. 11
( 13)


what he is required or obligated to do? If it is capable of generating these
conclusions, then what becomes of the fabled priority of the right over
the good in liberal theory?
There are several ways in which we may understand a claim that the
right has priority over the good. (1) It may be understood as a claim
about the deontological character of political morality “ as a denial of
consequentialism, for example. The idea here would be that we do not
construct our political morality by ¬guring out, ¬rst, what is valuable
and, second, how best to promote it; instead, we establish certain moral
absolutes “ rights, for example “ deontologically, without reference to the
goals that it might be worthwhile for individuals or societies to pursue.69
Or (2) the priority of the right over the good may be understood as a
claim about the relation between individual aspirations and the social
demands of morality. People have their own individual conceptions of
the good, but these are subject to the demands of right conceived as a
system of morally reconciling individual ends:

The principles of right, and so of justice, put limits on which satisfactions have
value; they impose restrictions on what are reasonable conceptions of one™s good.
In drawing up plans and in deciding on aspirations men are to take these con-
straints into account.70

In principle, these two versions of the priority are distinct. One might
have a teleological conception of social morality and still insist on (2) “
that is, on the priority of that conception over individual ends: a utilitar-
ian might say, for example, that an individual is not entitled to pursue
his own happiness when the exigencies of the general happiness require
otherwise. Or there might be deontological elements in the conceptions
Moral Autonomy and Personal Autonomy 323

that individuals pursue and identify with, but these might still be subor-
dinated to social morality (however that is con¬gured).
This latter possibility is of particular interest. Suppose that individ-
ual conceptions of the good yield elements of deontological right or
requirement. What is the relation of these personal deontological ele-
ments to the demands of morality conceived as a social enterprise of
reconciling individual interests? Or suppose that personal autonomy is
not conceived primarily in terms of “good” at all, but ab initio in terms
of right or requirement? What is the relation between the moralism that
personal autonomy might exhibit and the social morality of a theory of
Another way of approaching the same issue is to ask about the rela-
tion between autonomy and conscience. For most of this chapter, I have
presented personal autonomy as a matter of self-determination in the
pursuit of value in one™s life: one chooses to pursue knowledge, for ex-
ample, or love or aesthetic excellence. On another account, however,
personal autonomy is a matter of responding to something that presents
itself in one™s inner life as an imperative. “Here I stand; I can do no
other” has quite a different ¬‚avor from (say) “I have chosen to ¬nd
value and enjoyment in a life of literary achievement.” Perhaps the two
may be assimilated if we associate “I can do no other” with the ethics
of authenticity: “I just have to respond to my inner urgings. I have no
choice but to be a poet.”71 But if conscience is understood as a subjective
representation of law-like requirements rather than the subjective repre-
sentation of personal destiny, then it is a little more dif¬cult to relate it
to the modern notion of autonomy. For then it is no longer just a matter
of self-determination, or of what I make of myself. Instead it is a matter
of my heeding or not heeding an inner representation of “what is to be
done.” I am not saying it is utterly independent of the notion of the self:
conscience associates itself with integrity, and there is a sense that failing
to heed its moralistic demands amounts to a form of self-betrayal. Still,
the relation of the self to these demands is now much more complex. The
demands are not thought of as originating from the self, for all that they
are bound up with integrity. Instead, the demands represent the self™s
participation in a moral order de¬ned in a way that is independent of
it.72 If personal autonomy takes on this character, it is going to be much
harder to keep it from usurping the role played by a social theory of right.
It is tempting to respond that this sort of autonomy “ the sense of
autonomy associated with conscience “ is to be ¬led under “moral auton-
omy” and kept strictly apart from the modern liberal notion of “personal
Jeremy Waldron

autonomy.” And then one would say that a constructive theory of the right
is supposed to model the claims of conscience, or to stand in re¬‚ective
equilibrium with them, but not to accept them as inputs, in the way that it
accepts individual conceptions of the good. One can concede that in the
real world, the broad moral powers of actual individuals will tangle these
things together “ plan of life, commitment to values, views about right-
ness, and deliverances of conscience. But “ according to the response I
am now considering “ that just makes it all the more important to draw
the theoretical distinction, even if some quite delicate dissection is nec-
essary to distinguish those elements that are properly analyzed under
the auspices of “individual conception of the good” and those elements
that are properly analyzed under the auspices of “individual views about
morality and justice.”
But I don™t think this response works; it is no longer enough to defuse
the challenge posed here by the phenomenology of conscience. For the
question now is whether the distinction between personal and moral
autonomy actually stands up “ that is, whether the phenomenon that the
liberal describes as forming and pursuing a conception of the good can
be kept apart from the phenomenology of experiencing and responding
to moral demands. John Rawls™s theory, for example, requires there to
be not only a verbal distinction here, but a real distinction in terms of
modes of construction and criteria of validity. Apart from anything else,
he says, we do not want the sort of proliferation in the realm of right that
we have “ and welcome “ in the realm of individual good:

[I]t is, in general, a good thing that individuals™ conceptions of their good should
differ in signi¬cant ways, whereas this is not so for their conceptions of the
right. . . . In a well-ordered society, . . . the plans of life of individuals are differ-
ent in the sense that these plans give prominence to different aims, and persons
are left free to determine their good, the views of others being counted as merely
advisory. . . . But the situation is quite otherwise with justice: here we require not
only common principles but suf¬ciently similar ways of applying them . . . so that
a ¬nal ordering of con¬‚icting claims can be de¬ned.73

Rawls acknowledges that some have taken a different approach: “They
have suggested that autonomy is the complete freedom to form our moral
opinions and that the conscientious judgment of every moral agent ought
absolutely to be respected.”74 But he says this is mistaken. When people
make assertions of right based on conscience, “[h]ow do we ascertain
that their conscience and not ours is mistaken . . . ?”75 In matters of the
good, a view gets some credentials from the mere fact that it is held by
an individual: individuals are, after all, self-originating sources of moral
Moral Autonomy and Personal Autonomy 325

claims.76 But in matters of the right, an individual™s personal attachment
to a view counts for nothing at all, on Rawls™s approach. In matters of
right, correctness is associated exclusively with what people would agree
to in the original position, not with the conscience or intuitions of the
participants. Unless Rawls can hold the line between conceptions of the
right and conceptions of the good, he cannot stem this proliferation,
nor can he justify holding conceptions of the right to quite the same
standards as those to which conceptions of the good are held. But the
normative implications of “the good” seem to threaten this distinction,
all the way down the line.

We seem to have identi¬ed the horns of a dilemma. On the one hand,
we have the position that if too sharp a distinction is drawn between
personal autonomy and moral autonomy, we cannot explain how the
former is subordinated to the latter. On the other hand, we have the
position that if the line between them is blurred, then there is the prospect
of a proliferation of standards of right “ one per person “ with individuals
committed to them in a way that seems to command the same respect as
their commitment to their conceptions of the good. Clearly what is called
for is some sort of moderate or intermediate position. This chapter has
not sought to identify that third way, but I hope it has helped illuminate
some of the perils as well as some of the advantages of a distinction
between personal and moral autonomy.
A sharp distinction between the two seems indispensable for analytic
clarity. But if we erect too high a wall of separation, we conceal the com-
mon features and analogies that have prompted the use of the same term
“autonomy” in both cases. We will miss the various ways in which the two
ideas are interrelated “ not only the fact that personal autonomy is often
the subject-matter of moral autonomy, as morality attempts to reconcile
one person™s autonomous pursuit of his ends with others™ autonomous
pursuit of theirs, but also the fact that individuals™ exercise of personal
autonomy must be amenable to the demands of morality, and their per-
sonal autonomy must be capable of being integrated with the exercise
of their moral autonomy, normally understood as equally indispensable
to their individual being. And we will miss, too, the damage that per-
sonal autonomy may do to moral autonomy when, through conscience
or through the normativity of its own value-conceptions, it challenges
Jeremy Waldron

