm that forbids John to be in this state if p is false.
30
See, for a start, Allan Gibbard, Thinking how to live (Cambridge, Mass.: Harvard
University Press, 2003) 48-53.
31
Cf. her comments about causal judgments as being essentially third-personal as opposed to
rational ones which are essentially first-personal. Korsgaard, "Normativity of Instrumental
Reason," 222 n. 23. Again the view cannot be that we cannot say, “I caused him pain”.
24
giving a descriptively adequate account of the world. It is only from within a normative practice
in which we make normative judgments that we can think that someone has willed something.
The contrast between “first-person” and “third-person” is the contrast between engaged
participant and disengaged observer. Surely this is what Korsgaard is getting at. However, this
is of course precisely what our non-cognitivist would say: talk of willing is normative talk, thus
only someone normatively engaged can genuinely talk of willing. A disengaged theorist purely
in the business of giving a descriptively adequate account of the world cannot say either (2) or
(4)—at least if we agree with Korsgaard that talk of willing is essentially normative.32
Our discussion above has been restricted to Korsgaard’s article, “The Normativity of
Instrumental Reason”; however, we think that the strategy used for developing a non-cognitive
interpretation of her views here can be extended to her other texts. The conclusion we draw from
our discussion of non-cognitivism here and our discussion elsewhere of Korsgaard’s rejection of
realism is that her claims are compatible with both metaethical interpretations precisely because
they are actually claims in normative ethics, not metaethics. That said there may be things to be
said in favour of a non-cognitivist over a non-reductive realist interpretation of Korsgaard. A
noncognitivist reading holds out promise that it can respond to some of the problems for the non-
reductive realist position that we suspect lie behind Korsgaard's official objections. The problems
that we have in mind are the familiar ones: non-reductive realism does not provide any
illumination of normative content, leaves us with no account of what it is for a property to be a
normative property, and provides no substantive epistemology that explains how we come to
know normative facts. The noncognitivist does attempt to give a detailed account of the content
of normative thought that will help answer epistemological and metaphysical questions about the
normative. The issues here are far too complex to discuss briefly, but it is, we think, fair to say
that noncognitivism requires nothing beyond an ontology of natural properties and relations, so it
32
The fact that Korsgaard's account can be interpreted so easily in non-cognitivist terms is
not at all surprising. R. M. Hare quipped, "It may seem a far a cry from Kant to Professor
Stevenson", but he went on famously to insist repeatedly across his works that his universal
prescriptivism was a plausible interpretation of Kant and could account for Kant's insistence that
moral claims were not empirical claims; that, as Hume insisted, one cannot get from ought to is;
and that moral concepts were essentially practical and not theoretical. The quote is from R. M.
Hare, The Language of Morals (Oxford: Clarendon Press, 1964) 70. A brief summary of Hare
on Kant is R. M. Hare, "Objective Prescriptions," in Objective prescriptions, and other essays
(Oxford and New York: Clarendon Press, 1999), 11-16. See also R. M. Hare, Sorting out ethics
(Oxford and New York: Clarendon Press; Oxford University Press, 1997).
25
can avoid the metaphysical problems the non-reductive realist faces in postulating a layer of sui
generis normative properties. Its epistemology, if it has one, is not one of tracking intrinsically
normative entities and so it also can avoid the epistemological problems the non-reductive realist
faces in explaining how we can know normative truths.33
3.2 Korsgaard's Official Rejection of Noncognitivism
However, Korsgaard clearly does not share our view about the potential usefulness of adding
a non-cognitivist metaethics to her theory. There are indeed worries about non-cognitivism:
whether, for example, it can indeed deal with the Frege-Geach problem, for example. Our
discussion above has simply assumed that it can and if it turns out that it cannot, then that would
have severe consequences for the plausibility of the above non-cognitivist version of Korsgaard’s
view. Thus it would not have surprised us if Korsgaard had rejected non-cognitivism on the
basis of an argument that in fact it cannot address the range of objections raised against it within
traditional metaethics. Her official grounds though for rejecting non-cognitivism are not these
traditional problems. Her most explicit statements about noncognitivism are expressed in two
articles, “Normativity, Necessity, and the Synthetic a priori: A Response to Derek Parfit” and
“Realism and Constructivism”, that also contain, in some ways, the most explicit positioning of
her own view relative to other metaethical views. In these articles she claims that nothing is
"gained" by introducing noncognitivism, that the distinction between cognitivism and non-
cognitivism leaves no space for Kant, that the distinction itself rests on a theory of normative
concepts that she wants to present an alternative to, and finally that noncognitivism may be true
in a way but "boring".34 Korsgaard's position towards noncognitivism is thus complicated even
if it is fairly consistently negative in one way or another. We will argue in this section that her
reasons for rejecting non-cognitivism seem to betray a misunderstanding of non-cognitivism and
metaethics and fail to show how she can manage to go beyond metaethics.
Korsgaard wants to put into question an idea that she claims lies behind the distinction
between cognitivism and noncognitivism in metaethics, namely, the idea that "what all of our
concepts are for" is the description of reality, "that their cognitive job, so to speak, is to describe
33
See Gibbard, Thinking how to live pp. 221-35.
34
Christine Korsgaard, "Normativity, Necessity, and the Synthetic a priori: A Response to
Derek Parfit" (MS, 2003), 8 and Korsgaard, "Realism and Constructivism," 105, 06, 22 n. 49.
26
reality".35 The implication is that what we need is the possibility that some concepts—the
normative ones no doubt—have a cognitive job that is not to describe reality. Unless this
assumption is dislodged she says, "it will continue to appear that moral realism is the only
possible alternative to relativism, scepticism, subjectivism, and all of the various ways that ethics
might seem hopeless".36 We wonder though what is being built into the use of "cognitive" here.
After all, a contemporary non-cognitivist might well find Korsgaard's description of the missing
position a pretty compelling statement of what the non-cognitivist takes to be the intuitive heart
of his position. How is Korsgaard’s anti-descriptivism supposed to be different than the
traditional anti-descriptivism that looks very much like the heart of noncognitivism? Our only
