L )�:-· b.� ..e. r��.·a. . I 1·1·.

s· 1·�1,·1 I
I..

�!
I ....
·· I ·..

_
. I II

_ �'
.
I _1
.
I. _I _
� ..
..1
.
I
_
I
_I,
·

_·I
.

a1'1 dl�i t·�1 ]�1 e

I - :,�
.
-' • - I

L,,I 111 1 ts () fI
. . - lr -
I

J us,tice
_ .,

. \ lit lta4·1 J �:und,.,l . .

- - -
Published by the Press Syndicate of the University of Cambridge
The Pitt Building, Trumpington Street, Cambridge CB2 1RP
32 Eas't 57th Street, New York, NY 10022, t.:SA
296 Beaconsfield Parade, Middle Park, Melbourne 3206, Australia

© Cambridge University Press 1982

First published 1982

Reprinted 1983

Printed in the United States of America

Library of Congress catalogue card number:82-4394

British Library Cataloguing in Publication Data

Sandel, Michael J.

Liberalism and the limits of justice.

I. Justice (Philosophy)
I. Title
179 B 105.]87

ISBN o 521 24501 x hard covers

ISBN o 521 27077 4 paperback
To rrry parents
Contents

A cknowledgments ix
Introduction : Li beralism and the Primacy of] ustice
The foundations ofliberalism: Kant versus Mill 2
The transcendental subj ect 7
The sociological obj ection 11
Deontology with a H umean face 13

1 Justice and the Moral Subj ect 15

The primacy ofj ustice and the priority of the self 15
Liberalism without metaphysics : the original position 24
The circumstances ofj ustice: empiricist objections 28
The circumstances ofjustice : deontological rejoinder 40
In search of the moral subj ect 47
The self and the other: the priority ofplurality 50
The self and its ends: the subj ect of possession 54
I ndividualism and the claims of community 59

2 Possession , Desert, and Distributivejustice 66

Libertarianism to egalitarianism 66
M eritocracy versus the difference principle 72
Defending common assets 77
The basis of desert 82
I ndividual and social claims: Who owns what? g6

3 Contract Theory andJ ustification 1 04

The morality of con tract 1 o5
Con tracts versus con tractarian arguments 1 og
Liberalism and the priority of proced ure 1 13
What really goes on behind the veil ofignorance 1 22
Vll
C ontents

4 Justice and the Good 1 33

The unity of the self 133
The case of affirmative action 135
Three conceptions of community 1 47
Agency and the role of reflection 154
Agency and the role of choice 161
The status of the good 165
The moral epistemology of justice r68
Justice and community 173

Conclusion: Liberalism and the Limits ofjustice

Deontology's liberating project
Character, self-knowledge, and friendship

B ibliography
Index

Vlll
Acknowledgments

This book began life in Oxford in the late 1 970s, a stimulating time
for the study of political philosophy, especially so at Balliol College.
I am indebted to teachers and friends of those years and since, who
taught me much of what I have written here. Thanks are due first
of all to William Connolly, Richard Fallon, Donald Herzog, Steven
Lukes, David Miller, Alan Montefiore, Judith Shklar, and Charles
Taylor, all of whom read an earlier version of this essay and offered
valuable comments and criticisms. Their advice was all to the
good, and does not account for whatever weaknesses remain. Some,
however, are implicated a little more deeply. Among my contempo­
raries while abroad, Ronald Beiner, Richard Fallon, and Scott
Matheson bear a special responsibility. Discussions and travels
with them helped shape these thoughts, and made for intellectual
comradeship of the highest order . I owe my deepest debt to those
among my teachers whose influence is most present in this work:
Ronald Dworkin, whose arguments demanded a better answer
than I could offer at the time; Charles Taylor, who broadened
Anglo-American horizons and taught the relevance of Aristotle and
Hegel; and especially Alan Montefiore, who made philosophy
inescapable, and set me on this path six years ago.

IX
Introduction: Liberalism and the primacy ofjustice

THE FOUNDATIONS OF LIBERALISM: KANT VERSUS MILL

The primacy of justice can be understood in two different but

related ways. The first is a straightforward moral sense. It says that
justice is primary in that the demands of justice outweigh other
moral and political interests, however pressing these others may be.
On this view, justice is not merely one value among others, to be
weighed and considered as the occasion arises, but the highest of all
social virtues, the one that must be met before others can make their
claims. If the happiness of the world could be advanced by unjust
means alone, not happiness but justice would properly prevail. And
when justice issues in certain individual rights, even the general
welfare cannot override them.
But the primacy of justice, in its moral sense alone, hardly
distinguishes this liberalism from other well-known varieties. Many
liberal thinkers have emphasized the importance of justice and
insisted on the sanctity of individual rights. john Stuart Mill called
justice 'the chief part, and incomparably the most sacred and
binding part, of all morality' ( 1863: 465 ), and Locke held man's
natural rights to be stronger than any commonwealth could override
(16go ). But neither was a deontological liberal in the deeper sense
that concerns us here. For the full deontological ethic is not only
about morals but also about the foundation of morals. It concerns
not just the weight of the moral law, but also the means of its
derivation, what Kant would call its 'determining ground' ( 1788).
On the full deontological view, the primacy of justice describes
not only a moral priority but also a privileged form of justification;
the right is prior to the good not only in that its claims take
precedence, but also in that its principles are independently derived.
This means that, unlike other practical injunctions, principles of
justice are justified in a way that does not depend on any particular
vision of the good. To the contrary: given its independent status, the
right constrains the good and sets its bounds. 'The concept of good
and evil is not defined prior to the moral law, to which, it would
seem, the former would have to serve as foundation; rather the
concept of good and evil must be defined after and by means of the
law' (Kant q88: 65 ).
From the standpoint of moral foundations, then, the primacy of
justice amounts to this: the virtue of the moral law does not consist in

2
T he foundations of liberalism

the fact that it promotes some goal or end presumed to be good. It is

instead an end in itself, given prior to all other ends, and regulative
with respect to them. Kant distinguishes this second-order, found­
ational sense of primacy from the first-order, moral sense as follows:
By primacy between two or more things connected by reason, I understand
the prerogative of one by virtue of which it is the prime ground of
determination of the combination with the others. In a narrower practical
sense it refers to the prerogative of the interest of one so far as the interest of
the others is subordinated to it and is not itself inferior to any other ( 1788:
1 24) .
The contrast might also be drawn in terms of two different senses
of deontology. In its moral sense, deontology opposes c on sequent ial ism;
it describes a first-order ethic containing certain categorical duties
and prohibitions which take unqualified precedence over other
moral and practical concerns. In its foundational sense, deontology
opposes te le ology; it describes a form of justification in which first
principles are derived in a way that does not presuppose any final
human purposes or ends, nor any determinate conception of the
human good.
Of the two strands of the deontological ethic, the first is no doubt
the more familiar. Many liberals, not only deontological ones, have
given special weight to justice and individual rights. This raises the
question of how the two aspects of deontology are related. Can
liberalism of the first kind be defended without recourse to the
second? Mill, for one, thought so, and argued for the possibility,
indeed for the necessity, of detaching the two.
To have a right, says Mill, is 'to have something which society
ought to defend me in the possession of ' (I 863: 459). So strong is
society's obligation that my claim 'assumes that character of abso­
luteness, that apparent infinity, and incommensurability with all
other considerations, which constitute the distinction between the
feeling of right and wrong and that of ordinary expediency and
inexpediency' (I863: 460). But if it be asked why society must meet
this obligation, it is for 'no other reason than general utility' (I 863:
459). Justice is properly regarded as 'the chief part, and incompara­
bly the most sacred and binding part, of all morality', not by reason
of abstract right, but simply because the requirements of justice
'stand higher in the scale of social utility, and are therefore of more
paramount obligation, than any others' (I863: 465, 469).

3
Introduction: Liberalism and the primacy ofjustice

It is proper to state that I forego any advan tage which could be derived to
my argument from the idea of abstract right, as a thing independent of
u tility. I regard utility as the ultimate appeal on all ethical questions ; but it
must be u tility in the largest sense, grounded on the permanent interests of
man as a progressive being ( 1 849: 485 ) .

The overriding importance of justice and rights makes them

'more absolute and imperative' than other claims, but what makes
them important in the first place is their service to social utility, their
ultimate ground. 'All action is for the sake of some end, and rules of
action, it seems natural to suppose, must take their whole character
and color from the end to which they are subservient' ( 1 863: 402 ).
On the utilitarian view, principles of justice, like all other moral
principles, take their character and color from the end of happiness.
For 'questions of ends are . . . questions about what things are
desirable', and happiness is desirable, in fact 'the only thing desir­
able as an end', because 'people do actually desire it' ( 1863: 438). In
this the teleological foundation and psychological assumptions of
Mill's liberalism become clear.
For Kant, by contrast, the two aspects of deontology are closely
connected, and his ethics and metaphysics argue powerfully against
the possibility of having one without the other. Against a position
such as Mill's (and that of modern-day 'rule utilitarians' ) the
Kantian view suggests at least two compelling objections. One says
that utilitarian foundations are unreliable, the other that unreliable
foundations, where justice is concerned, can be coercive and unfair.
Utilitarianism is unreliable in that no merely empirical founda­
tion, uti litarian or otherwise, can secure absolutely the primacy of
justice and the sanctity of individual rights. A principle that must
presuppose certain desires and inclinations can be no less condition­
al than the desires themselves. But our desires and the means of
satisfying them typically vary, both between persons and, over time,
within individual persons. And so any principle that depends on
them must be similarly contingent. Thus 'all practical principles
which presuppose an object (material ) of the faculty of desire as the
determining ground of the will are without exception empirical and
can furnish no practical laws' (Kant 1 7 88: 19 ) . Where utility is the
determining ground - even 'utility in the largest sense' - there
must in principle be cases where the general welfare overrides justice
rather than secures it.

4
T he fou ndatio ns of liberal ism

Mill in effect concedes the point, but would question whether

justice should be that unconditionally privileged anyhow. He ac­
knowledges that the utilitarian account does not make justice abso­
lutely prior, that there may be particular cases 'in which some other

maxims of justice' ( I 863 : 46 9 ) . But if, by this qualification, the

social duty is so important as to overrule any one of the general

happiness of mankind is advanced , what grounds could there be for

affirming the primacy of justice more completely?'
Kant's answer would be that even exceptions in the name of
human happiness must be rejected , for the failure to affirm absolute­
ly the primacy of justice leads to unfairness and coercion . Even if the
desire for happiness were universally shared , it could not serve as
basis for the moral law. Persons would still differ in their conceptions
of what happiness consists in, and to install any particular concep­
tion as regulative would impose on some the conceptions of others ,
and so deny at least to some the freedom to advance their own
conceptions . It would create a society where some were coerced by
the values of others , rather than one where the needs of each
harmonized with the ends of all . ' Men have different views on the
empirical end of happiness and what it consists of, so that as far as
happiness is concerned , their will cannot be brought under any
common principle nor thus under any external law harmonizing
with the freedom of everyone' (Kant I 7 9 3 : 73-4) .
For Kant, the priority of right is 'derived entirely from the concept
of freedo m in the mutual external relationships ofhuman beings, and
has nothing to do with the end which all men have by nature ( i . e . the

ing this end ' ( I 7 9 3 : 73 ) . As such, it must have a basis prior to all
aim of achieving happiness) or with the recognized means of attain­

empirical ends. Even a union founded on some common end which

all members share will not do. Only a union 'as an end in itself which
they all ought to share and which is thus an absolute and primary
duty in all external relationships whatsoever among human beings '
can secure justice and avoid the coercion of some by the convictions
of others . Only in such a union can no one 'compel me to be happy in
I Mill goes on to claim that j ustice j us t is whatever utility requires. Where the general
maxims ofj ustice are outweighed, 'we usually say, not that j ustice must give way to some
other moral principle, but that what is j ust in ordinary cases is, by reason of that other
principle, not j ust in the particular case. By this useful accommodation of language, the
character of indefeasibility attributed to j ustice is kept up, and we are saved from the
necessity of maintaining that there can be laudable inj ustice' ( 1 863: 469) .

5
Introduction: Liberalism and the primacy ofjustice

accordance with his conception of the welfare of others' (I793: 73-4).

Only when I am governed by principles that do not presuppose any
particular ends am I free to pursue my own ends consistent with a
similar freedom for all.
On the Kantian view, the two strands of the deontological ethic
hang together. The moral priority of justice is made possible (and
necessary) by its foundational priority. Justice is more than just
another value, because its principles are independently derived.
Unlike other practical principles, the moral law is not implicated in
advance in various contingent interests and ends; it does not presup­
pose any particular conception of the good. Given its basis prior to all
merely empirical ends, justice stands privileged with respect to the
good, and sets its bounds.
But this raises the question what the basis of the right could
possibly be. If it must be a basis prior to all purposes and ends,
unconditioned even by 'the special circumstances of human nature'
(I 785: 92 ), where could such a basis conceivably be found? Given the
stringent demands of the deontological ethic, the moral law would
seem almost to require a foundation in nothing, for any material
precondition would undermine its priority. ' Duty!' asks Kant at his
most lyrical, 'What origin is there worthy of thee, and where is to be
found the root of thy noble descent which proudly rejects all kinship
with the inclinations?' (I788: 89).
His answer is that the basis of the moral law is to be found in the
subject, not the object of practical reason, a subject capable of an
autonomous will. No empirical end but rather 'a subject of ends,

maxims of action' ( I 785: 1 05 ). Nothing other than 'the subject of all

namely a rational being himself, must be made the ground for all

possible ends himself ' can give rise to the right, for only this subject is
also the subject of an autonomous will. Only such a subject could be
that 'something which elevates man above himself as a part of the
world of sense' and enables him to participate in an ideal, uncon­
ditioned realm wholly independent of our social and psychological
inclinations. And only this thoroughgoing independence can afford
us the detachment we need if we are ever freely to choose for
ourselves, unconditioned by the contingencies of circumstance. On
the deontological view, what matters above all is not the ends we
choose but our capacity to choose them. And this capacity, being
prior to any particular end it may affirm, resides in the subject. 'It is

6
T he t ranscende nt al su bj ect

nothing else than personality, i. e . , the freedom and independence

from the mechanism of nature regarded as a capacity of a being
which is subject to special laws (pure practical laws given by its own
reason) ' ( 1 788: 8g) .
The concept of a subject given prior to and independent of its
objects offers a foundation for the moral law that, unlike merely
empirical foundations, awaits neither teleology nor psychology . I n
this way, i t powerfully completes the deontological vision. A s the
right is prior to the good , so the subj ect is prior to its ends. For Kant,
these parallel priorities explain 'once and for all the reasons which
occasion all the confusions of philosophers concerning the supreme
principle of morals. For they sought an object of the will in order to
make it into the material and the foundation of a law . ' But this was
bound to leave their first principles mired in heteronomy. ' I nstead ,
they should have looked for a law w hich directly determined the will
a priori and only then sought the obj ect suitable to it' ( 1 788: 66 ) . Had
they done so, they would have arrived at the distinction between a
subject and an object of practical reason , and thus a basis of right
independent of any particular object.
If the claim for the primacy ofjus tice is to succeed , if the right is to
be prior to the good in the interlocking moral and foundational
senses we have distinguished, then some version of the claim for the
primacy of the subject must succeed as well. This much seems clear.
What remains to be shown is whether this las t claim can be
defended . How do we know that there is any such subject, identi­
fiable apart from and prior to the objects it seeks? This question
assumes special interest once it is recalled that the claim for the
priority of the subject is not an empirical claim . If it were, it could
hardly do the work that recommends it to the deontological ethic in
the first place.

THE TRANSCENDENTA L SUBJ ECT

Kant offers two arguments in support of his notion of the subject

- one epistemological, the other practical . Both are forms of 'trans­
cendental' arguments, in that they proceed by seeking out the
presuppositions of certain apparently indispensable features of our
experience . The epistemological argument inquires into the presup­
positions of self-knowledge . It begins with the thought that I cannot
7
Introduction: Liberalism and the primacy ofjustice

know everything there is to know about myself j ust by looking, or

introspecting. For when I introspect, all I can see are the deliver­
ances of my senses; I can know myself only qua object of experience,
as the bearer of this or that desire, inclination, aim, disposition , and
so on . But this kind of self-knowledge is bound to be limited . For it
can never enable me to get behind the stream of appearances to see
what they are appearances of ' So far as man is acquain ted with
himself by inner sensation . . . he cannot claim to know what he is in
himself' (Kan t I 785: I I g) . Introspection , or 'inner sense' alone
could never deliver knowledge of anything standing behind these
appearances, for any such deliverance would ins tan tly dissolve into
yet another appearance . None the less, we must presume something
further. 'Beyond this character of himself as a subj ect made up, as it
is, of mere appearances he must suppose there to be something else
which is its ground - namely his Ego as this may be constituted in
itself' (Kant I 785 : I I g) .
This something further, which we cannot know empirically but
must none the less pres uppose as the condition of knowing anything
at all, is the subj ect itself. The subject is the something 'back there ' ,
antecedent t o any particular experience, that unifies o u r diverse
perceptions and holds them together in a single consciousness. I t
provides the principle o f unity without which our self-perceptions
would be nothing more than a stream of disconnected and ever­
changing representations, the perceptions of no one. And while we
cannot grasp this principle empi rically, we must pres ume i ts validity
if we are to make sense of self-knowledge at all .
The though t that the representations given in intuition one and all belong to
me, is therefore equivalent to the thought that I unite them in one
self-consciousness, or can at least so unite them; and although this thought
is not itself the consciousness of the synthesis of the representations, it
presupposes the possibility of that synthesis. I n other words, only in so far
as I can grasp the manifold of the representations in one consciousness, do I
call them one and all mine. For otherwise I should have as many-coloured

myself ( Kant I 787: I 54) .

and diverse a self as I have representations of which I am conscious to

The discovery that I must understand myself as a subject as well

as an object of experience suggests two different ways of conceiving
the laws that govern my actions . It thus leads from the epistemolo­
gical argument to a further, practical argument for the priority of the

8
T he tranrcendental subject

subject. Qua obj ect of experience, I belong to the sensible world; my

actions are determined , as the movements of all other objects are

effect. Qua subj ect of experience, by contrast, I inhabit an intelligible

determined , by the laws of nature and the regularities of cause and

or super-sensible world ; here, being independent of the laws of

nature, I am capable of autonomy, capable of acting according to a
law I give myself.
Only from this second stand point can I regard myself as free, 'for
to be independent of determination by causes in the sensible world
( and this is what reason must always attribute to itself) is to be free'
(Kant I 785: 1 20) . Were I wholly an empirical being, I would not be
capable offreedom, for every exercise of will would be conditioned by
the desire for some obj ect. All choice would be heteronomous choice,
governed by the pursuit of some end . My will could never be a first
cause, only the effect of some prior cause, the instrument of one or
another impulse or inclination . In so far as we think of ourselves as
free, we cannot think of ourselves as merely empirical beings . 'When
we think of ourselves as free, we transfer ourselves into the intelli­
gible world as members and recognize the autonomy of the will'
(Kant 1 785 : 1 2 1 ). And so the notion of a subj ect prior to and
independent of experience, such as the deontological ethic requires ,
appears not only possible but indispensable, a necessary presupposi­
tion of the possibility of self-knowledge and of freedom .
We can now see more clearly what, on the deontological ethic, the
claim for the primacy ofjustice consists in. On the Kantian view, the
priority of right is both moral and foundational . It is grounded in the
concept of a subj ect given prior to its ends, a concept held indispens­
able to our understanding ourselves as freely choosing, autonomous
beings . Society is best arranged when it is governed by principles
that do not presuppose any particular conception of the good , for any
other arrangement would fail to respect persons as beings capable of
choice; it would treat them as objects rather than subjects , as means
rather than ends in themselves.
Deon tologi cal themes find similar expression in much con tem­
porary li beral thought. Thus 'the rights secured by j ustice are not
subj ect to the calculus of social interests ' ( Rawls 1 9 7 1: 4) , but instead
'function as trump cards held by individ uals' ( Dworkin 1 9 78: 1 36)
against policies that would impose some particular vision of the good
on society as a whole. 'Since the citizens of a society differ in their

9
Introduction: Liberalism and the primacy ofjustice

conceptions ' , the government fails to respect them as equals 'if it

prefers one conception to the other, either because the officials
believe that one is intrinsically superior, or because one is held by the
more numerous or more powerful group' ( Dworkin 1 978: 1 2 7 ) . By
comparison with the good , the concepts of right and wrong ' have an
independent and overriding status because they establish our basic
position as freely choosing entities '. More important than any
choice, the val ue of personhood 'is the presupposition and substrate
of the very concept of choice . And that is why the norms surrounding
respect for person may not be compromised , why these norms are
absolute in respect to the various ends we choose to pursue' ( Fried
1 9 78: 8- 9, 2 9 ) .
By virtue of its independence from ordinary psychological and
teleological assumptions, this liberalism, at least in its contemporary
versions , typically presents itself as immune from most controversies
to which political theories have traditionally been vulnerable, espe­
cially on questions ofhuman nature and the meaning of the good life .
T h u s it is claimed that 'liberalism does n o t rest o n a n y special theory
of personality' ( Dworkin 1 9 78: 1 42 ) , that its key assumptions involve
'no particular theory of human motivation' (Rawls 1 9 7 1: 1 2 9 ) , that
'liberals, as such , are indifferent' to the ways of life ind ividuals
choose to pursue ( Dworkin 1 9 78 : 1 43 ) , and that, in order to accept
li beralism, one 'need not take a position upon a host of Big Questions
of a highly controversial character' (Ackerman 1 9 80: 36 1 ) .
But if certain ' big questions' of philosophy and psychology are
beside the point for deontological li beralism, it is only because it
locates its con troversy elsewhere. As we have seen , this liberalism
avoids reliance on any particular theory of the person, at least in the
trad itional sense of attributing to all human beings a determinate
nature, or certain essential desires and inclinations , such as selfish­
ness or sociability, for example. But there is another sense in which
this liberalism does imply a certain theory of the person . It concerns
not the obj ect ofhuman desires but the subject of desire, and how this
subj ect is constitu ted.
For justice to be primary, certain things must be true of us. We
must be creatures of a certain kind , related to human circu mstance
in a certain way . In particular, we must stand to our circumstance
always at a certain distance, conditioned to be sure, but part of us
always antecedent to any conditions. Only in this way can we view

10
T he sociological objection

ourselves as subjects as well as objects of experience, as agents and

not just instruments of the purposes we pursue. Deontological
liberalism supposes that we can , indeed must, understand ourselves
as independent in this sense. I shall argue that we cannot, and that,
in the partiality of this self-image, the limits ofj ustice can be found .
Where, then, does the deontological theory of the person go
wrong? How do its shortcomings undermine the primacy ofj ustice,
and what rival virtue appears when the limits of justice are found?
These are the questions this essay seeks to answer. To set the stage
for my argument it will be helpful first to consider two other
challenges that might be made to the Kantian view.