the latter™s attempt to monopolize the realm of the right and the

1. There is a useful discussion of this distinction in David Johnston, The Idea of
a Liberal Theory (Princeton, NJ: Princeton University Press, 1994), pp. 72“7.
2. See, for example, Joseph Raz, The Morality of Freedom (Oxford: Clarendon
Press, 1986), p. 370 on: “Personal autonomy, which is a particular ideal of
individual well-being should not be confused with the only very indirectly
related notion of moral autonomy.”
3. “On the Common Saying: That May Be Correct in Theory But It Is of No Use
in Practice” (hereinafter “Theory and Practice”), in Immanuel Kant, Practical
Philosophy, translated and edited by Mary Gregor (Cambridge: Cambridge
University Press, 1996), 277, at p. 291. For convenience, I will add to all
citations from Kant a parenthetical page-reference to the relevant volume of
the standard Prussian Academy edition of Kant™s works. The reference for
this passage is p. 290 of volume 8, or, as I shall abbreviate it (8: 290).
4. Immanuel Kant, Groundwork of the Metaphysics of Morals, in Kant, Practical
Philosophy, 41, at pp. 83 ff. (4: 433 ff ).
5. Immanuel Kant, Critique of Practical Reason, in Kant, Practical Philosophy, 133,
at p. 168 (5: 25).
6. Kant, Groundwork, p. 59 (4: 405).
7. Immanuel Kant, The Metaphysics of Morals, in Kant, Practical Philosophy, 353,
at p. 393 (6: 237).
8. Kant, “Theory and Practice” p. 291 (8: 290); my emphasis.
9. Ibid., p. 285 (8: 283).
10. See Paul Guyer, Kant on Freedom, Law and Happiness (Cambridge: Cambridge
University Press, 2000), p. 98.
11. The same is true of a number of other ways in which happiness and morality
are connected in Kantian moral philosophy. In the Groundwork, p. 49 (4:
393), Kant suggests that one has to have a morally good will to be worthy of
happiness. A little later “ ibid., p. 52 (4: 396) “ he de¬nes “moral happiness”
as “satisfaction with one™s person and one™s own moral conduct.” Third, there
is happiness conceived of as the reward for the blessed in the life to come;
see Kant, Metaphysics of Morals, p. 519 (6: 387). These suggestions, however,
are not what we are looking for. Though they do connect autonomy with the
pursuit of happiness, they do so in a backhanded way.
12. See Alexander Kaufman, Welfare in the Kantian State (Oxford: Clarendon
Press, 1999), p. 39, for the suggestion that Kant™s principle of freedom in
“Theory and Practice” is to be read as an attack on contemporary cameralism,
which held that the ultimate aim of every republic is the common happiness.
13. See Isaiah Berlin, “Two Concepts of Liberty,” in his collection Four Essays
on Liberty (Oxford: Oxford University Press, 1969). For a slightly different
contrast between positive and negative freedom in moral philosophy, see
Kant, Groundwork, p. 94 (4: 446).
Moral Autonomy and Personal Autonomy 327

14. Kant, Metaphysics of Morals, p. 387 (6: 230).
15. Kant, Critique of Practical Reason, p. 159 (5:25)
16. However, as Kant points out, “[t]o assure one™s own happiness is a duty (at
least indirectly); for, want of satisfaction with one™s condition, under pressure
from many anxieties and amid unsatis¬ed needs, could easily become a great
temptation to transgression of duty.” [Kant, Groundwork, p. 54 (4: 399); emphasis
in original.]
17. For “assertoric imperative,” see Kant, Groundwork, p. 68 (4: 415“6).
18. Kant, Metaphysics of Morals, p. 371 (6:215“6). See also Kant, Critique of Practical
Reason, p. 159 (5: 25).
19. Allen D. Rosen, Kant™s Theory of Justice (Ithaca: Cornell University Press,
1993), pp. 70“1.
20. Kant, “Theory and Practice,” p. 291 (8: 290“1).
21. Guyer, Kant on Freedom, Law and Happiness, pp. 111“2.
22. Kant, Groundwork, pp. 74“5 (4: 422“3).
23. Kant, Metaphysics of Morals, p. 565 (6: 444).
24. Ibid., p. 566 (6: 445).
25. Cf. John Stuart Mill, On Liberty, ed. Currin V. Shields (Indianapolis: Bobbs
Merrill, 1956), pp. 69“72.
26. Kant, Groundwork, p. 51 (4: 395).
27. Kant, Critique of Practical Reason, p. 189 (5: 61). See also the discussion in John
Rawls, Lectures on the History of Moral Philosophy (Cambridge, MA: Harvard
University Press, 2000), p. 232.
28. Cf. David Hume, A Treatise of Human Nature, second edition, eds. L. A. Selby-
Bigge and P. H. Nidditch (Oxford: Clarendon Press, 1978), pp. 413 ff. For
“hypothetical imperatives,” see Kant, Groundwork, pp. 67“8 (4: 414“5).
29. This is partly because of the indeterminacy of the ends that constitute our
happiness, and partly because of the unpredictable vicissitudes of the empir-
ical world. Kant says in Groundwork, pp. 70“1 (4: 418):

[I]t is impossible for the most insightful and at the same time most powerful but still
¬nite being to frame for himself a determinate concept of what he really wills here. If
he wills riches, how much anxiety, envy, and intrigue might he not bring upon himself
in this way! . . . If he at least wills health, how often has not bodily discomfort kept
someone from excesses into which unlimited health would have let him fall, and so
forth. . . . One cannot therefore act on determinate principles for the sake of being
happy, but only on empirical counsels . . .

30. Christine M. Korsgaard, Creating the Kingdom of Ends (Cambridge: Cambridge
University Press, 1996), pp. 111“3.
31. Kant, Groundwork, p. 51 (4:395). The passage from Conjectural Beginnings of
Human History is cited in Korsgaard, Creating the Kingdom of Ends, p. 112
32. For example, Kant, Critique of Practical Reason, p. 199 (5:73).
33. Kant, Groundwork, p. 54 (4: 399). See also the discussion in Victoria S. Wike,
Kant on Happiness in Ethics (Albany: State University of New York Press, 1994),
pp. 6“13.
34. Ronald Dworkin, “Liberalism,” in A Matter of Principle (Cambridge, MA:
Harvard University Press, 1985), 181, at p. 191.
Jeremy Waldron

35. The state may be incompetent to make decisions about what makes life worth
living: see Bruce Ackerman, Social Justice in the Liberal State (New Haven: Yale
University Press, 1980), p. 12.
36. There may be aspects of the state™s other duties that make neutrality
impossible: see the discussion in Jeremy Waldron, “Legislation and Moral
Neutrality,” in Liberal Rights: Collected Papers 1981“1991 (Cambridge: Cam-
bridge University Press, 1993), 143, at pp. 149“50.
37. Dworkin, “Liberalism,” p. 191. See also Stephen Macedo, Liberal Virtues: Citi-
zenship, Virtue, and Community in Liberal Constitutionalism (Oxford: Clarendon
Press, 1990), p. 253: “Explicitly or not, liberal regimes endorsee and promote
autonomy. But we still respect the non-autonomous: people have the right
to lead lazy, narrow-minded lives . . .”
38. Raz, The Morality of Freedom, p. 204; see also ibid., p. 369.
39. Ibid., p. 371.
40. Ibid., pp. 290“1.
41. Ibid., p. 382. I have adapted part of this paragraph from Jeremy Waldron,
“Autonomy and Perfectionism in Raz™s The Morality of Freedom,” Southern
California Law Review, 62 (1989), 1097.
42. Harry G. Frankfurt, “Freedom of the Will and the Concept of a Person,”
Journal of Philosophy, 68 (1971), 5“20, at p. 7.
43. Gerald Dworkin, “The Concept of Autonomy,” in John Christman (ed.) The
Inner Citadel: Essays on Individual Autonomy (New York: Oxford University
Press, 1989), 54“62, at p. 60.
44. See Raz, The Morality of Freedom, pp. 391“4.
45. See note 2.
46. John Rawls, A Theory of Justice, (Oxford: Oxford University Press, 1971),
p. 505.
47. Ibid., p. 93.
48. See ibid., pp. 423“4.
49. For an account of the role played in a theory of justice by this theorizing
about individuals™ conceptions of the good, see ibid., pp. 396“8.
50. Ibid., p. 432.
51. Ibid., p. 515. The Kantian character of the account is explicit at ibid.,
pp. 252 ff.
52. Sandel, Liberalism and the Limits of Justice, p. 157, states the problem as follows:

[W]e need to know in greater detail how the constraint of justice makes itself felt,
how exactly it enters into the deliberation of the agent. Are the constraints of right
somehow built into the activity of deliberation such that only just desires or conceptions
of the good can arise in the ¬rst place, or does the agent form values and aims based
on certain unjust desires only to suppress them in practice or set them aside once it
becomes clear that they violate justice?

53. Rawls, A Theory of Justice, pp. 561“3.
54. I say “intriguingly” because in his later work, Rawls is usually at pains to
minimize the distinctive or philosophically controversial aspects of the con-
ceptions he deploys. Thus, for example, he rejects some of the distinctive
features of Joseph Raz™s conception of autonomy because they are part of a
Moral Autonomy and Personal Autonomy 329

comprehensive philosophical conception and thus unsuitable for a political
theory of justice: see John Rawls, Political Liberalism, revised edition (New
York: Columbia University Press, 1996), p. 135n.
Rawls, Political Liberalism, p. 72.
Ibid., pp. 75“81.
Ibid., pp. 73“4.
Or quasi-objectively: the argument here is presented without prejudice to
the debate about moral realism. [See Jeremy Waldron, “The Irrelevance of
Moral Objectivity,” in Robert George (ed.) Natural Law Theory: Contemporary
Essays (Oxford: Clarendon Press, 1992), 158, at pp. 165“7.]
Raz, The Morality of Freedom, p. 381.
Ibid., p. 380.
Ibid., p. 379.
Ibid., pp. 140“2.
Ibid., p. 417: “The autonomy principle permits and even requires govern-
ments to create morally valuable opportunities, and eliminate repugnant
ones.” See Waldron, “Autonomy and Perfectionism,” pp. 1127 ff.
Ibid., p. 381. (It™s an impression he mitigates only because of the plurality of
goods: there are multiple valuable ways of living one™s life.)
Johnston, The Idea of a Liberal Theory, p. 78.
Raz, The Morality of Freedom, pp. 161 and 213“6.
See note 2.
See R. M. Hare, The Language of Morals (Oxford: Clarendon Press, 1952),
chapter 8.
See Robert Nozick, Anarchy, State and Utopia (New York: Basic Books, 1974),
pp. 28 ff.; for an attack on this proposition, see Charles Taylor, “Atom-
ism,” in his collection Philosophy and the Human Sciences: Philosophical Papers 2
(Cambridge: Cambridge University Press, 1985), pp. 187“210.
Rawls, Theory of Justice, p. 31.
Cf. Charles Taylor, The Ethics of Authenticity (Cambridge, MA: Harvard Uni-
versity Press, 1991), pp. 26 ff.
In this connection, we must not neglect the fact that in the real world, people
associate this conscientious aspect of their autonomy with the demands of
their religion “ the demands of God, even “ and that they see this as a
consummation of their integrity, not as something that detracts from it.
Rawls, Theory of Justice, pp. 447“8
Ibid., p. 518.
Ibid., p. 515.
Rawls, Political Liberalism, pp. 32“33 .

