. 10
( 13)


interference is justi¬ed by reasons that rational others would recognize
(Section IIIe).
The Place of Autonomy within Liberalism 291

Insofar as liberalism is to function as a public morality regulating life
among diverse strangers, justi¬cations of interferences must be addressed
to all members of the relevant public, as morally autonomous agents: all
have moral reasons to accept this limitation on the pursuit of their goals.
And that, we have seen, implies the claim that if they were fully rational “
displaying the excellences of evidence gathering, deliberation, and ac-
tion that constitute our concept of practical rationality “ they would
all recognize such reasons and act on them. The liberal principle, to-
gether with what I have called the post-Enlightenment insight, thus
implies a non-trivial commitment to principles of social morality that
are justi¬ed through reasons we share as rational agents “ public

IVb Universal Laws and Moral Autonomy
In explaining a morality based on autonomy Kant writes:

If we now look back upon all previous attempts which have ever been undertaken
to discover the principle of morality, it is not to be wondered that they all had
to fail. Man is seen to be bound to laws by his duty, but it was not seen that he is
subject only to his own, yet universal, legislation, and that he is only bound to act
in accordance with his own will. . . . For if one thought of him as a subject only to
a law (whatever it may be), this necessarily implied some interest as a stimulus or
compulsion to obedience because the law did not arise from his will.62

Kant goes on to insist that all moralities moved by “some interest as a stim-
ulus or compulsion to obedience” are heteronomous. An autonomous
morality, in contrast, conceives “each rational being as a being that must
regard itself as giving universal law through the maxims of its will.”63
As I have depicted it, under a social morality justi¬ed through public
reason, each rational autonomous individual has (an internal) reason
to act on that morality. And not because of “some interest as a stimu-
lus or compulsion to obedience because the law did not arise from his
will,” as in a heteronomous morality, but because the reason to accept
and act on the moral principle is one to which the agent qua rational
is committed (and this in the non-trivial sense). As rational, then, the
agent wills the moral principle and the acts it requires, even though it
requires a limitation on the pursuit of his goals. Thus it is the case that
under public reason, moral principles are willed by all rational agents
in the relevant public, and only moral principles so willed are justi¬ed
under public reason. We can see, then, how an autonomous morality
both limits freedom and is itself an expression of freedom. Insofar as it
Gerald F. Gaus

limits the ways in which we can pursue our ends and goals and opens
us to moral claims by others that we must do as they insist, it is a re-
straint on freedom (and that is why moral regulation must be justi¬ed
under the fundamental liberal principle); but because these demands
do not confront us simply as external requirements but are con¬rmed
by our own reasons to act, they are freely willed by all. No concep-
tion of morality that does not account for this Janus-headed nature of
morality “ as both a restriction and expression of our freedom “ can be
We can now interpret the link between Kantian autonomy, contractu-
alism, and the idea of the general will. Recall that for Kant, the “test of
the rightfulness of every public law” is the “idea of reason,” that there
is “an original contract by means of which a civil and thus a completely
lawful constitution and commonwealth alone can be established.”64 An
original contract, Kant tells us, is “based on a coalition of all the pri-
vate wills in a nation to form a common, public will, for the purposes
of rightful legislation.”65 Contractualism, understood as a justi¬catory
device, requires that justi¬ed principles be those that all rational indi-
viduals would accept. The hypothetical or counterfactual nature of this
claim has led some critics to object that such contracts cannot bind.66
This, though, misses the justi¬catory role of the contractual device given
a commitment to a (non-trivial) concept of public reasoning. Only princi-
ples that could be accepted by all rational, morally autonomous persons
can identify the reasons we share. If R could not be accepted by each
and every rational, morally autonomous agent, it could not be a moral
reason that each wills, and so could not qualify as part of an autonomous
public morality.
Kant believed that the very idea of an original contract has the “prac-
tical reality” of obliging “every legislator to frame his laws in such a way
that they could have been produced by the united will of a whole nation,
and to regard each subject, in so far as he can claim citizenship, as if
he had consented within the general will.”67 Thus conceived, the idea of
the general will is not only fully consistent with liberalism, but is implied by the
fundamental liberal principle. Although in the hands of Hegelians such as
Bernard Bosanquet, the notion of the general will implies a collectivistic,
and at least arguably, an illiberal understanding of the state, interpreted
as an ideal according to which all just laws are rationally willed by all
citizens, it expresses the fundamental liberal principle that interferences
must be justi¬ed, conjoined with the ideal of public reason “ that they
must be justi¬ed to all.68
The Place of Autonomy within Liberalism 293

V Personal Autonomy
Thus far, my concern has been the place of moral autonomy within
liberalism. Section III argued that the fundamental liberal principle
presupposes that agents are morally autonomous. Section IV then
maintained that this conception of moral autonomy, relying on an inter-
nalist conception of moral reasons, leads us to an ideal of public reason-
ing, and together these endorse a substantively Kantian (though, again,
not Kant™s own) autonomous morality, according to which the justi¬ed
principles of social morality must be such as to be rationally willed by all,
and we saw how this gives rise to the ideal of legislation that expresses the
general will.69 I now turn to the implications of this analysis for placing
personal autonomy within liberalism.

Va Ultra-Minimal Personal Autonomy
The fundamental liberal principle supposes an ultra-minimal conception of
personal autonomy. Thus far I have not sought to justify the fundamental
liberal principle. Of the justi¬cations that have been advanced, though,
the most compelling maintains that self-directed agents (who are in addition
morally autonomous persons in a world of morally autonomous persons)
will necessarily be led to endorse it. Consider again Benn™s story of Alan
on the beach. Suppose that Betty continues to frustrate his actions, in the
sense that every time he seeks to act, she seeks to interfere, by handcuf¬ng
him, taking the pebbles from the beach, or whatever. Why would Alan
object? Basic to any plausible answer is that Alan conceives of himself as an
agent whose deliberations about what he should do normally determine
his own actions. It is not morally neutral to him whether he or Betty
decides what he is to do; the moral default is that he decides what he is to do,
and some special case needs to be made for letting another™s deliberations determine
his actions. It seems impossible that Alan could conceive of himself as a
self-directed agent (who is also morally autonomous) without claiming
this basic moral default. Suppose that he renounces this default “ as
a utilitarian acquaintance of mine purports to do. When such an agent
decides to ±, he entertains no moral presumption that he should ± rather
than, say, β, which is what another has decided he is to do. Should he
be made to β without justi¬cation, it would be inappropriate for him to
experience resentment, indignation, blame “ none would be called for,
since he really has no claim to ± rather than β.70 They are both competing
judgments about what he should do, neither having any intrinsic moral
privilege. That he has decided that he should ± in itself provides no more
Gerald F. Gaus

guidance about what he should than that another has decided that he
should β.
The dissent of my utilitarian acquaintance notwithstanding, such a
denial undermines one™s sense of one™s own agency as a self-directed
person. Crucial to one™s own self-conception is that one™s practical reason
is just that: one reasons about what to do because it has the practical
consequence of determining what one does. This seems not to be the
case with schizoid personalities, who apparently see others as controlling
their activities, and so conceive of the deliberating self as alienated from
the acting self.71 If the deliberating self is not to be similarly alienated
from its activities, it must suppose that, in lieu of special considerations,
its deliberations guide its activity.
Underlying this argument for the fundamental liberal principle
(which I have only sketched here) is the supposition that we are indeed
self-directing agents in this sense. As I said, this supposes that we are not
schizoid; it also supposes that we are not “role-directed” personalities, for
whom all actions are required by social scripts, and the proper perfor-
mance of these scripts is determined by the audience, not each of us qua
actors.72 Imagine we were such people (as Clifford Geertz suggests the
traditional Balinese might be).73 I need not advance a moral claim to act
as I have decided, for as I would conceive of myself, there is nothing spe-
cial about my deliberations in deciding what I should do. The script, and
the audience™s reaction, is what counts. The fundamental liberal princi-
ple would be as alien to such people as many philosophers would have
us believe it is to us.
My claim, then, is that the fundamental liberal principle only gets its
grip on those who are self-directed in the minimal sense I have been
discussing. This can be understood as the ultra-minimal conception of
personal autonomy on which liberalism is founded. It is not in itself a
notion of moral autonomy. Although to advance liberalism™s basic moral
claim the agent must be morally autonomous, before he is even interested
in such a claim he must possess the non-moral characteristic of conceiving
of himself as self-directed or what Benn calls a “natural person”:

The use of expressions such as “decision making,” “making a choice,” “forming
an intention,” suggest a kind of creativity in personal causation, in which the
relation between agent and process is initiated by his decision is more like that
between a potter and his pot or an architect and his plan, than like the relation
between a skidding car and the resulting accident. . . .
. . . It is this consciousness of one™s own thought as the prolegomenon to in-
tended action that underlies a person™s conviction that he makes decisions “ that,
unlike skids or lightening strikes, they do not just happen to him.74
The Place of Autonomy within Liberalism 295

In this ultra-minimal sense, liberalism supposes that people are “self-
ruled” “ they are in charge of themselves.