clue here is that she seems to want a "cognitive" job for these concepts other than describing
reality. And we take it she would think that Stevenson and Hare, and their contemporary
followers, are committed to the job not being cognitive. We worry though that this may just be a
misunderstanding about the term ‘cognitivism’. Non-cognitivism and cognitivism are without
doubt terms of art and so susceptible to the vagaries of individual philosophers' deployments.
That said, there is still a reason why these terms are introduced and used, which has to do with
the association between cognitivism and knowledge. The point of calling a theory of normative
concepts non-cognitive is that the theory rejects the assumption that the role of normative
concepts is primarily that of helping with knowledge—cognitive accounts are rejected precisely
"for theorizing practical reason", for failing, again as Korsgaard herself puts it, to understand
"the difference between knowledge and action".37
So why would she want to insist in the face of her own rejection of descriptivism and her
insistence that it is not knowledge that normative concepts are for that she wants a cognitive job
for normative concepts? A cognitive job that is not primarily for description or knowledge?
This almost sounds like a contradiction in terms.
We suspect that there is more being associated with "non-cognitivism" then just, one wants
to say, the non-cognitivism. This becomes apparent in the paragraph after her declaration that
35
36
Korsgaard, "Realism and Constructivism," 105. We should note that there are many
different kinds of relativism and subjectivism and it is not immediately clear that they all entail
that ethics is “hopeless”.
37
The first quotation is from Korsgaard, "Normativity, Necessity", 9 and the second is from
Korsgaard, "Realism and Constructivism," 110. In the latter article she repeatedly attacks the
idea that normative thought is primarily a matter of knowledge.
27
she wants a non-descriptive cognitive role for normative concepts. She claims that even before
we get to her own alternative account we can see that there is "a problem with the distinction
between cognitivism and non-cognitivism in ethics" because it does not leave space for a theory
like Kant's "according to which moral judgments are the conclusions of practical reasoning".38
Here is her reason for claiming this: "A conclusion of practical reasoning is not obviously a
description of a fact about the world, but it hardly seems like some sort of emotional expletive,
either".39 This is not helpful since either the phrase "some sort of emotional expletive" is meant
to suggest some crude, simplistic theory—as surely this phrase would in ordinary language—or,
if the phrase is meant to acquire its content from its reference to the complex and subtle theories
of meaning and thought that non-cognitivists have provided, then the "hardly seems like" is
unjustified.
It is hard to resist the conclusion that cognition is being confused with cogitation. A non-
cognitivist theory is not a non-cogitative theory. All the major non-cognitivists—Stevenson,
Hare, Blackburn, Gibbard—have complex theories about how practical reasoning proceeds in all
its complexity. Indeed the traditional worries about noncognitivism do focus on whether they
can capture the complexity of practical reasoning—thus the focus on the problem of embedding.
We grant that the embedding problem is a real concern, but what is needed is a serious
engagement with it, something that Korsgaard does not even attempt.40
Surprisingly, she grants on the next page the following:
Alternatively, someone who takes seriously Kant's thesis that moral laws are the laws of
autonomy, legislated by the agent's own will, may read him, as Hare sometimes seems to
do, as a prescriptivist and so a non-cognitivist. So although practical reason theories might
at first seem to fall between the cracks, there are ways of making them fit the mold.41
Now we assume that she is not simply taking back her claim on the previous page that there is a
problem with the distinction between cognitivism and non-cognitivism because Kant does not fit.
Presumably she thinks that Hare is simply getting Kant wrong. Is she taking back the claim that
the distinction between cognitivism and non-cognitivism forces one to interpret conclusions of
reasoning as either descriptions of fact or emotional expletives and that the second option is
38
Korsgaard, "Realism and Constructivism," 105. She also mentions Aristotle.
39
40
Such an attempt would also be hard to square with her claim later that expressivism is true.
41
28
clearly unacceptable? Even if he gets Kant wrong, does she think that Hare does provide us with
an account that could indeed adequately explain practical reasoning? Again presumably not, but
why not? Hare's account is precisely the kind of non-descriptivist, non-knowledge based theory
that Korsgaard seems to be calling for. It thus is difficult to see what the grounds are for her
negative attitude towards non-cognitivism. In fact it would seem preferable to accept the label of
non-cognitivism rather than adopt the incoherent sounding position of a cognitivist non-
descriptivism that does not take knowledge as primary.
In a footnote in "Realism and Constructivism", Korsgaard claims that expressivism is true
but “boring” because it “describes moral language from the outside, as if we were not ourselves
the creatures who face practical problems, but only someone else making anthropological
observations about them”(note 49, 122). And in her “Response to Derek Parfit” she says “Nor is
anything gained, in my view, by saying that the first person use (of normative concepts) is an
expression of your endorsement rather than merely a statement describing your endorsement,
since that is still describing the situation from the outside” (8-9). It is true that expressivism
attempts to describe our moral practices from a vantage point that is external to that practice. It is
also true that no prescriptions about what to do fall out of expressivism since, according to the
expressivist, pronouncing on what to do requires engaging in the normative practice, not
describing it. But if expressivism is true, as Korsgaard claims it is, it gives us a full-blown
account of what it is to make a normative judgment, giving us answers to our semantic,
metaphysical, and epistemological questions about moral discourse. It might seem, then, that
Korsgaard just is not interested in these questions, and thus finds answers to them “boring” and
unhelpful. However, she also claims that “Behind this stance is the idea that so long as we are
reasoning we must remain at this anthropological level, and behind that view is the same error
that animates moral realism––the view that the business of cognition is describing the world” (9).
As to the claim that according to expressivism, reasoning can only take place at the
anthropological level, this need not be true of all expressivisms. As we have indicated, Gibbard’s
expressivism is aimed at giving a metaethical account of the process of practical reasoning.
According to him, judgments of reasons, judgments of what to do, are made from within the