Autonomy, Self-Knowledge, and Liberal Legitimacy

John Christman

In the Enlightenment tradition of the justi¬cation of political authority,
institutions of state power are seen as legitimate only if such institutions
can be freely supported by those living under them. Liberal legitimacy,
then, assumes that autonomous citizens can endorse the principles that
shape the institutions of political power. The conception of autonomy
functioning in such a picture, moreover, requires that such citizens uni-
formly enjoy the capacity to rationally re¬‚ect upon and critically appraise
their own values, moral commitments, and political convictions. In this
way, political power is an outgrowth of autonomous personhood and
This traditional understanding of political legitimacy has been chal-
lenged from any number of directions, most notably from those who
charge that the picture of the autonomous person underlying the mech-
anism of authority is parochial, exclusionary, and in tension with the
sought-for legitimacy it is used to support.1 In this last vein, it can be
charged that the requirements of general support for principles of jus-
tice in a modern, pluralistic society are in tension with the assumptions
concerning individual autonomy underlying that concept. For the prob-
lem facing liberal conceptions of justice and legitimacy is that politi-
cal power can be seen as justi¬ed only when supported by autonomous
citizens, but the requirements of autonomy, in many construals of that
term, are too stringent to be met by the majority of citizens bound by
political institutions. Or, in other versions of this critique, the condi-
tions set out for autonomy refer at best only to some in the population
and not others, thereby valorizing certain personality types, value per-
spectives, and social positions over others. So modern institutions fail
Autonomy, Self-Knowledge, and Liberal Legitimacy 331

to achieve the desired legitimacy. It is this concern that I will deal with
More speci¬cally, the dif¬culty I want to examine here is that for polit-
ical institutions to be legitimate, citizens living under them must achieve,
for example, a level of self-knowledge and re¬‚ective self-endorsement
that most fail to meet and that, in fact, would run counter to other pro-
cesses of value commitment and moral obligation that motivate our moral
choices. Two parallel questions arise at this point: what exactly are the
conditions of autonomy that best support the role that concept plays in
principles of justice and legitimacy? And what reasons are there for as-
suming that citizens expressing endorsement of political institutions (of
the sort required by liberal legitimacy) be autonomous in this way, espe-
cially when the conditions of such autonomy do not obtain universally
for all in the population?
I will approach these issues by ¬rst focusing on the concept of au-
tonomy, where I will examine the general pattern that theorists of that
notion have followed, and propose a particular view on the concept™s
meaning, at least as it might be used in the context of liberal political
theory. The problems that have been raised about seeing autonomy in
this way “ in particular that it would demand certain capacities and prac-
tices that are at once dif¬cult to achieve for most of us as well as being
disruptive of our most basic value commitments “ will be noted. Indeed, I
will add to the usual chorus of complaints on this score, pointing out the
ways that some understandings of autonomy may require levels of self-
understanding and re¬‚ection that few of us ever achieve (or would want
to achieve). Nevertheless, I want to suggest that the process of legitimat-
ing principles of justice in the liberal tradition require seeing autonomy
in this way. That is, despite the fact that people generally do not exhibit
levels of self-knowledge that some conceptions of autonomy assume, it is
nonetheless important to treat them as the fundamental representatives
of their own values and commitments, and it is correspondingly impor-
tant to ask them to re¬‚ectively appraise those commitments as part of the
process of giving reasons that political legitimacy demands.
To keep track of the rather circuitous route I will be taking through
these issues, let me lay out the plan: I will ¬rst discuss the concept of
autonomy; in doing so, I will propose a version of that concept that takes
competence and the capacity for self-re¬‚ection as central. Then I will con-
sider problems with such requirements, in that understanding autonomy
this way appears to assume a level of self-knowledge that most people can-
not achieve. Moreover, acts of re¬‚ection can in some ways disturb moral
John Christman

commitment and manifest aspects of personhood that are not de¬nitive
of the most settled aspects of the self. With these challenges laid out, I
then turn to political theory, in particular to the requirements for the
legitimacy of political authority in the liberal tradition. In doing so, I will
make some general claims about the nature of liberalism, in particular its
commitment to pluralism, rejecting certain forms of perfectionism, and
in requiring citizen endorsement for all legitimate state institutions and
the principles that guide them (the so-called “endorsement constraint”).
I then distinguish two importantly different strains in liberal thinking “
one in which legitimacy is established as a result of self-interested bargain-
ing for the purposes of establishing stable social environments (within
which citizens can pursue valued projects), and the other in which legiti-
macy is seen as grounded in a moral commitment to political institutions
resting on mutual respect and reciprocity. And I support the latter view
of political justi¬cation over the former. I then return to the question
of the nature of autonomy, where I will claim that the mechanisms for
establishing legitimacy in the strand of liberalism worth defending need
not attribute levels of self-knowledge to citizens that they are unable sys-
tematically to meet (or if they are, they must be treated as meeting them
nonetheless). And re¬‚ective self-appraisal of the sort demanded by lib-
eral legitimacy is not problematic in the ways that our earlier concerns
pointed to.
In the end, then, the kind of autonomy assumed in the mechanisms of
liberal legitimacy does not assume levels of self-knowledge or capacities
of re¬‚ection that citizens either cannot or would not want generally to

I The Conditions of Autonomy
Various conceptualizations of autonomy have been put forward, and the
contrasts among these highlight differences in the way that this concept
operates in different theoretical terrains.2 In certain contexts, stress has
been placed on the way that autonomy has traditionally rested on a single
and parochial conception of the self “ one, for example, that assumed
a “true” or “core” self residing inside of us like an “inner citadel.”3 But
as many have pointed out, there are several reasons to avoid reference
to a singly conceived notion of a self in models of autonomy. For there
are far too many contrasting conceptualizations of our selves relevant
in various settings and relative to various needs for any one of them to
unproblematically count as our authentic core. Our embodiment, for
Autonomy, Self-Knowledge, and Liberal Legitimacy 333

instance, is sometimes the most prominent aspect of our person (in
medical settings, for example), whereas in others our identi¬cation as
a member of a particular group, religion, culture, or ethnicity is salient.
Moreover, as communitarian critics of liberalism have repeatedly stressed,
our identities are often constituted by our deepest value commitments.4
But these foci of selfhood vary from context to context and hence can-
not, singly, play the role of the “true self” of which autonomy is meant
to be an expression. So insofar as a conception of autonomy assumes a
model of selfhood that features one of these aspects to the exclusion
of the others, it can rightly be labeled as overly narrow and hence
Some theorists have therefore approached autonomy, not as the op-
eration of a core set of identity-creating characteristics, but rather as
a range of capacities, competences, and functions. This “functional” ac-
count of autonomy may be in a better position to avoid charges of narrow-
ness that have plagued more traditional notions.5 Such accounts focus
on a number of conditions that manifest the “self-government” of the
person, while at the same time acknowledging the deeply embedded,
interpersonally constructed, and historically situated nature of the self.
The ¬rst set picks out those characteristics by which a person effectively
makes competent decisions: rationality, self-control, freedom from psy-
chosis and other pathologies, access to minimally accurate information,
motivational effectiveness, and the like. The second set refers to require-
ments that the person™s values and decisions are truly her own; these
most often include the condition that persons re¬‚ect on their personal
characteristics6 and identify with (or at least not feel deeply alienated
from) them. Whereas the ¬rst family of requirements ensures that the
autonomous person effectively acts (rules), the second guarantees that
the ruling is truly her own. Thereby the self-rule promised by the etymol-
ogy of the word “autonomy” is established.
So on the view offered here, autonomy requires that the person be
able to submit the factors of her personality to critical self-re¬‚ection.7
This requires that factors relevant to identity, decision, and choice be
such that, hypothetically, the person could re¬‚ect upon them without
repudiation in light of how they came about. In this way, the autonomous
person is competent (in the ways described) as well as authentic in the
sense of being moved by values that would withstand self-scrutiny.
Note also the reference to the history of the agent relative to the trait
in question. I have argued in earlier work that the processes by which a
person develops a trait are relevant to her autonomy vis-` -vis that trait.8
John Christman