Vb The Ultra-Minimal View Contrasted with Personal Autonomy
as Self-Authorship
The case I have sketched for the fundamental liberal principle is
grounded on an ultra-minimal conception of personal autonomy, and
is to be distinguished from accounts of liberalism that accord primacy to
a thicker conception of personal autonomy.75 Ultra-minimal personal au-
tonomy is consistent with many of the character traits that proponents of
personal autonomy deem heteronomous. A self-directed person may be
guided by superstitious beliefs, be totally unre¬‚ective about his commit-
ments, have con¬‚icting desires and inconsistent beliefs, or live according
to traditional rules simply because he has been brought up to. All these
traits are consistent with being an agent who sees his actions as following
from his own deliberations (based, to be sure, on unre¬‚ective, traditional,
or superstitious considerations).
Steven Wall advances a “perfectionist” conception of “personal auton-
omy” according to which it is an “ideal of people charting their own
course through life, fashioning their character by self-consciously choos-
ing projects and taking up commitments from a wide range of eligible
alternatives, and making something out of their lives according to their
own understanding of what is valuable and worth doing.”76 In a similar
vein, Joseph Raz maintains that “[t]he autonomous person is one who
makes his own life,” while Robert Young tells us that “[t]he fundamental
idea in autonomy is that of authoring one™s own world.”77 Although these
formulations are by no means identical, all identify autonomy with being
the author of one™s life. An autonomous life is chosen by the agent rather
than, say, dictated by tradition; most articulations of this ideal require a
wide range of choices through which “one makes something” out of one™s
Personal autonomy as self-authorship is a controversial ideal that is
dif¬cult to publicly justify. The self-authorship metaphor points to an
aesthetic view of life in which one™s life is a creation and the agent the
artist. The metaphor is not misleading; such conceptions of autonomy
are offered by “perfectionist liberals.”78 The very idea of perfection indi-
cates a quasi-artistic attitude towards one™s life, as a work to be perfected.
It hardly seems that all agents have reason to adopt such a view. Con-
sider one whose goals are entirely focused on bringing about states of
affairs that do not include the perfection of human beings but, say, the
protection of nature or scienti¬c discovery. The latter may involve the
Gerald F. Gaus

perfection of human nature (the former may well call for thwarting it),
but the point is that these goals are not about human nature and its
excellences; they concern the production of certain states of affairs that
do not make necessary reference to the ¬‚ourishing of humans.79 If these
states of affairs can be brought about without perfecting human nature,
or without a life of self-authorship, that in no way detracts from their
value. Agents pursuing such states of affairs are not self-focused; they do
not take up an attitude of creative authorship to their lives, but possess
practical reasons to investigate and change the world in a variety of ways.
As such, they are not committed to personal autonomy as self-authorship
or perfectionism.
To be sure, the perfectionist can argue that they should be: he can insist
that there is a reason for them to care, and they should see it. I am not
seeking to refute such arguments, but to show that they are controversial,
and certainly make claims that go far beyond ultra-minimal autonomy.
Our enviromentalist or scientist, I have argued, is committed to seeing
himself as an agent with reasons to act, so he must conceive of himself
as a self-directed agent; many do not “ and as far as I can see, rationally
so “ conceive of themselves as authors or creators of their own lives,
seeking to make something out of them through their chosen modes of
Still, it might be thought that all self-directed agents must possess per-
sonal (or what Forst in Chapter 10 of the present volume calls “ethical”)
autonomy in the sense that they have and exercise the capacity to, as
Waldron says, “defy desires and inclinations” that are alien to their con-
ception of the good (Chapter 13 in the present volume). And certainly
achieving some minimal degree of integration and consistency is neces-
sary for self-direction; it must be the case that one has enough of a self for
one to be able to make decisions, as opposed to merely giving in to one
inclination after another. But a self-directed person may not be one who
af¬rms a way of life, or who sees himself as following personal imperatives
about what is to be done. Self-directed agents may fall well short of fully in-
tegrated personalities.80 They may possess nothing so grand as a concep-
tion of the good life, much less an examined life: the much-derided beer-
drinking81 football fan “ whose week is, unre¬‚ectively, centered around
Sunday™s Buffalo Bills game “ possesses ultra-minimal personal autonomy.

Vc Personal and Moral Autonomy
If I am right about this, the fundamental liberal principle does not rest on
a commitment to a “perfectionist” conception of “personal autonomy.”
It would, however, be wrong to conclude that conceptions of personal
The Place of Autonomy within Liberalism 297

autonomy more demanding than the ultra-minimal notion have no place
within liberal morality and politics. We have seen that the appeal to the
fundamental liberal principle presupposes that both the person appeal-
ing to the principle, and the person to whom the appeal is directed, pos-
sess moral autonomy (Section III). That is, both are supposed to possess
the ability to distinguish her own speci¬c wants and aims from the re-
quirements of public morality. Now this moral capacity is bound up with
fairly sophisticated cognitive skills. As the work of Lawrence Kohlberg
and others has shown, to be able to distinguish what you want or prefer,
or your goals and aspirations, from what can be universalized and thus
accepted by rational others, requires a cognitive ability to take up the
perspectives of others.82 One must be able to put oneself in their place,
consider what reasons they have, and so whether they have reasons to act
in certain ways. Moreover, because the fundamental liberal principle is
open-ended insofar as it only requires that justi¬cations be provided but
does not provide a canonical list of those justi¬cations, liberal citizens
have an ongoing commitment to examine proposed justi¬cations and
enter into justi¬catory argument.
The skills required by moral autonomy overlap with those that are of-
ten identi¬ed with personal autonomy.83 Unless a citizen is self-re¬‚ective
about her own reasons to act, and so understands whether her reasons
stem from personal commitments or can be shared from a public per-
spective, she will be unable to determine what is required by a publicly
justi¬ed morality. A person who is unable to distinguish her goals and
personal commitments from moral reasons will not be able to grasp the
idea that moral reasons may require her to put aside her goals, for she
will insist that her beliefs and values are a seamless web. It is no defense
to say of such people that the basic premises of their moral thinking lie in
their personal “ say, religious “ convictions and so they are, understand-
ably, unable to contemplate the possibility that the demands of public
secular morality can be distinguished from, much less override, their re-
ligious convictions.84 Because they are insuf¬ciently re¬‚ective about the
nature of their reasons, and have an insuf¬ciently developed capacity to
see things from the perspectives of others, they are apt to press morally
unjusti¬ed demands, and fail to recognize the requirements of public
This points to the error of hyper-ecumenical versions of “political lib-
eralism.” Political liberalism seeks to identify liberal principles endorsed
by public reason. Yet many versions have been especially accommodating
to religious reasoning, often including versions of fundamentalism. The
idea has been that, insofar as we seek truly public reasoning, these
Gerald F. Gaus

religious reasoners must be brought into public justi¬cation. To be sure,
political liberals insist on limits: to qualify as “reasonable,” citizens must
tolerate the competing views of others, and be willing to seek, and abide
by, fair terms of social cooperation with them. But toleration of others
is not suf¬cient to exercise moral autonomy. Consider a fundamentalist
religion that is internally committed “ committed simply in virtue of its
own tenets “ to tolerating other religions and embracing a fair scheme
of cooperation with others. If this is the sole source of these commit-
ments, when contemplating objections from others that some policy was
not tolerant, or not fair, members of this religion would still appeal to
their religious convictions in deciding what constitutes toleration and
fairness. They would not concern themselves with providing other citi-
zens with public reasons in support of their interpretations “ they would
have failed to exercise moral autonomy. Only if they have developed the
cognitive ability to distinguish what is a reason to them from what is a
reason for others can they justify a substantively autonomous morality
and laws that express the general will.
We can now understand the ambivalent stance towards religious rea-
soning that, I think, has characterized most modern liberal thinking. On
the one hand, liberals insist that people be free to pursue religious con-
victions as a matter of personal liberty. Yet because they are not public
reasons, and further because many religions insist on the superiority of
religious to public shared reasoning (in terms of the force of their respec-
tive reasons for actions), liberals (perhaps especially outside of America)
are typically wary of appeals to religion in public life. Moreover, insofar as
some religious communities are totalistic, seeking to provide a pervasive
religious structure for every member™s personal and intellectual life, lib-
erals object that such communities undermine citizens™ moral autonomy.
This is especially troubling if communities seek to raise their young in
ways that undermine their children™s personal autonomy by thwarting de-
velopment of their skills of self-re¬‚ection and role-taking. We might say,
then, that moral autonomy requires minimal personal autonomy: the ability
to re¬‚ect on the adequacy of one™s own moral reasons, and to distinguish
one™s own reasons from the reasons of others.
It might be objected that this analysis endorses intolerance towards reli-
gious groups such as the Protestant fundamentalists, some types of devout
Catholics, or the Amish. Such objections are based on a common but nev-
ertheless erroneous simple inference from general philosophical princi-
ples to public policies. Public policy is the realm of complex and com-
peting considerations, including problems of abuse of power, incentives
The Place of Autonomy within Liberalism 299