29
engaged participant’s point of view, whereas his expressivist account of these judgments is given
from within the disengaged anthropological point of view.42
3.3 Constitution, Inescapability, and Necessity
It may seem that attempting to articulate an interpretation of her positive position by focusing
on how it can be distinguished from traditional positions in metaethics is to force Korsgaard to
play a game that she thinks should not be played. After all, as we have repeatedly insisted, she at
least sometimes clearly indicates that she wants to "break with the platitudes of twentieth century
ethics" and reject the thought that there is "a difference between doing 'meta-ethics' and doing
'normative' or practical ethics".43 This “break”, we have been presuming, needs to be argued for
by showing what is wrong with traditional metaethical theories or the questions they ask.
Furthermore, in order for her own position to count as a genuine alternative to traditional
metaethical theories, it must not be the case that it is consistent with a version of one of the
theories it is being presented as an alternative to. We have argued that Korsgaard has failed to
demonstrate any problems with traditional metaethics and that all of the statements of her own
view are consistent with traditional metaethical views. Nonetheless, in the hopes of seeing some
other way in which she might reject the traditional distinction between metaethics and normative
ethics, we return in this section to the constitution claims we began with and consider other ways
in which these might be used to reject the traditional distinction.
Again our focus will be the principle of instrumental reason. When we first introduced her
claim that the "instrumental principle is constitutive of an act of the will" we immediately asked
whether there was anything here that the non-reductive realist or non-cognitivist has to reject.44
Let us now instead see whether there is anything here that allows us to bypass the traditional
metaethical project altogether.
42
And as to the alleged error about cognition committed by both realists and expressivists,
until Korsgaard explains what the opposing non-descriptivist conception of cognition might be, it
is difficult to make sense of this criticism. We take it that her account of normative concepts is
supposed to provide this alternative account of cognition. Presumably this positive account of
normative concepts is supposed to be provided by what she calls “constructivism” or “procedural
realism”. We discuss this part of her view in Hussain and Shah, "Misunderstanding Metaethics,"
.
43
44
Korsgaard, "Normativity of Instrumental Reason," 244.
30
Here is one suggestion.45 Showing that a principle is constitutive of the will, and thus of
agency, is to show that no sense to be made of acting that does not involve following the
principle of instrumental reason.46 The correct theory in the philosophy of action thus shows that
an agent necessarily follows the principle of instrumental reason. We take this to entail the claim
that as a matter of logical necessity an agent takes the principle of instrumental reason to have
rational necessity. Since, when I deliberate about what to do, according to Korsgaard, I must
regard myself as an agent, I thus must regard myself as following the instrumental principle. It is
unclear what type of necessity Korsgaard thinks is involved in regarding oneself as an agent or as
regarding oneself as following the instrumental principle, but we will put this issue to the side.
To engage in practical deliberation, then, I must regard myself as following the principle of
instrumental reason, which entails that I regard the principle of instrumental reason as justified or
rationally necessary.47 No metaethical claim is apparently required to draw this conclusion.
Let us assume that the argument goes through. Would this show that traditional metaethics is
somehow beside the point or involves some deep misunderstanding? We think not. The
argument, for all we have said so far, would show that I have to think that the principle of
instrumental reason is rationally required but it would not show what it is to think this. Now
there may well be an argument that nothing more can be said about what it is to think this, but
notice that the argument so far does not show that more cannot be said. It would show that we
do not need to do metaethics in order to see why all practical deliberators must conclude that
they should follow the principle of instrumental reason. Metaethics, if this argument worked,
would then be irrelevant to the justification of the principle of instrumental reason to practical
deliberators. This would show that metaethics would not have practical significance, but it does
nothing to show that metaethics would not be of theoretical interest.
When Korsgaard talks about expressivism and realism as "boring" we wonder whether she
just means to point out that, if the above argument works, metaethics is not required to justify
45
Our thanks to David Velleman for emphasizing this.
46
It is important to bear in mind that "following" here is being used in a normative sense. In
following the principle of instrumental reason I am, as Korsgaard puts it, committed to realizing
the end and this committing of myself is "an essentially first-personal and normative act.”
Korsgaard, "Normativity of Instrumental Reason," 245.
47
Well, not exactly. The conclusion Korsgaard’s argument actually entitles us to is that one
must regard oneself as regarding the principle of instrumental reason as rationally necessary.
31
normative principles.48 Despite her suggestion though it is hard to see how this shows that
metaethics is something that should be left behind with other rejected "platitudes of twentieth
century ethics".49 It is odd to hear a philosopher deploy the charge that a theory is boring as an
objection—what at least looks for all the world like an objection. Confused, mistaken,
incoherent, false, yes, but boring? We admit that our eyes tend to glaze over during talks in
mereology and we even suspect that mereology has no practical significance, but some of our
best friends are mereologists and we would be hard put to raise our being bored, or the lack of
practical significance, as an objection to any particular theory they held, let alone as grounds to
give up on mereology altogether.
As far as we can see, then, even if this argument went through it would undermine neither the
presuppositions of twentieth-century metaethics nor the difference between metaethics and
normative ethics.
Furthermore, we do not think that the argument actually quite succeeds in showing that
metaethics is irrelevant to the question of justification. The reason for this is that the argument is
assuming that no error theory is true. An error theory about normative judgments would be the
view that normative judgments express truth-conditional beliefs about normative properties and
that all atomic normative judgments that attribute normative properties are false. The falsity of
error theoretic accounts is of course compatible with a wide range of metaethical theories.
Recall how we put it. If I am engaged in practical deliberation, I have to regard myself as an
agent. Talk of "regarding" does not quite tell us the attitude involved. There are other options,
but the natural thought is that I have to believe that I am an agent. Believing that I am an agent,
if Korsgaard is correct, entails believing that I ought to take the necessary means to my ends. The
claim then is that there is no way that I can, clear-headedly, think about what to do without
believing that I ought to take the necessary means to my ends. But this does not show that this
belief is true. The argument does not show that this belief is true even if, when I think about what
to do, I have to believe that I am an agent and so believe that I ought to take the necessary means
to my ends. One could interpret the attitude of "regarding … as …" in, say, expressivist terms,
thereby dismissing error-theories without having to show that the normative judgments
constitutive of agency are true, but then, as far as we can see, one would be doing traditional
metaethics.
48