THE S O C I O LO G I C A L OBJ ECTION

The first might be called the sociological objection , for it begins by

emphasizing the pervasive influence of social conditions in shaping
individual values and political arrangements . It claims that liberal­
ism is wrong because neutrality is impossible, and that neu trality is
impossible because try as we might we can never wholly escape the
effects of our conditioning. All political orders thus em body som e
values; the question is whose values prevail, and who gains and loses
as a result. The vaunted independence of the deontological subject is
a liberal illusion . It misunderstands the fundamentally 'social'
nature of man, the fact that we are conditioned beings 'all the way
down ' . There is no point of exemption , no transcendental subject
capable of standing outside society or outside experience. We are at
every moment what we have become, a concatenation of desires and
inclinations with nothing left over to inhabit a noumenal realm . The
priority of the subj ect can only mean the priority of the individual,
thus biasing the conception in favor of individualistic values familiar
to the liberal trad ition . Justice only appears primary because this
individ ualism typically gives rise to conflicting claims . The limits
of justice would therefore consist in the possibility of cultivating
those co-operative virtues , such as altruism and benevolence, that
render conflict less pressing. But these are precisely the virtues least
likely to flourish in a society founded on individ ualistic assumptions.
In short, the ideal of a society governed by neutral principles is
li beralism ' s false promise. It affirms individ ualistic values while
pretending to a neutrality which can never be achieved.

I I
Introduction: Liberalism and the primacy ofjustice

But the sociological obj ection fails in various ways to appreciate

the force of the deontological view . First, it misunderstands the
neutrality this liberalism claims to offer. What is neutral about the
principles of right is not that they admit all possible values and ends
but rather that they are derived in a way that does not depend on any
particular values or ends. To be sure, once the principles ofj ustice,
thus derived , are on hand , they rule out certain ends - they would
hardly be regulative if they were incompatible with nothing- but only
those that are unj ust, that is, only those inconsistent with principles
which do not themselves depend for their validity on the validity of
any particular way of life. Their neutrality describes their founda­
tion, not their effect.
But even their effect is in important ways less restrictive than the
sociological obj ection suggests . Altruism and benevolence, for exam­
ple, are wholly compatible with this liberalism, and there is nothing
in its assumptions to discourage their cultivation . The priority of the
subj ect does not say that we are governed by self-interest, only that
whatever interests we have must be the interests of some subj ect.
From the standpoint of the right, I am free to seek my own good or
the good of others , so long as I do not act unj ustly. And this
restriction has not to do with egoism or altruism but rather with the
overriding interest in assuring a similar liberty for others . The
co-operative virtues are in no way inconsistent with this liberalism.
Finally, it is unclear how the sociological obj ection proposes to
deny the deontological notion of independence. If it means to offer a
psychological obj ection, then it cannot reach the deontological view,
which makes an epistemological claim. The independence of the
subj ect does not mean that I can , as a psychological matter, summon
at any moment the detachment required to overcome my prejudices
or step outside my convictions, but rather that my values and ends
do not define my identity, that I must regard myself as the bearer of a
self distinct from my values and ends, whatever they may be.
I f, on the other hand , the sociological obj ection means to challenge
this epistemological claim, it is unclear what the basis for this
challenge could be. H ume perhaps came closest to portraying a
wholly empirically-conditioned self, such as the sociological view
requires, when he described the self as ' a bundle or collection of

able rapidity, and are in a perpetual flux and movement' ( 1 73 9 : 2 52 ) .

different perceptions , which succeed each other with an inconceiv­

12
D eo nto logy with a Hum ean f ace

But as Kant would later argue, 'no fixed and abiding self can present
itself in this flux of inner appearances ' . To make sense of the
continuity of the self through time, we must presume some principle

possible' ( I 78 I: I 36) . Indeed Hume himself anticipated this difficul­

of unity which ' precedes all experience, and makes experience itself

ty when he admitted that he could not in the end account for those

consciousness' ( I 7 39: 636) . Problematic though the Kantian trans­

principles ' that unite our successive perceptions in our thought or

cendental subj ect may be, the sociological obj ection seems ill­
equipped to offer an effective critique. The epistemology it must
presuppose is hardly more plausible.

D E O N T O L O GY W I T H A H U M EAN FA CE

The second challenge poses a deeper difficulty with the Kantian

subj ect. Like the first, it comes from an empiricist direction . But
unlike the first, it seeks to secure deontological liberalism rather than
oppose it. In fact this second challenge is less an objection to the
Kantian view than a sympathetic reformulation . It embraces the
priority of the right over the good , and even affirms the priority of the
self over its ends. Where this view departs from Kant is in denying
that a prior and independent self can only be a transcendental, or
noumenal subj ect, lacking altogether an empirical foundation . This
' revisionist' deontology captures the spirit of much contemporary
liberalism , and finds its fullest expression in the work ofjohn Rawls.
'To develop a viable Kantian conception ofjustice,' he writes , ' the
force and content of Kant's doctrine must be detached from its
background in transcendental idealism' and recast within the
'canons of a reasonable empiricism' ( Rawls I 9 7 7: I 65) .
For Rawls, the Kantian conception suffers from obscurity and
arbitrariness, for it is unclear how an abstract, disembodied subj ect
could without arbitrariness produce determinate principles of j us­
tice, or how in any case the legislation of such a subj ect would apply
to actual human beings in the phenomenal world. The idealist
metaphysic, for all its moral and political advantage, cedes too much
to the transcendent, and in positing a noumenal realm wins for
justice its primacy only at the cost of denying it its human situation .
And so Rawls takes as his proj ect to preserve Kant's deontological
teaching by replacing Germanic obscurities with a domesticated
Introduction: Liberalism and the primacy ofjustice

metaphysic less vulnerable to the charge of arbitrariness and more

congenial to the Anglo-American temper. His proposal is to derive
first principles from a hypothetical choice situation ( the 'original
position ' ) , characterized by conditions meant to yield a determinate
outcome fit for actual human beings. Not the kingdom of ends but
the ordinary circumstances of j ustice - as borrowed from Hume­
prevail there. Not an ever-receding moral future but a present firmly
planted in human circumstance provides justice its occasion . I f
deontology b e t h e result, it will b e deontology with a Humean face !
The theory of j ustice tries to present a natural procedural rendering of
Kant's conception of the kingdom of ends, and of the notions of autonomy
and the categorical imperative. In this way the underlying structure of

can be seen more clearly and presen ted relatively free from obj ection ( 264) .'
Kant's doctrine is detached from its metaphysical surroundings so that it

Whether Kant's metaphysics are detachable 'surroundings ' or

inescapable presuppositions of the moral and political aspirations
Kant and Rawls share - in s hort, whether Rawls can have liberal
politics without metaphysical embarrassment - is one of the central
issues posed by Rawls' conception. This essay argues that Rawls'
attempt does not succeed , and that deontological liberalism cannot
be rescued from the difficulties associated with the Kantian subj ect.
Deontology with a Humean face either fails as deontology or recre­
ates in the original position the disembodied subj ect it resolves to
avoid . Justice cannot be primary in the deontological sense, because
we cannot coherently regard ourselves as the kind of beings the
deon tological ethic - whether Kantian or Rawlsian - requires us to
be. But attending to this liberalism is of more than critical interest
alone. For Rawls' attempt to situate the deontological self, properly
reconstructed , carries us beyond deontology to a conception of
community that marks the limits of j ustice and locates the incom­
pleteness of the liberal ideal .

2 I am indebted to Mark Hulbert for suggesting this phrase.

3 All page numbers given alone in round brackets refer to Rawls 1 9 7 1 . A Theory ofJustice,
Oxford.

14
I

Justice and the Moral Subject

Like Kant, Rawls is a deontological liberal . His book takes the main
thesis of the deontological ethic as its central claim. That this claim
has received little direct discussion in the voluminous critical litera­
ture on A Theory ofjustice may attest to i ts fixed place in the moral and
political assumptions of the time. It concerns not the principles of
j ustice but the status of j ustice itself. It is the assertion that both
opens the book and concludes it, the core conviction Rawls seeks
above all to defend . I t is the claim that 'j ustice is the first virtue of
social institutions' , the single most important consideration in asses­
sing the basic structure of society and the overall direction of social
change.
Justice is the first virtue of social institutions, as truth is of systems of
thought. A theory however elegant and economical must be rej ected or
revised if it is untrue; likewise laws and institutions no matter how efficient
and well-arranged must be reformed or abolished if they are u njust. . . .
Being first virtues of human activities, truth and j ustice are uncompro­
mising ( 3-4).

I have tried to set forth a theory that enables us to understand and to assess
these feelings about the primacy of j ustice. Justice as fairness is the out­
come: it articulates these opinions and supports their general tendency
(sB6) .
I t is this claim for the primacy ofjustice that I propose to examine.

T H E PRIMACY O F J USTICE AND T H E PRIORITY O F T H E S E L F

Now the primacy ofj ustice is a powerful claim, and there is a danger
that the familiarity of the thought is apt to blind us to its boldness. To
understand why it is intuitively appealing but at the same time
deeply puzzling and problematic, we might consider the following
reconstruction of the claim, designed to capture both its familiarity
and its force: j ustice is not merely one important value among others ,
IS
justice and the moral subject

to be weighed and considered as the occasion requires , but rather the

m eans by which values are weighed and assessed . It is in this sense the
'value of values ',' so to speak, not subject itself to the same kind of
trade-offs as the values it regulates . justice is the standard by which
conficting values are reconciled and competing conceptions of the
good accommodated if not always resolved . As such, it must have a
certain priority with respect to those values and those goods . No
conception of the good could possibly defeat the requirements of
justice, for these requirements are of a qualitatively different order;
their validity is established in a different way . With respect to social
values generally, justice stands detached and aloof, as a fair decision
procedure stands aloof from the claims of the disputants before it.
But what exactly is the sense in which j ustice, as the arbiter of
values, 'must' be prior with respect to them? One sense of this
priority is a mo ral 'must' which emerges from Rawls' critique of
utilitarian ethics . From this point of view, the priority ofj ustice is a
requirement of the essential plurality of the human species and the
integrity of the individuals who comprise it. To sacrifice j ustice for
the sake of the general good is to violate the inviolable, to fail to
respect the distinction between persons .
Each person possesses an inviolability founded on justice that even the
welfare of society as a whole cannot override. For this reason justice denies
that the loss of freedom for some is made right by a greater good shared by
others . It does not allow that the sacrifices imposed on a few are outweighed
by the larger sum of advantages enjoyed by many. Therefore in a j ust
society the liberties of equal citizenship are taken as settled ; the rights
secured by j ustice are not subject to political bargaining or to the calculus of
social interes ts ( 3-4) .

But there is another sense in which j ustice 'must' be prior to the

values it appraises - prior in the sense of independently derived - and
this has to do with a problematic feature of standards of j udgment
generally. It is an epistemological rather than a moral requirement,
and arises from the problem of distinguishing a standard of assess­
ment from the thing being assessed . As Rawls insists, we need an
1 The phrase is that of Alexander Bickel, who attributes to law a primacy comparable to the
status accorded justice here. 'The irreducible value, though not the exclusive one, is the
idea oflaw. Law is more than j ust another opinion; not because it embodies all right values
. . . but because it is the value of values. Law is the principle institution through which a
society can assert its values' ( 1 9 75: 5 ) .

r6
T he priority of the se(f

'Archimedean point' from which to assess the basic structure of

so ciety. The problem is to give an account of where such a point
could conceivably be found . Two possibilities seem to present them­
selves, each equally unsatisfactory: if the principles of justice are
derived from the values or conceptions of the good current in the
so ciety, there is no assurance that the critical standpoint they pro­
vide is any more valid than the conceptions they would regulate,
since, as a product of those values, j ustice would be subj ect to the
same contingencies . The alternative would seem a standard some­
how external to the values and interests prevailing in society . But if
our experience were disqualified entirely as the source of such princi­
ples , the alternative would seem to be reliance on a priori assump­
tions whose credentials would appear equally suspect, although for
opposite reasons . Where the first would be arbitrary because contin­
gent, the second would be arbitrary because groundless . Where
j ustice derives from existing values, the standards of appraisal blur
with the obj ects of appraisal and there is no sure way of picking out
the one from the other. Where j ustice is given by a priori principles,
there is no sure way of connecting them up.
These then are the perplexing and difficult demands of the
Archimedean point - to find a standpoint neither compromised by
its i mplication in the world nor dissociated and so disqualified by
detachment. 'We need a conception that enables us to envision our
obj ective from afar' ( 2 2 ) , but not too far; the desired standpoint is 'not
a perspective from a certain place beyond the world, nor the point of
view of a transcendent being; rather it is a certain form of thought
and feeling that rational persons can adopt within the world' ( 587 ) .
Before we consider Rawls' response to this challenge, i t may be
worth noting how Rawls' case for the primacy ofj ustice is related to
several parallel claims throughout his theory which, taken together,
reveal a structure of argument characteristic of the deontological
ethic as a whole. Closely tied to the primacy of justice is the more
general notion of the priority of the right over the good . Like the
primacy of j ustice, the priority of the right over the good appears
initially as a first-order moral claim, in opposition to utilitarian
doctrine, but comes ultimately to assume a certain meta-ethical
status as well , particularly when Rawls argues more generally for
deontological ethical theories as opposed to teleological ones .
As a straightforward moral claim the priority of right over good

I7
Justice and the moral subject

means that principles of right invariably outweigh considerations of

welfare or the satisfaction of desire, however intense, and constrain
in advance the range of desires and values properly entitled to
satisfaction.
The principles of right, and so ofj ustice, put limits on which satisfactions
have value; they impose restrictions on what are reasonable conceptions of
one's good . . . . We can express this by saying that in j ustice as in fairness
the concept of right is prior to that of the good . . . . The priority ofjustice is
accounted for, in part, by holding that the interests requiring the violation

override its claim ( 3 1 ) .

of j us tice have no value. Having no merit in the first place, they cannot

I n j ustice as fairness, unlike utilitarianism, the individual ' s right to

equal liberty in the face of maj ority preferences to the contrary is
unconditionally affirmed .
The intense convictions of the maj ority, if they are indeed mere preferences
without any foundation in the principles ofjustice antecedently established,
have no weight to begin with . The satisfaction of these feelings has no value
that can be put in the scales against the claims of equal liberty . . . . Against
these principles neither the intensity of feeling nor its being shared by the
majority counts for anything. On the contract view, then, the grounds of
liberty are completely separate from exis ting preferences (450 ) .

Although Rawls argues first against utilitarian conceptions, his

overall proj ect is more ambitious , for j ustice as fairness stands not
only against utilitarianism, but against all teleological theories as
such. As a second-order, meta-ethical claim, the priority of right
means that, of the ' two concepts of ethics ' , the right is derived
independently from the good, rather than the other way around .
This foundational priority allows the right to stand aloof from pre­

ception deontological rather than teleological ( 24-5, 30) .

vailing values and conceptions of the good , and makes Rawls' con­

One of the first-order consequences of the deontological ethic is a

firmer foundation for the equal liberty of individuals than could be
available on teleological assumptions . I n this, the importance of
deontology to familiar liberal concerns mos t dearly appears . Where
the right is instrumental to the advancement of some end held to be
prior, the denial ofliberty for some may be j ustified in the name of an
overriding good for others . The liberties of eq ual citizenship are thus
'insecure when founded upon teleological principles . The argument

18
T he priority of the self

for them relies upon precarious calculations as well as controversial

and uncertain premises' ( 2 I I ) . On the deontological view, 'equal
liberties have a different basis altogether' . No longer mere means for
maximizing satisfactions or realizing some overriding aim, ' these
rights are assigned to fulfill the principles of co-operation that
citizens would acknowledge when each is fairly represented as a
moral person' ( 2 I I ) , as an end in himself.
But the fail ure to secure the righ ts of eq ual li berty betrays a deeper
flaw in the teleological conception. In Rawls' view, teleology con­
fuses the relation of the right to the good because it misconceives the
relation of the self to its ends . This leads Rawls to assert yet another
deontological priority . Teleology to the contrary, what is most
essential to our personhood is not the ends we choose but our
capacity to choose them . And this capacity is located in a self which
must be prior to the ends it chooses .
The structure of teleological doctrines is radically misconceived ; from the
start they relate the right and the good in the wrong way. We should not
attempt to give form to our life by first looking to the good independently
defined . It is not our aims that primarily reveal our nature but rather the
principles that we would acknowledge to govern the background conditions
under which these aims are to be formed and the manner in which they are to be
pursued. For the selfis prior to the ends which are affirmed by it; even a dominant end
must be chosen from among numerous possibilities . . . . We should therefore
reverse the relation between the right and the good proposed by teleological
doctrines and view the right as prior. The moral theory is then developed by
working in the opposite direction [emphasis added] (560).

The priority of the self over its ends means that I am not merely the
passive receptacle of the accumula ted aims, attri butes, and purposes
thrown up by experience, not simply a prod uct of the vagaries of
circumstance, but always , irred ucibly, an active , willing agent,
distinguishable from my surroundings , and capable of choice . To
identify any set of characteristics as my aims, ambitions, desires , and
so on, is always to imply some subject 'me' standing behind them,
and the shape of this ' me' must be given prior to any of the ends or
attri butes I bear . As Rawls writes, 'even a dominant end must be
chosen from among numerous possibilities ' . And before an end can
be chosen, there must be a self around to choose it.
But what exactly is the sense in which the self, as an agent of
choice, 'must' be prior to the ends it chooses? One sense of the

I9
Justice and the moral subject

priority is a moral 'must' which reflects the imperative to respect

above all the autonomy of the individual, to regard the human
person as the bearer of a dignity beyond the roles that he inhabits
and the ends he may pursue. But there is another sense in which the
self 'must' be prior to the ends it affirms - prior in the sense of
independently identifiable - and this is an epistemological require­
ment.
Here the account of the self reproduces the perplexities we encoun­
tered in the case ofjustice. There, we needed a standpoint of apprais­
al independent of prevailing social values. In the case of the person,
we need a notion of the subject independent of its contingent wants
and aims. As the priority ofjustice arose from the need to distinguish
the standard of appraisal from the society being appraised, the
priority of the self arises from the parallel need to distinguish the
subj ect from its situation . Although this account is not offered by
Rawls himself, I believe it to be implicit in his theory, and a reason­
able reconstruction of the perplexities he seeks to address .
I f all the self consisted in were a concatenation of various contin­
gent desires, wants, and ends , there would be no non-arbitrary way,
either for the self or for some outside observer, to identify these
desires, interests, and ends, as the desires of any particular subject.
Rather than be of the subject, they would be the subj ect. But the
subj ect they would be would be indistinguishable from th e sea of
undifferentiated attributes of an unarticulated situation, which is to
say it would be no subj ect at all, at least no subject we could
recognize or pick out as resembling a human person .
Any theory of the self of the form 'I am x,y, and z', rather than 'I
have x,y, and z' , (where x, y, and<: are desires , etc . ) collapses the
distance between subject and situation which is necessary to any
coherent conception of a particular human s ubj ect. This space, or
measure of detachment, is essential to the ineliminably possessive
aspect of any coherent conception of the self. The possessive aspect of
the self means that I can never fully be constituted by my attributes,
that there must always be some attributes I have rather than am .
Otherwise, j ust any change in my situation, however slight, would
change the person I am. But taken literally, and given that my
situation changes in some respect at least with every passing mo­
ment, this would mean that my identity would blur indistinguish­
ably into ' my' situation . Without some distinction between the

20
T he prio rity of the self

subj ect and the object of possession, it becomes impossible to disting­

uish what is me from what is mine, and we are left with what might be
called a radically situated subject.
Now a radically situated subj ect is inadequate to the notion of the
person in the same way a standard of appraisal thoroughly impli­
cated in existing values is inadequate to the notion of j ustice; the
impulse to priority reflected in the search for an Archimedean point
is the response to both predicaments.
But in both cases, the alternatives are seriously constrained, the
possible sites for the Archimedean point severely restricted. In the
case ofj ustice, the alternative to a situated conception would seem an
appeal beyond experience to a priori principles . But this would be to
assert the desired priority with a vengeance, and would achieve the
necessary detachment only at the price of arbitrariness. A similar
difficulty arises in the case of the subj ect. For a self totally detached
from its empirically-given features would seem no more than a kind
of abstract consciousness ( conscious of what?) , a radically situated
subj ect given way to a radically disembodied one. Here again, 'we
need a conception that enables us to envision our obj ective from
afar' , but not so far that our obj ective fades from view and our vision
dissolves into abstraction.
Thus we can see, in rough outline at least, how the argument
hangs together, how the primacy ofj us tice, the rej ection of teleology,
and the priority of the self are related, and finally, how these claims
lend support to familiar liberal positions. The connection between
the meta-ethical view and the conception of the self can be seen in the
fact that teleological and d eontological conceptions account for the
unity of the self in different ways . Where, according to Rawls,
teleological conceptions suppose that the unity of the self is achieved
in the course of experience - in the case of hedonism, through
maximizing the sum of pleasurable experiences within its 'psychic
boundaries' (56 1 ) - j ustice as fairness reverses this perspective and
conceives the unity of the self as something antecedently established,
fashioned prior to the choices it makes in the course of its experience.
The parties [in the original position] regard moral personality and not the
capacity for pleasure and pain as the fundamental aspect of the self . . . . The
main idea is that given the priority of right, the choice of our conception of
the good is framed within definite limits . . . The essential uniry of the self is
.

already provided by the conception of right [emphasis added) (563 ) .