The way that attention to personal history should be captured is that a
person cannot be labeled autonomous if some aspect of the manner in
which a characteristic is developed would, if known, cause her to disavow
that trait, to become deeply alienated from it. Let us say that a person
discovered that the only reason she remains so devoted to her revolu-
tionary activities is that she was kidnaped and tortured at an earlier time
(a memory she had suppressed until now). Her autonomy is clearly in
question if, were she to realize how these attitudes came about, she would
disavow them. However, what matters is the person™s relation to the atti-
tude or characteristic given its etiology rather than her attitude toward
that etiology simpliciter : I might think that the way I was raised was
too restrictive, but I accept the way I turned out nonetheless, because
it wasn™t so restrictive that I want to reject or disavow the character traits
that developed from it.
The requirement of self-re¬‚ection demands that the person is au-
tonomous (relative to some factor) if, were piecemeal re¬‚ection in light
of the history of the factor™s development to take place, she would not feel
deeply alienated from the characteristic in question. To be alienated from
some aspect of oneself is to experience negative affect relative to it, and
to experience diluted or con¬‚icted motivation stemming from it, and to
feel constricted by it, as though by an external force. It is, moreover, to
feel a need to repudiate that desire or trait, to reject it and alter it as much
as possible, and to resist its effects. If I re¬‚ect on some addiction I have,
for example “ one that I did not bring upon myself voluntarily “ I view it
as distanced from me, as something about which I feel regret or dismay
and that is less than fully motivating (relative to non-alienated desires).9
Moreover, the re¬‚ection required of autonomous agents is considered to
be piecemeal, requiring that agents re¬‚ect on particular aspects of their
character without ever presupposing the ability to look at the whole of
themselves from a completely disembodied perspective.
Further, a mere capacity to re¬‚ect is too weak: if a person has a ca-
pacity to re¬‚ect on herself but never does, and some of her ¬rst-order
traits would be unacceptable to her if she did, we would not call her au-
tonomous as she continues blithely to act on the basis of those traits.10 It
is not merely that the person can re¬‚ect but that, were she to do so, she
would not feel alienated in the manner described. Moreover, the capacity
to re¬‚ect alone, even if exercised, seems insuf¬cient to pick out a mean-
ingful conception of autonomy. An unwilling addict, who may be unable
to resist the debilitating grip of his destructive cravings, but nevertheless
retains a tragically robust ability to re¬‚ect on his life and take in all of
Autonomy, Self-Knowledge, and Liberal Legitimacy 335

its de¬ciencies, is not autonomous despite this tragic self-knowledge. So
an autonomous person must be able to alter those characteristics toward
which she feels resistance, alienation, and repugnance.11
Non-alienation is also a different condition from the familiar require-
ment of identi¬cation, which one typically ¬nds in discussions of auton-
omy. On the one hand, I can feel no alienation toward a characteristic
but not fully identify with it, in the sense of wholehearted endorsement
without regret.12 We all contain some measure of internal con¬‚ict and
complexity, and an attitude of ironic acceptance of the tensions of our
own psyches is inevitable, and perhaps healthy, in a multi-dimensional
and perplexing world. But to be alienated in the sense I mean here is
to be actively derisive of some aspect of the self, to want to reject and
resist it. An alienated person feels no af¬nity with such traits, wants to
change or, if that is not feasible, distance herself from them; she is a
divided and con¬‚icted person, and is unable to present a minimally set-
tled sense of herself to others in practical discourse. On the other hand,
non-alienation is stronger than identi¬cation when the latter is consid-
ered as mere acknowledgment: I can admit that a trait is, alas, part of
my identity (especially in my motivational structure), but still not want to
repudiate and distance myself from it. Therefore, on the present view, a
person is not autonomous relative to those aspects of herself that would
produce such feelings of self-repudiation were she to re¬‚ect on them in
light of how they came about. (Notice also how non-alienation adds an
affective element to autonomy, in contrast to the picture of the disengaged
cognizer described in our earlier discussion of reasons-responsiveness.)13
One ¬nal point: for a person to be autonomous on this model, the
hypothetical re¬‚ection being considered cannot itself operate under
the in¬‚uence of factors that effectively prevent normal self-awareness.
This prevents the possibility of a regress when considering the ways in
which manipulative factors constrain both choice and re¬‚ection.14 So
self-re¬‚ection “ even the hypothetical re¬‚ection being considered here “
cannot be the result of distorting factors that guarantee that the self-
appraisal in question has a particular result. Such factors include the
in¬‚uence of drugs or substances that prevent settled concentration, tor-
ture or intimidation that prevents the person from considering alterna-
tive ideas, educational backgrounds that severely limit opportunities to
raise questions and come to minimally independent conclusions, and the
like. As we will notice later, this condition will need to be re¬ned in light
of the ways that we all engage in “distorted” self-re¬‚ection in systematic
John Christman

Interesting challenges, however, have been raised about the conceptu-
alization (and related valorization) of autonomy, challenges that concern
both the “competence” conditions and the “authenticity” conditions. For
example, critics have claimed that autonomy problematically assumes
herculean powers of self-knowledge, that the competence assumed in
such accounts demands that agents have understandings of their motives
and inner selves that few, if any, tend to realize. Moreover, such compe-
tency requirements have tended to emphasize the intellectual capacities
over the emotional and affective.16 This is shown in the characterization of
competence as “rationality” and re¬‚ective self-endorsement in terms akin
to the justi¬cation of belief. Concerning re¬‚ection, critics have charged
that second-order appraisals of ¬rst-order motives and habits often reveal
less authentic aspects of the self and, worse, cause dangerous disruption in
people™s deepest commitments, disrupting settled and authentic agency
rather than securing it. Such emphasis on re¬‚ective re-evaluation and
revision of the self both causes and re¬‚ects an unmerited valuation of
change, instability, and hyper-mobility.17
I want to investigate these charges in greater detail, and indeed I will
emphasize and support versions of these claims. For the sake of brevity,
we can examine these concerns as focused on the general requirement of
“competent self-re¬‚ection” assumed in models of autonomy. We will fur-
ther discuss the speci¬c conceptual conditions of autonomy later; for now,
we can assume that the conditions of autonomy at issue involve the com-
petent self-re¬‚ection and inner endorsement just described. The idea is
that autonomy requires that the agent in question be competent in the
sense that she suffers from none of the disabilities that would system-
atically hamper re¬‚ective decision-making and that she exhibit minimal
abilities to re¬‚ect, choose, and act. As a result of such re¬‚ection, the agent
must not repudiate the characteristic in question to be autonomous. Let
us survey, then, problems raised about such a model.

II Dif¬culties With Self-Re¬‚ection
There are many initially compelling reasons to resist taking the re¬‚ective
functions of the person as centrally indicative of her autonomy. Two fam-
ilies of reasons can be given on this score: one is that re¬‚ection itself is
often costly, and carries with it effects on commitment and devotion that
raise questions about its role in self-determination; a second is that the
re¬‚ective voice in all of us often does not speak for our most settled and
authentic personae in that such voices can cover over or mis-diagnose
Autonomy, Self-Knowledge, and Liberal Legitimacy 337

the inner workings of our psyches. Let us look at these concerns more
The ¬rst set of problems involve the way in which re¬‚ectively question-
ing our commitments and motivations can often disrupt and undercut
those very commitments. This problem of ¬rst-order motivational dis-
tortion can best be brought out in a two-person case: consider longtime
spouses or romantic partners. One day, one of them enters the breakfast
room to announce that she has lately been re¬‚ecting on the value of the
relationship for her and on her commitment to it. Now, even if the re-
sult of such re-thinking is to redouble the strength of her commitment,
the partner hearing this may well be disappointed and shocked, and the
ties between the two deeply shaken. Now if we collapse this dynamic
into a single mental life, we have cases where self-evaluation leads to self-
doubt and diminished motivation.18 The paradox is that if a person re-
¬‚ects, she loses the autonomy she seemed to enjoy before the moment of
Second, critics have charged that in many ways, our introspective judg-
ments fail to re¬‚ect our settled, authentic selves. Such re¬‚ections merely
give voice to a rationalizing super-ego attempting to quash the more cen-
tral elements of our motivational system, elements that, if allowed to move
us, would issue in action that is more truly our own. For an illustration
of such a phenomenon, consider the character Jude in Thomas Hardy™s
Jude the Obscure. For a good part of the novel, Jude is clearly in love with
his cousin Sue, though he is still married to his estranged wife Arabella,
to whom he still feels a strong obligation of ¬delity (backed by all the
force of his North Wessex Christian upbringing). But Jude™s most basic
motivational drive is clearly his love for Sue, evidenced by the cold sweats
he experiences at the thought of her leaving, and his ¬ts of jealousy at
the sight of her with another man. Re¬‚ecting on these emotions, driven
by the thought that he is still of¬cially married and that Sue is, after all,
his cousin, Jude mis-characterizes these emotions as merely those of a
platonic concern of a friend toward a family member. As the events in
the novel soon bear out, Jude™s true nature is not revealed by his re¬‚ective
voices but by those ¬rst-order affective drives.
Now, in addition to revealing the important place that emotions have
in the speci¬cation of our authentic selves, this case indicates how the
voice of re¬‚ection may distort rather than clarify our self-conceptions.
Re¬‚ection, for Jude, produces profound alienation from his emotions
and destroys whatever authentic motivation he might experience were
he, as he eventually does, to allow his feelings of love to move him to
John Christman