32
We take it that the Korsgaardian response would be some form of the objection that any such
approach "describes moral language from the outside, as if we were not ourselves the creatures
who face practical problems, but only someone else making anthropological observations about
them."50 But why cannot I both deal with my practical problems and make anthropological
observations—take up theoretical problems—about myself? When I am deliberating what to do,
let us grant, I have to believe that I ought to take the necessary means to me ends. But when
engaged in theoretical deliberation I can also wonder whether my belief is true. Only some
implausibly strong restriction on the ability of a thinker to engage in reflective assessments of his
own beliefs would show that this cannot be done.
Our practical deliberator, though forced to regard himself as an agent and thus as accepting
the instrumental principle, can recognize that the ability of the argument to force him to this is
compatible with the claim that it is false that he is an agent. Thus, for all that has been said, it
would be perfectly sensible to wonder whether these claims were true and theoretically
investigate the matter. If these claims turned out not to be true, our deliberator would then no
doubt be in an awkward position. As Kant worried long ago, there would then be a
"contradiction of reason with itself".51 Perhaps if one thought that it was likely that error theory
would be true, then one would have a practical reason not to investigate the truth of the belief in
order to avoid the potential awkwardness.52 But none of this shows that the theoretical
investigation somehow does not make sense.
What Korsgaard says elsewhere about this issue is not that easy to interpret. In her article on
Parfit, "Personal Identity and the Unity of Agency", she makes the same argument. We must
regard ourselves as agents "when we occupy the standpoint of practical reason—that is, when we
are deciding what to do." "[T]his fundamental attitude is forced upon us by the necessity of
making choices, regardless of the theoretical or metaphysical facts."53 In a footnote to these
claims she writes:
49
50
51
AK 5:6 note. See also, of course, AK 4:455-56.
52
Though the locution of "regarding … as …" may suggest an escape clause. Perhaps full-
blown belief is not required. See footnote on Velleman below.
53
Christine Korsgaard, "Personal Identity and the Unity of Agency: A Kantian Response to
Parfit," Philosophy and Public Affairs 18, no. 2 (1989): 120.
33
Some people suppose that this means that freedom and agency are an illusion produced by
the practical standpoint. But this presupposes the primacy of the theoretical standpoint,
which is in fact the point at issue. Free agency and, according to my argument, unified
personal identity are what Kant calls "Postulates of Practical Reason”.54
This does not quite help since we need to know more about what postulates of practical reason
are and why the theoretical standpoint could not show that they are false. On one traditional
interpretation of Kant, he was concerned about precisely such a potential contradiction in reason
and it is only the machinery of transcendental idealism that allows the practical standpoint to
coexist with the theoretical standpoint. This is not the place to engage in Kant exegesis, but we
think something like this traditional approach needs to be deployed.
Showing that the theoretical standpoint and the practical standpoint are not in conflict
requires hard work and here we think we find ourselves in agreement with Kant as opposed to
some contemporary Kantians. And the work it requires is precisely the kind of work that we take
it traditional metaethics, and associated discussions in mind and the philosophy of action, are
engaged in. It is the work of deciding what commitments about the world are involved in
believing that I am an agent, in believing that I am following, say, the principle of instrumental
reason, and in believing that a normative principle is rationally required. Or how the attitude
involved in "regarding … as …" is somehow different from straightforward belief. Most
traditional metaethical accounts, like noncognitivism, are attempts to provide accounts of the
nature of normative thought that are meant to show why there is no contradiction in reason.55
There may well be many ways to avoid error theory and practically speaking we may well not
take the likelihood of error theory seriously, but the point remains that avoiding it, for all we
54
Korsgaard, "Personal Identity and the Unity of Agency: A Kantian Response to Parfit,"
120 n. 31.
55
This is not to commit ourselves to the claim that a contradiction in reason would result in
complete paralysis. There may still be options. If Korsgaard is right that, in some sense, we
have to take ourselves as agents—that we cannot help taking ourselves as agents—then the
position Velleman expressed about value judgments quite some time ago might well fit this
situation: "Instrumentalism says that we can be perfectly justified in thinking about the good as if
it were real despite our awareness that it isn't. … Instrumentalism about the good … explains the
psychological instability of evaluative skepticism. Our doubts about the existence of value never
deter us from thinking about what's valuable. And the explanation is that our motive for thinking
about what's valuable is quite independent of the assumption that anything really is" (J. David
Velleman, Practical Reflection (Princeton: Princeton University Press, 1989) 296). For some
worries about such strategies, see Hussain, "Moral Fictionalism," .
34
have seen, still requires engaging in traditional metaethics even if, given some people's
assessments of likelihoods, this may not be practically important and so may be boring.
It might be replied on Korsgaard’s behalf that there is nothing more to showing that the
principle of instrumental reason is correct than demonstrating that it is a commitment that is
constitutive of being an agent. The metaethicist’s error, it might be claimed, is to think that there
is an intelligible question about the truth of the instrumental principle left over once it has been
demonstrated that the principle is constitutive of agency. The question whether it is true that one
ought to take the necessary means to one’s ends is answered precisely by showing that the
commitment to the instrumental principle is inescapable for agents as such. Therefore it makes
no sense to accept that a commitment to the instrumental principle is constitutive of agency and
yet think that an error theory is so much as an intelligible option.
First of all, even if we allow that such a demonstration counts as providing a justification of a
normative principle, and thus as ruling out an error-theory about that normative principle, it does
not provide a complete metaethical account of the normative principle. One might accept that
demonstrating that a commitment to a normative principle is constitutive of agency suffices to
show that the principle is correct without yet knowing anything about the semantics or
metaphysics of normative judgments. Second, this method for ascertaining the correctness of a
normative principle can hardly settle all of our epistemological questions about normative
principles. Just as a full epistemological account of our beliefs about middle-sized physical
objects will go beyond saying merely that we know facts about these objects by looking to
explain what it is about these objects and our visual faculties that makes looking a means of
gaining such knowledge, so too a full epistemological account of our normative judgments will
explain why demonstrating that the commitment to a normative principle is constitutive of
agency is a means of showing it to be correct. As far as we can see, both non-cognitivists and
realists, if they so choose, can adopt this method of justification as part of their epistemology of
normative judgment. Realists might argue that demonstrating that a commitment is constitutive
of agency is the way to discover non-natural normative facts; non-cognitivists might argue that
making normative judgments involves accepting a norm that requires one to accept any norm
that is constitutive of agency.
Of course, one might insist that our understanding of the unique way in which normative
judgments are justified exhausts our understanding of their nature; nothing more can be said
about the epistemological, semantic, and metaphysical issues. This would be to take a quietist