of government agents and legislators, epistemic limits of government,
dif¬culty of framing adequate legislation and safeguards, undesirable
side-effects, and so on. Nothing directly follows about what laws ought
to be enacted from a general philosophical conclusion. It does follow
from this analysis, though, that ways of life that seek to undermine min-
imal personal autonomy, and so ultimately the moral autonomy of their
members, are illiberal. How a liberal state should deal with illiberal ways
of life is a complex and dif¬cult issue, but little headway can be made
without recognizing that they are indeed illiberal in the sense that they
are based on practices and beliefs that are hostile to the very capacities
and dispositions that render liberal public morality possible.
While the capacity for, and exercise of, critical self-re¬‚ection is re-
quired for a liberal moral order, the conception of personal autonomy
is still minimalist insofar as it is not part of the justi¬ed conception that
citizens re¬‚ect on their own plans and projects, except insofar as they dis-
tinguish these from public morality. Thus the commitment to personal
autonomy necessary for a liberal moral order does not require that all cit-
izens be self-re¬‚ective about their goals, aims, or projects. The requisite
personal autonomy can be fully achieved by those who embrace tradi-
tional, customary, or religious ways of life, not out of explicit choice, but
because they have been reared in them. However, the traditionalist can-
not be so immersed in traditional culture that he is unable to distinguish
the reasons it provides him from the moral reasons that apply to all.
To many this seems a precarious compromise: liberal morality allows
one to be an unre¬‚ective traditionalist in many aspects of one™s life, but
not to become so immersed in one™s traditions that one confuses them
with public reason. While indeed precarious, this is precisely the line that
liberal political culture walks. It can admit traditionalism, and need not
seek to turn all citizens into liberal-Millian individualists “ up to a point.
The point is that citizens must be suf¬ciently liberal to re¬‚ect on their
traditions and observe that they do not form the basis of public reasoning,
and so they must be prepared to also live in a public world that, because
religious reasons are not reasons for all citizens, must be a secular world.
More than that, they must understand that these public reasons will often
override their important goals.

VI Conclusion: Walking the Liberal Tightrope
My aim has been to show, ¬rst, that the most plausible understanding
of the fundamental liberal principle presupposes a Kantian conception
Gerald F. Gaus

of moral autonomy. Showing this required inquiry into the nature of
practical reasons and morality. It is fashionable nowadays to claim that
one can engage in political philosophy without such investigations, that
we can have a purely political theory of liberalism. It is impossible to see
how this can be done; if liberal principles are to be practical, they must
provide us with practical reasons. But then we need to know what practical
reasons are, and how they relate to liberal principles. Whether or not we
need a metaphysics of liberalism, we certainly require a metaethics of it.
When we do develop such a metaethics, I argued in Section III, we are
led to a Kantian conception of moral autonomy.
Section IV linked this conception of moral autonomy to public laws,
freedom, public reason, and the general will. Kant™s basic intuition,
that our capacity for moral autonomy leads to a substantive universal-
istic morality was, in its essence, vindicated, though not, of course, simply
Having argued in favor of a Kantian understanding of moral auton-
omy, I then turned in Section V to consider the relationship of liberalism
to conceptions of personal autonomy. The results were not quite so neat.
Although an ultra-minimal conception of personal autonomy underlies
the basic liberal principle, autonomy understood as self-authorship does
not; indeed it seems a controversial and rationally rejectable view. How-
ever, I have just argued that the very commitment of liberalism to moral
autonomy itself leads to a public commitment to minimal personal au-
tonomy as a capacity the exercise of which is necessary to a moral order
based on the fundamental liberal principle. A liberal moral and political
order, I have claimed, walks a tightrope. On one side is immersion into
traditional cultures and religions, which insist that their reason is the rea-
son of all; on the other is the public proclamation of the liberal ideal of
individuality as part of the public morality, and so the illegitimacy of most
traditional and religious ways of life. Only societies composed of citizens
who are suf¬ciently self-re¬‚ective to recognize the distinction between
their personal (or subcultural) and public reasons, and who embrace
diverse communities while recognizing their non-public character, can
walk the liberal tightrope.85 Happily, our modern liberal societies seem
reasonably adept at this particular balancing act.

On the idea of a liberal public morality, see my Value and Justi¬cation: The
Foundations of Liberal Theory (Cambridge: Cambridge University Press, 1990),
The Place of Autonomy within Liberalism 301

pp. 323ff; Justi¬catory Liberalism (New York: Oxford University Press, 1996),
pp. 120ff; Social Philosophy (Armonk, NY: M. E. Sharpe, 1999).
I consider these in Section V.
John Christman, after discussing the ideal of personal autonomy, notes that it
seems an “abrupt departure from the traditional Kantian notion,” though he
adds that “despite. . . . [the] radical differences, there remain crucial aspects
that our core idea of autonomy shares with its Kantian ancestor.” “Introduc-
tion” to his edited collection, The Inner Citadel: Essays on Individual Autonomy
(New York: Oxford University Press, 1989), p. 14. Kant scholars have agreed;
see Roger J. Sullivan, Immanuel Kant™s Moral Theory (Cambridge: Cambridge
University Press, 1989), pp. 46“47. Gerald Dworkin, however, sees moral
autonomy as a particular case of the wider notion of autonomy as critical
self-re¬‚ection. The Theory and Practice of Autonomy (Cambridge: Cambridge
University Press, 1988), p. 48. Compare my argument in Section V.
In the end, the difference between our positions may be modest: the closer
one connects the two ideas, the more blurred the distinction becomes.
See Steven Wall, Liberalism, Perfectionism and Restraint (Cambridge: Cam-
bridge University Press, 1998), Part II.
As attested to by the work of the most important contemporary liberal au-
tonomist, Joseph Raz. See his The Morality of Freedom (Oxford: Clarendon
Press, 1986), chapter 14.
Harry Brighouse, “Is there a Neutral Justi¬cation for Liberalism?” Paci¬c
Philosophical Quarterly, vol. 77 (September 1996): 193“215, at p. 209. This, I
shall argue, gets things almost precisely backwards.
See, for example, Joel Feinberg, Harm to Self (New York: Oxford University
Press, 1986), chapter 18.
The importance of Kantian autonomy is clearest in Rawls™ “Kantian Construc-
tivism in Moral Theory” in Samuel Freeman (ed.), John Rawls: Collected Papers
(Cambridge, MA: Harvard University Press, 1999), pp. 303“358. Kantian
autonomy remains a basic feature of political liberalism. See Rawls™ Political
Liberalism, paperback edition (New York: Columbia University Press, 1996),
Lecture II.
Stanley Benn, A Theory of Freedom (Cambridge: Cambridge University Press,
1988), p. 87.
Feinberg, Harm to Self, p. 9. Benn is talking about interference, a wider notion
than Feinberg™s “coercion,” so these are not identical formulations.
Feinberg, Harm to Self, pp. 14ff.
John Locke, Second Treatise of Government, in Peter Laslett (ed.), Two Trea-
tises of Government (Cambridge: Cambridge University Press, 1960), p. 287
(Section 4).
It is sometimes objected that this view requires that there be an uncontro-
versial, basic, negative conception of liberty, which is at the foundation of
morality, and so does not itself presuppose any moral claims. If, it is charged,
liberty is itself a moralized concept, or one that involves moral ideas (as,
for example, some concepts of positive liberty seem to), then a basic claim
to liberty cannot be the presupposition of all other moral claims. There
is something to this charge: for the fundamental liberal principle to make
Gerald F. Gaus

sense, there must be some sensible liberty claims that are claims to non-
interference, and that do not themselves presuppose justi¬ed moral norms.
Thus it must make sense “ and it does “ to say that in an amoral Hobbesian
state of nature, people interfere with each other, and in that sense limit each
other™s liberty. What is not supposed by this account, however, is that this use
of “liberty” exhausts all sensible liberty claims (positive or norm-based liberty
claims still might make sense and be important) or that “interference” is an
uncontroversial idea, such that we never disagree about what constitutes an
interference. We do, of course, disagree, which means we disagree about the
interpretation and application of the fundamental liberal principle.
Unless speci¬ed to the contrary, “reasons” throughout this chapter means
“reasons for action.”
Geoffrey Brennan and Loren Lomasky, Democracy and Decision: The Pure Theory
of Electoral Preference (Cambridge: Cambridge University Press, 1993), p. 9.
Footnote in original text omitted.
Robert Nozick, The Nature of Rationality (Princeton: Princeton University
Press, 1993), p. 164.
See my essay “Why All Welfare States (Including Laissez-Faire Ones) Are Un-
reasonable,” Social Philosophy and Policy, vol. 15 (June, 1998): 1“33; “Goals,
Symbols, Principles: Nozick on Practical Rationality” in David Schmidtz (ed.),
Robert Nozick (Cambridge: Cambridge University Press, 2002), pp. 83“130;
“The Limits of Homo Economicus” in Gerald F. Gaus, Julian Lamont, and
Christi Favor (eds.), Values, Justice, and Economics (Amsterdam: Rodopi, forth-
This de¬nes instrumental reason simply as effective action, with no regard
to justi¬ed belief, a view that I argue against in “The Limits of Homo Eco-
nomicus.” This simple characterization suf¬ces for our present purposes; a
more adequate conception would push the analysis of rationality even more
towards the internalist position defended in Sections IIId“e.
See David Gauthier, Morals By Agreement (Oxford: Clarendon Press, 1986).
The classic paper on this issue is, of course, William K. Frankena, “Obliga-
tion and Motivation in Recent Moral Philosophy” in K. E. Goodpaster (ed.),
Perspectives on Morality: Essays by William K. Frankena (Notre Dame: University
of Notre Dame Press, 1976), pp. 49“73. See also Philippa Foot, Virtues and
Vices (Berkeley: University of California Press, 1978), pp. 157“173. For an
effective criticism of Foot™s externalism, see Michael Smith, The Moral Problem
(Oxford: Balckwell, 1994), chapter 3.
This view is suggested by Wall, Liberalism, Perfectionism, and Restraint, p. 118.
I contrast internalism and externalism as accounts of obligation and of rea-
sons, in Value and Justi¬cation, pp. 153ff, 261ff.
See Smith, The Moral Problem, pp. 7ff.
Judith Jarvis Thomson, “Killing, Letting Die and The Trolley Problem,” The
Monist, vol. 59 (1976): 204“217; “A Defense of Abortion,” Philosophy & Public
Affairs, vol. 1 (1971): 47“66.
See, for example, Michael Tooley, “An Irrelevant Consideration: Killing ver-
sus Letting Die” in B. Steinboch (ed.), Killing and Letting Die (Englewood
Cliffs, NJ: Prentice-Hall, 1980), pp. 56“62.
The Place of Autonomy within Liberalism 303