act. Only without the self-re¬‚ection that autonomy demands (under self-
re¬‚ection views) can Jude, and those like him, act authentically.19
The other set of problems for requiring re¬‚ection of this sort concerns
the inaccuracies (so to speak) of the judgments made from the higher-
order perspective of our re¬‚ective selves. For it is clear that only a marginal
proportion of the self implicated in behavior and social interaction can
ever be said to be available to conscious re¬‚ection, both generally and
at any particular time. Factors connected with embodiment, demeanor,
habit, and the emersion of the self in the ongoing ¬‚ow of events operate
outside of the purview of re¬‚ection, and often completely beyond its
scope. Hence, a person™s inner picture of her motivational matrix can be
highly incomplete and, in many other ways, innacurate.
Psychoanalysis provides one of the starkest models of the self™s mis-
understanding of itself.20 The fundamental theoretical commitment of
psychoanalytic theory is the postulate that mental contents that are not
integrated into the dominant “ that is, consciously available “ schema of
self-organization exert in¬‚uences on thought and behavior. The picture
that emerges is, of course, of a con¬‚icted and non-rational psychic mech-
anism whose operations are accessible to conscious re¬‚ections only in
distorted form or through the mediation of therapeutic intervention or
other complex self-interpretive techniques.
Of course, psychoanalysis is controversial, and many rightly raise ques-
tions about the reliability of (at least the details of) the postulates it pro-
duces concerning sub-conscious mechanisms. But evidence of systematic
self-misunderstanding can be gleaned from several other traditions in
individual and social psychology.21 Cognitive dissonance theory, for ex-
ample, trades on the postulate that a fundamental operation of mental
re¬‚ection is to embrace propositions that accord with established self-
conceptions and resist those that destabilize them, independent of the
epistemological ground of such information. Internal coherence trumps
probable truth. In addition, any number of (often self-serving) biases
shape judgements about internal states, capabilities, and traits. A general
tendency has been observed toward attributing responsibility for positive
outcomes to the self while explaining failures in terms of environmental
factors (the “self-serving attribution bias”). Also, people will ¬nd ¬‚aws in
evidence that portrays them in an un¬‚attering light, and they are selec-
tive in sorting through memory when considering evidence for desirable
More generally, what psychologists call the “fundamental attribution
error” refers to the systematic tendency to mis-estimate the role of either
Autonomy, Self-Knowledge, and Liberal Legitimacy 339

personal characteristics or environmental factors in explaining one™s own
or others™ behaviors.23 People routinely and predictably attribute behav-
ior to character traits or dispositions when overwhelming evidence (avail-
able to them) indicates otherwise. And agents™ perceptions of themselves
and their own motives follow the same pattern as observation of others:
the tendency to misidentify motives and the causes of our behavior is
psychologically ubiquitous.24 And our appraisals of our own emotions
and attitudes tends to be equally prone to “error,” indicating little, if
any, advantage we have in having direct (introspective) access to such
Material of this sort, admittedly presented here only selectively,
does much to bolster skepticism concerning the possibilities of self-
transparency. What emerges from these several angles is a picture of
systematic self-delusion or, at best, a fundamental disconnect between in-
trospective understanding and actual structures of motivation, thought,
and behavior. In these ways, to the extent that autonomy demands that
we re¬‚ect accurately on our motivations, desires, and reasons, most of us
are systematically heteronomous in identi¬able ways.
Indeed, this is a more empirically-minded way of expressing what com-
mentators writing in a post-modern mode have been saying about the lib-
eral conception of the self for some time “ namely, that such a conception
wrongly assumes a transparent, uni¬ed, fully rationalized self-conception
of a sort no one realistically can realize.26 Even when avoiding the psycho-
analytic models mentioned earlier, such critics decry the ¬ction of a fully
self-transparent consciousness as a basic presupposition of the model of
the (autonomous) person at the heart of liberal theories of justice.
With all these reasons for questioning the reliability of our re¬‚ec-
tive functions in capturing and representing ourselves, why should we
continue to require re¬‚ective endorsement of any kind for autonomy?
Answering such a question involves two complicated steps. The ¬rst is
to examine the role the concept of autonomy plays in various theoret-
ical and practical contexts, here the context of liberal political theory,
thereby locating the manner in which self-re¬‚ection ¬gures in that dy-
namic. The second is returning to the concept of autonomy and re¬ning
the conditions of self-knowledge that (1) capture what is required by
the concept™s role in those political/theoretical settings, and (2) squares
with the information just outlined concerning the systematic limits of the
typical person™s self-understanding. What I will suggest is that autonomy,
when viewed a certain way, plays a role in the legitimation of political
principles in such a manner that re¬‚ective self-appraisal will be a crucial
John Christman

requirement, despite (and in come cases because of) the complex effects
that such re¬‚ection has on motivation, self-understanding, and social

III Autonomy and Varieties of Liberalism
The context in which the conception of autonomy at issue here will be
tested is that of modernist liberal theories of justice, ones in which po-
litical authority is generated by way of citizen endorsement of collective
social values. Liberal views reject a metaphysically ordered hierarchy of
values, and thereby embrace a degree of value pluralism. No single over-
riding value and no ¬xed ordering of values can be determined to be
objectively valid for all agents, on this view.27 Liberalism rests on the idea,
then, that political power is legitimate only if it is endorsed or accepted
by citizens living under it “in light of their common human reason” (as
Rawls puts it).28 This implies that the principles expressive of this power
rest on respect for citizens™ abilities to rationally endorse the content
of those principles. Therefore, liberalism rests on respect for individual
autonomy as conceived generally as the “moral power” of judging both
principles of justice and conceptions of value.29 This respect is afforded
equally to all and is re¬‚ected in the manner in which both basic principles
and more speci¬c social policies are derived (that is, democratically).
Justice, then, is formulated in a way that expresses this respect, where
people are considered ultimately able to re¬‚ect upon and embrace (or
reject or revise) conceptions of value for themselves.30 Liberalism can
be seen to rest on the fundamental valuation of persons as having a ba-
sic interest in pursuing their own conceptions of what is valuable, and
doing so “from the inside.” This conception of justice as the set of princi-
ples claimed as legitimate by those living under them utilizes what some
have labeled the “endorsement constraint” on value assumed in liberal
It is important to note how liberalism, in this general sketch, is funda-
mentally opposed to certain kinds of perfectionism (in both moral and
political theory). Although there are varieties of perfectionist liberals,
most of those views take it that the fundamental (perfectionist) value
that just institutions must respect is autonomy itself.32 What liberalism of
all these sorts opposes is the view that there are values or moral impera-
tives that are valid (for a person) independent of that person™s subjective
appraisal, and hence ¬rst-person endorsement, of that value. Not only the
(European) medieval worldview concerning a metaphysically structured
Autonomy, Self-Knowledge, and Liberal Legitimacy 341

value scheme in which humans played only a part, but contemporary views
of the objectivity of value must be put to one side here, at least as a means
to provide foundations for political principles.33 The tradition of political
thought in which autonomy plays a crucial role and is the subject of exami-
nation here is one that contrasts deeply with that perfectionist standpoint.
But we must recognize a sharp distinction in liberal views of the author-
ity of the state. In one, which we can call the Hobbesean variant, collec-
tive choice (via either the original social contract or ongoing democratic
mechanisms) is seen as an aggregation of individual rational desires. The
purpose of political institutions, on this view, is to provide stability and
peace in order that citizens may pursue their own rational life plans,
separately and for their own reasons. The ground of political authority,
in this tradition, is self-interested rationality manifested in strategic in-
teraction with others.34 In the second tradition, emanating from Locke,
Rousseau, and Kant, citizens are understood to have a moral connection to
the authority of the state, insofar as such authority is a collective manifes-
tation of their own autonomy. Collective choice, on this model, is simply
the social version of the independent self-government that grounds all
morality and obligation. Political authority, then, is grounded in a moral
obligation (rather than simply rational bargaining).35
Indeed, we can generalize this distinction to apply to any social inter-
action whose purpose is to generate norms that will, in turn, constrict,
guide, or constitute the resultant activities of the participants. In the
Hobbesean case, agents view each other in a purely strategic manner,
where knowledge and empathic understanding of the other™s experi-
ences or perspective are, at best, of instrumental importance to the in-
teracting parties. There is no constitutive relation between recognition
of the thoughts, preferences, and experiences of others and the binding
nature of the outcome of such an interaction. Whatever one™s social com-
patriots think or feel, on this model, one relates to them as instrumental
to the achievement of the outcome of the exchange. Call this the purely
strategic relation.
In the other case, the interpersonal exchange involves at least a respect-
ful understanding of the other™s perspective (at some level of abstraction
or description), an understanding that is a crucial component of the
reciprocity involved in this kind of social dynamic. And this attempt at
understanding forms an ineliminable part of the normative grounding
of the outcome. That is, participants view both the process of collec-
tive deliberation and confrontation, as well as the result, as normatively
signi¬cant in part because of their shared understanding and projected
John Christman