35
position with respect to metaethics. In order to acquire the right to this position, however, one
needs to explain why there is no further elucidation to be had of our normative judgments
beyond a demonstration of their method of justification. There are two ways of doing this. One
way is to argue that the best explanation of the failures of all current and past metaethical
theories is that normative judgments are sui generis and cannot be given a substantive
metaethical analysis. We sympathize with this sentiment, but it is unclear why we should accept
this pessimistic conclusion as opposed to the conclusion that there is a correct metaethical
account, but that it just hasn’t been discovered yet.
The other way to argue that nothing substantive can be said at the metaethical level is by
demonstrating that metaethical questions rest on a false presupposition, and that once this
presupposition is removed, no substantive metaethical distinctions remain. An example of this
kind of argument can be gleaned from discussions of minimalism about truth. What follows is
the barest sketch of such an argument. Minimalists claim that the truth predicate is mainly a
grammatical device. Minimalists differ about which grammatical functions are fundamental to
our understanding of the truth predicate, but they are united in their denial that the truth predicate
expresses a substantive relation that some sentences bear to the world. Traditional realist and
noncognitivist theories alike, though, seem to presuppose just such a substantive, metaphysical
conception of truth. Realists claim, and noncognitivists deny, that normative judgments are
capable of truth in this substantive sense. But even noncognitivists accept that other kinds of
judgments, namely empirical ones, are capable of substantive truth. This is what allows them to
draw a contrast between normative judgments and those other judgments. If minimalism about
truth is correct and there is no substantive notion of truth to be had, then it would seem that the
very contrast between realism and noncognitivism lapses.56 We take no position on the
correctness of minimalism about truth or on the question whether metaethical distinctions can be
maintained if minimalism about truth is correct. Our point is just that this is the kind of argument
that, if successful, would justify quietism. We have argued that although Korsgaard does seem to
think that metaethics rests on some kind of mistake, she fails to give the kinds of grounds that
could justify such a claim.
56
For a much fuller presentation of this line of thought see Drier [2004].
36
4. Conclusion
We began by taking Korsgaard as a prime example of a certain kind of discontent with