27. See further my essays: “Goals, Symbols, and Principles,” “Why All Welfare
States (Including Laissez-Faire Ones) Are Unreasonable,” and “The Limits of
Homo Economicus.”
28. I analyze this rough idea of one reason “outweighing” another in “Why
All Welfare States (Including Laissez-Faire Ones) Are Unreasonable” and in
“The Limits of Homo Economicus.” The rough idea suf¬ces for our present
29. See Donald Davidson, “Actions, Reason and Causes” in his Essays on Actions
and Events (Oxford: Oxford University Press. 1980).
30. Foot, Virtues and Vices, p. 179. See also David Copp, Morality, Normativity and
Society (New York: Oxford University Press, 1995).
31. See Stanley Benn and G. F. Gaus, “Practical Rationality and Commitment,”
American Philosophical Quarterly, vol. 23 (1986): 255“266.
32. See further my Value and Justi¬cation, pp. 266ff.
33. See Smith, The Moral Problem.
34. For a position along these lines, see David A. J. Richards, A Theory of Reasons
for Action (Oxford: Oxford University Press, 1971).
35. See Henry E. Allison, Kant™s Theory of Freedom (Cambridge: Cambridge
University Press, 1990), p. 98.
36. “The idea that autonomy is responsiveness to reasons is of course not new. A
version of this idea is central to Kant™s ethical theory. . . . ” George Sher, Beyond
Neutrality: Perfectionism and Politics (Cambridge: Cambridge University Press,
1997), p. 48.
37. See Allison, Kant™s Theory of Freedom, p. 98.
38. Susan Wolf, Freedom Within Reason (Oxford: Oxford University Press, 1990),
pp. 67ff.
39. Ibid., p. 68.
40. See Christine Korsgaard, The Sources of Normativity (Cambridge: Cambridge
University Press, 1996).
41. See Sullivan, Immanuel Kant™s Moral Theory, chapters 4 and 5; Allison, Kant™s
Theory of Freedom, chapter 5.
42. I have tried to explicate this distinction in detail in Value and Justi¬cation. See
also my “What is Deontology?” Parts 1 and 2, Journal of Value Inquiry, vol. 35
(2001): 27“42, 179“193.
43. See, however, Jeremy Waldron™s insightful chapter (13) in the present vol-
ume. I follow Charles Larmore in understanding the contrast between attrac-
tive and imperitival moralities as dividing pre-modern and modern ethics.
See Larmore™s The Morals of Modernity (Cambridge: Cambridge University
Press, 1996), chapter 1.
44. Immanuel Kant, Foundations of the Metaphysics of Morals, Lewis White Beck
(trans.) (Indianapolis: Bobbs-Merrill, 1959), pp. 60ff [pp. 442ff of the
Prussian Academy edition].
45. See Rawls, Political Liberalism, pp. 50ff.
46. See my Value and Justi¬cation, pp. 292“330. Compare Rawls, Political Liberal-
ism, p. 51.
47. Jan Narveson, The Libertarian Idea (Philadelphia: Temple University Press,
1988), p. 66.
Gerald F. Gaus

48. I have considered this view in more depth in my “Property, Rights and Free-
dom,” Social Philosophy & Policy, vol. 11 (Summer 1994): 209“40.
49. This claim is argued for by Hillel Steiner, An Essay on Rights (Cambridge, MA:
Blackwell, 1994), chapters 2 and 3.
50. This could, I suppose, be interpreted as a sort of moral dilemma produced by
imcompossible oughts. Although not unintelligible, I have argued that such
conceptions of deontic logic are by no means compelling. See my essay “Dirty
Hands” in R. G. Frey and Christopher Heath Wellman (eds.), A Companion
to Applied Ethics (Oxford: Basil Blackwell, 2003), pp. 167“179.
51. “Central to understanding paternalism is the conjunction of two factors: an
imposition and a particular rationale. X acts to diminish Y™s freedom, to the
end that Y™s good is secured.” John Kleinig, Paternalism (Totowa, NJ: Rowman
and Allenheld, 1983), p. 18.
52. See my Social Philosophy, chapter 11.
53. For Hohfeld™s classic analysis, see his “Some Fundamental Legal Conceptions
As Applied in Judicial Reasoning,” Yale Law Review, vol. 23 (1913): 16“59.
54. See Allison, Kant™s Theory of Freedom, pp. 94“106.
55. This view was famously upheld by Thomas Nagel in his The Possibility of Altru-
ism (Princeton: Princeton University Press, 1979). Cf. Christine Korsgaard,
“The Reasons We Can Share” in her Creating the Kingdom of Ends (Cambridge:
Cambridge University Press, 1996), pp. 275“310.
56. See, for example, Korsgaard, The Sources of Normativity, pp. 136ff; Philip Pettit,
The Common Mind (Oxford: Oxford University Press, 1993).
57. See my Contemporary Theories of Liberalism: Public Reason as a Post-Enlightenment
Project (London: Sage Publications, 2003), pp. 104“113.
58. See ibid., chapter 1. See also my entry on “Public Reason,” International Ency-
clopedia of the Social and Behavioural Sciences (Oxford: Elsevier Scienti¬c Pub-
lishers, 2002); John Gray, Enlightenment™s Wake: Politics and Culture at The Close
of the Modern Age (London: Routledge, 1995), pp. 122ff.
59. Rawls, Political Liberalism, p. 36.
60. Kant, Metaphysical Elements of Justice, Lewis White Beck, trans. (Indianapolis:
Bobbs-Merrill, 1959), p. 76 (Section 44).
61. “Non-trivial” because even on the view according to which all must share
all the same reasons, it is trivially true that only justi¬cations that provide
everyone with reasons justify moral impositions.
62. Kant, Foundations of The Metaphysics of Morals, p. 51 [p. 433 of the Prussian
Academy edition].
63. Ibid.
64. Immanuel Kant, “On the Common Saying: ˜This May be True in Theory,
but it Does not Apply in Practice™,” in Hans Reiss (ed.), Kant™s Political Writ-
ings (Cambridge: Cambridge University Press, 1970), p. 79. Emphasis in
65. Ibid.
66. I consider this objection more fully in my Social Philosophy, chapter 5.
67. Kant, “Theory and Practice,” p. 79.
68. See Bernard Bosanquet, “The Philosophical Theory of the State,” in Gerald Gaus
and William Sweet (eds.), The Philosophical Theory of the State and Related
The Place of Autonomy within Liberalism 305