21
Justice and the moral subject

Here, as in the case of equal liberty and the priority of right

discussed above, deontological assumptions can be seen to generate
familiar liberal conclusions and to ground them more firmly than
traditional empiricist or utilitarian metaphysics allowed . The theme
common to much classical liberal doctrine that emerges from the
deontological account of the unity of the self is the notion of the
human subj ect as a sovereign agent of choice, a creature whose ends
are chosen rather than given, who comes by his aims and purposes
by acts of will, as opposed, say, to acts of cognition . 'Thus a moral
person is a subj ect with ends he has chosen, and his fundamental
preference is for conditions that enable him to frame a mode of life
that expresses his nature as a free and equal rational being as fully as
circumstances permit' (56 1 ) .
The antecedent unity of the self means that the subj ect, however
heavily conditioned by his surroundings, is always, irreducibly,
prior to his values and ends, and never fully constituted by them .
Though there may be times when conditions are formidable and
choices few, man 's sovereign agency as such is not dependent on any
particular conditions of existence, but guaranteed in advance. On
the deontological conception, we could never be so thoroughly con­
ditioned that our self became fully constituted by our situation, our
ends fully determined in such a way that the self ceased to be prior
with respect to them . The consequences of such a view for politics
and j ustice are considerable. As long as it is assumed that man is by
nature a being who chooses his ends rather than a being, as the
ancients conceived him, who discovers his ends, then his fun­
damental preference must necessarily be for conditions of choice
rather than, say, for conditions of self-knowledge.
The full force of the claim for the primacy of j ustice in both its
moral and epistemological dimensions can more clearly be appreci­
ated in the light ofRawls ' discussion of the self. Since the self owes its
constitution, its antecedent status, to the concept of right, we can
only express our true nature when we act out of a sense of j ustice.
This is why the sense of j ustice cannot be regarded as merely one
desire among others but must be seen as a motivation of a qualita­
tively higher order, why j ustice is not merely one important value
among others but truly the first virtue of social institutions.

The desire t o express o u r nature a s a free and equal rational being can be

22
T he priority oft he se(f

fulfilled only by acting on the principles of right and j ustice as having first
priority . . . . I t is acting from this precedence that expresses our freedom
from contingency and happenstance. Therefore in order to realize our
nature we have no alternative but to plan to preserve our sense ofj ustice as
governing our other aims. This sentiment cannot be fulfilled if it is com­
promised and balanced against other ends as but one desire among the rest.
I t is a desire to conduct oneselfin a certain way above all else, a striving that
contains within itself its own priority ( 574) .

What we cannot do is express our nature by following a plan that views the
sense of j u s tice as but one desire to be weighed against others . For this
sentiment reveals what the person is, and to compromise it is not to achieve
for the self free reign but to give way to the contingencies and accidents of
the world ( 575) .

The connection between the primacy ofj ustice and other central
features of Rawls' conception - the priority of right, the deontologi­
cal meta-ethic, and the antecedent unity of the self - reveals some­
thing of the overall structure of Rawls' theory, and indicates how
deep and powerful a claim the primacy ofj ustice is intended to be. I t
also suggests how this cluster o f assertions, i f they can b e defended ,
provides an impressive foundation , at once moral and epistemologic­
al, for certain central liberal doctrines . We have sought to under­
stand these claims and to clarify their connections by seeing them as
answers to perplexities posed by two related reconstructions ; the first
seeks a standard of appraisal neither compromised by existing stan­
dards not arbitrarily given , and the second seeks an account of the
self as neither radically situated and therefore indistinguishable from
its surroundings nor radically disembodied and therefore purely
formal . Each reconstruction poses a set of unacceptable alternatives
and requires for its solution a kind of Archimedean point that
manages to detach itself from the contingent without lapsing into
arbitrariness.

To this point, Rawls' project looks much like Kant' s . But notwith­
standing their roughly common agenda and deontological affinities ,
Rawls' proposed solu tion departs radically from Kant's. The differ­
ence reflects Rawls ' concern to establish the required deontological
priorities - including the priority of the self - without recourse to a
transcendent or otherwise disembodied subj ect. This contrast
assumes special interest given that Kant's idealism - the dimension
23
Justice and the moral subject

Rawls seeks above all to avoid - would lead much nineteenth- and
twentieth-century Continental philosophy in a direction largely
alien to the Anglo-American tradition of moral and political thought
in which Rawls' work is firmly installed .
For Kant, the priority of right, or the supremacy of the moral law,
and the unity of the self, or the synthetic unity of apperception, could
only be established by means of a transcendental deduction and the
positing of a noumenal or intelligible realm as the necessary presup­
position of our capacity for freedom and self-knowledge. Rawls
rej ects Kant's metaphysics, but believes he can preserve their moral
force 'within the scope of an empirical theory' ( Rawls I 9 79 : I 8) . This
is the role of the original position.

L I B E R A L I S M WIT H O U T M E TA P H YS I C S : T H E O R I G I N A L P O SITION

The original position is Rawls' answer to Kant; it is his alternative to

the route represented by the Critique of Pure Reason and the key to
Rawls' solution to the perplexities we have considered . It is the
original position that 'enables us to envisage our obj ective from afar' ,
but not so far as to land us in the realm of transcendence. I t aims to
satisfy these demands by describing an initial situation of fairness
and defining as j ust those principles that rational parties subject to
its conditions would agree to.
Two crucial ingredients equip the original position to solve the
dilemmas described by the reconstructions and to answer the need
for an Archimedean point. Each takes the form of an assumption
about the parties to the original position: one says what they do not
know, the other, what they do know. What they do not know is any
information that would distinguish any one of them from any other
as the particular human beings they are. This is the assumption of
the veil of ignorance. It means that the parties are assumed to be
deprived of any knowledge of their place in society, their race, sex, or
class, their wealth or fortune, their intelligence, strength, or other
natural assets and abilities . Nor even do they know their conceptions
of the good, their values, aims, or purposes in life. They know that
they do in fact possess such conceptions and deem them worthy of
advancement, whatever they are, but must choose the principles of
j ustice in temporary ignorance of them. The purpose of this restric­
tion is to prevent the choice of principles from being prej udiced by
The original position

the contingency of natural and social circumstances, to abstract from

all considerations deemed irrelevant from a moral point of view. It is
the veil of ignorance that assures that the principles ofj ustice will be
chosen under conditions of equality and fairness. Since the parties to
the contract are not distinguished by different interests, a further
consequence of the veil of ignorance is to assure that the initial
agreement be unanimous.
What the parties do know is that they, like everyone else, value
certain primary social goods. Primary goods are 'things which it is
supposed a rational man wants whatever else he wants ' , and include
such things as rights and liberties, opportunities and powers, income
and wealth. Regardless of a person's values, plans, or ultimate aims,
it is assumed there are certain things of which he would prefer more
rather than less, on the grounds that they are likely to be useful in
advancing all ends, whatever ends they happen to be. So while the
parties to the original position are ignorant of their particular ends,
they are all assumed to be motivated by the desire for certain
primary goods.
The precise content of the list of primary goods is given by what
Rawls calls the thin theory of the good . It is thin in the sense that it
incorporates minimal and widely shared assumptions about the
kinds of things likely to be useful to all particular conceptions of the
good, and therefore likely to be shared by persons whatever their
more specific desires. The thin theory of the good is distinguished
from the full theory of the good in that the thin theory can provide no
basis for j udging or choosing between various particular values or
ends. So while the veil of ignorance provides that the parties deliber­
ate in conditions of fairness and unanimity, the account of primary
goods generates the minimal motivations necessary to get a problem
of rational choice going, and to make possible a determinate solu­
tion. Together, the two assumptions assure that the parties act only
on those interests that are common interests, that is, common to all
rational persons , the foremost of which turns out to be an interest in
establishing terms of social co-operation such that each person will
have the fullest liberty to realize his aims and purposes compatible
with an equal liberty for others .
The principles ofj ustice emerge from the original position in what
can be seen as a three-stage procedure. First comes the thin theory of
the good embodied in the description of the initial choice situation.

25
Justice and the moral subject

From the thin theory are derived the two principles ofjustice, which
define, in turn, the concept of good and provide an interpretation of
such values as the good of the community. lt is important to note that
although the thin theory of the good is prior to the theory of right and
the principles of j ustice, it is not substantial enough a theory to
undermine the priority of the right over the good that gives the
conception its deontological character. The priority of right on
which the theory depends is with respect to the full theory of the good
- the one having to do with particular values and ends - and the full
theory of the good only appears after the principles ofj ustice and in
the light of them . As Rawls explains,
To establish [the principles of right] it is necessary to rely on some notion of
goodness, for we need assumptions about the parties' motives in the original
position. Since these assumptions must not jeopardize the prior place of the
concept of right, the theory of the good used in arguing for the principles of
j ustice is restricted to the bare essentials . This account of the good I call the
thin theory: its purpose is to secure the premise about primary goods
required to arrive at the principles ofj ustice. Once this theory is worked out
and the primary goods accounted for, we are free to use the principles of
j ustice in the further development of what I shall call the full theory of the
good (396) .
This three-stage procedure seems to meet Rawls' deontological
requirements in the following way: The priority of the right over the
( full theory of the) good satisfies the requirement that the standard of
appraisal be prior to and distinguishable from the obj ects of apprais­
al, uncompromised by implication in existing wants and desires .
And the fact that the principles of right come not from nowhere but
from a thin theory of the good related to actual (if very general)
human desires gives the principles a determinate ground and pre­
vents their being arbitrary and detached from the world . And so,
without recourse to transcendental deductions , it seems possible to
find an Archimedean point that is neither radically situated nor
radically disembodied, neither 'at the mercy of existing wants and
interests' nor dependent on a priori considerations .
The essential point is that despite the individualistic features of j ustice as
fairness, the two principles ofjustice are not contingent upon existing desires or present
social conditions. Thus we are able to derive a conception of a j ust basic
structure, and an ideal of the person compatible with it, that can serve as a
standard for appraising institutions and for guiding the overall direction of
The original position

s ocial change. I n order to find an Archimedean point it is not necessary to

appeal to a priori or perfectionist principles. By assuming certain general desires,
su ch as the desire for primary social goods , and by taking as a basis the
agreements that would be made in a suitably defined situation , we can
achieve the requisite independence from existing circumstances [emphasis
added] ( 2 63 ) .

This, i n brief, i s the procedure by which the two principles of

j ustice are derived . As Rawls notes , j ustice as fairness, like other
contract views, consists of two parts, the first being the interpretation
of the initial situation and the problem of choice posed there, the
second involving the two principles of j ustice which, it is argued,

some variant thereof) , but not the other, and conversely' ( 1 5 ) . Even
would be agreed to. 'One may accept the first part of the theory (or

short of taking up the actual principles of j ustice Rawls believes

would be chosen, it is possible to identify two kinds of objections that
would be likely to arise on the move from the first half of the theory to
the second .
One set of obj ections would question whether the original position
achieves genuine detachment from existing wants and desires . This
sort of obj ection would be likely to fix on the account of primary
goods or some other aspect of the thin theory of the good and argue
that it is biased in favor of particular conceptions of the good and
against others . It might contest Rawls' claim that the list of primary
goods really is equally or nearly equally valuable to all ways oflife. I t
might question the thinness o f the thin theory of the good, claiming
that it undermines the fairness of the initial situation, that it intro­
duces assumptions not universally shared , that it is implicated too
deeply in the contingent preferences of, say, Western liberal
bourgeois life plans, and that the resulting principles are the product
of prevailing values after all.
A second set of objections would argue, on the other hand, that the
original position achieves too much detachment from human cir­
cumstances, that the initial situation it describes is too abstract to
yield the principles Rawls says it would , or for that matter, any
determinate principles at all. Such an objection would most likely
take issue with the veil of ignorance on the grounds that it excludes
morally relevant information, information necessary to generate any
meaningful results . It would argue that the notion of the person
embedded in the original position is too formal and abstract, too

27
justice and the moral subject

detached from contingency to account for the requisite motivations .

Where the first obj ection complains that the thin theory of the good is
too thick to be fair, the second contends that the veil of ignorance is
too opaque to yield a determinate solution .
I shall not pursue either of these obj ections here. Given our
concern with the deontological proj ect as a whole, our interest in the
original position is a more general one. Simply stated , it is this : If the
original position is Rawls ' answer to Kant, is it a satisfactory answer?
Does it succeed in its aspiration to reformulate Kantian moral and
political claims ' within the scope of an empirical theory'? Can it
provide a foundation for deontological liberalism while avoiding the
metaphysically contentious 'surroundings' of Kant's theory? More
specifically , can the description of the original position accommo­
date and support the claim for the primacy of j ustice in the strong
sense Rawls seeks to advance?
On one reading of the original position , a straightforward empiri­
cist interpretation that Rawls himself invites , it cannot support the
deontological claim . To see why this is so, we must examine its
conditions more closely, in part with a view to understanding the
kind of claim these conditions embody. We will not be concerned ,
therefore, with the q uestion of whether the thin theory of the good is
too thick or too thin to generate the principles ofj u stice Rawls says it
does, but rather with the question of what m akes the theory of the
good thick or thin, and how this account fits with the account of what
m akes j ustice primary . But perhaps we had best get down to examin­
ing the conditions that characterize the original position as Rawls
describes them . For this we must turn to the circumstances ofj ustice .

T H E C I R C U M S T A N C E S OF J USTIC E : E M P I RICIST O BJ E CTIONS

The circumstances of j ustice are the conditions that engage the

virtue of j ustice. They are the conditions that prevail in human
societies and make human co-operation both possible and necessary .
Society is seen as a co-operative venture for mutual advantage,
which means that it is typically marked by a conflict as well as an
identity of interests - an identity of interests in that all stand to gain
from mutual co-operation, a conflict in that, given their divergent
interests and ends, people differ over how the fruits of their co­
operation are to be distribu ted . Principles are needed to specify
Empiricist objections

arrangements by which such claims can be sorted out, and it is the

role of j ustice to provide them . The background conditions that
make such sorting-out arrangements necessary are the circum­
stances of j ustice.
Following Hume, Rawls notes that these circumstances are of two
kinds - obj ective and subj ective. The objective circumstances of
justice include such facts as the moderate scarcity of resources,
whereas the subjective circumstances concern the subj ects of co­
operation, most notably the fact that they are characterized by
different interests and ends. This means that each person has a
distinctive life plan, or conception of the good, which he regards as
worthy of advancement. Rawls emphasizes this aspect by assuming
that, as conceived in the original position at least, the parties are
mutually disinterested, that they are concerned to advance their own
conception of the good and no one else's, and that in advancing their
ends they are not bound to each other by prior moral ties . The
circumstances of j ustice are thus summarized :
One can say, in brief, that the circumstances of j ustice obtain whenever
mutually disinterested persons put forward conflicting claims to the divi­
sion of social advantages under conditions of moderate scarcity. U nless
these circumstances existed there would be no occasion for the virtue of

be no occasion for physical courage ( 1 28) .

j ustice, j us t as in the absence of threats of inj u ry to life and limb there would

The circumstances ofjustice are the circumstances that give rise to

the virtue ofj ustice. In their absence, the virtue ofj ustice would be
nugatory; it would not be required nor for that matter even possible.
'But a human society is characterized by the circumstances ofj ustice'
[emphasis added] ( I 29-30) . Therefore the virtue of j ustice is re­
quired .
The conditions that occasion the virtue of j ustice are empirical
conditions . About this Rawls is clear and unabashed . 'Moral phi­
losophy must be free to use contingent assumptions and general facts
as it pleases ' . It can proceed in no other way. What matters is that
the premises be 'true and sufficiently general' (5 I , I 58) .
The fundamental principles of j ustice quite properly depend upon the
natural facts about men in society. This dependence is made explicit by the
description of the original position : the decision of the parties is taken in
the light of general knowledge. Moreover, the various elements of the
Justice and the moral subject

original position presuppose many things about the circumstances of hu­

man life . . . . I f these assumptions are true and suitably general, everything

and empty ( 1 59, 1 6o ) .

is in order, for without these elements the whole scheme would be pointless

But an empiricist understanding of the original position seems

deeply at odds with deontological claims . For if j ustice depends for
its virtue on certain empirical preconditions, it is unclear how its
priority could unconditionally be affirmed . Rawls says that he bor­
rows his account of the circumstances ofj ustice from Hume ( 1 26-8) .
But Hume's circumstances cannot support the priority of right in the
deontological sense. They are after all empirical conditions. To
establish the primacy ofj ustice in the categorical sense Rawls' claim
requires, he would have to show not only that the circumstances of
j ustice prevail in all societies, but that they prevail to such an extent
that the virtue ofj ustice is always more fully or extensively engaged
than any other virtue. Otherwise, he would be entitled to conclude
only that j ustice is the first virtue of certain kinds of societies , namely
those where conditions are such that the resolution of conflicting
claims among mutually disinterested parties is the most pressing
social priority.
To be sure, a sociologist might argue, for example, that given the
increasing scarcity of energy and other basic resources facing mod­
ern advanced industrial societies, combined with the breakdown of
consensus and the loss of common purposes ( the obj ective and
subjective circumstances respectively) , the circumstances ofj ustice
have come to prevail with such intensity that j ustice has become, for
these societies , the first virtue. But if Rawls means the primacy of
j ustice to depend on a generalization such as this, he would need at
least to provide the relevant sociological support . Just asserting that
'a human society is characterized by the circumstances of j ustice'
would not be enough [emphasis added] ( 1 2 9 -30) .
The notion that the primacy of j ustice could be grounded empir­
ically becomes all the more implausible when we consider how
unlikely the necessary generalization must be, at least when applied
across the range of social institutions . For while we can easily enough
imagine that certain large-scale associations such as the modern
nation-state might meet its req uirements in many cases , we can
readily imagine a range of more intimate or solidaristic associations
in which the values and aims of the participants coincide closely
Empiricist object ions

enough that the circumstances ofj u stice prevail to a relatively small

degree. As H ume himself observes , we need not have recourse to
uto pian visions or the fiction of poets to imagine such conditions, but
' may discover the same truth by common experience and observa­
tio n ' ( I 7 39 : 495 )
·

I n the present disposition of the human heart, it would , perhaps, be difficult

to find complete ins tances of such enlarged affections; but still we may
observe that the case of families approaches towards it; and the stronger the
mu tual benevolence is among the individuals, the nearer it approaches; till
all distinction of property be, in a great measure, los t and confounded
among them . Between married persons, the cemen t of friendship is by the
laws supposed so strong as to abolish all division of possessions; and has
often, in reality, the force ascribed to i t ( 1 7 7 7 : 1 7- 1 8) .

While the institution of the family may represent an extreme case

in this respect, we can easily imagine a range of intermediate cases of
social institutions, a continuum ofhuman associations characterized
in varying degrees by the circumstances of j ustice. These would
include, at various points along the spectrum, tribes, neighbour­
hoods , cities, towns, universities,trade unions, national liberation mo ­
vements and established nationalisms, and a wide variety of ethnic,
religious, cultural, and linguistic communities with more or less
clearly-defined common identities and shared purposes , precisely
those attributes whose presence signifies the relative absence of the
circumstances of j ustice. Although the circumstances of j ustice
might well exist in all of these cases , they would not likely predomi­
nate, at least not to such an extent that j ustice was engaged in all
cases in greater measure than any other virtue . On the empiricist
interpretation of the original position , j ustice can be primary only for
those societies beset by sufficient discord to make the accommoda­
tion of conflicting interests and aims the overriding moral and
political consideration; j ustice is the first virtue of social institutions
not absolutely, as truth is to theories, but only conditionally, as
physical courage is to a war zone.
But this formulation suggests a further sense in which the primacy
of j ustice is undermined by the empiricist account of the circum­
stances ofj ustice. It fixes on the sense in which j ustice appears as a
remedial virtue, whose moral advantage consists in the repair it
works on fallen conditions . But if the virtue ofj ustice is measured by
Justice and the moral subject

the morally diminished conditions that are its prerequisite, then the
absence of these conditions - however this state of affairs might be
described - must embody a rival virtue of at least commensurate
priority, the one that is engaged in so far as j ustice is not engaged . If
physinl courage is a virtue only in the face of inj urious conditions,
then the peace and tranquillity that would deny courage its occasion
must surely be virtues of at least equivalent status. And so it is with
j ustice. As Hume's account confirms, the remedial character of
j ustice entails another set of virtues of at least a comparable order.