metaethics. The discontent expresses itself as not merely a set of objections to existing theories
in metaethics but as the view that there is something wrong with the very way in which the
debate is set up, the very way in which the questions of metaethics are posed. As we have
argued, it is not plausible, to think that Korsgaard is merely doing something unconnected to
traditional metaethical concerns—after all she sets her position up in opposition to realism and
empiricism and presents herself as showing how despite the fact that realism and expressivism
might be true they nonetheless, in some other sense, miss the point of moral philosophy. We
have tried in the space available here to make sense of these claims. However, we have argued
that in fact Korsgaard fails to distinguish her own positive position from either realism or
noncognitivism—noncognitivism being the focus of this paper, the point about realism having
been argued elsewhere. This makes it hard to see how her view can be alternative to these
traditional views let alone a position from which we can see what was confused with the
presuppositions lying behind these traditional metaethical theories. We also attempted to see
how her claims about the inescapability of certain presuppositions of agency could be used to
show that these traditional metaethical positions were somehow fundamentally confused. We
argued that the claims about inescapability did not show that the questions traditional metaethical
theories were designed to answer were somehow illegitimate or that the answers provided by
these traditional theories were philosophically irrelevant to anything that might be of concern to
us in moral philosophy.
The fundamental problem, we have hoped to have shown, occurs at the very beginning. In
framing her inquiries about the “source of normativity,” Korsgaard fails to distinguish the
metaethical question of what it is to make a normative judgment from normative questions about
which normative judgments to make or even which normative judgments we cannot help but
make. This leads her to mistakenly think that by making a strong case for the position that certain
normative judgments are constitutive of agency, she has given an alternative to the non-reductive
normative realist’s and non-cognitivist’s metaethical positions about the meaning, metaphysics,
and epistemology of these normative claims. Her opposition to non-reductive realism and non-
cognitivism similarly suffers from a misunderstanding of the metaethical tasks these theories
seek to accomplish. We suspect that the problems with Korsgaard’s form of discontent can be
generalized to other discontents—there is a systematic failure to appreciate how a range of
37
questions in normative ethics can be asked, and even answered, while leaving open metaethical
issues. The discontent with metaethics is often based on misunderstandings of its tasks and
goals. Careful assessments of other versions of discontent will have to be done elsewhere;
however, we hope that the attempt in this paper to clarify the different tasks of normative ethics
and metaethics and the detailed analysis of Korsgaard’s version of discontent goes some way
towards indicating how such assessments of other versions of discontent would proceed.57
57
We would like to thank Allan Gibbard, Pamela Hieronymi, Elijah Millgram, Tamar
Schapiro, Yonatan Shemmer, Kenneth Stalzer, and Sharon Street for very useful conversations
about the issues raised in this paper. Thanks to Sarah Buss, Kieran Setiya, David Velleman and
Allen Wood for written comments on an earlier draft of this paper.
38
References
Broome, John. "Normative Requirements." In Normativity, edited by Jonathan Dancy, 78-99.