Essays (Indianapolis: St. Augustine Press, 2001). Cf. John W. Chapman,
Rousseau “ Totalitarian or Liberal? (New York: Columbia University Press,
I consider the relationship between legislation, public reason, and Rosseau™s
theory of the general will in “Does Democracy Reveal the Will of the People?
Four Takes on Rousseau,” Australasian Journal of Philosophy, vol. 75 (June
1997): 141“162.
On the appropriateness of emotions, see my Value and Justi¬cation, chapter II.
See also Wolf, Freedom Within Reason, especially chapter 1.
See further my Value and Justi¬cation, p. 388.
Ibid, pp. 385“386.
See Clifford Geertz, “Person, Time, and Conduct in Bali” in his The Interpre-
tation of Cultures (New York: Basic Books, 1973).
Benn, A Theory of Freedom, p. 91.
They are often run together. In the quote from Benn (fn 74), in which he
is explicating self-direction, he likens a person™s relation to his life as that
between a pot and a potter, thus moving to self-authorship. Sharon Hayes
argues that Benn™s liberalism ultimately is based on a robust conception of
(personal) autonomy. See her “Autonomy and Rights in S. I. Benn™s A Theory
of Freedom,” Ph.D. thesis, School of Humanities, Queensland University of
Technology, 2000.
Wall, Liberalism, Perfectionism and Restraint, p. 203. Cf. Dworkin: “What makes
an individual the particular person he is in his life plan, his projects. In
pursuing autonomy, one shapes one™s life, one constructs its meaning. The
autonomous person gives meaning to his life.” The Theory and Practice of
Autonomy, p. 31
Raz, The Morality of Freedom, p. 375; Robert Young, Personal Autonomy: Beyond
Negative and Positive Liberty (New York: St. Martin™s Press, 1986), p. 19
Raz and Wall, for example.
On the tension between this conception of autonomy and environmen-
tal ethics, see my “Respect for Persons and Environmental Values” in Jane
Kneller and Sidney Axin (eds.), Autonomy and Community: Readings in Contem-
porary Kantian Social Philosophy (Albany, NY: SUNY Press, 1998), pp. 239“264.
Benn, A Theory of Freedom, chapter 10.
The implication, perhaps obvious to most academics, is that a wine-drinking
fan would achieve an altogether higher level of autonomy.
See, for example, Lawrence Kohlberg, The Philosophy of Moral Development
(New York: Harper & Row, 1981); B¨ rbel Inhelder and Jean Piaget, “The
Growth of Logical Thinking from Childhood to Adolescence” in Howard
E. Gruber and J. Jacques Von´ che (eds.), The Essential Piaget (London:
Routledge & Kegan Paul, 1977), pp. 403“444, at 440“41. I discuss other
literature in Value and Justi¬cation, p. 260.
Interestingly, Wall distances his perfectionist conception of autonomy
from autonomy as self-re¬‚ection. See Liberalism, Perfectionism and Restraint,
pp. 128“129.
See Christopher Eberle, Religious Convictions in Liberal Politics (Cambridge:
Cambridge University Press, 2002).
Gerald F. Gaus

85. I thus concur with Steven Macedo: “Liberal citizens should be committed to
honoring the public demands of liberal justice in all departments of their
lives. They should be alert to the possibility that religious imperatives, or even
inherited notions of what it means to be a good parent, spouse, or lover, might
in fact run afoul of equal freedom. A basic aim of liberal education should
be to impart to all children the ability to re¬‚ect critically on their personal
and public commitments for the sake of honoring our shared principles
of liberal justice and equal rights for all.” And Macedo is clear that “[t]he
point is not to promote a comprehensive philosophical doctrine of autonomy
or individuality.” Diversity and Distrust (Cambridge, MA: Harvard University
Press, 2000), pp. 238“239.

Moral Autonomy and Personal Autonomy

Jeremy Waldron

Modern philosophers distinguish between personal autonomy and moral
autonomy.1 Talk of personal autonomy evokes the image of a person in
charge of his life, not just following his desires but choosing which of his
desires to follow. It is not an immoral idea, but it has relatively little to
do with morality. Those who value it do not value it as part of the moral
enterprise of reconciling one person™s interest with another™s; instead,
they see it as a particular way of understanding what each person™s interest
consists in. Moral autonomy, by contrast, is associated speci¬cally with the
relation between one person™s pursuit of his own ends and others™ pursuit
of theirs. This is particularly true of its Kantian manifestations. A person
is autonomous in the moral sense when he is not guided just by his own
conception of happiness, but by a universalized concern for the ends of
all rational persons.
Modern proponents of personal autonomy are anxious to emphasize
the distance between their conception and the moral conception.2 But I
think it is worth considering some of the overlaps and af¬nities between
them. We all know that autonomy in the moral sense is supposed to en-
gage very speci¬c capacities of rational deliberation and self-control. And
these might seem out of place in a conception of autonomy oriented to-
wards the pursuit of the good life at an individual level. In fact, modern
theorists of personal autonomy have also tended to emphasize the en-
gagement of speci¬c capacities “ the capacity for re¬‚ection, for example,
and for what some have called “second-order” motivation “ and these

Jeremy Waldron

turn out to be similar in many respects to the capacities implicated in the
Kantian account.
So a sharp distinction between moral autonomy and personal auton-
omy may not be available. Is it, in any case, desirable? I am not sure. Rawl-
sian liberals emphasize the importance of subjecting individuals™ pursuit
of the good to moral principles of justice: the right has priority over the
good, they say. Now, the sharper the distinction between personal au-
tonomy and moral autonomy, the more challenging it is to explain how
this priority is supposed to work, for the more alien the requirements of
morality will seem from the personal point of view. On the other hand, if
we blur the distinction between pursuing a conception of the good and
following principles of right, we open the possibility that each person has
his own moral standards implicated already in his personal view about
what makes life worth living. And this too seems unsatisfactory, because
it undermines the idea of the right as something shared rather than as
something intensely personal.
In what follows, I shall consider these issues from two angles. I shall
look ¬rst at the contrast in Kant™s moral philosophy between the pur-
suit of individual happiness and the realm of autonomy and free moral
agency. That distinction looks clear enough; but as we shall see, Kant
blurs it somewhat by characterizing an individual™s entitlement to pursue
his own happiness in his own way as a fundamental principle of freedom.
Then, having complicated the Kantian picture, I would like to look more
squarely at the positions held by modern liberals, and consider how sharp
they need the distinction between the right and the good to be, and
how sharp they can afford it to be, both in light of the tasks of morality
and in light of the actual characteristics of people™s moral and ethical

“No one,” said Kant, “can coerce me to be happy in his way (as he thinks of
the welfare of other human beings); instead each may seek his happiness
in the way that seems good to him. . . .” This, Kant said, was the ¬rst half of
“the principle of freedom” for the constitution of a commonwealth. (The
second half added the familiar proviso: “. . . provided he does not infringe
upon that freedom of others to strive for a like end which can coexist with
the freedom of everyone in accordance with a possible universal law.”)3
“Each may seek his happiness in his own way.” For us “ I mean for us
modern liberals “ this sounds like a principle of autonomy. But on Kant™s
Moral Autonomy and Personal Autonomy 309

account, autonomy and happiness are supposed to operate in utterly
different realms. In the Groundwork, Kant associated autonomy with the
will™s ability to determine itself in accordance with the form of universal-
ity, unconstrained by nature or inclination.4 Autonomy is “the supreme
principle of morality,” and morality is “[t]he direct opposite of . . . the
principle of one™s own happiness [being] made the determining ground
of the will.”5 Happiness is about needs and inclinations, and as such it
must be regarded as “a powerful counterweight to all the commands of
duty.”6 Kant says that I become an autonomous being only when I rise
above any concern for happiness and follow the moral law for its own
Yet Kant de¬nes the principle with which we began as a principle of
freedom, and freedom “ while not exactly synonymous with autonomy in
his system “ is not as far from it as the dissonance between happiness and
morality might suggest. He says it is “the only original right belonging to
man by virtue of his humanity,”7 and humanity is supposed to be some-
thing “holy” in us, something characteristic of us as ends in ourselves. So
what is going on? If there is a place for the free pursuit of happiness in a
Kantian system, does that pursuit have its own dimension of autonomy?
If it does, what is the relation between that autonomy and the autonomy
associated with the rigors and severity of the moral law?
Another way of posing this issue is to ask: why does it matter, from
the perspective of autonomy, who lays down the conception of happiness
that I follow. Why, on Kant™s account, is it important that my pursuit of
happiness be determined by me rather than by another person or by the
state? Certainly the latter would involve heteronomy. But heteronomy is
involved in the pursuit of happiness anyway, on Kant™s account. Where
one ¬nds happiness is an empirical matter; it is a question of the relation
between one™s needs and inclinations and the circumstances in which one
¬nds oneself. From this point of view, the in¬‚uence of another person “
or the coercive in¬‚uence of the state “ is just one empirical contingency
among others. If by chance I grow up in Iowa, it is less likely that I will
acquire a taste for sur¬ng than if I grow up in San Diego. And if I grow
up in a Lutheran community, I may not learn to dance. Why is the (soci-
ological) fact that my community frowns on dancing any different from
the (geographical) fact that I live a thousand miles from the ocean? Both
are contingent features of the empirical world; and happiness, on Kant™s
account, is not supposed to be a matter of the existence or non-existence
of empirical determinants but rather of how one™s needs and desires
are satis¬ed in relation to them. So, in this matter of the pursuit of
Jeremy Waldron