Justice takes it rise from human conventions . . . and these are intended as a
remedy to some inconveniences , which proceed from the concurrence of
certain qualities of the human mind with the situation of external obj ects .
The qualities of the mind are selfishness and limited generosity; and the
situation of external obj ects is their easy change, join'd to their scarcity in
comparison of the wants and desires of men . . . . Encrease to a sufficient
degree the benevolence of men, or the bounty of nature, and you render
justice useless, by supplying its place with much nobler virtues, and more
favourable blessings ( I 739: 494-5) .

To invoke the circumstances of j ustice is simultaneously to con­

cede, implicitly at least, the circumstances of benevolence, or
fraternity, or of enlarged affections, whatever the description might
be; such are the circumstances that prevail in so far as the circum­
stances of justice do not prevail, and the virtue to which they give
definition must be a virtue of at least correlative status.
One consequence of the remedial aspect of j ustice is that we
cannot say in advance whether, in any particular instance, an in­
crease in justice is associated with an overall moral improvement.
This is because a gain injustice can come about in one of two ways; it
can arise where before there was inj ustice, or it can occur where
before there was neither j ustice nor inj ustice but a sufficient measure
of benevolence or fraternity such that the virtue of j ustice had not
been extensively engaged . Where j ustice replaces inj ustice, other
things being equal, the overall moral improvement is clear. On the
other hand, where an increase in j ustice reflects some transformation
in the quality of pre-existing motivations and dispositions, the over­
all moral balance might well be diminished .
When fraternity fades, more j ustice may be done, but even more
may be required to restore the moral status quo. Furthermore, there
Empiricist objections

i s no guarantee that j u s tice and its rival virtues are perfectly com­
m ens urable. The breakdown of certain personal and civic attach­
m e nts may represent a moral loss that even a full measure ofj ustice
ca n not redeem . Does it go without saying that a rent in the fabric of
i m p licit understandings and commitments is fully morally repaired
so lon g as everyone 'does what he ought' in the aftermath?
Consider for example a more or less ideal family situation , where
relations are governed in large part by spontaneous affection and
where, in consequence, the circumstances of j ustice prevail to a
relatively small degree . I ndividual rights and fair decision proce­
dures are seldom invoked , not because injustice is rampant but
because their appeal is pre-empted by a spirit of generosity in which
I am rarely inclined to claim my fair share. Nor does this generosity
necessarily imply that I receive out of kindness a share that is equal
to or greater than the share I would be entitled to under fair
principles ofj ustice. I may get less . The point is not that I get what I
would otherwise get, only more spontaneously, but simply that the
questions of what I get and what I am due do not loom large in the
overall context of this way of life .
Now imagine that o n e day t h e harmonious family comes t o be
wrought with dissension . Interests grow divergent and the circum­
s tan ces ofj us tice grow more acute. The affection and s pontaneity of
previous days give way to demands for fairness and the observance of
rights . And let us further imagine that the old generosity is replaced
by a j udicious temper of unexceptionable integrity and that the new
moral necessities are met with a full measure ofjustice, such that no
i nj ustice prevails. Parents and children reflectively equilibriate,
dutifully if sullenly abide by the two principles of j ustice, and even
manage to achieve the conditions of stability and congruence so that
the good ofj ustice is realized within their household . Now what are
we to make of this? Are we prepared to say that the arrival ofjustice,
however full, restores to the situation its full moral character, and
that the only difference is a psychological one? Or consider again the
parallel of physical courage. I magine a society once tranquil but
with little courage ( not out of cowardice but quietude) , now turned
violent and precarious, but where the virtue of courage is on bold,
even plentiful display. Is it obvious we would prefer the second from
a moral point of view?
To be sure, the incommensurabilities , if they exist, could pull in

33
justice and the moral subject

the opposite direction as well. I t may be that despite the harshness of

the circumstances of courage, there is a certain nobility that
flourishes in the new way of life unavailable to the human spirit
under more protected conditions and that this goes uncompensated
by even the most blissful peace . And if the demise of familial or
communal Gemeinschaft reflects not the onset of material meanness
but the flowering of diversity, or the children outgrowing the paro­
chial ways of their parents ' home, we might be inclined to view the
advent of j ustice in a more favorable light. The general point re­
mains . An increase in j ustice can fail to be associated with an overall
moral improvement in at least two different ways : either by failing
fully to meet an increase in the circumstances of j ustice, or by an
inability, however full, to compensate the loss of certain 'nobler
virtues , and more favourable blessings ' .
If a n increase i n justice does not necessarily imply a n unqualified
moral improvement, it can also be shown that in some cases, j ustice
is not a virtue but a vice. This can be seen by considering what we
might call the reflexive dimension of the circumstances of j ustice.
The reflexive dimension refers to the fact that what the parties know
about their condition is an ingredient of their condition. Rawls
acknowledges this feature when he writes, 'I shall, of course, assume

stances of j ustice obtain' ( 1 28) .

that the persons in the original position know that these circum­

The circumstances ofj ustice, and more specifically the subj ective
aspect of these circumstances , consist partly in the motivations of the
participants and in the way they perceive their motivations . If the
parties one day came to regard their circumstances differently, if
they came to believe that the circumstances ofj ustice (or of benevo­
lence) obtained to a greater or lesser extent than before, this very
shift would amount to a change in those circumstances . As Rawls
points out in his discussion of the good ofjustice, acting out of a sense
ofj ustice can be contagious ; it reinforces the assumptions it presup­
poses and enhances its own stability by encouraging and affirming
like motivations in others .
But what is the effect of this 'contagion' when it is applied to a
situation where, or in so far as, the circumstances of j ustice do not
obtain? When I act out of a sense of j ustice in inappropriate cir­
cumstances , say in circumstances where the virtues of benevolence
and fraternity rather than j ustice are relevantly engaged , my act may

34
Empiricist objections

not merely be superfluous, but might contribute to a reorientation of

prevailing understandings and motivations, thereby transforming
the circumstances of j ustice in some degree. And this can be true
even where the 'act' I perform out ofj ustice is ' the same act' as the
one I would have performed out of benevolence or fraternity, except
in a different spirit. As in Rawls' account of stability, my act and the
sense ofjustice that informs it have the self-fulfilling effect ofbringing
about the conditions under which they would have been appropriate.
But in the case of the inappropriate act of j ustice, the result is to
render the circumstances ofj ustice more pressing without necessari­
ly evoking an increase in the incidence ofj ustice to a similar degree.
Gratuitous displays of physical courage in the midst of tranquil
conditions can prove disruptive of the very tranquility they fail to
appreciate and quite possibly can fail to replace. It is similar with
j ustice. If, out of a misplaced sense of j ustice, a close friend of
long-standing repeatedly insists on calculating and paying his pre­
cise share of every common expenditure, or refuses to accept any
favor or hospitality except at the greatest protest and embarrass­
ment, not only will I feel compelled to be reciprocally scrupulous but
at some point may begin to wonder whether I have not misunder­
stood our relationship. The circumstances of benevolence will to this
extent have diminished, and the circumstances of j ustice grown .
This follows as a consequence of the reflexive dimension of the
(subjective aspect of the) circumstances of j ustice. But as we have
already seen, there is no guarantee that the new sense ofjustice can
fully replace the old spontaneity, even in those cases where no
inj ustice results . Since the exercise ofj ustice in inappropriate condi­
tions will have brought about an overall decline in the moral charac­
ter of the association, j ustice in this case will have been not a virtue
but a vice.

So the circumstances ofj ustice fit badly with the primacy ofj ustice
and the related deontological themes Rawls seeks to defend . Given
the contrasting philosophical pedigrees of the two accounts, it is little
wonder that the inconsistencies arise. Where the circumstances of
j ustice are explicitly Humean - 'Hume's account of them is especial­

his much fuller discussion' ( 1 2 7-8) - the deontological conception at

ly perspicuous and the preceding summary adds nothing essential to

the heart of Rawls' theory finds its primary formulation in Kant,

35
justice and the moral subject

whose epistemology and ethics were directed in large part against

the very empiricist and utilitarian tradition that Hume represents.
For Kant, the deontologically-given notion of right which Rawls
seeks to recapture derives its force from a moral metaphysic that
rules out precisely the appeal to contingent human circumstances on
which Hume's account of the virtue of j ustice is based .
For Hume, j ustice is the product of human conventions and
'derives its existence entirely from its necessary use to the intercourse
and social state of mankind' .
Thus, the rules o f equity o r justice depend entirely o n the particular state
and condition in which men are placed , and owe their origins and existence
to that utility, which results to the public from their strict and regular
observance. Reverse, in any considerable circumstance, the condition of
men: Produce extreme abundance or extreme necessity: Implant in the
human breast perfect moderation and humanity, or perfect rapaciousness
and malice: By rendering j ustice totally useless, you thereby totally destroy
its essence, and suspend its obligation upon mankind ( I 7 7 7 : 2 0 ) .

For Kant, by contrast,

Empirical principles are always unfitted to serve as a ground for moral laws .
The universality with which these laws should hold for all rational beings
without exception - the unconditioned practical necessity which they thus
impose - falls away if their basis is taken from the special constitution ofhuman
nature or from the accidental circumstances in which it is placed ( I 785: I og) .

If, as it appears, a Humean account of the circumstances ofj ustice

can neither support nor accommodate the privileged status ofj ustice
and right required by Rawls and derived from Kant, the question
naturally arises why Rawls does not adopt instead a Kantian
account of the circumstances of j ustice. The answer is that Kant,
strictly speaking, has none, at least none that situates the virtue of
j ustice in circumstances characteristic of human society. Nor is it
obvious that he could provide one. To do so would be to contradict
the essential point of Kantian ethics - that man acts morally only in
so far as he is able to rise above the heteronomous influences and
contingent determinations of his natural and social conditions and
act according to a principle given by pure practical reason. For
Kant, the circumstances ofj ustice do not consist in those conditions
of human society that make j ustice necessary, but rather in an ideal
realm abstracted from human society that makes j ustice, and moral-
Empiricist objections

ity in general, a possibility. This realm is the kingdom of ends. I t is a

realm beyond the phenomenal world - as Kant acknowledges, 'it is
ce rtainly only an ideal' - where human beings are admitted not as
p ermanent residents but more likely as fleeting visitors . Admission
to these circumstances ofj ustice is not a precondition of moral virtue
but a measure of its achievement, a place where human beings arrive
only in so far as they are able to act out of conformity to the
autonomously-given moral law, in so far, that is, as they can abstract
from their situation to will and act as unsituated beings, from a
universal point of view. This is why the categorical imperative can
enjoin only that man act as though he were a legislating member of the
kingdom of ends.
We shall be able - if we abstract from the personal differences between
ra tional beings, an d also from all the content of their p riv a te ends - to
conceive a whole of all ends in systematic conj unction . . . that is, we shall be
able to conceive a kingdom of ends which is possible in accordance with the
above principles (Kant I 785: 1 00- I ) .

Now a kingdom of ends would actually come into existence through maxims
which the categorical imperative prescribes as a rule for all rational beings,
if these maxims were universal!J followed. Yet even if a rational being were
himself to follow such a maxim strictly, he cannot coun t on everybody else
being fai thful to it on this ground . . . . But in spite of this the law 'Act on the
maxims of a member who makes universal laws for a merely possible
kingdom of ends' remains in full force, since its command is categorical
( Kant I 785: 1 06) . ·

As we have already seen, Rawls parts company with Kant where

ideal realms and transcendental subjects enter the conception, and
the account of the kingdom of ends is one such point. Rawls finds
such a notion unsatisfactory as a basis for human justice, since it
seems to apply to human beings only in so far as they are detached
from actual human circumstances , which is to say only in so far as
they cease to be human beings . Such a notion is at least obscure, and
may fall subject to the strictures against a priori standards of
appraisal and radically disembodied conceptions of the self. Rawls
expresses these worries, especially concerning the problem of arbi­
trariness , by suggesting with Sidgwick that Kant's doctrine requir­
ing abstraction from all contingency may be unable to distinguish
between the lives of the saint and the scoundrel, as long as both are

37
justice and the moral subject

lived by a consistent set of principles freely chosen and conscien­

tiously acted upon. The choice of the noumenal self may - in fact
might necessarily - be arbitrary in this sense. 'Kant did not show
that acting from the moral law expresses our nature in identifiable
ways that acting from contrary principles does not' ( 2 5 5 ) . This
criticism reflects Rawls' more general difference with Kant on the
role of the empirical and the a priori in moral theory, in particular
Rawls' view that ' the analysis of moral concepts and the a priori,
however traditionally understood , is too slender a basis' for a sub­
stantive theory of j ustice. 'Moral philosophy must be free to use
contingent assumptions and general facts as it pleases' ( 5 1 ) .
To overcome these difficulties , while at the same time preserving
the priority of right, Rawls seeks to reformulate the notion of the
kingdom of ends in a way that accommodates an empirical account
of the circumstances of j ustice but rules out the contingent differ­
ences between persons that would otherwise prevail there.
The description of the original position interprets the point of view of
noumenal selves , of what it means to be a free and equal rational being. Our
nature as such beings is displayed when we act from the principles we would
choose when this nature is reflected in the conditions determining the
choice . Thus men exhibit their freedom, their independence from the
contingencies of nature and society, by acting in ways they would acknow­
ledge in the original position ( 255-6) .

Like the kingdom of ends, the original position, with the veil of
ignorance, has the effect of ' abstract [ing] from the personal differ­
ences between rational beings , and also from all the content of their
private ends ' . But unlike the Kantian version , it has the purported
advantage of applying to actual human beings subj ect to the ordin­
ary conditions of human circumstance.
The original position may be viewed , then, as a procedural interpretation of
Kant's conception of autonomy and the categorical imperative. The princi­
ples regulative of the kingdom of ends are those that would be chosen in this
position, and the description of this situation enables us to explain the sense
in which acting from these principles expresses our nature as free and equal
rational persons. No longer are these notions purely transcendent and lacking explic­
able connections with human conduct, for the procedural conception of the original
position allows us to make these ties [emphasis added] ( 2 56) .

The aim of the original position is to provide a means of deriving

Empiricist objections

p rinciples of justice that abstracts from contingent and therefore

morally irrelevant social and natural influences - this is the Kantian
as p iration - without having to rely on a noumenal realm or on the
notion of a transcendent subject wholly beyond experience. Rawls'
solution is to restrict the description of the p arties in the original
p osition to those characteristics which all human beings share as free
and equal rational beings . Roughly speaking, these are that each is a
being who chooses his ends and who values certain p rimary goods as
instrumental to their realization, whatever those ends might be.
These features are assumed to be common to all human beings as
such, and are in this sense non-contingent.

Thus, given human nature, wanting them [primary goods] is part o f being
rational . . . . The preference for primary goods is derived , then, from only
the most general assumptions about rationality and the conditions of
human life . To act from the principles of justice is to act from categorical
imperatives in the sense that they apply to us whatever in particular our
aims are. This simply reflects the fact that no such contingencies appear as
premises in their derivation (253 ) .

Rawls acknowledges that notwithstanding its Kantian affinities ,

the original position departs from Kant's views in several respects
( 2 5 6) . Among them mus t surely be the reliance of Rawls on certain
generalized human preferences or desires for the derivation of the
principles ofj ustice. For Kant, to found the moral law on generalized
preferences and desires , however widespread across human beings ,
would be merely to substitute a wider heteronomy for a narrower one
( 1 7 88 2 5 -8) . It would not be to escape from contingency in his more
exacting sense, which applies to the constitution of human nature as
well as to the constitution of particular human beings . Even the ' thin
theory of the good ' would be too thick to satisfy the Kantian
conception of autonomy.
For Rawls, this more global contingency is not a problem . His
concern is to develop a theory ofj ustice that is fair between persons ,
and so only those contingencies that differentiate persons from each
other need be ruled out . Con tingent attri butes common to human
beings as such are not only not a problem for Rawls , but are essen tial
ingredients of his moral theory. ' Moral philosophy mus t be free to
use con tingent assumptions and general facts as it pleases . ' Among
these general facts are the facts of the circumstances of j ustice.

39
Justice and the moral subject

Although Kant's view does not admit them, Rawls' theory relies on
them; these facts assure that the principles of justice the theory
produces apply to human beings in the real world rather than to
disembodied or transcendent beings beyond the world .

Being in the circumstances ofj ustice [the parties] are situated in the world
with other men who likewise face limitations of moderate scarcity and
competing claims . Human freedom is to be regulated by principles chosen
in the light of these natural restrictions. Thus j ustice as fairness is a theory of
human j ustice and among its premises are the elementary facts about
persons and their place in nature ( 2 5 7 ) .

We can see, then, why Rawls cannot simply adopt a Kantian

account of the circumstances of justice to go conveniently with his
other Kantian positions, why he is pressured instead to resort to the
notion of an original position which includes as part of its description
an empirical account of characteristic human circumstances . It is
this uneasy combination that gives rise to the objections we have
considered . As a Kantian conception of the moral law and the
kingdom of ends seems to deny j ustice its human situation, the
Humean account of the human situation seems unable to accommo­
date strong claims on behalf of the primacy of j ustice. But under­
standing how the inconsistencies arise is not to dissolve them but if
anything to confirm them . And so it would appear that the two
aspirations of Rawls' theory, to avoid both the contingency of ex­
isting desires and the alleged arbitrariness and obscurity of the
transcendent, are uncombinable after all, the Archimedean point
wiped out in a litany of contradictions .

T H E C I R C U M S T A N C E S OF J U STI C E : D E O N T O L O G I C A L R EJ O I N DE R

To all of this, Rawls might make the following reply: The apparent
incompatibilities between the primacy of j ustice and the circum­
stances of j ustice are based on a misunderstanding of the original
position and the role it plays in the conception as a whole. The
obj ections are too quick. They fail to appreciate that the account of
the circumstances ofj ustice is an account within the account of the
original position, which, it must be recalled, is hypothetical to begin
with. The conditions described there are meant to be the conditions
in which the parties to the original position carry out their delibera-
Deontological rejoinder

tions, not the actual conditions in which ordinary human beings live
their lives .
The obj ections depend in large part on the mistaken assumption
that the facts of the circumstances ofj ustice are meant to be facts of
life in the real, phenomenal world in which the principles ofj ustice
would actually apply, and that their validity therefore depends on
the same empirical considerations on which all ordinary factual
claims depend . But the description of the circumstances of j ustice
cannot be regarded as a straightforward empirical generalization , to
be established or refuted by the best evidence of sociology, psycholo­
gy, and so on. Since the entire account of the circumstances ofj ustice
is located within the account of the original position, the conditions
and motivations it describes are asserted only of the parties to the
original position and not necessarily of real human beings . Once
installed as a premise of the original position , the account of the
circumstances ofjustice ceases to work as a simple empirical account
which can be checked for accuracy against actual human conditions .
I ts validity depends instead on the extent to which the conception of
which it is a part yields principles ofj ustice that successfully capture
our settled convictions in reflective equilibrium . ' I t seems best to
regard these conditions simply as reasonable stipulations to be
assessed eventually by the whole theory to which they belong . . . .
Justification rests upon the entire conception and how it fits in with
and organizes our considered j udgments in reflective equilibrium'
( 5 78-g) .
The description of the circumstances of j ustice, then, does not
need to be true in any literal, empirical sense. The original position
that contains it is in any case an admitted fiction , a heuristic device
designed to constrain our reasoning about j ustice in certain ways .
The distinction between provisions of the circumstances of j ustice
and the motivations that actually prevail in human societies is for
Rawls a constant theme; 'We must keep in mind that the parties to
the original position are theoretically defined individuals' ( 1 47 ) .

involves no particular theory of human motivations ' ( 1 30) . 'The

'The account of these conditions [i.e. the circumstances of j ustice]

motivation of the persons in the original position must not be

confused with the motivation of persons in everyday life who accept
the principles that would be chosen and who have the corresponding
sense of j ustice' ( 1 48) .
Justice and the moral subject

Rawls emphasizes in particular that the assumption of mutual

disinterest and the absence ofprior moral ties in the original position
does not im ply a j udgment that people really are mutually disin­
terested or without moral ties .

We need not suppose of course that persons never make substantial sac­
rifices for one another, since moved by affection and ties of sentiment they
often do ( 1 78) .

There is no inconsistency in supposing that once the veil of ignorance is

removed , the parties find that they have ties of sentiment and affection, and
want to advance the interests of others and to see their ends attained ( 1 29) .

Although the assumption of mutual disinterest and the presumed

preference for primary goods are the main motivational premises of
the original position, this implies neither that these motivations hold
for persons in real life, nor that they would hold for persons living in a
well-ordered society governed by the two principles of justice.
As for the scope of the motivation assumption , one must keep in mind that it
holds oni:J for the parties in the original position; they are to deliberate as if they
prefer more rather than less primary goods . . . . The assumption may not
characterize the general motivation ofpeople in sociery, however, and in particular it may
failfor the citizens ofa well-ordered sociery (a society effectively regulated by the
public principles adopted in the original position) [emphasis added]
( Rawls 1 9 75: 543-4) .