Oxford: Blackwell Publishers, 2000.
Darwall, Stephen. Philosophical Ethics. Dimensions of Philosophy Series. Boulder: Westview
Press, 1998.
Darwall, Stephen, Allan Gibbard, and Peter Railton, eds. Moral Discourse and Practice: Some
Philosophical Approaches. New York and Oxford: Oxford University Press, 1997.
Drier, James. "Meta-ethics and the Problem of Creeping Minimalism." Philosophical
Perspectives 18, no.1 (2004): 23-44.
Fine, Kit. "The Varieties of Necessity." In Conceivability and possibility, edited by Tamar
Gendler and John O'Leary-Hawthorne, 253-81. Oxford: Clarendon Press, 2002.
Gibbard, Allan. Wise Choices, Apt Feelings: A Theory of Normative Judgement. Cambridge:
Harvard University Press, 1990.
———. Thinking how to live. Cambridge, Mass.: Harvard University Press, 2003.
Hare, R. M. The Language of Morals. Oxford: Clarendon Press, 1964.
———. Sorting out ethics. Oxford and New York: Clarendon Press; Oxford University Press,
1997.
———. "Objective Prescriptions." In Objective prescriptions, and other essays, 1-18. Oxford
and New York: Clarendon Press, 1999.
Hussain, Nadeem J. Z. "The Return of Moral Fictionalism." Philosophical Perspectives 18, no. 1
(2004): 149-87.
Hussain, Nadeem J. Z., and Nishi Shah. "Misunderstanding Metaethics: Korsgaard's Rejection of
Realism." Oxford Studies in Metaethics (2005): forthcoming.
Kagan, Shelly. Normative Ethics. Boulder: Westview Press, 1998.
Korsgaard, Christine. "Personal Identity and the Unity of Agency: A Kantian Response to
Parfit." Philosophy and Public Affairs 18, no. 2 (1989): 101-32.
———. The Sources of Normativity. Edited by Onora O'Neill. Cambridge: Cambridge
University Press, 1996.
———. "The Normativity of Instrumental Reason." In Ethics and Practical Reason, edited by
Garrett Cullity and Berys Gaut, 215-54. Oxford: Oxford University Press, 1997.