happiness, what is so special “ so specially bad “ about coercion as an
empirical determinant?
These questions invite us to reconsider the idea of a non-negotiable
separation in Kant™s theory between the rational capacities involved in
morality and the capacities (whatever they are) that are involved in the
pursuit of happiness.
One possibility is that the element of autonomy does enter the picture,
but only in the second half of Kant™s principle of freedom “ that is, in
the proviso about respecting the freedom of others: “[E]ach may seek
his happiness in the way that seems good to him, provided he does not
infringe upon that freedom of others to strive for a like end which can coexist
with the freedom of everyone in accordance with a possible universal law.”8 It
is by seeing the force of, and following this part of, the principle that
one reveals oneself as an autonomous being. This need not mean that
the element of autonomy plays no role in the ¬rst part of the principle.
Perhaps it folds back into the exercise of freedom in the following way:
Kant suggests that one shouldn™t (and a good person won™t) regard as
part of his happiness something that con¬‚icts with morality. Happiness,
he says, is not just a matter of satisfying the preferences I happen to have. “I
must ¬rst be sure that I am not acting against my duty; only afterwards am
I permitted to look around for happiness . . .”9 Or perhaps my autonomy
is revealed not by my egocentric interest in my own happiness but only
by my moral interest in the happiness of all “ my interest in what Paul
Guyer refers to as a “systematic distribution of happiness, both in one™s
own life as a whole and in the whole community of human beings.”10
These are surely important themes in Kant™s practical philosophy. But
they do not actually implicate autonomy in the choices I make among the
morally acceptable options for my happiness.11 And so the question remains:
why should the pursuit of happiness by individuals command the sort of
respect that Kant™s principle of freedom requires? Why exactly is someone™s
pursuit of happiness an appropriate source of moral constraint for me (or
for anyone)? I can see why the exercise of a good will is something that
commands respect: in that case, the autonomy that answers the command
is an echo of the autonomy that elicits it. But there are all sorts of things
about a person “ the course of his dreams, his involuntary movements, or
the rate of his heartbeat “ that command no respect at all, because they
are empirically determined. Why does my pursuit of happiness belong in
the former rather than the latter category?
Maybe there is no answer to this question. Kant™s principle of freedom
is presented as a principle of external right, as a feature of his political
Moral Autonomy and Personal Autonomy 311

philosophy, not his moral philosophy.12 In that context the psychologi-
cal or metaphysical nature of its exercise may be irrelevant. Freedom in
Kant™s external principle may be purely negative (in Berlin™s sense):13
it protects a space for choice, but it offers little in the way of a positive
account of what ideally should go on in that space once the interference
of others is cleared away. As “the sum of the conditions under which the
choice of one can be united with the choice of another in accordance
with a universal law,” external right may take no interest in the content
or the character of the choices it protects.14
About the pursuit of happiness, then, there may not be much more
to be said than that I ¬nd myself making choices “ pursuing my own
happiness “ and that the formalities of universalization require that I re-
spect this also in all others. Kant recognizes that the pursuit of happiness
is more or less unavoidable as a feature of the human condition: “To be
happy is necessarily the demand of every rational but ¬nite being . . . ,
satisfaction with one™s whole existence is . . . a problem imposed upon
him by his ¬nite nature itself, because he is needy. . . .”15 It may not be a
moral end,16 but it is an end we have by natural necessity, an assertoric
imperative.17 So perhaps one could argue that any sort of respect for per-
sons is bound to include respect for their pursuit of happiness, inasmuch
as it is a necessary incident of their (human) being.
I wonder, though, whether it might be possible to go beyond this,
in the interpretation of Kant™s position. Granted that the principle of
freedom protects only choice as such, without any reference to its content
or character, and granted that the pursuit of happiness is not the main
¬eld in which we display our moral powers “ still, is there not, on Kant™s
account, anything remotely like autonomy involved in our ¬guring out
what our happiness consists in? I said at the outset that modern liberals
distinguish between personal autonomy and moral autonomy in order
to distance themselves from the latter, which they associate with Kant.
But I want to know whether there is anything approximating personal
autonomy in Kant™s account of happiness.
One thing that Kant emphasizes is the uniqueness (or at least the
idiosyncracy) of each person™s happiness:

Only experience can teach us what brings joy. Only the natural drives for food,
sex, rest, and movement, and (as our natural predispositions develop) for honor,
for enlarging our cognition and so forth, can tell each of us, and each only
in his particular way, in what he will ¬nd those joys; and, in the same way, only
experience can teach him the means by which to seek them. All apparently a priori
reasoning about this comes down to nothing but experience raised by induction
Jeremy Waldron

to generality, a generality still so tenuous that everyone must be allowed countless
exceptions in order to adapt his choice of a way of life to his particular inclinations
and his susceptibility to satisfaction. . . . 18

Allen Rosen invites us to take this as the basis of Kant™s principle of
freedom: “[B]ecause . . . no one else can decide what will make [a person]
happy, his right to pursue his own happiness cannot be usurped by a self-
appointed proxy (for example, a paternalistic government), but must
instead be exercised by the individual concerned.”19 Letting each pursue
happiness in his own way may not be respect for moral personality, but it
is respect for something like identity “ for each person™s uniqueness and
the particularity of his situation and experience.
Kant also puts weight in one or two places on a normative distinction
between action and passivity in the pursuit of happiness. As he articulates
the principle of freedom in “Theory and Practice,” he deplores the pas-
sivity of people who are subject to an of¬cial conception of happiness:
“the subjects, like minor children . . . , are constrained to behave only pas-
sively, so as to wait only upon the judgment of the head of state as to how
they should be happy. . . .”20 Paul Guyer has drawn attention to some re-
marks in Kant™s Re¬‚exionen that suggest that we are more content when
we view ourselves actively as authors of our happiness, rather than simply
having contentment wash over us.21 In this regard, we should also not
neglect the importance Kant accords to self-cultivation. Though this is
presented in the Groundwork as a moral duty,22 it is not just a matter of
moral perfectibility. It is a moral duty in relation to all aspects of one™s
natural potential:

He owes it to himself (as a rational being) not to leave idle and, as it were, rusting
away the natural predispositions and capacities that his reason can some day
use. . . . [A]s a being capable of ends . . . , he must owe the use of his powers not
merely to natural instinct but rather to the freedom by which he determines their

This responsibility covers physical as well as mental self-improvement,
though, as Kant goes on to say,

[w]hich of these natural perfections should take precedence, and in what pro-
portion one against the other it may be a human being™s duty to himself to make
these natural perfections his end, are matters left for him to choose in accor-
dance with his rational re¬‚ection about what sort of life he would like to lead and
whether he has the powers necessary for it (e.g., whether it should be a trade,
commerce, or a learned profession).24

This remarkable passage puts Kant almost in the company of those like
Humboldt and J. S. Mill who emphasize the importance of a person™s
Moral Autonomy and Personal Autonomy 313

taking responsibility for his own individuality and for the overall shape
of his life and career.25
Finally, we should consider Kant™s views about the role of reason in the
pursuit of individual happiness. His observation that bestial contentment
would be better secured by instinct than by reason is well known, though
what is usually inferred from this is that the function of reason has nothing
to do with happiness.26 In fact, however, Kant suggests that reason does
have a speci¬c role in this regard:

Certainly, our well-being and woe count for a very great deal in the appraisal of our
practical reason. . . . The human being is a being with needs, insofar as he belongs
to the sensible world, and to this extent his reason certainly has a commission for
the side of his sensibility which it cannot refuse, to attend to its interests, and to
form practical maxims with a view to happiness. . . . 27

It is important to see that reason™s commission here is not just a matter of
prudential calculation “ that is, of the ef¬cient relation of means to ends.
That would be a modest and familiar function that even Humean reason
could discharge, a function highlighted in Kant™s talk of “hypothetical
imperatives.”28 But Kant is quite skeptical about reason™s ability to deliver
in this regard. He suspects that the instrumentalities of happiness more
or less defy rational calculation.29 Christine Korsgaard has suggested that
the familiar picture in which nature supplies the ends and reason the
means may have to be given up as an interpretation of Kant.30 She says
we should take at face value a remark in the Groundwork where Kant
suggests that reason has “the presumption . . . , to think out for itself a
plan for happiness,” and that we should attend also to some remarks
in the essay Conjectural Beginnings of Human History that associate the
Biblical story of the Fall with man™s discovery in himself of “a power of
choosing . . . a way of life, of not being bound without alternative to a
single way, like the animals.”31 Certainly Korsgaard is right that it would
be a mistake to see Kant as conceding the realm of happiness to Bentham.
Though Kant talks in several places about happiness as comprising the
systematic satisfaction of inclinations,32 he denies that this is just a matter
of scheduling satisfactions on the utilitarian model. For one thing, the set
of possible desires that has to be taken into account in any calculus of one™s
future happiness is, if not in¬nite, then certainly radically indeterminate.
And for another, humans can pick and choose in an non-quantitative way
which desires they wish to give priority to in their pursuit of happiness. A
man who suffers from gout, Kant says, may choose intelligibly to opt for
the pleasures of port even at the cost of physical agony, which in quantity
and extent, far outweighs those pleasures on any utilitarian calculation.33
Jeremy Waldron