So Rawls might reply, and to considerable advantage. To re­

nounce the straightforward empiricist reading of the circumstances
of justice is to rescue the claim for the primacy of justice from the
more obvious empiricist obj ections at least. It would also go some
way toward making sense of such enigmatic locutions as 'a human
society is characterized by the circumstances of j ustice' [emphasis
added] ( 1 2 9 -30 ) , which in the context of Rawls' account seems more
than a merely empirical generalization and yet something short of a
stipulative definition . But the rej ection of the empiricist interpreta­
tion raises a more difficult question : if the descriptive premises of the
original position are not subj ect in any straightforward sense to
empirical tests, to what sort of test are they subj ect? If the constraints
on the motivational assumptions are not empirical constraints, then
what kinds of constraints are they?
All we have been able to say so far about the grounds for a premise
Deontological rejoinder

of the original position is that its validity depends on whether, or to

what extent, the conception of which it is a part yields principles of
j ustice that successfully capture our considered j udgments in reflec­
tive equilibrium. But this does not tell us enough. For what keeps the
method of reflective equilibrium from being circular is the availabil­
ity of independent criteria of j udgment at each end , however pro­
visional, in the light of which we adj ust and correct the other. I n the
case of j ustice, this means that we must have some (independent if
p rovisional) way of judging both the desirability of the principles of
justice a particular description may yield and the plausibility or
reasonableness of the motivational assumptions that generate them.
( ' Each of the presumptions should by itselfbe natural and plausible'
( 1 8) . ) The independent yet provisional criteria on the side of the
desirability of the principles are given by our intuitions about what is
just. But what is the corresponding ground on the descriptive side?
What we are looking for is that with reference to which the plausibility of
the premises of the original position might be assessed . One is
tempted to say, in line with the normative side, that the criteria of
p lausibility are given by our 'intuitions' about what is empirically
true. But as we have found , the empiricist temptation to think that
the actual conditions and motivations of human beings provide the
standard of plausibility leads to unacceptable consequences .
To put our problem another way: as the account of reflective
equilibrium makes clear, the conditions of the original position
cannot be so immune from actual human circumstance that j ust a'V'
assumptions producing attractive principles of j ustice would do.
Unless the premises of such principles bear some resemblance to the
condition of creatures discernibly human, the success of the equilib­
rium is, to that extent, undermined . I f we could match our convic­
tions about justice only by appealing to premises that struck us as
eccentric or outlandish or metaphysically extreme, we would rightly
be led to question the convictions those principles happened to fit.
This, after all, comes close to being Rawls' case against Kant, that he
was able to establish morally compelling conclusions only at the
expense of an account of moral circumstance that bore little resem­
blance to anything discernibly human .
To summarize: the validity o f a premise of the original position is
not given empirically, but by a method of j ustification known as
reflective equilibrium. This method involves two different kinds of

43
justice and the moral subject

j ustification coming together to provide mutual correction and sup­

port. One aspect of the justification appeals to our considered con­
victions about j ustice; the other appeals to a standard of descriptive
but not strictly empirical plausibility which we are searching to
define.
Rawls himself seems unclear on what he takes the descriptive
standard to be. Both in his general remarks onj �stification and in his
defense of specific premises of the original position, his language
reflects his equivocation, and is worth examining closely:
But how are we to decide what is the most favored interpretation [of the
initial situation] ? I assume, for one thing, that there is a broad measure of
agreement that principles ofjustice should be chosen under certain condi­
tions. To j ustify a particular description of the initial situation one shows
that it incorporates these commonry shared presumptions. One argues from
widery accepted hut weak premises to more specific conclusions . Each of the
presumptions should by itself be natural and plausible; some of them may
seem innocuous or even trivial [emphasis added] ( 1 8) .

I n searching for the most favored description of this situation we work from
both ends . We begin by describing it so that it represents generally shared and
preferably weak conditions. We then see if these conditions are strong enough to
yield a significant set of principles. If not, we look for further premises
equalry reasonable [emphasis added] ( 2o) .

In his specific defense of the assumption of mutual disinterest, Rawls

appeals to similar criteria:
The postulate of mutual disinterest in the original position is made to insure
that the principles ofj ustice do not depend upon strong assumptions. Recall
that the original position is meant to incorporate widely shared andyet weak
conditions. A conception ofj ustice should not presuppose, then, extensive ties
of natural sentiment. At the basis of the theory, one tries to assume as little as
possible [emphasis added] ( 1 29) .

I n arguing for mutual disinterest over benevolence as the

appropriate motivational premise, Rawls argues that the combined
assumptions of mutual disinterest and the veil of ignorance have 'the
merits of simplicity and clarity' while at the same time insuring the
beneficent aspects of seemingly more generous motivations . If it is
asked why not postulate benevolence with the veil of ignorance, 'the
answer is that there is no need for so strong a condition. Moreover, it
would defeat the purpose of grounding the theory ofjustice on weak

44
Deontological rejoinder

stipulations, as well as being incongruous with the circumstances of

justice' [emphasis added] ( 1 49) . Finally, Rawls states in his conclud­
ing remarks on j ustification, 'I have several times noted the minimal
nature of the conditions on principles when taken singly. For example,
the assumption of mutually disinterested motivation is not a demanding
stipulation. Not only does it enable us to base the theory upon a
reasonably precise notion of rational choice, but it asks little of the
parties' [emphasis added] ( 583) .
Rawls seems to offer two basic answers to the question of how,
from the descriptive point of view, an assumption of the original
position is to be j ustified, neither of which takes us very far. The first
answer is that it should be widely accepted and commonly shared ,
the second that it should be a weak as opposed to a strong assump­
tion, and if possible, natural, reasonable, innocuous and even trivial .
Yet it is unclear what these considerations amount to, or how in any
case they help us know whether to describe the parties as mutually
disinterested or benevolent.
First, it is unclear what about a motivational assumption should be
commonly shared or widely accepted , and why this should count in
its favor. Should we look for the motive that is most commonly shared
(in which case we would have to generalize about the motives people
have) ? Or the one that is most widely regarded to be the prevailing
motive (in which case we would have to generalize about the gener­
alizations people make about other people's motives ) ? Or the one
that is most widely agreed to be an appropriate condition on princi­
ples ofj ustice ( in which case we would have to generalize about how
people are likely to interpret the requirement of common agreement
that we are struggling to interpret) ? But these interpretations are
either empirical or question-begging or both, and in any case are not
obviously relevant to the validity of a premise such as m utual
disinterest or benevolence as a condition of the original position .
The requirement that the assumption be weak rather than strong
begs precisely the question we seek to answer: weak or strong with
respect to what? We might say an assumption is weak from a concep­
tual point of view, and therefore likely to be innocuous or trivial or
otherwise unobj ectionable, when it depends for its validity on the
validity of relatively few related propositions, and where those it does
rely on are themselves weak and uncontroversial . An assumption is
strong in these terms when, for it to be true, many other things ,

45
Justice and the moral subject

including controversial things, have also to be true. But surely the

assumptions of mutual disinterest and benevolence cannot be dis­
tinguished on the grounds that either is weaker or stronger an
assumption in the conceptual sense; neither relies on a premise
which is conceptually more contentious or problematic than the other.
I f neither is more nor less conceptually demanding than the other,
the alternative sense would seem to refer to statistical probabilities .
When welfare economists, for example, refer to motivational
assumptions as strong or weak, they mean to describe the likelihood
that the motivation applies to a large proportion of the population .
Much of Rawls' language seems to suggest this general probabilistic
usage . What else could he mean when he says that assuming the
parties to be mutually disinterested is assuming 'less' than to assume
that extensive ties of natural sentiment prevail? How does he know
that assuming mutual disinterest is not a demanding stipulation in
that 'it asks little of the parties' ? Does this assume that we are
inclined by nature toward selfishness rather than benevolence?
Maybe for some it is asking much more to ask that they act selfishly
rather than benevolently . Is it then a matter of statistical probabili­
ties which way people are naturally inclined to behave? And how
could such a question even be formulated precisely enough to get a
reasonable estimate without specifying the range of situations in­
volved? In any case, if all Rawls means by the distinction between
weak and strong assumptions is that the premises should be more
realistic rather than less realistic, then we are back to the merely
empirical understanding of the conditions in the original position
which has already been rejected .
Although Rawls' own account of the original position and the
status of its descriptive premises is unclear, some account of these
matters is essential if we are to make sense of his theory at all. U nless
we can overcome the apparent obstacles to explicating the original
position and its premises, the coherence of the entire conception will
be in doubt. What we need is an account of what exactly constrains
the descriptive assumptions appropriate to the initial situation, that
is, what constrains them besides the constraints imposed from the
normative side, which consist of our considered convictions about
j ustice. More generally, we need to know something more precise
about the status of the original position, phenomenal or otherwise;
what is the original position , anyhow?
In search of the moral subject

IN S E A R C H OF T H E M O R A L SUBJ E CT

I believe these questions can be answered , if not in Rawls' explicit

la nguage at least in terms consistent with his conception as a whole.
Fin ding the answers may therefore involve certain departures from
th e text in order to make sense of the text. The justification for this
in terpretation will be found in hints and traces of evidence scattered
throughout the text and , more importantly, in the sense it enables us
to make of Rawls' theory as a whole, and in particular its ability to
resolve certain problematic features of the original position we have
been unable to resolve in any other way .
But our point of departure is a point firmly within the text, in the
notion of reflective equilibrium as the method of j ustification that
governs the conception as a whole. The key is to see the original
position as the fulcrum of reflective equilibrium, in so far as it can be
achieved . The original position is the fulcrum of the j ustificatory
process in that it is the device through which all justification must
pass, the place at which all arguments must arrive and from which
they must depart. This is why a premise of the original position can
be defended or attacked from either of two directions, on grounds of
its plausibility (in a sense yet to be determined) or on grounds of its
fit with our considered convictions about j ustice.
In searching for the most favored description of this situation we work from
both ends . . . . By going back and forth, sometimes altering the conditions of
the contractual circumstances, at others withdrawing our j udgments and
c o n for m ing them to principle, I assume that eventually we shall find a

description of the initial situation that both expresses reasonable conditions

a n d yields principles which match our cons idered j udgments duly pruned
and adj usted . This state of affairs I refer to as reflective equilibrium (20) .

The description of the original position is the product of two basic

i ngredients : our best j udgments of 'reasonableness and plausibility'
( yet to be explicated ) from one side, and our considered convictions
of j u stice from the other. From the raw materials of our intuitions,
properly filtered and shaped by the original position, a final product
e merges . But it is a final product of dual dimensions, and in this lies
the key to our account. For what issues at one end in a theory of
justice must issue at the other in a theory of the person, or more
precisely, a theory of the moral subject. Looking from one direction
through the lens of the original position we see the two principles of

47
justice and the moral subject

j ustice; looking from the other direction we see a reflection of

ourselves. If the method of reflective equilibrium operates with the
symmetry Rawls ascribes to it, then the original position mus t
produce not only a moral theory but also a philosophical anthropology.
Throughout most of his book, Rawls is concerned primarily with
the former. His aim is to produce a theory ofj ustice, and so most of
his attention is devoted to the argument from the original position to
the principles ofj ustice, and to a description of the original position
which adequately anticipates the requirements of j ustice. He is
understandably less concerned to pursue the argument in the oppo­
site direction, and accordingly less explicit about what would be
found there. This may account in part for his greater clarity on the
basis of our moral intuitions than on the origin and status of our
'descriptive' intuitions (what makes them reasonable or unreason­
able, strong or weak, etc. ) . If this reconstruction is correct, then the
independent yet provisional standard by which the reasonableness
of our descriptive assumptions is assessed is given not by the laws of
empirical psychology or sociology but instead by the nature of the
moral subj ect as we understand it, which is to say by the constitutive
understanding we have of ourselves .
Given that the aim of the conception is to produce a theory of
j ustice, there is a tendency to dismiss the sometimes unattractive
motivational assumptions of the original position as part of a merely
heuristic device of no independent or continuing interest once the
principles of justice have been fully worked out. But if reflective
equilibrium truly works both ways, then the account of human
circumstance that emerges once reflective equilibrium is achieved
can no more be dispensed with as the incidental product of a fictive
contrivance than can the principles ofj ustice themselves . Given the
methodological symmetry of the original position, we cannot regard
one of its products as chaff to the other's wheat, to be chucked away
once the flour has been ground . We must be prepared to live with the
vision contained in the original position, mutual disinterest and all,
prepared to live with it in the sense of accepting its description as an
accurate reflection of human moral circumstance, consistent with
our understanding of ourselves .

Finally, we may remind ourselves that the hypothetical nature of the

original posi tion invites the q uestion: why should we take any interest in it,
In search of the moral subject

of this situation are ones that we do in fact accept. Or if we do not, then we can be
mor a l or otherwise? Recall the answer: the conditions embodied in the description

in troduced [emphasis added] (587 ) .

p ers u aded to do so by philosophical considerations of the sort occasionally

A s th e concepts o f pure speculative reason are t o Kant's moral

th eo ry, so the conditions embodied in the original position are to
Rawls ' theory ofj ustice. 'They are not like the props and buttresses

they are rather true members making the structure of the system
which usually have to be put behind a hastily erected building, but

plai n' ( Kant 1 788: 7 ) .

theory ofj ustice is a conception of the moral subject that both shapes
These considerations powerfully suggest that implicit i n Rawls'

the principles of j ustice and is shaped in their image through the

medium of the original position . It is this conception I propose to

help not only to resolve the perplexities concerning the status of the
illuminate and explore. lfit can somehow be made explicit, it should

original position, but also to assess the central claims of the concep­
tion as a whole. So where Rawls' main discussion tends to take the

po s ition to the principles ofj ustice, I propose to work in the opposite

nature of the moral subj ect as given and argue through the original

direction, to take the principles ofjustice as provisionally given and

argue back to the nature of the moral subject. In doing so, I take
myself to be tracing the lineaments of an argument of the following
kind : assuming we are beings capable ofj ustice, and more precisely,
beings for whom j ustice is primary, we must be creatures of a certain
kind , related to human circumstances in a certain way. What then
must be true of a subject for whom j ustice is the first virtue? And how
is the conception of such a subject embodied in the original position?
Now the description of this subject will have a distinctive logical

to an y particular experience - the 'must' in the formulation is not for

s tat us. It will in some sense be necessary, non-contingent, and prior

nothing - but it will not of course be an analytic claim . I t will in some

sense be empirical, but not 'merely' empirical . Given the reflexive
ch ara cter of such descriptions, they are not merely descriptive but
al so partly constitutive of the kind of beings we are. Our knowing
th em is part of what makes them true, and makes us the reflexive,
s elf-i nte rpreting creatures that we are.
O ne might describe a general account of these constituent features

49
justice and the moral subject

of our self-understanding by a variety of names : a theory of the

person , a conception of the self, a moral epistemology, a theory of
human nature, a theory of the moral subj ect, a philosophical anthro­
pology. These descriptions carry differing, sometimes conflicting
connotations , usually associated with the philosophical traditions
from which they derive . To speak of human nature, for example, is
often to suggest a classical teleological conception, associated with
the notion of a universal human essence, invariant in all times and
places . Talk of the self, on the other hand , tends to bias the issue in
favor of ind ivid ualistic notions and to suggest that the self­
understanding involved amounts merely to the coming to awareness
of an individ ual person, as in psychotherapy, for example. These
associations pose certain difficulties, for they threaten to beg the very
question we seek to answer, namely, how the subject is constituted ,
in what terms and on what scale it is properly conceived . To avoid
confusion on this count, I should say in advance that the account I
have in mind is a philosophical anthropology in the broadest sense;
philosophical in that it is arrived at reflectively rather than by
empirical generalization, anthropology in that it concerns the nature
of the human subj ect in its various possible forms of identity .

TH E S E L F AND T H E O T H E R : T H E P R I O RITY OF P L U RALITY

With these qualifications as background , and with some reference to

our earlier discussion of the problem of the self, we can proceed to
reconstruct Rawls' reasoning about the nature of the moral subj ect
as follows . For Rawls, the first feature of any creature capable of
justice is that it be plural in number. Justice could not apply in a
world where only one subj ect existed . I t could only have place in a
society of beings who were in some sense distinguishable one from
another. ' Principles of j ustice deal with conflicting claims upon the
advantages won by social co-operation; they apply to the relations
among several persons or groups. The word "contract" suggests this
plurality' ( 1 6) . For there to be j ustice, there must be the possibility of
conflicting claims, and for there to be conflicting claims, there must
be more than a single claimant. I n this way j the plurality of persons
can be seen as a necessary presupposition of the possibility ofj ustice.
Rawls insists on the essential plurality of the human subj ect when
he faults utilitarianism for extending to society as a whole the

so
The self and the other

p rinciples of rational choice for one man. This is a fallacy, he argues,

because it conflates diverse systems ofjustice into a single system of
desire, and so fails to take seriously the distinction between persons .
On utilitarianism, ' m any persons are fused into one ' , and 'separate
individuals are thought of as so many different lines ' . But utilitarian­
ism is mistaken , for ' there is no reason to suppose that the principles
which should regulate an association of men is simply an extension of
the principle of choice for one man . On the contrary: if we assume
that the correct regulative principle for anything depends on the
nature of that thing, and that the plurality of distinct persons with separate
system s of ends is an essential feature of hum an societies, we should not
expect the principles of social choice to be utilitarian' [emphasis
added] ( 2 8-g) .
Taking seriously ' the plurality and distinctiveness of individuals'
means more than defending liberty and freedom of thought, and
holding that the good of society consists in the advantages enjoyed by
individuals, as the utilitarians did. It means understanding the
plurality of persons to be an essential feature of any account of the
moral subj ect, a postulate of philosophical anthropology . In this
light, 'utilitarianism is not individualistic, at least when arrived at by
the more natural course of reflection ' , for by conflating all systems of
desires into one, it contradicts this essential postulate ( 2g) .
But in order for subj ects to be plural, there must be something that
differentiates them, some way of distinguishing one from another,
some principle of individuation . For Rawls , our individuating char­
acteristics are given empirically, by the distinctive concatenation of
wants and desires, aims and attributes, purposes and ends that come
to characterize human beings in their particularity . Each individual
is located uniquely in time and place, born into a particular family
and society, and the contingencies of these circumstances , together
with the interests and values and aspirations to which they give rise,
are what set people apart, what make them the particular persons
they are.
Within any group of persons, especially those of similar circumst­
ances , there will likely be found certain overlapping characteristics,
certain interests held in common . But notwithstanding even the
closest similarity of situation , no two persons could ever be said to be
identically situated , nor could it be that any two persons had identic­
al aims and interests in every respect, for if they did, it would no
justice and the moral subject

longer be clear how we could identify them as two distinguishable

persons. I n this way, the essential plurality of persons is assured , or
perhaps better, defined . For Rawls, the fact of our fundamental
plurality is a necessary presupposition of our being creatures capable
ofj ustice. What any particular person 's individuality actually con­
sists in, however, is an empirical matter. The fundamental feature of
the moral subject is its plurality, and given the means of individua­
tion, the number of its plurality corresponds to the number of
empirically-individuated human beings in the world . All of which is
to say that on Rawls ' view of the moral subject, every individual
human being is a moral subj ect, and every moral subj ect is an
individual human being.
At this point it might reasonably be asked whether, on Rawls '
conception o f the person , unity a s well a s plurality might appear as
an essential feature of the moral subj ect, equally necessary a presup­
position of the view that man is a creature capable ofj ustice. Indeed ,
Rawls describes the circumstances of j ustice as 'the normal condi­
tions under which human co-operation is both possible and neces­
sary ' , and from this it might be thought that where the essential
plurality of the moral subject makes human co-operation necessary,
some essential unity of persons makes human co-operation possible.
But this would misunderstand the logic of Rawls' conception, and
threaten to undermine the priorities on which the deontological ethic
depends. I believe he would answer roughly as follows :
While it is true that the principle of unity has an important place in
j ustice as fairness ( see in particular the account of the idea of social
union , section 79) , it is a mistake to accord it an equal priority with
plurality; it is not essential to our nature in the same way . This is
because any account of the unity of human subjectivity must presup­
pose its plurality, in a way that is not true in reverse. This can be seen
if we consider the notion of a human society as a co-operative venture
for mutual advantage, marked as it typically is by a conflict as well as
an identity of interests . Now the conflict of interests arises , as we
have seen , from the fact that the subjects of co-operation have
different interests and ends, and this fact follows from the nature of a
being capable of j ustice at all. The identity of interests, however,
expresses the fact that the parties happen to have suitably similar
needs and interests such that co-operation among them is mutually
advantageous . And this fact, that their needs and interests happen to

52
The self and the other

coincide in such a way, does not follow from the nature of their
subj ectivity, but merely from the happy accident of their circum­
ances . That they are able to come together to co-operate for mutual
advantage presupposes an antecedent plurality . Co-operation is by
its very nature co-operation between or among agents, whose plural­
ity must therefore be antecedent to the identity of interests they
realize in co-operative association .
The essential idea is that we want to account for the social values, for the
intrinsic good of institu tional, community, and associative activities, by a
conception of j ustice that in its theoretical basis is individualistic. For
reasons of clarity among others, we do not want to rely on an undefined
concept of community, or to suppose that society is an organic whole with a
life of its own distinct from and superior to that of all its members in their
relations with one another. Thus the con tractual conception of the original
position is worked out first . . . . From this conception , however individualis­
tic it might seem, we must eventually explain the value of community
(264-s ) .

That w e are distinct persons, characterized by separate systems of

ends , is a necessary presupposition of a being capable of justice.
What in particular our ends consis t in, and whether they happen to
coincide or overlap with the ends of others , is an empirical question
that cannot be known in advance. This is the sense - epistemological
rather than psychological - in which the plurality of subjects is given
prior to their unity . We are distinct individuals first, and then
(circumstances permitting) we form relationships and engage in
co-operative arrangements with others . The point is not that persons
co-operate out of selfish motives alone, but rather that our know­
ledge of the basis of plurality is given prior to experience, while our
knowledge of the basis of unity or co-operation can only come in the
light of experience. In any particular instance, we just have to see
whether or not the basis for co-operation exists .
The priority of plurality over unity, or the notion of the antecedent
individuation of the subject, describes the terms of relation between
the self and the other that must obtain for j ustice to be primary . But
before our reconstruction of Rawls' conception of the person can be
complete, we must consider a parallel issue, and this is the relation of
the self to its ends.