39
———. "Normativity, Necessity, and the Synthetic a priori: A Response to Derek Parfit." MS,
2003.
———. "Realism and Constructivism in Twentieth-Century Moral Philosophy." The Journal of
Philosophical Research APA Centennial Supplement (2003): 99-122.
Miller, Alexander. An Introduction to Contemporary Metaethics. Cambridge: Polity Press, 2003.
Shah, Nishi, and J. David Velleman. 'Doxastic Deliberation.' Philosophical Review
(forthcoming).
Snare, Francis. The Nature of Moral Thinking. London and New York: Routledge, 1992.
Velleman, J. David. Practical Reflection. Princeton: Princeton University Press, 1989.

MD

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MD

Uploaded by

Copyright:

Available Formats

Metaethics and Its Discontents: A Case Study of Korsgaard

DRAFT: PLEASE DO NOT QUOTE OR ATTRIBUTE WITHOUT PERMISSION

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 1 July 24, 2005

2. Metaethics and Normative Ethics

In order to go beyond metaethics, Korsgaard has to engage with traditional metaethical

presupposing some kind of account—either a developed one or at least a working

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 July 24, 2005

3.1 The Non-Cognitivist Interpretation of “The Normativity of Instrumental Reason”

immediately help us disambiguate. The distinction we have been emphasizing is plausibly a

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 July 24, 2005

3.2 Korsgaard's Official Rejection of Noncognitivism

Version 1.8, Rev. 3 July 24, 2005

3.3 Constitution, Inescapability, and Necessity

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 July 24, 2005

We began by taking Korsgaard as a prime example of a certain kind of discontent with

Broome, John. "Normative Requirements." In Normativity, edited by Jonathan Dancy, 78-99.

Version 1.8, Rev. 3 July 24, 2005

Version 1.8, Rev. 3 July 24, 2005

You might also like