This is an intriguing example, for it seems to present “ in the domain of
earthly pleasures “ some sort of prototype or analog of the renunciation
of desire that our moral powers involve. To be sure, for the gout-sufferer,
desire (for relief from pain) is renounced for the sake of desire (for the
pleasures of port). But the renunciation is not dictated mechanically by
any calculus of inclination. It is dictated by a choice that controls and
disciplines inclination even for the sake of other inclinations that are
treated as incommensurate with the ¬rst.
Well, we should not exaggerate the signi¬cance of all this. What Kant
makes of the gout-sufferer example is murky, to say the least. We are not
presented here with a well worked-out conception of the role of reason in
the choice of ends, and the comments we have considered do not add up
to a theory of personal autonomy in the sense used by modern liberals.
Still, they point a little bit in that direction; I mean by this that they
point to something in the Kantian pursuit of happiness that is somewhat
more rigorous and somewhat more worthy of respect than (say) the mere
indulgence of appetites or the prudent satisfaction of inclinations. And
most intriguingly, from our point of view, they seem to do so by implicating
in the pursuit of individual happiness some of the capacities of practical
reason that are more commonly associated with the exercise of moral

We have spent same time considering how far something like the modern
notion of personal autonomy is implicated in Kant™s account of the pur-
suit of happiness. I now want to turn my attention in the other direction,
and consider how far something approximating Kantian moral autonomy
is implicated in modern liberal conceptions of personal autonomy.
In liberal philosophy, the principle that corresponds to Kant™s princi-
ple of freedom is the principle that individuals are entitled to form and
pursue their own conceptions of what makes life worth living. Sometimes
we express this negatively as a principle of state neutrality: the state must
be neutral on the question of what makes life worth living, or, as Ronald
Dworkin puts it “political decisions must be, so far as possible, indepen-
dent of any particular conception of the good life, or of what gives value to
life.”34 But there are reasons for holding the state to a neutrality principle
that are not centered on individual freedom or autonomy,35 and there
may be reasons for opposing neutrality that are not reasons for opposing
the af¬rmative principle with which I began this section.36 So I think it
Moral Autonomy and Personal Autonomy 315

better just to state the liberal principle straightforwardly: individuals are
entitled to form and pursue their own conceptions of what makes life
worth living.
Now the idea referred to in this principle “ a person™s conception of
what makes life worth living “ may be understood narrowly or gener-
ously. In its most generous sense, it includes anyone™s overall orientation
towards life no matter how crude or inchoate. Dworkin puts it like this:

Each person follows a more-or-less articulate conception of what gives value to
life. The scholar who values a life of contemplation has such a conception; so
does the television-watching, beer-drinking citizen who is fond of saying “This is
the life,” though he has thought less about the issue and is less able to describe
or defend his conception.37

On the most generous account, a person™s conception of the good is sim-
ply something revealed by his behavior. He drinks beer while watching
television, he does it cheerfully, and he does little else: therefore we im-
pute to him the thought that that is the good life. A principle that allowed
people to pursue their conception of the good in this most generous sense
amounts to little more than a principle of free action (subject of course
to the other-regarding proviso): “Each may do what he likes (so long as
that doesn™t impact upon similar freedom for others).” And the notion
that there is something important about conceiving what one likes as good
seems to play little role in this.
Compare with this the much narrower account of personal autonomy
in Joseph Raz™s book, The Morality of Freedom. For Raz, autonomy is not
just a matter of having values and revealing them in one™s choices. It is a
quite speci¬c notion of self-authorship:

An autonomous person is part author of his own life. His life is, in part, of his
own making. The autonomous person™s life is marked not only by what it is but
also by what it might have been and by the way it became what it is. A person is
autonomous only if he had a variety of acceptable options to choose from, and
his life became as it is through his choice of some of these options. A person who
has never had any signi¬cant choice, or was not aware of it, or never exercised
choice in signi¬cant matters but simply drifted through life is not an autonomous

Although he de¬nes autonomy in a way that contrasts with “drifting
through life without ever exercising one™s capacity to choose,”39 Raz con-
cedes that many of the most important things in our lives may be projects
we have grown up with, aspirations we discover we already have when
we ¬rst undertake autonomous deliberation.40 What matters for Razian
Jeremy Waldron

autonomy is not the genesis of our projects, but, ¬rst, that we recognize
the possibility now of abandoning or continuing to embrace them; sec-
ond, that when we choose among these options, we do so for reasons that
play a conscious role in our continuing practical deliberations; and third,
that we identify in good faith with the choices we have made.41
That notion of identi¬cation is a deliberate echo of an idea developed
by Harry Frankfurt and others. Frankfurt attributes great importance to
the capacity to stand back from one™s occurrent volitions, to consider
whether one wants to be in¬‚uenced by them, and to act at least some of
the time on the basis of these “second-order desires.”42 Though Frankfurt
does not explicitly associate this re¬‚ective self-evaluation with the idea of
autonomy, others have used it in their accounts. Gerald Dworkin, for
instance, identi¬es autonomy with the following condition:

It is only when a person identi¬es with the in¬‚uences that motivate him, assimi-
lates them to himself, views himself as the kind of person who wishes to be moved
in particular ways, that these in¬‚uences are to be identi¬ed as “his.”43

Dworkin™s account also makes it clear that in the context of a theory of
autonomy, the relevant second-order desires must refer the ¬rst-order
desires to the concept of one™s self. It is not enough for a second-order
desire to be motivated (say) by the frisson of pleasure that one knows
is characteristic of acting on a certain ¬rst-order desire. The conception
of who one is “ the sort of life one wants to lead, the sort of person one
wants to be “ is essential to the second-order re¬‚ection that constitutes
autonomy. Raz, I think, sees this too when he associates autonomy with
self-authorship. Though the autonomous person need not live a highly-
scripted existence, he is nevertheless a person who can relate the choices
he makes to some sense of the overall course of his life. He not only has
options and can carry them out, but he also understands their course and
signi¬cance on the matrix of “this life of mine,” and he chooses among
them on that basis. I guess it is possible that Ronald Dworkin™s “television-
watching, beer-drinking citizen who is fond of saying ˜This is the life,™”
quali¬es as autonomous under this criterion, but it™s also possible that he
does not. To qualify, “This is the life” would have to refer “ as it does not
often refer in colloquial discourse “ to some valued feature of the whole
shape of the person™s existence, not just the comfort of the moment.
Now the point here is not that Raz™s narrower conception of auton-
omy might exclude some of those who qualify as having a conception of
the good in Ronald Dworkin™s sense. The normative direction of Raz™s ac-
count is quite different: he is considering autonomy as a speci¬c ideal that
Moral Autonomy and Personal Autonomy 317

the state ought to foster, rather than as a principle that exhausts the state™s
duties in regard to the values held by its citizens.44 What is remarkable,
however, about this emphasis on second-order desire and identi¬cation
in modern liberal accounts of personal autonomy is its resonance with
conceptions of more strictly moral deliberation. It is characteristic of a
common Kantian conception of morality that I show my credentials as
a moral agent not by renouncing desire altogether but by being able to
stand back from my desires and consider whether they are the sort of
thing that I ought to be motivated by. It is characteristic, too, of this sort
of conception that morality is associated with authenticity: the delibera-
tion that considers whether a given desire is the sort of thing I ought to be
motivated by is deliberation which accesses my true self, a core of moral
authenticity that can defy desires and inclinations that are judged to be
alien to me. So personal autonomy is like moral autonomy in the kind of
deliberation and commitment that it emphasizes. In both cases, there is
an achievement of critical distance, in both cases there is re¬‚ection, and
in both cases this re¬‚ection involves the idea of “who I really am.”
I am not saying that the liberal conception of personal autonomy is
set up with this consonance in view. We have already seen that Raz, for
one, wants a sharp distinction between personal autonomy and moral
autonomy.45 But the fact that the two sorts of autonomy privilege capac-
ities of the same kind is helpful nonetheless. For it means that when the
demands of morality do enter the picture, we are not calling upon individ-
uals to engage in an utterly different kind of exercise from that involved
in their autonomous self-authorship. The introduction of moral consid-
erations is not the ¬rst moment at which desire is checked or spontaneous
inclination subjected to scrutiny. As personally autonomous, liberal indi-
viduals are already familiar with the idea of disciplining their inclinations
in the light of the sort of person they would like to be. All that happens
in the moral phase is that the image of such a person is conceived as
“one who lives on fair terms with others”: it is now in the light of this
self-image, rather than merely the self-image of “one who would like his
life to have such-and-such a character,” that inclination is checked and
desire subjected to scrutiny. Morality already has a toehold, and there is
no radical discontinuity between the modes of re¬‚ection and self-control
appropriate to the pursuit of happiness and the modes of re¬‚ection and
self-control required for submission to the right.
It is interesting to consider John Rawls™s account of individual concep-
tions of the good in light of these considerations. A theory of justice, on
Rawls™s account, de¬nes a framework within which each person will be
Jeremy Waldron

able to pursue his own conception of the good; and indeed the moral
equality of persons and their entitlement to justice is de¬ned partly in
terms of their capacity for a conception of the good.46 Now, we distin-
guished earlier between narrower and more generous understandings of
“conception of the good.” In Rawls™s theory, the pursuit of a conception
of the good seems to be de¬ned in a quite speci¬c way:

We are to suppose . . . that each individual has a rational plan of life drawn up
subject to the conditions that confront him. This plan is designed to permit
the harmonious satisfaction of his interests. It schedules activities so that various
desires can be ful¬lled without interference. It is arrived at by rejecting other
plans that are either less likely to succeed or do not provide for such an inclusive
attainment of aims.47