53
justice and the moral subject
T H E S E L F AND ITS E N D S : T H E S U BJ E C T O F POSSESSION

On the deontological ethic, 'the self is prior to the ends which are
affirmed by it' ( 560) . For Rawls, giving an account of this priority
poses a special challenge, for his proj ect rules out a self that achieves
its priority by inhabiting a transcendent or noumenal realm. I n
Rawls' view, any account o f self and ends must tell u s not one thing
but two things : how the self is distinguished from its ends , and also
how the self is connected to its ends. Without the first we are left with
a radically situated subj ect; without the second, a radically disem­
bodied subject.
Rawls' solution , implicit in the design of the original position, is to
conceive the self as a subject of possession, for in possession the self is
distanced from its ends without being detached altogether. The
notion of the self as a subj ect of possession can be located in the
assumption of mutual disinterest. This assumption looks on the
surface like a psychological assumption - it says the parties take no
interest in one another's in terests - but given its place in the original
position it works instead as an epistemological claim, as a claim
about the forms of self-knowledge of which we are capable. This is
why Rawls can coherently maintain that the assumption of mutual

tion' ( 1 8g) , and yet 'involves no particular theory of human motiva­

disinterest is ' the main motivational condition of the original posi­

tions ' ( 1 30) .

We can now see how this is so. The ass umption of mu tual disin­
terest is not an assumption about what motivates people, but an
assumption about the nature of subjects who possess motivations in
general . It concerns the nature of the self ( that is, how it is consti­

nature of the self's desires or aims . It concerns the subject of interes ts

tuted , how it stands with respect to its situation generally) , not the

and ends, not the content of those interests and ends, whatever they
may happen to be . As Kant argues that all experience must be the
experience of some subject, Rawls' assumption of mutual disinterest
holds that all interests must be the interests of some subject.
Although the in terests advanced by these plans are not assumed to be
in terests in the self, they are in terests ofa self that regards its conception of the
good as worthy of recog ni tion [emphasis added J ( 1 2 7 ) .
I make n o restrictive assu mptions about the parties' conceptions of the good
except that they arc rational long-term plans. While these plans determine

54
The self and its ends
the aims and interes ts ofa self, the aims and in terests are not presumed to be
egoistic or selfish . Whether this is the case depends upon the kinds of ends
which a person pursues. If wealth, position, and influence, and the acco­
lades of social prestige are a person's final purposes , then surely his concep­
tion of the good is egois tic. His dominan t in terests are in himself, not merely,
as they must always be, interests of a self [emphasis added ] ( 1 29) .

I n the assumption of mutual disinterest, we find the key to Rawls'

conception of the subject, the picture of the way we must be to be
subjects for whom j us tice is primary. But the notion of the self as a
subject of possession, taken alone, does not complete the picture. As
the account of plurality suggests, not just any subject of possession
will do, but only an antecedently individ uated subject, the bounds of
whose self are fixed prior to experience. To be a deontological self, I
must be a subject whose identity is given independently of the things
I have, independently, that is, of my interests and ends and my
relations with others . Combined with the idea of p ossession, this
notion of individuation powerfully completes Rawls' theory of the
p erson . We can appreciate its full consequences by contrasting two
aspects of possession - two different ways an interest can be 'of a self'
- and seeing how the notion of antecedent individuation commits the
deontological self to one of them .
I n so far as I possess something, I am at once related to it and
distanced from it. To say that I possess a certain trait or desire or
ambition is to say that I am related to it in a certain way - it is m ine
rather than yours - and also that I am distanced from it in a certain
way - that it is m ine rather than m e . The latter point means that if I
lose a thing I possess, I am still the same ' I ' who had it; this is the
s ense, p aradoxical at first but unavoidable on reflection, in which the
notion of possession is a distancing notion. This distancing aspect is
essential to the continuity of the self. It preserves for the selfa certain
dignity an d integrity by saving it from transformation in the face of
the slightest contingency . Preserving this distance, and the integrity
i t implies, typically req uires a certain kind of self-knowledge. To
preserve the distinction between what is me and what is ( merely)
mine, I must know, or be able to sort out when the occasion demands,
something about who I am. Thus, Odysseus was able to survive his
treacherous journey home by donning various disguises , and his
abili ty to do so pres upposed an unders tanding of who he was , to

55
justice and the moral subject

being with , so to speak. Since his self-knowledge preceded his

experience in this sense, he was able to return home the same person
who had left, familiar to Penelope, untransfigured by his journey,
unlike Agamemnon, who returned a stranger to his household and
met a different fate!
I t is a consequence of the dual aspect of possession that it can fade
or diminish in two different ways. I grad ually lose possession of a
thing not only as it is distanced from my person, but also as the
distance between my self and the thing narrows and tends toward
collapse. I lose possession of a desire or an ambition as my commit­
ment to it fades, as my hold on it becomes more attenuated, but also,
after a certain point, as my attachment to it grows , as it gradually
becomes attached to me. As the desire or ambition becomes in­
creasingly constitutive of my identity, it becomes more and more me,
and less and less mine. Or as we might say in some cases , the less I
possess it, and the more I am possessed by it. Imagine that a desire,
held tentatively at first, gradually becomes more central to my
overall aims , until finally it becomes an overriding consideration in
all I think and do. As it grows from a desire into an obsession , I
possess it less and it possesses me more, until finally it becomes
indistinguishable from my identity.
A different sort of example: in so far as the American Declaration
ofl ndependence is correct, that man is endowed by his Creator with
certain inalienable rights that among them are life, liberty, and the
pursuit of happiness, its famous litany describes not what we have as
free men but rather what we are. The endowment is less a possession
than a nat ure of a certain kind ; he who would abnegate his liberty or
pursue a miserable existence would experience these endowments
not as possessions but as constraints . In so far as these rights are
truly inalienable, a man is no more entitled to do away with them in
his own case than to take them from another. Suicide is on a par with
murder, and selling oneself into slavery is morally equivalent to
enslaving another.
As these images suggest, possession is bound up with human
agency and a sense of self-command . Dispossession, from both
points of view, can be understood as a kind of disempowering. When
my possession of an obj ect fades, whether because it slips from my
grasp or looms so large before me that I am overwhelmed , disem-
2 I am indebted for this example to Allen Grossman.
The self and its ends

powered in the face of it, my agency with respect to the object is

diminished as well. Each challenge is associated with a different
notion of agency, which implies , in turn, a different account of the
relation of the self to its ends. We can think of the two dimensions of
agency as different ways of repairing the drift toward dispossession,
and distinguish them by the way they work to restore a sense of
self-comm and .
The first kind of dispossession involves the distancing of the end
from the self whose end it once was . It becomes increasingly unclear
in what sense this is my end rather than yours , or somebody else' s, or
no one's at all . The self is disempowered because dissociated from
those ends and desires which, woven gradually together into a
coherent whole, provide a fixity of purpose, form a plan oflife, and so
account for the continuity of the self with its ends. Where the self is
regarded as given prior to its ends, its bounds fixed once and for all
such that they are impermeable, invulnerable to transformation by
experience, such continuity is perpetually and inherently problema­
tic; the only way it can be affirmed is for the self to reach beyond
itself, to grasp as an obj ect of its will the ends it would possess, and
hold them, as it always must, external to itself. '
The second kind of dispossession disempowers in another way .
Here, the problem is not to overcome the distance created by the drift
of the end from the self, but rather to recover and preserve a space
that increasingly threatens to collapse . Crowded by the claims and
pressures of various possible purposes and ends, all impinging indis­
criminately on my identity, I am unable to sort them out, unable to
mark out the limits or the boundaries of my self, incapable of saying
where my identity ends and the world of attributes, aims, and desires
begins . I am disempowered in the sense of lacking any clear grip on
who, in particular, I am. Too much is too essential to my identity.
Where the ends are given prior to the self they constitute, the bounds
of the subj ect are open, its identity infinitely accommodating and
ultimately fluid . Unable to distinguish what is mine from what is me,
I am in constant danger of drowning in a sea of circumstance.
We might understand human agency as the faculty by which the

3 Compare Kant ( 1 797: 62 ) : 'Therefore, the relation of having something external to oneself
as one's own (property) consists of a purely dejure union of the Will of the subject with that
object, independently of his relationship to it in space and time and in accordance with the
concept of intelligible possession. '

57
Justice and the moral subject

self comes by its ends. This acknowledges its close connection with
the notion of possession without begging the question which dimen­
sion of possession is at stake, nor the q uestion of the relative priority
of self and ends. For if l am a being with ends, there are at least two
ways I might 'come by' them : one is by choice, the other by
discovery, by 'finding them out' . The first sense of 'coming by' we
might call the voluntarist dimension of agency, the second sense the
cognitive dimension . Each kind of agency can be seen as repairing a
different kind of dispossession .
Where the self is disempowered because detached from its ends ,
dispossession i s repaired by the faculty o f agency i n its voluntarist
sense, in which the self is related to its ends as a willing subject to the
obj ects of choice . The relevant agency involves the exercise of will,
for it is the will that is able to transcend the space between the subject
and its object without requiring that it be closed .
Where the self is disempowered because undifferentiated from its
ends, dispossession is repaired by agency in its cognitive sense, in
which the self is related to its ends as a knowing subject to the obj ects
of understanding. Where the ends of the self are given in advance, the
relevant agency is not voluntarist but cognitive, since the subject
achieves self-command not by choosing that which is already given
( this would be unintelligible) but by reflecting on itself and inquiring
into its constituent nature, discerning its laws and imperatives, and
acknowledging its purposes as its own . Where the faculty of will
seeks to reverse the drifting apart of self and ends by restoring a
certain continuity between them , reflexivity is a distancing faculty,
and issues in a certain detachment. It succeeds by res toring the
shrunken space between self and ends. In reflexivity, the self turns its
lights inward upon itself, making the self its own obj ect of inquiry
and reflection . When I am able to reflect on my obsession, able to
pick it out and make it an obj ect of my reflection, I thereby establish
a certain s pace between it and me, and so diminish its hold . I t
becomes more a n attribute and less a constituent o fm y identity, and
so dissolves from an obsession to a mere desire.
Where the subject is regarded as prior to its ends, self-knowledge is
not a possibility in this sense, for the bounds it would define are taken
as given in advance, unreflectively, by the principle of antecedent
individuation . The bounds of the self are fixed and within them all is
transparent. The relevant moral q uestion is not 'Who am I ?' (for the
Individualism and the claim s of com m uniry

answer to this question is given in advance) but rather 'What ends

shall I choose?' and this is a question addressed to the will.
For the self whose identity is constituted in the light of ends
already before it, agency consists less in summoning the will than in
seeking self-understanding. The relevant question is not what ends
to choose, for my problem is precisely that the answer to this
question is already given, but rather who I am, how I am to discern
in this clutter of possible ends what is me from what is mine. Here,
the bounds of the self are not fixtures but possibilities, their contours
no longer self-evident but at least partly unformed . Rendering them
clear, and defining the bounds of my identity are one and the same.
The self-command that is measured in the first case in terms of the
scope and reach of my will is determined in the second by the depth
and clarity of my self-awareness.
We can now see how the cluster of assumptions associated with the
voluntarist notion of agency and the distancing aspect of possession
fill out Rawls ' theory of the person . The notion of a subject of
possession, individuated in advance and given prior to its ends,
seems just the conception required to redeem the deontological ethic
without lapsing into transcendence . In this way, the self is distin­
guished from its ends - it stands beyond them , at a distance, with a
certain priority - but is also related to its ends, as willing subject to
the obj ects of choice .
The voluntarist notion of agency is thus a key ingredient in Rawls'
conception, and plays a central role in the deontological ethic as a
whole. ' I t is not our aims that primarily reveal our nature' ( 560) , but
rather our capacity to choose our aims that matters most, and this
capacity finds expression in the principles ofjustice. 'Thus a moral
person is a subj ect with ends he has chosen, and his fundamental
preference is for conditions that enable him to frame a mode of life
that expresses his nature as a free and equal rational being as fully as
circums tances permit' ( 56 I ) . This , finally, is why we cannot regard
j ustice as j ust one value among others . ' I n order to realize our nature
we have no alternative but to plan to preserve our sense ofjustice as
governing our other aims' ( 5 74) .

I N DIVIDUALISM AND T H E C LAIMS O F C O M M U N ITY

In our reconstruction of the deontological subject we find at last the

59
Justice and the moral subject

standard by which the descriptive premises of the original position

may be assessed , the counterweight to our moral intuitions that
provides Rawls' reflective equili brium with a test at both .ends. It is
this conception of the subject, and no particular account of human
motivations, that the assumption of mutual disinterest conveys .
We may recall that on Rawls ' account, 'the postulate of mutual
disinteres t in the original position is made to insure that the prin­
ciples ofjustice do not depend upon strong assumptions ' ( 1 29) , and
the point of avoiding strong assumptions is to make possible the
derivation of principles that do not presuppose any particular
conception of the good . 'Liberty in adopting a conception of the good
is limited only by principles that are deduced from a doctrine which
imposes no prior constraints on these conceptions . Presuming
mutual disinterest in the original position carries out this idea' ( 2 54) .
Strong or controversial ass umptions would threaten to impose a
particular conception of the good , and so bias the choice of principles
in advance .
How strong or weak, then, are the assumptions that form Rawls'
conception of the person? With what range of values and ends are
they compatible? Are they weak and innocent enough to avoid ruling
out any conceptions of the good in advance? We have already seen
that the empiricist reading of the original position produces a litany
of obj ections on this score; the circumstances ofj ustice and especially
the assumption of mutual disinterest are thought to introduce an
individ ualistic bias, and to rule out or otherwise devalue such
motives as benevolence, altruism , and communitarian sentiments .
As one cri tic has written , the original position contains 'a strong
individualistic bias , which is further strengthened by the motivation­
al assumptions of mutual disinterest and absence of envy . . . . The
original position seems to presuppose not j ust a neutral theory of the
good , but a liberal, individualistic conception according to which the
best that can be wished for someone is the uni mpeded pursuit of his
own path , provided it does not interfere with the rights of others'
(Nagel 1 9 7 3 : g- r o) .
But as Rawls rightly insists , his theory is not the 'narrowly
individ ualistic doctrine' that the empiricist objection supposes .
' Once the point of the assumption of mutual disinterest is under­
stood , the obj ection seems misplaced ' ( 584) . Notwithstanding its
individualist dimension, j ustice as fairness does not defend private

6o
Individualism and the claim s of com m uniry

society as an ideal ( 5 2 2 f ), or presuppose selfish or egoistic motiva­

tions ( 1 29) , or oppose communitarian values . 'Although justice as
fairness begins by taking the persons in the original position as
individuals . . . this is no obstacle to explicating the higher-order
moral sentiments that serve to bind a community of persons
together' ( 1 92 ) .
Rawls has emphasized i n particular that the assumption of mutual
disinterest does not bias the choice of principles in favor of indi­
vidualistic values at the expense of communitarian ones . Those who
suppose that it does overlook the special status of the original posi­
tion, and mistakenly assume that the motives attributed to the
parties are meant to apply generally to actual human beings or to
persons in a well-ordered society. But neither is the case. The
motives attributed to the parties in the original position neither
reflect the actual motivations current in society nor determine direct­
ly the motives of persons in a well-ordered society.
Given the restricted scope of these assumptions, Rawls argues,

well-ordered society should be predominantly individualistic' ( 1 9 75:

' there seems to be no reason oflhand why the ends of people in a

544) . Communitarian values, like any other values individuals

might choose to pursue, would likely exist, and possibly even flourish
in a society governed by the two principles of justice .

There is no reason why a well-ordered society should encourage primarily

individualistic values if this means ways of life that lead individuals to
pursue their own way and to have no concern for the interest of others
(although respecting their rights and liberties ) . Normally one would expect
most people to belong to one or more associations and to have at least some
collective ends in this sense. The basic liberties are not intended to keep
persons in isolation from one another, or to persuade them to live private
lives, even though some no doubt will, but to secure the right of free
movement between associations and smaller communities ( 1 9 75: 550) .

On Rawls' conception of the person , my ends are benevolent or

communitarian when they take as their obj ect the good of another, or
of a group of others with whom I may be associated , and indeed there
is nothing in his view to rule out communitarian ends in this sense.
All interests, values, and conceptions of the good are open to the
Rawlsian self, so long as they can be cast as the interests of a subject
individuated in advance and given prior to its ends, so long, that is,
Justice and the moral subject

as they describe the obj ects I seek rather than the subject I am. Only
the bounds of the self are fixed in advance .
But this suggests a deeper sense in which Rawls' conception is
individualistic. We can locate this individualism and identify the
conceptions of the good it excludes by recalling that the Rawlsian self
is not only a subj ect of possession, but an anteceden tly individuated
subj ect, standing always at a certain distance from the interests it
has . One conseq uence of this distance is to put the self beyond the
reach of experience, to make it invulnerable, to fix its identity once
and for all . No commitment could grip me so deeply that I could not
understand myself without it. No transformation oflife purposes and
plans could be so unsettling as to disrupt the contours of my identity .
No project could be so essential that turning away from it would call
into question the person I am . Given my independence from the
values I have, I can always stand apart from them ; my public
identity as a moral person 'is not affected by changes over time' i n my
conception of the good ( Rawls r g8o : 544 - 5 ) . '
But a self so thoroughly independent as this rules out any concep­
tion of the good (or of the bad ) bound up with possession in the
constitu tive sense. It rules out the possibility of any attachment ( or
obsession ) able to reach beyond our values and sentiments to engage
our identity itself. It rules out the possibility of a public life in which,
for good or ill, the identity as well as the interests of the participants
could be at stake. And it rules out the possibility that common
purposes and ends could inspire more or less expansive self­
understandi ngs and so define a community in the constitu tive sense,
a community describing the subject and not just the objects of shared
aspirations . More generally, Rawls' account rules out the possibility
of what we might call 'intersubjective' or 'intrasubjective' forms of
self-understanding, ways of conceiving the subj ect that do not
assume its bounds to be given in advance. Unlike Rawls' conception,
intersubj ective and in trasubjective conceptions do not assume that
to speak of the self, from a moral point of view, is necessarily and
unproblematically to speak of an anteceden tly-individuated self.
Intersubjective conceptions allow that in certain moral circum­
stances, the relevant description of the self may em brace more than a
single, individual human being, as when we attribute responsibility
4 Rawls suggests at one point that my private identity as a moral person might not be similarly
immune from constitutive at tachments ( r g8o : 545). See p. r 82 , below.
Individualism and the claim s of com m unity

or affirm an obligation to a family or community or class or nation

rather than to some particular human being. Such conceptions are
presumably what Rawls has in mind when he rejects, 'for reasons of
clarity among others ' , what he calls 'an undefined concept of com­
munity' and the notion that 'society is an organic whole' ( 2 64) , for
these suggest the metaphysically troubling side of Kant which Rawls
is anxious to replace .
lntrasubj ective conceptions, on the other hand , allow that for
certain pu rposes, the appropriate description of the moral subj ect
may refer to a plurality of selves within a single, individual human
being, as when we account for inner deliberation in terms of the pull
of competing identities , or moments of introspection in terms of
occluded self-knowledge, or when we absolve someone from respon­
s i bility for the heretical beliefs 'he' held before his religious
conversion. On intrasubjective conceptions, to speak of selves within
a(n) (antecedently-individuated empirical) self is not merely meta­
phorical but sometimes of genuine moral and practical import.
While Rawls does not rej ect such notions explicitly, he denies
them by implication when he assumes that to every individual
person there corresponds a single system of desires , and that utilitar­
ianism fails as a social ethic in mistakenly applying to society the
principles of choice appropriate for one man . Since he takes for
granted that every individual consists of one and only one system of
desires, the problem of conflating desires does not arise in the
individual case, and the principle of rational prudence can properly
govern one' s conduct toward oneself. 'A person quite properly acts,
at least when others are not affected , to achieve his own greatest
good , to advance his rational ends as far as possible' ( 2 3 ) . Whereas
society consists of a plurality of subjects and so requires j ustice, in
private morality, utilitarianism seems to suffice; where others are not
involved , I am free to maximize my good without reference to the
principle of right.' Here again Rawls departs from Kant, who

5 In his discussion of deliberative rationality, Rawls stops j ust short of acknowledging an

in trasubj ective dimension and admitting the concept of right as a constraint on private
moral choice: 'One who rej ects equally the claims of his future self and the in terests of
others is not only irresponsible with respect to them but in regard to his own person as wel l .
He does n o t see himself a s o n e enduring individual. Now looked at in this way, t h e principle
of responsibility to self resembles a principle of right . . . . The person at one time, so to speak,
must not be able to complain about actions of the person at another time' [emphasis
added ] (42 3).
justice and the moral subject

emphasized the concept of 'necessary duty to oneself, and applied

the category of right to private as well as public morality ( Kant 1 785 :
8g-go, g6- 7 , 1 0 1 , 1 05) .
The assumptions of the original position thus stand opposed in
advance to any conception of the good requiring a more or less
expansive self-understanding, and in particular to the possibility of
community in the constitutive sense. On Rawls' view, a sense of
community describes a possible aim of an tecedently individ uated
selves, not an ingredient or constituent of their identity as such . This
guarantees its subordinate status. Since 'the essential unity of the self
is already provided by the concept of right' (563 ) , community must
find its virtue as one contender among others within the framework
defined by justice, not as a rival account of the framework itself. The
question then becomes whether individuals who happen to espouse
communitarian aims can pursue them within a well-ordered society,
antecedently defined by the principles of justice, not whether a
well-ordered society is itself a community (in the constitutive sense) .
'There is, to be sure, one collective aim supported by state power for
the whole well-ordered society, a j ust society wherein the common
conception of j ustice is publicly recognized ; but within this
framework communitarian aims may be pursued , and quite possibly
by the vast majority of persons ' ( Rawls 1 975 : 550) .
We can see now more clearly the relation between Rawls' theory of
the person and his claim for the primacy of j ustice. As a person's
values and ends are always attributes and never constituents of the
self, so a sense of community is only an attribute and never a
constituent of a well-ordered society . As the self is prior to the aims it
affirms, so a well-ordered society, defined by justice, is prior to the
aims - communitarian or otherwise - its members may profess . This
is the sense, both moral and epistemological, in which justice is the
first virtue of social institutions .