moral judgment. Such an interaction must involve mutual respect and a
sense of reciprocity in the familiar Kantian sense, where one attributes
basic moral weight to the capacities of one™s co-citizens to deliberate and
decide. But it also includes an attempt at empathic comprehension of
the subjectivity and motivations of others. This need not involve a ¬‚aw-
less or even accurate understanding of another™s deliberative processes,
but it does require an attempt to see the point of view (together with
affective and subjective elements of it when relevant) of those with whom
one shares a common relation to a collectively formulated outcome, in-
complete though this process will inevitably be. Again, this can take place
at virtually any level of abstraction, rising, say to the point of merely say-
ing, “I think I can understand what it is like being a motivated human
who is passionate about a cause such as that.” We will call this empathic
The normative hold that the outcomes of this type of interaction has on
participants will be constitutively related to this emphatic respect. Such
a theory will be grounded much more ¬rmly in one™s own perceived
value-commitments. In the case of strategic interaction “ the Hobbesean
model “ one™s commitment to outcomes of interchange extends only as
deep as one™s occurrent self-interest, and that outcome and commitment
remain stable only as a function of the initial power relations that made
the compromise with the objecti¬ed other possible. We will return to this
point later.
Liberal political theory, then, presupposes a conception of the (au-
tonomous) person that is both the object of respect (upon which those
principles are built) and the model for basic interests that those prin-
ciples protect. (Rawls™s use of the index of social primary goods as a
measure of just distributive shares is an example of this, based as it
is on the projection of persons as capable of forming and embracing
conceptions of both justice and the good.) Parallel to this commit-
ment, though, is the liberal presupposition of value pluralism noted
earlier. Liberal theory developed (historically) by rejecting various me-
dieval and Scholastic metaphysical conceptions that postulated a teleo-
logically structured order of the universe. These rejected pictures of the
world served to specify completely the virtues and values for both in-
dividuals and societies. Liberalism, in both its Hobbesean and Kantian
varieties, replaced this metaphysical framework with (what would later
galvanize into) a conception of moral commitment with the human
will at its center. Political principles, then, and the sense of obligation
binding citizens to them, are seen as grounded in the individual and
Autonomy, Self-Knowledge, and Liberal Legitimacy 343

collective judgments of the people involved, expressed by their rational
Value pluralism is the understanding that various individuals will em-
brace irreducibly divergent, but equally valid, moral conceptions. And
political principles must take into account what Rawls calls the “fact of
reasonable pluralism” “ citizens pursue divergent comprehensive moral
conceptions but recognize this divergence itself and accept it as a per-
manent fact of modern life.38 Social values and the political principles
re¬‚ective of them are generated (in part, at least) by way of collective
choice and deliberation, and not given fully formed from above.
Such public endorsement of dominant political values must also occur
against the backdrop of the inevitabilities of social existence. That is,
contrary to the traditional assumption of a state of nature by which to
measure the bene¬ts of a speci¬c political arrangement (a pre-social
arena to which disgruntled citizens can retreat), political principles are
judged by citizens who take the ongoing, historically embedded dynamics
of social existence as an unavoidable fact (along with the pluralism of
value-conceptions this brings with it).39
Therefore, interaction and collective deliberation among divergent
viewpoints is fundamental to the process of legitimation and justi¬ca-
tion of social power. This view has been the dominant theme in the re-
cent work of both Rawls and Habermas. The latter has developed the
most complex picture of the centrality of discursive communicative ac-
tion in the justi¬cation of both moral and political principles (indeed
all of the claims to validity that underlie the use of language itself).40
Indeed, on Habermas™s view of the development of individualized iden-
tity (individuation), the person (child) internalizes the social meanings
and normative structures of the surrounding, usually parental, voices.41
The dialogic interaction with a “generalized other” takes the place of
the assumption of a disembodied and objective viewpoint of Enlighten-
ment (that is, purely Kantian) thinking. Intersubjective validity replaces
depersonalized objectivity, and such intersubjectivity is established by on-
going, linguistically mediated social interaction with surrounding others.
Normative (moral) validity is ¬xed in reference to a principle whereby
all affected by a decision could freely accept the consequences of its
general observance given their needs and interests.42 Individuation oc-
curs with the development of capacities of questioning, re¬‚ection, and
critique as a component of the participation in dialogue and internal-
ization of social meanings that such a test of validity requires.43 Hence,
as a view of personal development, this model mirrors the requirements
John Christman

of legitimacy that liberal theories require of social principles. That is,
the ability to re¬‚ectively negotiate collectively generated norms and to
present critical points of view in the dynamic of such deliberations is
central to the legitimacy of the political principles expressive of those
norms. The relevance of this comparison will arise presently when we dis-
cuss the requirements of self-understanding in the social negotiations so
While the social contract tradition has expressed the establishment of
the collective endorsement of political principles as a hypothetical agree-
ment among rational parties, recent developments in liberal theory have
underscored the need for actual social interaction and ongoing negotia-
tion to be seen as constitutive of political legitimacy.44 The traditional view
of a hypothetical and philosophically determined ground for agreement
has been rightly challenged by those who insist that these abstract con-
ceptions of social and individual life utilized in such hypothetical models
very likely betray actual biases and exclusionary tendencies inherent in
the contemporary social milieu out of which they arise (valorizing certain
middle-class, white, male value-conceptions to the relative denigration
of other, marginalized groups).45 Even standard liberal theorists have
claimed the centrality of democratic deliberation in the determination
of the principles of justice, at least in their ¬nal form.46
Therefore, insofar as actual public deliberation and communication
must occur for the principles of liberal justice to be settled upon and
political legitimacy to be established, self-expression will be crucial in the
functions of the citizen acting in this process. The ability to settle upon
and give, publically, reasons for claims will function as an ineluctable el-
ement in the determination of just principles. Final determination for
the order of values that will be represented in the principles of a just
society must be given to citizens themselves, and such values must be de-
¬ned by way of ongoing, open, discussion among autonomous citizens
(and/or their representatives) in a diachronic process of re¬ning justice
and maintaining legitimacy. Thus, persons themselves must be in a posi-
tion to re¬‚ect upon, and report in public settings, the value commitments
that they wish to receive weight in such political deliberations.
But there are importantly different positions one can occupy with re-
gard to the expression to others of one™s own experiences, ideas, and pref-
erences. In cases where the person speaks for herself in expressing her
beliefs, desires, values, and experiences, there are two ways that this ¬rst-
person representational “authority” can be understood. In the ¬rst way,
we view the person as the epistemic authority on what she is representing:
Autonomy, Self-Knowledge, and Liberal Legitimacy 345

she speaks for herself because she knows herself (perhaps more than
others or perhaps absolutely). In the second way, she has personal au-
thority where she is designated as the expressive voice of those value
commitments, independent of her actual ongoing hold on the content
of those expressions. The point of this distinction is that one can enjoy
the second type of self-representational authority without claiming the
¬rst. I might be assigned a role of expressing some material for reasons
independent of my epistemic position in regards to it.
Epistemic authority over self-expressions is grounded in the assump-
tion of a “truth of the matter” regarding the content of what is to be rep-
resented. That is, granting someone expressive authority on epistemic
grounds makes sense only if (1) there is a settled truth concerning which
representation is appointed or accepted, and (2) the representing agent
is in the best epistemic position to know that content.47 However, in cases
where these two provisions fail to hold, the case for granting represen-
tative authority on epistemic grounds becomes weaker. And as we will
see later, the assumptions about the kind of social deliberation involved
in processes of liberal legitimacy cannot presuppose that there is a “fact
of the matter” concerning value-statements independent of the person™s
own internal grasp and endorsement of that value; being in the best po-
sition to know what is “true” is less important than being in a position to
adopt for oneself that to which one is committed.
Now let us return to the conception of the autonomous person to which
liberal theory relies so that we can connect our endorsement of self-
re¬‚ection as a component of autonomy while acknowledging the dif¬-
culties in its operation outlined earlier.