It is not at all clear that Ronald Dworkin™s “television-watching, beer-
drinking citizen” would have a conception of the good in this sense.
However, the speci¬city of this part of Rawls™s theory is misleading. Rawls
says he is not imposing the plan-like aspects of his theory of the good
as either an ideal for individuals or a condition of their entitlement to
freedom. The theory is constructive and hypothetical.48 He says that we
are to suppose that each individual has a rational plan of life, and I think
that means that something answering to Rawls™s elaborate description of
a plan of life can be imputed to each person for the purposes of a theory
of justice49 “ imputed to him on the basis of facts about his abilities,
circumstances, tastes, and so on. But in the end, for each individual the
imputation of a substantive plan of life is always subject to brute facts
about where he ¬nds happiness:

Thus imagine someone whose only pleasure is to count blades of grass in various
geometrically shaped areas such as park squares and well-trimmed lawns. . . . The
de¬nition of the good forces us to admit that the good for this man is indeed
counting blades of grass, or more accurately, his good is determined by a plan
that gives an especially prominent place to this activity.50

So the entitlement of each to justice, on Rawls™s account, does not de-
pend on a their ends having any particular substantive character, nor
does it depend on their attachment to those ends being distinguished by
any narrowly-de¬ned features of rational commitment. We may impute
certain features of rationality to a person in order to work out, for the
purposes of a theory of justice, what respecting each person™s pursuit
of his particular ends requires. But that is not the same as conditioning
respect for his ends on the rationality of his attachment to them.
Moral Autonomy and Personal Autonomy 319

Now, in A Theory of Justice, Rawls does not use the term “autonomy” to
characterize a person™s pursuit of a conception of the good. As we have
seen, the idea of personal autonomy in other theories is often associated
with a particular mode of connection between a person and his ends:
the pursuit of one™s ends represents a form of self authorship, and the
connection amounts to a person™s higher-order evaluation of his ends,
af¬rmative identi¬cation with them, and so on. But Rawls does not use
this language in his theory of the good. Instead, he associates autonomy
with morality or justice “ that is, with a person™s acceptance of something
like the proviso in Kant™s principle of freedom “ “provided he does not
infringe upon that freedom of others to strive for a like end,” and so on.51
Autonomy, for Rawls, is a way of characterizing the ability and willingness
of a person to submit his pursuit of his conception of the good to the
conditions necessary for the similar pursuit of conceptions of the good
by others.
The fact that Rawlsian autonomy is supposed to refer to a capacity
quite different from those implicated in an individual™s pursuit of the
good means that Rawls does face the dif¬culty we mentioned earlier. He
has to explain why individuals are willing to subordinate their pursuit of
the good to principles of right, and how that subordination is possible
given the discontinuity between moral autonomy and the pursuit of an
individual conception of the good.52 I don™t think Rawls ever gives an
adequate account of this in A Theory of Justice. He just asserts that peo-
ple are willing to subordinate their good to the demands of right, even
though there is nothing about their good “ or even about the construc-
tive character that Rawls imputes to individual conceptions of the good “
that explains this ability. At the very end of A Theory of Justice, Rawls says
that people will regard their moral powers as the most fundamental as-
pect of their selves.53 But this solves the problem (if it actually does) only
by reversing it. Now it is the formation of a personal conception of the
good that looks mysterious. If the true ground of one™s being lies in the
exercise of one™s moral autonomy, why would quite different capacities
be exercised in choosing a plan of life?
Intriguingly, we ¬nd a more complex “ and somewhat more adequate “
account of the relation between moral and personal autonomy in Rawls™s
later work.54 In Political Liberalism, Rawls introduces the idea of “rational
autonomy” to represent a person™s “moral power to form, to revise, and
rationally to pursue a conception of the good.”55 A person shows his
rational autonomy, according to the argument in Political Liberalism, not
just by identifying with his own particular ends, but by having and valuing
Jeremy Waldron

the ability to stand back from any ends he happens to have “ in other
words, by using the same capacities that Joseph Raz, Harry Frankfurt, and
Gerald Dworkin emphasized. Rawls contrasts this rational autonomy with
“full autonomy,” which continues to represent a person™s overall ability
to deliberate about, adopt, and comply with principles of justice.56 Now,
that may seem to add up to a contrast between personal autonomy and
moral autonomy. But, in fact, the sort of critical re¬‚ection that Rawls™s
rational autonomy involves is not valued primarily for its connection with
authenticity or self-authorship. Instead, it seems to be oriented mainly
towards the priority of the right over the good. The capacity to distance
onself from one™s ends is connected to a person™s ability to question his
attachment to his ends in the light of their implications for justice.57
In other words, it is calculated to have the effect that we noted in the
case of conceptions of autonomy that made use of Frankfurt™s idea. The
difference is that in Rawls™s case, this seems to be a deliberate strategy:
Rawls evidently now feels the need to use the form of personal autonomy
to explain the ef¬cacy of the demands of morality and justice.

So far we have considered what personal autonomy may require so far as
the form of an individual™s engagement with his ends is concerned. But
what about the substantive idea, which we also ¬nd in modern liberal the-
ory, that autonomy might be seen as engagement with the good? We talk
of an individual™s pursuing a conception of the good, and, although this
phrase may be used casually, still in its literal sense it does connote a cri-
terion of ethical if not moral judgment. Certainly it evokes the idea of the
subjection of a individual™s life to the discipline of objectively58 approp-
riate or inappropriate responses to the presence or absence of value.
Joseph Raz™s account places particular emphasis on this connotation,
with his insistence that “autonomy is valuable only if exercised in pursuit
of the good.”59
No one would deny that autonomy should be used for the good. The question
is, has autonomy any value qua autonomy when it is abused? Is the autonomous
wrongdoer a morally better person than the non-autonomous wrongdoer? Our
intuitions rebel against such a view. It is surely the other way round. The wrong-
doing casts a darker shadow on its perpetrator if it is autonomously done by

It follows, says Raz, that something can hardly be an abrogation of auton-
omy if it interferes only with the choice of valueless options: “a choice
Moral Autonomy and Personal Autonomy 321

between good and evil is not enough” for autonomy.61 From the subjec-
tive point of view as well, Raz thinks the exercise of autonomy represents
an individual™s attempt to engage with the good. A person P™s experience of
choosing between two ways of life, A and B, is P™s experience of judging
which of them it would be better for him to have; and Raz thinks it follows
that anyone who chooses A on this basis is necessarily committed to the
thought that if it turns out that A is undesirable (that is, if it turns out that
A lacks the value that he thought it had when he chose it), then he no
longer has a reason to pursue it. And this remains true, Raz insists, even
if P is still convinced (mistakenly) that A is valuable, for no one thinks it
is a good idea to pursue a course of life under a misapprehension about
its value.62
Now, this is not the place to discuss the perfectionism that Raz es-
tablishes on these premises.63 For our purposes, what is signi¬cant is
the bridge that this seems to establish between personal autonomy and
morality. Raz acknowledges that his account of personal autonomy ends
up sounding like “a very rigoristic moral view,”64 and that a person who
lacks the power of moral discernment will not have what it takes to be
personally autonomous. So now it™s not just the form of commitment “
critical self-re¬‚ection “ but the substance of evaluation that links the
two forms of autonomy. One may go even further. On David Johnston™s
interpretation, “Raz believes that a sense of justice is part of personal au-
tonomy in the sense that a person who is personally autonomous would
want to avoid doing things that are unjust.”65 I am not sure that I see
this explicitly in The Morality of Freedom. But I™m also not sure that Raz can
avoid it, given the moral dimension of personal autonomy on his account
and his view that “all aspects of morality derive from common sources.”66
If the proper use of autonomy is to choose between good options, and
not between good options and evil options, then it is hard to see how an
unjust choice can be regarded as a genuine exercise of personal auton-
omy. This means that in the exercise of one™s personal autonomy, one
may already be making judgements about justice. And it may follow in
turn from that that each individual associates his personal autonomy with
the criteria of justice that he uses in making these judgments. If this is
true, then “ despite Raz™s insistence on a distinction between personal
and moral autonomy67 “ we are going to have to rethink the relation
between them. The exercise of personal autonomy can no longer be con-
ceived merely as the subject-matter of moral autonomy “ in the sense
that moral autonomy is about reconciling one individual™s personal au-
tonomy with another™s. The task of reconciliation has now become more
Jeremy Waldron

complicated and re¬‚exive, inasmuch as my exercise of personal auton-
omy may now already involve a view about the appropriate way to recon-
cile it with yours.
Raz™s theory has its peculiarities, but I think the points I have made
can be generalized to any liberal view that associates personal autonomy
with the pursuit of a conception of the good. After all, the good does not
lose its normativity or its connection to other normative ideas by being
involved in an individual conception of what makes life worth living. It
would be odd for persons to inform their choices with a conception of
the good without inferring some signi¬cant consequences as to what it is
appropriate for them to do.68 And then we have to ask: how far does this
go? Is an individual™s conception of the good capable also of generating
conclusions about what he ought to do, what it is right for him to do,


. 10
( 13)