Our reconstruction of Rawls' conception of the person now com­

plete, it remains to assess this conception and the deontological ethic
it must support . We have seen that the assumptions contained in the
original position are strong and far-reaching rather than weak and
innocuous, though not for the reasons the empiricist obj ection sug­
gests. These assumptions do not admit all ends, but rule out in
advance any end whose adoption or pursuit could engage or trans-
Individualism and the claims of communiry

form the identity of the self, and they rej ect in particular the
possibility that the good of community could consist in a constitutive
dimension of this kind .
If it therefore cannot be said that Rawls' principles derive 'from a
doctrine which imposes no prior constraints' on conceptions of the
good , still it might be argued that the conceptions he exdudes are
somehow dispensable, that it is possible to account for j ustice and to
arrive at a conception of a well-ordered society without them . Rawls'
theory ofj ustice is j ust such an attempt. I n order to assess it, we must
descend from deontological meta-ethics to consider first-order prin­
ciples. In the chapters that follow, I shall argue that Rawls' concep­
tion of the person can neither support his theory of j ustice nor
plausibly account for our capacities for agency and self-reflection ;
j ustice cannot be primary in the way deontology requires, for we
cannot coherently regard ourselves as the sort of beings the deonto­
logical ethic requires us to be.
2

Possession, Desert, and Distributive

Justice

Having clarified the status of Rawls' motivational assumptions, we

can now put his theory of the person and his theory ofj us tice side by
side to check for the fit between them . In this way we can work within
the argument from reflective equilibrium, by asking whether the
theory of the person contained in the original position corresponds to
the principles of justice it must both shape and reflect. Of special
interest for this purpose is the difference principle, the principle that
permits only those inequalities that work to the benefit of the least
advantaged members of society. We shall see in this chapter that an
adequate defense of the difference principle must presuppose a
conception of the person unavailable on deontological assumptions,
that we cannot be subjects for whom j ustice is primary and also be
subj ects for whom the difference principle is a principle ofj ustice. A
central focus will be on the role of desert in distributive j ustice, and
on the conception of possession it requires . To explore these issues,
we begin by con trasting Rawls' views with various competing dis­

strikingly similar theory defended by Robert Nozick ( I 9 7 4 ) .

tributive theories, and in particular the rival, but in some ways

L I B E R T A R I A N I S M TO E GA L I T A R I A N I S M

From a practical political point of view, the positions of Rawls and

Nozick are clearly opposed . Rawls, the welfare-state liberal, and
Nozick, the libertarian conservative, define between them the
clearest alternatives the American political agenda has to offer, at
least where issues of distributive j ustice are concerned . And yet, from
a philosophical point of view, they have much in common . Both
define their positions in explicit opposition to utilitarianism, which
each rej ects on the grounds that it denies the distinction between
persons . Both offer instead a rights-based ethic said to secure the
liberty of individuals more completely. Although Nozick's account of
rights owes much to Locke, both appeal to Kant's precept to treat

66
Libertarianism to egalitarianism

every person as an end and not merely as a means, and seek princi­
ples ofj ustice that embody it. Both d eny that there exists any social
entity above or beyond the individu als who comprise it. As Nozick
writes, echoing Rawls in both principle and rhetoric,
Side constraints upon action [ that is, unqualified prohibitions] reflect the
underlying Kantian principle that individuals are ends and not merely
means . . . . Side constraints express the inviolability of other persons. But
why may not one violate persons for the greater social good? I ndividually,
we each sometimes choose to undergo some pain or sacrifice for a greater
benefit or to avoid a greater harm . . . . But there is no social entiry with a good
that undergoes some sacrifice for its own good . There are only individual
people, different individual people, with their own individual lives . Using
one of these people for the benefit of others, uses him and benefits the others .
Nothing more . . . . To use a person in this way does not sufficiently respect

life he has ( 1 974: go-g ) .

and take account of the fact that he is a separate person, that his is the only

The moral side constraints upon what w e may do, I claim, reflect the fact of
our separate existences. They reflect the fact that no moral balancing act
can take place among us; there is no moral outweighing of one of our lives by

sacrifice of some of us for others ( 1 974: gg) .

others so as to lead to a greater overall social good . There is no j ustified

Both theorists emphasize what Rawls calls ' the plurality and
distinctness of persons' and what Nozick calls 'the fact of our sepa­
rate existences' . This is the central moral fact that utilitarianism
denies and that an individualistic, rights-based ethic affirms . On this
moral fact, and on the importance of rights, Rawls and Nozick
emphatically agree. And yet Rawls arrives at a theory ofj us tice on
which social and economic inequalities are permitted only in so far as
they benefit the least well off, while Nozick holds justice to consist in
voluntary exchanges and transfers alone, ruling out redistributive
policies altogether. How then do their theories of j ustice come so
sharply to diverge? Fortunately, the point of divergence can be
located with some precision, since Rawls, in developing his second
principle ofj ustice ( the one containing the difference principle) , lays
out a line of reasoning that begins with a position similar to Nozick's
and ends with his own .
Rawls considers three possible principles by which the distribu­
tion of social and economic benefits might be regulated or assessed :
natural liberty ( similar to Nozick's 'entitlement theory' ) , liberal
Possession, desert, and distributive justice

equality ( akin to a standard meritocracy) , and democratic equality

( based on the difference principle) . The system of natural liberty
defines as j ust whatever distribution results from an efficient market
economy in which a formal ( i . e . legal ) equality of opportunity
prevails , such that positions are open to those with the relevant
talents . Rawls finds this principle inadequate on the grounds that the
distribution it sanctions tends simply to reproduce the initial dis­
tribution of talents and assets ; those substantially endowed will wind
up with su bstantial shares , and those with meager assets will end
with meager results . Where the outcome tends simply to reproduce
the initial distribution, it is possible to call it j ust only on the
additional assumption that initial endowments were justly distri­
buted . But this assumption cannot be established . 'The initial
distribution of assets for any period of time is strongly influenced by
natural and social contingencies ' , and as such is neither just nor
unj ust but simply arbitrary . And since there is nothing to recom­
mend the j us tice of initial endowments, to install them in the name of
j ustice is to incorporate the arbitrariness of fortune, nothing more .
' I ntuitively, the most obvious injustice of the system of natural
liberty is that it permits distributive shares to be improperly in­
fluenced by these factors so arbitrary from a moral point of view'
(72) .
The principle of liberal equality seeks to remedy the injustice of
natural liberty by going beyond formal equality of opportunity and
correcting, where possible, for social and cultural disadvantages .
The aim is a kind of 'fair meritocracy ' , in which social and cultural
inequalities are mi tigated by equal educational opportunities ,
certain redistribu tive policies, and other social reforms. On the
principle ofliberal equality, the ideal is to provide all an 'equal start ' ,
so that those with similar native talents and capacities and a similar
willingness to exercise them would have ' the same prospects of
success regardless of their initial place in the social system, that is,
irrespective of the income class into which they are born . In all
sectors of society there should be roughly equal prospects of culture
and achievement for everyone similarly motivated and endowed .
The expectations of those with the same abilities and aspirations
should not be affected by their social class' ( 7 3 ) .
But while liberal equality represents an improvement over the
system of natural liberty, 'intuitively it still appears defective' . Fair

68
Libertarianism to egalitarianism

opportunity, however full, is too weak an assault on the arbitrariness

of fortune .

Even i f it works t o perfection i n eliminating the influence o f social contin­

gencies, it still permits the distribution of wealth and income to be deter­
mined by the natural distribution of abilities and talents. Within the limits
allowed by the background arrangements, distributive shares are decided
by the outcome of the natural lottery; and this outcome is arbitrary from a
moral perspective. There is no more reason to permit the distribution of

historical and social fortune ( 73-4) .

income and wealth to be settled by the distribution of natural assets than by

Once we are struck by the arbitrariness of initial endowments

determining life prospects generally, we are bound on reflection to be
as disturbed by the influence of natural contingencies as social and
cultural ones . ' From a moral standpoint, the two seem equally
arbitrary' ( 75) . The same reasoning that leads us to favor a 'fair
meritocracy' (as in liberal equality) over a purely formal equality of
opportunity ( as in natural liberty) naturally leads us further to seek
what Rawls calls the democratic conception . But it seems clear that
the democratic conception cannot be found in a simple extension of
the principle af fair opportunity. For one thing, it would be virtually
impossible to extend opportunity so completely as to eradicate even
those inequalities stemming from social and cultural conditions
alone. The institution of the family, for one, makes it 'impossible in
practice to secure equal chances of achievement and culture for those
similarly endowed' ( 74) . But even if compensatory education and
other reforms could fully, or even nearly, correct for social and
cultural deprivation, it is difficult if not vaguely forbidding to im­
agine what kind of social policies would be required to ' correct' in a
comparable way for the contingencies of natural fortune. What is
needed, then, is a conception that nullifies the effect of these differ­
ences while at the same time acknowledging their intractability .
Some com mentators, and particularly those hostile to the prin­
ciple of democratic equality, describe the next logical step as a move
from equality of opportunity to equality of result. In their view, any
theory ofj ustice that rej ects a meritocratic conception for the moral
arbitrariness of its distributive consequences must necessarily be
committed to a kind oflevelling equality requiring constant readj ust­
ment of distributive shares to correct for persisting differences of
6g
Possession, desert, and distributive justice

native talent and ability (Bell 1 9 7 3 : 44 1 -3 ) . But equality of result is

by no means the only democratic alternative to a meritocratic reg­
ime, nor is it the principle adopted by Rawls. The difference princi­
ple is not synonymous with equality of result, nor does it require the
levelling of all differences between persons. ' I t does not follow that
one should eliminate these distinctions,' writes Rawls. 'There is
another way to deal with them' ( 1 02 ) . Rawls' way is not to eradicate
unequal endowments but to arrange the scheme of benefits and
burdens so that the least advantaged may share in the resources of
the fortunate. This is the arrangement that the difference principle
seeks to achieve. It defines as j ust only those social and economic
inequalities that work to the benefit of the least advantaged members
of society . Taken together with the principle of offices and positions
open to all under fair equality of opportunity, the difference principle
defines Rawls' conception of democratic equality.
The difference principle is not simply a fuller version of the
principle of fair opportunity; it attacks the problem of arbitrariness
in a fundamentally different way. Rather than transform the condi­
tions under which I exercise my talents, the difference principle
transforms the moral basis on which I claim the benefits that flow
from them. No longer am I to be regarded as the sole proprietor of my
assets, or privileged recipient of the advantages they bring. 'The
difference principle represents, in effect, an agreement to regard the
distribution of natural talents as a common asset and to share in the
benefits of this distribution whatever it turns out to be. ' In this way,
the difference principle acknowledges the arbitrariness offortune by
asserting that I am not really the owner but merely the guardian or
repository of the talents and capacities that happen to reside in me,
and as such have no special moral claim on the fruits oftheir exercise.

Those who have been favored by nature, whoever they are, may gain from
their good fortune only on terms that improve the situation of those who
have lost out. The naturally advantaged are not to gain merely because they
are more gifted, but only to cover the costs of training and education and for
using their endowments in ways that help the less fortunate as well. No one

place in society ( 10 r - 2 ) .
deserves his greater natural capacity nor merits a more favorable starting

B y regarding the distribution o f talents and attributes a s a com­

mon asset rather than as individual possessions, Rawls obviates the
70
Libertarianism to egalitarianism

need to 'even out' endowments in order to remedy the arbitrariness

of social and natural contingencies . When ' men agree to share one
another's fate' ( 1 02 ) , it matters less that their fates, individually,
may vary. This is why, although the difference principle tends to
' redress the bias of contingencies in the direction of equality', it 'does
not require society to try to even out handicaps as if all were expected
to compete on a fair basis in the same race' ( 1 o 1 ) .
Rawls acknowledges that the difference principle and in particular
the notion of endowments as 'common assets' clash with traditional
conceptions of individual desert. 'There is a natural inclination to
obj ect that those better situated deserve their greater advantages
whether or not they are to the benefit of others' ( 1 03 ) . Rawls'
response is that this conception of individual desert is a mistake, as
the general argument from arbitrariness suggests . ' I t seems to be one
of the fixed points of our considered j udgments that no one deserves
his place in the distribution of native endowments, any more than
one deserves one's initial starting place in society' ( 1 04) . The claim
that a person deserves at least what he achieves through his own
effort is more intuitively plausible, but even the willingness to strive
conscientiously may largely be determined by social and natural
contingencies. 'The assertion that a man deserves the superior char­
acter that enables him to make the effort to cultivate his abilities is
equally problematic; for his character depends in large part upon
fortunate family and social circumstances for which he can claim no
credit. The notion of desert seems not to apply to these cases' ( 1 04) .
This is not to deny the role of individual entitlements altogether.
Certain kinds of entitlements are consistent with the difference
principle, but here it is necessary to distinguish between desert and
legitimate expectations. Since it is in the general interest that I
cultivate and exercise (some of) the talents and assets in my charge,
rather than have them lie dormant, society is typically arranged to
provide resources for their cultivation and incentives for their exer­
cise. To be sure, I am entitled to my share of these benefits when I
have qualified for them under the terms specified . What is important
to stress, however, is that these claims honor the legitimate expecta­
tions created by institutions designed to elicit my efforts, not a
primordial right or claim of desert in virtue of qualities I possess.

I t is perfectly true that given a just system of co-operation as a scheme of

71
Possession, desert, and distributive justice

public rules and the expectations set up by it, those who, with the prospect
of improving their condition, have done what the system announces that it
will reward are entitled to their advantages. In this sense the more fortunate
have a claim to their better condition; their claims are legitimate expecta­
tions established by social institutions, and the community is obligated to
meet them. But this sense of desert presupposes the existence of the co­
operative scheme; it is irrelevant to the question whether in the first place
the scheme is to be designed in accordance with the difference principle or
some other criterion ( 1 03 ) .

Although I a m entitled t o the benefits answering m y legitimate

expectations, I do not deserve them, for two reasons: first, given the
assumption of common assets, I do not really possess the attributes
that give rise to the benefits, or if I do possess them, it is only in the
weak, accidental sense rather than the strong, constitutive sense, and
this sense of possession is inadequate to establish desert in the strong,
pre-institutional sense. And second, while I am entitled to my fair
share under the rules, I am not entitled that these rules, rewarding
these attributes, be in force rather than some others . For these
reasons, the well-endowed 'cannot say that he deserves and therefore
has a right to a scheme of co-operation in which he is permitted to
acquire benefits in ways that do not contribute to the welfare of
others. There is no basis for his making this claim' ( I o4) .

M E RITOCRACY VERSUS T H E DIFFERENCE P R I N C I P L E

Before moving on to consider Nozick's answer to Rawls and his

defense of natural liberty, it may be helpful to clarify further some of
the contrasts between the difference principle ( as in Rawls' 'demo­
cratic equality') and the meritocratic conception (as in 'liberal
equality') . Perhaps the most striking difference is in the role of
individual merit or desert, which is central in the meritocratic con­
ception and absent, or at least seriously weakened, in j ustice as
fairness . ' In a 'fair meritocracy', that is, one in which discrimination
and class biases are overcome, those who achieve favored positions
have earned their status and so deserve the rewards that attach to it.
Unequal distributive shares are allocated in recognition of superior
1 I leave aside those versions of meritocracy that would allocate distributive shares for the
sake of creating incentives and attracting the relevant talents alone, without reference to

72
the moral worthiness of the recipients.
Meritocracy versus the difference principle

achievement and not merely in satisfaction of legitimate expecta­

tions . As one defender of the meritocratic ethic explains,
A meritocracy is made up of those who have earned their authority . . . .
Meritocracy, in the context of my usage, is an em phasis on individual
achievement and earned s tatus as confirmed by one's peers . . . . While all
men are entitled to respect, they are not all entitled to praise. The
meritocracy, in the best meaning of that word, is made up of those worthy of
praise (Bell 1 973: 453-4) .

A second , related difference concerns the distinction between

genetic and cultural advantages . For Rawls, this distinction is
virtually irrelevant to questions ofj ustice. On meritocratic assump­
tions, however, the distinction is crucial; hence the intense debate
among those committed to meritocratic ideals over the role of genetic
and cultural factors in determining intelligence and life prospects
generally. Where the justice of distributive arrangements is seen to
depend on the 'fair opportunity' of all to compete equally for
(ultimately) unequal rewards, the distinction of genetic from social
and cultural obstacles becomes central to any assessment of the
scheme. The more closely success can be traced to hereditary factors,
the fewer inequalities social institutions can be presumed capable of
(or responsible for) ameliorating, and the less scope seems available
for the kind of individual effort on which desert is said to depend .
In the nature of meritocracy, as it has been traditionally conceived, what is
central to the assessment of a person is the assumed relation of achievement
to intelligence, and of intelligence to its measurement on the I ntelligence
Quotient scale. The first question, therefore, is what determines intelli­
gence.

All this makes the question of the relation of intelligence to genetic

inheritance very touchy. Is intelligence largely inherited? Can one raise
intelligence by nurture? How does one separate native ability and drive
from improvements in skill acquired through education? (Bell 1 973: 4 1 1 )

For justice as fairness, the debate over the determinants ofintelligence

and the extensive social scientific literature it has produced are more or
less beside the poin t. Once one rej ects the notions of
individual desert and 'fair opportunity' as the primary bases of distribu­
tive shares, the distinction between genetic and cultural obstacles to

73
success loses much of its moral interest. Once we agree to regard the
Possession, desert, and distributive justice

distribution of talents as a common asset, it matters little how some

came to reside in you and others in me.
Defenders of the meritocratic conceptions are not always explicit
on the grounds for their distinction between natural and social
advantages, but one can imagine at least two possible arguments,
one moral and the other practical . The first would argue that genetic
endowments are inviolable in a way that social or cultural character­
istics are not, that a person's natural endowments are somehow more
essentially his, more deeply constitutive of his identity than his
socially-conditioned attributes. I nnate differences, however arbi­
trary, are not dispensable in the same way; they, and not the results
of social and cultural conditioning, mark out the traits in the absence
of which I would not be the particular person I am. On this view,
whether I deserve the intelligence with which I am born, for
example, is not the point; what matters is that my native intelligence
is a fact irreducibly about me, and the integrity of my person requires
that it not be tampered with, no matter how worthy the wider social
purpose.
But this argument seems flawed once it appears that those qual­
ities most plausibly regarded as essential to a person's identity -
one's character, values, core convictions, and deepest loyalties, for
example - are often heavily influenced by social and cultural factors ,
while many natural features - such as hair color and other trivial
physical characteristics - are more readily dispensable. Even if some
distinction between essential and merely accidental characteristics
of the person is valid, there seems no obvious reason why it must
correspond to the distinction between natural and social assets . Of
course, Rawls' obj ection to the argument would be stronger still, for
his theory of the person implies that no characteristic, whether social
or natural , can be essential in this way. Even those attributes, such as
a person 's character and values , that intuitively seem closest to
defining an essential self, are relegated to contingent status. As a
person's character 'depends in large part upon fortunate family and
social circumstances for which he can claim no credit' ( 1 04) , so our
values are accidental as well . 'That we have one conception of the
good rather than another is not relevant from a moral standpoint . In
acquiring it we are influenced by the same sort of contingencies that
lead us to rule out a knowledge of our sex and class' ( Rawls 1 9 75 :
537) .

74
Meritocracy versus the difference principle

The practical argument would distinguish natural and social

inequalities on the grounds that the first are intractable in a way that
the second are not, and that society can therefore be held responsible
for remedying social inequalities but not natural ones ; the more
inequality turns out to be genetic rather than culturally-induced, the
less society can 'do about it' . Given a fair system, some will advance
more successfully than others, and there comes a point when even
the most enlightened society can do nothing to alter this fact. At
some point, even the most dogged reformers must acknowledge that
life is unfair in a way that no social institutions can hope to set right.
People are different, and these differences, sooner or later, will
inevitably come to the fore, even - perhaps most assuredly - in a
society where fair opportunity prevails. 'What is important is that
the society, to the fullest possible extent, be a genuinely open one'
( Bell 1 9 7 3 : 454) .
To this Rawls would likely respond that society's role is naturally
limited in this way only if one assumes that its only role in promoting
justice consists in its efforts to even out the disadvantages of the least
fortunate so that they may compete more fairly. But this assumption
overlooks the equally significant social choice implicit in the aims
institutions pursue and the attributes they reward in the process .
Even if the vast maj ority o fdifferences between persons turned o u t to
be genetic rather than cultural, it would still remain for a society to
determine which of these differences, if any, should properly be made
the basis for differential distributive shares . To be sure, if the aim of
social institutions is taken to be fixed , as maximizing the overall
social product, for example, then the defenders of a 'fair meritocracy'
are right, that the only remaining question of j ustice is how well
people are equipped by the society to contribute to that purpose and
to reap the benefits of their contribution . But the question must
immediately arise as to why that aim should be primary, even ifit can
prevail only at the admitted expense of social inj ustice . In short, a
society not only equips or fails to equip its members with the endow­
ments relevant to its collective purpose, but also defines through its
institutional arrangements the nature of that purpose and , deriva­
tively, the attributes to be prized and installed as the basis of
distributive shares . In a passage of some eloquence, Rawls writes :

I n view of these remarks we may rej ect the contention that the ordering of

75
Possession, desert, and distributive justice

institutions is always defective because the distribution of natural talents

and the contingencies of social circumstances are unj ust, and this inj us tice
must inevitably carry over to human arrangements . Occasionally this
reflection is offered as an excuse for ignoring injustice, as if the refusal to
acquiesce in inj ustice is on a par with being unable to accept death . The
natural distribution is neither just nor unj ust; nor is it unjust that men are
born into society at some particular position . These are simply natural
facts. What is j ust and unj ust is the way that institutions deal with these
facts ( 1 02 ) .