IV Autonomy and Self-Re¬‚ection Revisited
In contexts where interaction with others brings about collective deci-
sions, the normative anchor that such outcomes provide for the par-
ticipants depends heavily on the acknowledgment of the autonomy of
one™s co-deliberators. But autonomy in what sense? The model I have
suggested here requires that the autonomous person exhibit minimal
cognitive competence and hypothetical self-endorsement (interpreted
as non-alienation) via self-re¬‚ection. That is, the authenticity that au-
tonomy requires obtains when, were one to turn a re¬‚ective eye toward
the motives, values, and concepts that structure one™s judgments (and
do so in a piecemeal manner), one would not feel deep self-alienation,
self-repudiation, and unresolvable con¬‚ict.
John Christman

An important point to note here is that the hypothetical self-re¬‚ection
involved in this test for authenticity does not imply accurate self-
knowledge or self-transparency. The test is purely subjective in that it
takes as its perspectival orientation the agent™s own point of view, inde-
pendent of any external account of the motives, values, and beliefs to
which she might turn her attention. Moreover, the non-alienation char-
acteristic of the autonomous person has both phenomenological and
affective elements: the agent would not feel a sense of self-repudiation in
the internal grasping of her sense of the motives and impulses that move
her to action.48
But in what way can this model of the liberal (autonomous) person
square with the accepted levels of self-misunderstanding that we outlined
above? To see if it does, we need ¬rst to recall the distinction between two
kinds of self-expressive authority “ epistemic and personal authority: the
lack of self-understanding we accepted only touches the assumption of
(some kind of) epistemic authority on the part of self-representing agents.
Insofar as the reasons for granting self-representational authority in col-
lective decisions are personal rather than epistemic, then failures of epis-
temic access to the content of one™s expressions “ one™s self-knowledge in
this case “ will be less serious. (Though they will by no means be irrelevant:
see later discussion.)
Second, we need to focus on the distinction between the two kinds of
liberalism noted earlier: In the Hobbesean variant, the point of granting
individual rights of self-expression and participation in the process of
legitimating state power is that such expression functions as a conduit
for the promotion of the rational interests of the parties. State power, re-
member, is justi¬ed as a coordination device for the maximal satisfaction
of such interests. Therefore, the authority granted to citizens to express
their own judgments is clearly epistemic authority: it is the authority to
judge and express their own interests, interests that are well de¬ned in-
dependently of the process of subjective grasping and endorsement. That
is, according to Hobbesean contract theory, state power is designed to
protect the idealized desires of the participants. Their own judgments of
what those desires are may well, for the reasons outlined in our examina-
tion of self-knowledge, be systematically distorted.
So for Hobbesean liberalism, full self-knowledge (as a condition of the
autonomy assumed in citizens) is a necessary condition for the validity of
outcomes of collective choice. Only when actual interests are expressed
in deliberation will the process of aggregating such interests “ which,
on the Hobbesean model, is the fundamental role of the state “ operate
Autonomy, Self-Knowledge, and Liberal Legitimacy 347

correctly. And the interests in question are determinable independent
(in principle) of the person™s judgment about them: a person can be
mistaken about her interests (as well as her motives, psychological states,
and the like, as we saw earlier).
However, one need not accept the Hobbesean account of political au-
thority. While I cannot argue for this here, there is good reason to avoid
seeing the legitimacy of political institutions as fundamentally the coor-
dination of individually determined interests of the citizens living under
it.49 The assumption of individualized self-interest, to take one exam-
ple of a core assumption of the Hobbesean view, is highly problematic;
moreover, it is not at all clear that even if citizens were self-interested and
motivated by individualized desires, social stability of the sort promised
by Hobbesean political theory would materialize or be maintained.50
On the other hand, the Kantian variant of liberalism grants individ-
uals powers of self-government for a different reason “ only when po-
litical principles are embraced authentically by those governed by them
are they valid for those people. Collective deliberation in order to le-
gitimate state power and to generate new legislation functions by way
of mutual respect and what I called “empathic respect” for others™ dif-
ferences. For the later Rawls, for example (who departs from the literal
Kantianism of his earlier view but remains in the category I am here la-
beling “Kantian”), the overlapping consensus that legitimates principles
of justice must be “af¬rmed” from within each citizen™s comprehensive
moral view, and hence must involve a moral commitment to cooperative
interaction with others whose views differ. For Habermas, valid (politi-
cal) claims presuppose sincere and free interchange among participants,
all of whom implicitly accept the normative presuppositions of discourse
itself. Sincerity involves not simply reporting what is in fact true but ex-
pressing what one deeply believes. One can be sincere but incorrect, and
it is sincerity that is presupposed in discursive interchange. Therefore,
personal (self-representational) authority (in my sense) is what is granted
in communicative action.
Thus, no presumption of epistemic authority over a person™s motives
and desires must be granted in this matrix; representational authority
is all that is needed to ground the mutual respect (and empathic un-
derstanding) that, I have argued, functions to legitimate state power.
It is as if we say to each other: you may be often mistaken about what
truly moves you and what is in your best interest, but nevertheless you
always get to speak for yourself on such matters. The reason for this po-
sition is moral/political, not epistemic: in order to ensure the personal
John Christman

endorsement necessary for the validity of value commitments one must
embrace and express for oneself such commitments; externally deter-
mined validity (a “fact of the matter” ¬xed independent of such endorse-
ment) of values is not recognized.
Democratic institutions that arise as part of (and, some would argue,
a constitutive part of) principles of justice require that citizens (perhaps
through their representatives) be in a position to advance reasons for the
interests they wish to see promoted collectively in their society. Demo-
cratic deliberation, then, also requires participants™ abilities to re¬‚ectively
endorse, indeed publically defend, the points of view, values, interests,
and opinions that are the inputs to such deliberative processes (the “out-
puts” of which are social principles and policies). This provides further
reason for the presupposition that the autonomous person is able to
re¬‚ectively grasp and present her values and perspective. This accords
her the kind of representational authority over those points of view but
also necessitates their capacity to re¬‚ect on their values as part of the dy-
namic of social interchange that produces collectively justi¬ed principles.
So autonomy as competent, self-re¬‚ective endorsement (non-alienation)
is central to this understanding of justice and politics.
Therefore, for reasons of social legitimation, interpreted as the lib-
eral principle of legitimacy for political institutions and principles, self-
re¬‚ection is a crucial mark of the autonomous citizen whose status is re-
spected and whose interests are protected in just political arrangements.
Only if a person is put in a position to speak for herself, can the collectively
generated principles of justice claim the legitimacy required by liberal
theory. Advancing her interests in a way that thoroughly bypasses re¬‚ec-
tive endorsement of them threatens to violate the requirement that values
promoted in a society obtain validity only by being subject to the citizens™
endorsement of them. So liberal legitimacy presupposes a model of the
(autonomous) person able to re¬‚ectively endorse her interests, respect
for which is re¬‚ected in the structure of the principles themselves.51
What, then, should be the standards of self-understanding and cogni-
tive competence that autonomy, used in this context, requires? To answer
this question, we must say a bit more about the epistemic standards of pub-
lic reason, within which autonomous self-expression plays such a crucial
role. This is a complex subject, but, beyond what has already been said,
we will have to be brief.52 First, in order for public justi¬cation to proceed
in a way consistent with the endorsement constraint, we must assume at
least a modest internalism as our epistemic standard of justi¬cation at
the individual level. That is to say, no value claim can be said to be valid
Autonomy, Self-Knowledge, and Liberal Legitimacy 349

for a person (or no belief about such a claim or its components) unless
there is an inferential relation between such a claim and other elements
of that person™s belief/value corpus. Pure externalism would deny this
and claim that some beliefs are justi¬ed for a person wholly independent
(in principle) of that person™s belief set. But the endorsement constraint
implies that, ideally at least, a person could come to embrace (or at least
not be deeply alienated from) the value in question. This is not possible
unless there is a hermeneutic or otherwise inferential relation between
that value and things the person already holds.53
Second, a person must have a level of understanding of her own psy-
che so that she is a relatively consistent representative of a viewpoint. If
manifest inconsistencies arise from or are involved already in her corpus
of desires and values, then the process of deliberation and negotiation
cannot fruitfully proceed. So absence of manifest inconsistencies “ where
fully contradictory beliefs or values are held in ways that could bring them
easily to mind “ is a necessary part of autonomy competence. But this is
compatible, it must be stressed, with sincere ambivalence and measured
changes of mind. I meet this requirement even if I am torn in two di-
rections on an issue or if I alter my view in light of new information and
deliberation itself. But a person who is notably pulled by inconsistent
desires in ways she does not admit “ acting on or expressing one at one
moment and doing the opposite the next “ is not a competent deliberator
and hence not autonomous in the requisite sense.
Third, mis-identi¬cation of motives as speci¬c as those described in
the various attribution errors described earlier need not disturb the self-
expressive authority assumed in autonomy-based liberalism. One need
not correctly identify the motivating reasons for action or decisions, as
long as one takes responsibility for such decisions once they are made. As
for the mis-labeling of either the character or the source of our emotions
and attitudes, public deliberation need not be seen as a process of discov-
ery of stable and independently existing attitudes that such deliberation
serves merely to coordinate or (as appropriate) aggregate. At least under
what I am calling the Kantian rubric, public discourse is itself a process
of moral importance not reducible to its revelatory role in uncovering
nascent internal states of the agent. We are not merely counting votes.
So when people™s interaction in public debate functions in ways that “dis-
tort” their reporting of their own attitudes, their public stance in that
debate thereby becomes the position they are committed to, indepen-
dent of its representational accuracy concerning the internal states of the
John Christman

A matter of some importance here is the manner in which commit-
ments to beliefs or value claims can be made valid upon the decision to
commit oneself to them. This is akin to the existentialist point that our
existence (and our choices) precedes our essence, and our commitments
follow in part from our choices themselves, thereby constituting our be-
ing. However, we should add (hence deviating from Sartrean doctrine)
that the validity of a norm for a person need not be understood as wholly
subjective. As Charles Taylor has argued, a fundamental aspect of human
agency is a commitment to “strong evaluations” “ value judgments whose
validity lies, in part at least, beyond the merely subjective choice to accept
them. Moreover, the process of public reason itself demands the giving
of reasons to others that are (1) sincere (so held as valid by the person
making the claim), and (2) grounded in considerations that could appeal
to those others, hence not wholly grounded in subjective choice.
But the endorsement constraint continues to operate here. For it im-
plies that subjective embrace of a value is a necessary component of its
validity (for a person). So a person™s act of embracing a view, or embrac-
ing a view as part of a process of publically expressing it in the dynamics
of public deliberation, makes it her own in this crucial sense. Even if I am
somewhat out of touch with my motives, or systematically mistaken about
the psychological sources of my opinions and values, I commit myself to
them as I advance them to others in public discourse. I, therefore, con-
struct myself (in part) by committing myself to this or that belief. At least
I construct and commit myself provisionally in that I am open to reasons
from others and, as a sincere and non-strategic communicator, I listen
to others in ways that may lead me to reconsider my own views. But as a
participant in this process, I commit myself to views I judge to be right
by expressing them, not (or not always) by simply discovering them as a
settled aspect of my nexus of other beliefs, desires, and values.
In this way, the fact that re¬‚ective self-appraisal tends to undercut the
person™s own commitments (or merely serve as a rationalization of some
of them) becomes less troubling: the public stand one takes in discourse
and deliberation becomes the position to which one is held responsible
in the process of generating valid social norms and the legitimacy they
enjoy. It is hoped (not entirely without reason) that the process of public
interchange itself can induce dynamic reconsideration of one™s own posi-
tion on various matters that will reduce whatever disconnect there might
exist between a public report and a private drive.
Hence, psychoanalytic, and indeed post-modern, pictures of the frag-
mented and decentered self do not con¬‚ict with this picture of liberal
Autonomy, Self-Knowledge, and Liberal Legitimacy 351

autonomy insofar as the requirements of self-understanding in the model
of autonomy at work make no demands of strict internal unity, stable emo-
tional or attitudinal matrices, opaque psychological mechanisms, and the
like. What it demands is that the person™s characteristics (values, desires,
and the like) be subject to her own re¬‚ective appraisal and, if not found
to be deeply repugnant, presented publicly as a position for which she is
held responsible. If the very act of discursive interchange in effect con-
structs the value position that is the focus of this responsibility, that is
consistent with the anti-perfectionism implicit in the version of liberal-
ism alluded to here: under this view, there is no pre-determined value
scheme that lies outside of human embrace and construction waiting to
be found.

V Summing Up
Several objectives were pursued in this chapter. One was to claim that
conceptions of autonomy should not rest on a single conception of the
“self,” since conceptualization of selves are (validly) understood to be
multiple and variable. Second, a model of autonomy was put forward
and (in part) defended, though problems with a central element of that
model (the requirement of self-re¬‚ection) were aired and expanded. But
we came back to the view that autonomy requires self-re¬‚ection because of
the role that the concept of autonomy plays in certain political principles
prominent in current theoretical constructions.
These constructions found no independent defense here, of course,
and those who reject them in whole or in part will not be particularly
satis¬ed with the chapter™s conclusions. But these constructions were
sketched in broad enough form (breadth that carried with it that de-
gree of vagueness and imprecision) that they should seem compelling to
many, if only because they are intended to represent a large current in
modern(ist) approaches to political legitimacy and justice. To show that
autonomy in something like the form defended here is necessary for the
acceptability of those broadly construed theoretical constructions is no
mean accomplishment, fragile though it is.

See, for example, Michael Sandel Liberalism and the Limits of Justice, 2nd ed.
(Cambridge: Cambridge University Press, 1999); Iris Young, Justice and the
Politics of Difference (Princeton, NJ: Princeton University Press, 1990); and
Daniel Bell, Communitarianism and its Critics. Oxford: Clarendon, 1993.
John Christman

For a survey of literature on autonomy, see John Christman, “Construct-
ing the Inner Citadel: Recent Work on the Concept of Autonomy” in Ethics
vol. 99, no. 1 (Fall, 1988), 109“24. See also Catriona Mackenzie and Natalie
Stoljar, “Introduction: Autonomy Recon¬gured” in Mackenzie and Stoljar,
eds., Relational Autonomy: Feminist Perspectives on Autonomy, Agency, and the So-
cial Self (New York: Oxford University Press, 2000), 3“31; and the essays in
The Inner Citadel: Essays on Individual Autonomy (New York: Oxford University
Press, 1989). Other discussions of note of the concept include Lawrence Ha-
worth, Autonomy: An Essay in Philosophical Psychology and Ethics (New Haven:
Yale University Press, 1986); Gerald Dworkin, The Theory and Practice of Auton-
omy (Cambridge: Cambridge University Press, 1990); Alfred Mele, Autonomous
Agents (New York: Oxford University Press, 1995); Diana T. Meyers, Self, Society
and Personal Choice (New York: Columbia University Press, 1989); and Bernard
Berofsky, Liberation from Self (Cambridge: Cambridge University Press, 1995).
Isaiah Berlin, “Two Concepts of Liberty” in Four Essays On Liberty (Oxford:
Oxford University Press, 1969), 118“72.
See, for example, Charles Taylor, The Ethics of Authenticity (Cambridge, MA:
Harvard University Press, 1991), 33ff.
Some, however, continue to insist on a close connection between autonomy
and the self: see Marina Oshana (Chapter 4 in the present volume). For a
focused argument for the separation between self and autonomy, see Bernard
Berofsky, Liberation from Self.
Typically, the focus of models of autonomy are speci¬cally the agent™s desires.
However, there is good reason to broaden this to include any aspects of the
person relevant to identity, action, and choice. One can lack autonomy rel-
ative to emotions, skills, physical factors, knowledge, and general states of
being as well as to desires per se. For discussion, see my “Liberalism, Auton-
omy, and Self-Transformation,” Social Theory and Practice 27, 2 (2001). See
also Richard Double, “Two Types of Autonomy Accounts,” Canadian Journal
of Philosophy 22, no. 1 (March, 1992), 66.
I defend this view also in “Liberalism, Autonomy, and Self-Transformation.”
In my original formulation of this idea, I claimed that the person must re-
¬‚ect upon and accept the processes of self development himself. I now see
the limitations of this formulation, and have amended this requirement as
indicated in the text. See “Autonomy and Personal History,” Canadian Jour-
nal of Philosophy 21, no. 1 (March, 1991), 1“24. For criticism of this version
of the view, see Alfred Mele, “History and Personal Autonomy,” Canadian
Journal of Philosophy 23 (1991), 271“80; and for a reply, see “Defending Per-
sonal Autonomy: A Reply to Professor Mele,” Canadian Journal of Philosophy 23
(1993), 281“90. For a discussion of historical views of autonomy, see Alfred
Mele, Autonomous Agents: From Self-Control to Autonomy (New York: Oxford Uni-
versity Press, 1995). For a historical account of moral responsibility, see John
Martin Fisher and Mark Ravizza, Moral Responsibility and Control (Cambridge:
Cambridge University Press, 1998).
The concept of self-alienation is analyzed in different form in certain areas
of psychoanalytic theory: see, for example, Karen Horney, Our Inner Con¬‚icts:
A Constructive Theory of Neurosis (New York: Norton, 1945). For a discussion of
Autonomy, Self-Knowledge, and Liberal Legitimacy 353

autonomy and personal integrity, see Diana T. Meyers, Self, Society, and Personal
Choice, 59“75. A parallel idea plays a role in Ronald Dworkin™s distinction
between those aspects of a person™s personality for which she should be held
responsible (for the purposes of distributive justice) and those that are part
of her “circumstances” and hence subject to egalitarian redistribution. See
Sovereign Virtue: The Theory and Practice of Equality (Cambridge, MA: Harvard
University Press, 2000), 286“91, 322“23.
Gerald Dworkin™s account of autonomy suffers from this (relatively minor)
weakness, I think, in that all his model requires is the “capacity to raise the
question of whether I will identify or reject the reasons for which I now act”
(The Theory and Practice of Autonomy, p. 15).
This does not imply, as some liberal theorists have been (rightly in some
cases) accused of claiming, that a person must be able to alter all aspects
of her values and convictions upon re¬‚ection. The requirement is that she
must be able to shed only those traits or commitments from which she feels
deeply alienated. For discussion, see my “Liberalism, Autonomy, and Self-
See, for example, Harry Frankfurt “Identity and Wholeheartedness” in The
Importance of What We Care About (Cambridge: Cambridge University Press),
Cf. Meyers, Self, Society, and Personal Choice, 72.
For further discussion, see my “Autonomy and Personal History”.
For further discussion and elaboration of this condition, see my “Relational
Autonomy, Liberal Individualism, and the Social Constitution of Selves,”
Philosophical Studies 117 (2004): 143“64.
Discussion of this issue can be found in Diana T. Meyers, Subjection and
Subjectivity: Psychoanalytic Feminism and Moral Philosophy (New York: Rout-
ledge, 1994) as well as Self, Society, and Personal Choice, pp. 28ff.
See, for example, Bernard Williams, “Persons, Character, and Morality”
in Moral Luck (Cambridge: Cambridge University Press, 1983), 1“19, and
Robert Bellah, et. al., Habits of the Heart (Berkeley, CA: University of California


. 11
( 13)