These considerations lead naturally to a third contrast between

the meritocratic conception and the democratic one, which concerns
the relation between the value of various assets and attributes on the
one hand and the institutions that prize and reward them on the
other. On the meritocratic conception, social institutions are bound
to reward certain attributes rather than others . The qualities that a
set of institutions calls forth have a worth antecedent to their institu­
tional valuation, and so provide an independent test of the j ustice of
the institutions themselves . Institutional arrangements that put a
premium on noble qualities rather than base ones are, in virtue of
that, the worthier, quite apart from other considerations relevant to
justice, such as the purposes they advance.
On Rawls' view, institutions are not constrained in this way, for
the virtues that would constrain them must themselves await institu­
tional definition. 'The concept ofmoral worth does not provide a first
principle of distributive j ustice ' , because it cannot be introduced
until after the principles ofj ustice are already on hand (3 1 2 ) . Since
no virtue has antecedent, or pre-institutional moral status , the
design of institutions is open with respect to the qualities it may
prize. As a result, the intrinsic worth of the attributes a society elicits
and rewards cannot provide a measure for assessing its j ustice, for
their worth only appears in the light of institutional arrangements to
begin with . Rawls ' rej ection of pre-institutional notions of virtue
reflects the priority of the right over the good and the refusal to
choose in advance between competing conceptions of the good . A
hunting society that rewards fleet-footedness over loquaciousness
( as prized in a litigious society, for example) is not less j ust or worthy
in virtue of that, for there are no antecedent grounds on which
swiftness of foot could be shown as more or less virtuous than
swiftness of tongue. The priority ofjust institutions wi th respect to
Defending common assets

virtue and moral worth provides a second reason why I cannot be

said to deserve the benefits flowing from my natural attributes . I n
order for me t o deserve the benefits associated with ' my' superior
intelligence, say, it is necessary both that I possess my intelligence
(in some non-arbitrary sense of possession) , and that I have a right
(in a strong, pre-institutional sense of right) that society value
intelligence rather than something else. But on Rawls' account,
neither condition holds. The argument from arbitrariness to com­
mon assets undermines the first, and the precedence of institutions
over moral worth denies the second.

DEFENDING C O M M O N ASSETS

In developing our critique of Rawls' theory of the person and the

difference principle, each in the light of the other, we shall take as our
points of departure two strands of Nozick's critique of justice as
fairness . The first attacks the difference principle and in particular
the notion of possession on which it relies, and the second defends a
version of natural liberty by fixing on the notions of desert and
entitlement. Playing Nozick off against Rawls in this way will enable
us to assess Rawls ' conception and also to clarify some of the
similarities and differences in their respective distributive theories .
Nozick's central obj ection to the argument producing the differ­
ence principle fixes on Rawls' view that the distribution of natural
talents is best regarded as a ' common' or 'collective' possession to be
shared across society as a whole. As Rawls writes ,
The difference principle represents, in effect, an agreement to regard the

of this distribution whatever it turns out to be ( 1 0 I ) .

distribution of natural talents as a common asset and to share in the benefits

The two principles are equivalent, as I have remarked , to an undertaking to

regard the distribution of natural abilities as a collective asset so that the

( I 79) .
more fortunate are to benefit only in ways that help those who have lost out

Rawls believes the notion of common assets as embodied in the

difference principle expresses the ideal of mutual respect deonto­
logical liberalism seeks to affirm .
By arranging inequalities for reciprocal advantage and by abstaining from

77
the exploitation of the contingencies of natural and social circumstance
Possession, desert, and distributive justice

within a framework of equal liberty, persons express their respect for one
another in the very constitution of their society . . . . Another way of putting
this is to say that the principles ofjustice manifest in the basic structure of
society men's desire to treat one another not as means only but as ends in
themselves ( 1 79) .

Nozick argues, to the contrary, that to regard people's natural

assets as common property is precisely to contradict all that deonto­
logical liberalism affirms in emphasizing the inviolability of the
individual and the distinction between persons .

People will differ in how they view regarding natural talents as a common
asset. Some will complain, echoing Rawls against utilitarianism, that this
'does not take seriously the distinction between persons' ; and they will
wonder whether any reconstruction of Kant that treats people's abilities
and talents as resources for others can be adequate. 'The two principles of
j ustice . . . rule out even the tendency to regard men as means to one
another's welfare . ' Only if one presses very hard on the distinction between
men and their talents, assets, abilities and special traits ( 1 974: 228) .

Here Nozick goes to the heart of Rawls' theory of the subj ect. For
as we have seen, Rawls presses very hard indeed on the distinction
between the self and its various possessions. The severity of this
distinction, problematic though it may be, is carefully fashioned to
the requirements of the deontological proj ect as a whole; it allows for
the priority of the self over its ends, which supports in turn the
priority of the right and the primacy ofj ustice . A further feature of
this conception is that it allows for the following defense against
Nozick's obj ection to the difference principle.
Regarding the distribution of natural talents as a common asset
does not violate the difference between persons nor regard some as
means to others' welfare, because not persons but only ' their' attributes
are being used as means to others' well-being. To say that I am
somehow violated or abused when ' my' intelligence or even effort is
used for the common benefit is to confuse the selfwith its contingent­
ly-given and wholly inessential attributes ( inessential, that is, to me
being the particular self ! am) . Only on a theory of the person that
held these endowments to be essential constituents rather than alien­
able attributes of the self could the sharing of assets be viewed as using
me as a means to others' end s . But on Rawls ' account all endowments
are contingent and in principle detachable from the self, whose
Defending common assets

p riority is assured by its ability constantly to recede before the swirl

of circumstance. This is the feature that preserves its identity, by
assuring its invulnerability to transformation by experience.
While this defense evades the inconsistency, it quickly invites a
related obj ection of incoherence, for if Rawls must invoke the distinc­
tion between the self and its possessions in this thoroughgoing way,
the question immediately arises whether, in avoiding a radically
situated subj ect, Rawls does not lapse into the opposite extreme of a
radically disembodied one. As Nozick presses the obj ection ,
'Whether any coherent conception of a person remains when the
distinction is so pressed is an open question. Why we, thick with

within us are not regarded as means is also unclear' ( 1 9 7 4: 2 2 8 ) .

p articular traits, should be cheered that (only) the thus purified men

Nozick thus anticipates Rawls' defense and shows it to be too

ingenious to redeem the theory. The notion that only my assets are
being used as a means, not me, threatens to undermine the plausibil­
ity, even the coherence, of the very distinction it invokes . It suggests
that on the difference principle, we can take seriously the distinction
between persons only by taking metaphysically the distinction be­
tween a person and his attributes . But this has the consequence of
leaving us with a subj ect so shorn of empirically-identifiable charac­
teristics (so 'purified' , in Nozick's word ) , as to resemble after all the
Kantian transcendent or disembodied subj ect Rawls set out to
avoid . It would seem that Rawls escapes the charge of inconsistency
only at the price of incoherence, and that Nozick's obj ection to the
difference principle therefore succeeds .
But Rawls has an alternate defense available, this one unantici­
pated by Nozick. Although it rescues the difference principle from
reliance on an apparently disembodied conception of the subj ect, it
comes at some expense to other asp icts of Rawls' doctrine, and so
would likely be resisted by Rawls himself. Still, I shall try to show
that only by somehow assuming it can the difference principle be
sustained . On this second defense, Rawls might deny that the differ­
ence principle uses me as a means to others' ends , not by claiming
t hat my assets rather than person are being used, but instead by
questioning the sense in which those who share in 'my' assets are
properly described as 'others ' . Where the first defense presses the
distinction between the self and its attributes, the second qualifies
the distinction between the self and the other by allowing that, in

79
Possession, desert, and distributive justice

certain moral circu mstances , the relevant description of the self may
em brace more than a single empirically-individuated human being .
The second defense ties the notion of common assets to the possibil­
ity of a common subject of possession . I t appeals , in short, to an
i ntersubj ective conception of the self.
That the difference principle commits Rawls to an intersubjective
conception he otherwise rejects seems the only way out of the
difficulties Nozick raises . It also serves to highlight an unargued
assumption in Rawls' theory of the subject . As we have seen , Rawls
conceives the self as a subj ect of possession , bounded in advance, and
given prior to its end s, and he assumes fu rthermore that the bounds
of the subj ect unproblematically correspond to the bod ily bounds
between individual human beings . But this claim is never defended
by Rawls , only assumed , its contestability hidden perhaps by its
affinity with our unreflective common-sense view of the matter!
Rawls' emphasis on plurality as an essential feature ofhuman society
may be thought to lend some support to the assumption , but this can
establish only that some principle of plurality or differentiation is
ess � ntial to an account of the human subject, not necessarily a
physical bodily one; nor can it show that the number of this plurality
must correspond in all cases to the number of individual human
beings in the world .
This i n any case is the assumption that must give way if Nozick's
obj ection to common assets is to be overcome. If the difference
principle is to avoid using some as means to others ' ends, it can only
be possi ble under circumstances where the subject of possession is a
'we' rather than an ' 1 ' , which circumstances imply in turn the
existence of a com munity in the constitutive sense.
The concl usion that Rawls' theory implicitly relies on an inter­
subj ective conception he officiii!Jly rejects finds further support in the
d iscussions of desert and j ust m-cation to follow, and is hin ted at by
various traces ofintersubjective language to be found throughout the
text. Such language first appears in the discussion of the difference
pri nciple when, as we have seen , the distribution of natural talents is
described al ternately as a 'common' or 'collective' or 'social ' asset to
be used for the 'common advan tage' ( 1 0 1, 1 79 , 1 0 7 ) . ' I n justice as
2 Still, Rawls' critique of utilitarianism makes especially perplexing his failure to defend his
principle of individuation more com pletely, for it is precisely on this issue that he claims
utilitarianism errs, by failing to recognize, or at any rate failing to take seriously, the
distinction between persons.

Bo
Defending common assets

fa irness men agree to share one another's fate' ( 1 02 ) . They resolve,

that is, not to take seriously the differences between persons as the
basis for differential life prospects, on the grounds that these differ­
ences arise from factors arbitrary from a moral point of view.
In his discussion of the idea of social union, Rawls carries his
intersubj ective language from common assets to common ends and
purposes, and in rhetoric that comes perilously close to the teleo­
logical, speaks of human beings realizing their common nature as
well. In his account of social union, Rawls shifts from distributive
issues to a concern for self-realization, and seeks to show that j ustice
as fairness can provide an interpretation of human sociability that is
neither trivial nor purely instrumental. 'Thus human beings have in
fact shared final ends and they value their common institutions and
activities as goods in themselves' (52 2 ) . The characteristics of social
union are 'shared final ends and common activities valued for them­
selves' (525 ) . Following Humboldt, a nineteenth century liberal
in the German idealist tradition, Rawls writes that 'it is through
social union founded upon the needs and potentialities of its mem­
bers that each person can participate in the total sum of the realized
natural assets of the others . . . . Onry in the social union is the individual
complete' [emphasis added] (523, 525n) . Social unions come in a
variety of shapes and sizes; 'they range from families and friendships
to much larger associations. Nor are there limits of time and space,
for those widely separated by history and circumstance can never­

(52 7) .
theless co-operate in realizing their common nature' [emphasis added]

By their intersubj ective dimensions, the difference principle and

the idea of social union counter individualistic assumptions in two
different ways, the difference principle by nullifying the arbitrariness
that arises when natural assets are seen as individual possessions, the
idea of social union by overcoming the partiality of persons that
a ppears when individuals are thought to be complete in themselves .
I n a social union, ' the members of a community participate in one
another's nature . . . [and] the selfis realized in the activities ofmany selves'
[emphasis added] (565) .
I t is a feature of human sociability that we are by ourselves but parts of what
we might be. We must look to others to attain the excellences that we must
leave aside, or lack altogether. The collective activity of society, the many
associations and the public life of the largest community that regulates
Possession, desert, and distributive justice

them, sustains our efforts and elicits our contribution. Yet the good attained
from the common culture far exceeds our work in the sense that we cease to he
merefragments: that part ofourselves that we directly realize is j oined to a wider
and j ust arrangement the aims of which we affirm [emphasis added] (529) .

T H E B A S I S OF D E S E R T

The notion of possession leads naturally to claims of desert and

entitlement. The argument over what people possess, and on what
terms, has a direct bearing on the question of what people deserve or
are entitled to as a matter ofj ustice. I t is to the issues of desert and
entitlement that we now turn , to consider the second strand of
Nozick's critique ofj ustice as fairness . Rawls rej ects the principles of
natural liberty and liberal equality on the grounds that they reward
assets and attributes which, being arbitrary from a moral point of
view, people cannot properly be said to deserve, and adopts the
difference principle on the grounds that it nullifies this arbitrariness .
Nozick attacks this line of reasoning by arguing first that arbitrari­
ness does not undermine desert, and second that, even if it did, a
version of natural liberty and not the difference principle would
emerge as the preferred result.
Stated in terms of possession, Rawls' objection to natural liberty
and liberal equality is that under these principles , persons are
allowed unfairly to benefit (or suffer) from natural and social endow­
ments that do not properly belong to them, at least not in the strong,
constitutive sense of belonging. To be s ure, the various natural assets
with which I am born may be said to ' belong' to me in the weak,
contingent sense that they reside accidentally within me, but this
sense of ownership or possession cannot establish that I have any
special rights with respect to these assets or any privileged claim to
the fruits of their exercise. I n this attenuated sense of possession, I
am not really the owner but merely the guardian or repository of the
assorted assets and attributes located ' here ' . By failing to acknow­
ledge the arbitrariness of fortune, the principles of natural liberty
and liberal equality go wrong in assuming that ' my' assets belong to
me in the strong, constitutive sense, and so allowing distributive
shares to depend on them .
Expressed in terms of desert, Rawls' objection to the principles of
natural liberty and liberal equality is that they reward assets and

82
The basis of desert

attributes that people cannot properly be said to deserve. Though

some may think the fortunate deserve the things that lead to their
greater advantage, ' this view is surely incorrect' .

I t seems to b e one o f the fixed points o f our considered j udgments that no

one deserves his place in the distribution of native endowments, any more
than one deserves one's initial starting place in society . The assertion that a
man deserves the superior character that enables him to make the effort to
cultivate his abilities is equally problematic; for his character depends in

claim no credit. The notion of desert seems not to apply to these cases ( 1 04) .
large part upon fortunate family and social circumstances for which he can

Because no one deserves his good luck in the genetic lottery , or his
favored starting place in society, or for that matter the s uperior
character that motivates him to cultivate his abilities conscientious­
ly, no one can be said to deserve the benefits these assets produce. I t
i s this deduction that Nozick disputes . ' I t i s not true, ' h e argues, ' that
a person earns Y ( a right to keep a painting he's made, praise for
writing A Theory ofJustice, and so on) only if he's earned (or otherwise
deserves) whatever he used (including natural assets) in the process of
earning Y. Some of the things he uses he j ust may have, not illegiti­
mately. It needn't be that the foundations underlying desert are
themselves deserved, all the Wl!Y down' ( 1 9 74: 2 25 ) .
Now what are we to make of this claim? If I do not necessarily have
to deserve everything I use in producing a thing in order to deserve the
thing, what does my desert depend on? Nozick says that some of the
things I use I 'j ust may have, not illegitimately' ( and, presumably,
possibly arbitrarily) . Once again, the notion of possession enters the
scene. To see whether my having a thing, not illegitimately, can
enable me to deserve what it helps me produce, we must explore in
greater detail the relation between possession and desert, and sort
out once more the sense of possession being appealed to.
For this purpose, it may be helpful to consider a recent discussion
of j ustice and personal desert by joel Feinberg, who analyzes the
bases of desert with an admirable clarity in terms suggestive for the
arguments before us ( 1 9 70) . Feinberg begins with the observation
that no one can deserve anything unless there is some basis for the
des ert. 'Desert without a basis is simply not desert'. But the question
immediately arises what kind of basis is necessary . As Feinberg
writes, ' Not any old basis will do' . Once again, the notion of posses-
Possession, desert, and distributive justice

s10n provides the key. ' I f a person is deserving of some sort of

treatment, he must, necessarily, be so in virtue of some possessed
charact erist ic or prior activity' [emphasis added) ( 1 9 70: 48) .

A characteristic of mine cannot be a basis for a desert of yours unless it

somehow reveals or reflects some characteristic of yours . I n general, the
facts which constitute the basis of a subject's desert must be facts about that
subj ect. If a student deserves a high grade in a course, for example, his
desert must be in virtue of some fact about him - his earlier performances ,
say, or his present abilities . . . . I t is necessary that a person 's desert have a
basis and that the basis consist in some fact about himself ( 1 9 70:
58-g, 6 1 ) .

Feinberg's analysis, tying a person' s desert to some fact about the

person, would appear to support Nozick's claim that ' the founda­
tions underlying desert need n ' t themselves be deserved , all t he way
do wn' . I n fact, the reliance of desert on some possessed characteristic
of the person s uggests a thesis even stronger than Nozick's : that the
foundations underlying desert canno t themselves be deserved , all t he
way do wn, any more than the foundations underlying possession can
themselves be possessed , all t he way do wn. We have already seen how
the notion of possession requires that somewhere, 'down there' , there
must be a subj ect of possession that is not itself possessed ( for this
would deny its agency ) , a subj ect 'doing the possessing' , so to speak.
The analogy for desert must be a basis of desert ultimately prior to
desert. For consider: if desert presupposes some possessed character­
istic, and if possessed characteristics presuppose some subj ect of
possession which is not itself possessed, then desert must presuppose
some subj ect of possession which is not itself possessed , and therefore
some basis of d esert which is not itself deserved . ] ust as there must be
some subject of possession prior to possession, so there must be some
basis of desert prior to desert . This is why the question whether
someone deserves ( to have) his sterling character, for example, is
notoriously difficult ( for i t is unclear who or what is left to j udge once
his character has been removed ) , and why, beyond a certain point,
asking j ust wholesale whether someone deserves to be the ( kind of)
person he is becomes incoherent altogether. Somewhere, 'down
there ' , there must be a basis of desert that is not itself deserved . The
foundations underlying desert cannot themselves be deserved, all the
way down .
The basis of desert

This result would seem amply to confirm Nozick's claim against

Rawls that I do not necessarily have to deserve everything I use in
producing a thing in order to deserve the thing, that some of what I
use I 'j ust may have, not illegitimately ' . And if this claim can be
established , then it would appear that Rawls' argument from arbi­
trariness fails to undermine desert after all. To say, as Rawls does,
that I do not deserve the superior character that led me to realize my
abilities is no longer enough. To deny my desert, he must show that I
do not have the requisite character, or alternatively, that I have it, but
not in the requisite sense.
But this is precisely the argument Rawls' theory of the person
allows him to make. For given his sharp distinction between the self,
taken as the pure subj ect of possession, and the aims and attributes it
possesses, the self is left bare of any substantive feature or character­
istic that could qualify as a desert base. Given the distancing aspect
of possession, the self itself is dispossessed . On Rawls' theory of the
person, the self, strictly speaking, has nothing, nothing at least in the
strong, constitutive sense necessary to desert. In a move similar to
the one invoked to show that the difference principle does not use a
person as a means, only a person's attributes, Rawls can accept that
some undeserved desert base is necessary to desert, only to claim
that, on an adequate understanding of the person, this condition
could never in principle be met! On Rawls' conception, the charac­
teristics I possess do not attach to the selfbut are only related to the self,
standing always at a certain distance. This is what makes them
attributes rather than constituents of my person; they are mine rather
than me, things I have rather than am.
We can see in this light how Rawls' argument from arbitrariness
undermines desert not directly, by claiming I cannot deserve what is
arbitrarily given, but indirectly, by showing I cannot possess what is
arbitrarily given, that is, that ' I ' , qua subj ect of possession, cannot
possess it in the undistanced, constitutive sense necessary to provide
a desert base. An arbitrarily-given asset cannot be an essential
constituent but only an accidental attribute of my person, for other­
wise my identity would hang on a mere contingency, its continuity
constantly vulnerable to transformation by experience, my status as
a sovereign agent dependent on the conditions of my existence rather
than epistemologically guaranteed . On Rawls' conception, no one
can properly be said to deserve anything because no one can proper-

8s
Possession, desert, and distributive justice

ly be said to possess anything, at least not in the strong, constitutive

sense of possession necessary to the notion of desert.
A theory of j ustice without desert would seem a dramatic depar­
ture from traditional conceptions, but Rawls is at pains to show that
it is not. In his opening pages , Rawls acknowledges that his
approach 'may not seem to tally with tradition' , but seeks to reassure
that in fact it does .
The more specific sense that Aristotle gives to j ustice, and from which the
most familiar formulations derive, is that of refraining from pleonexia, that is,
from gaining some advantage for oneself by seizing what belongs to another, his
property, his reward, his office, and the like, or by denying a person that
which is due to him . . . . A ristotle 's definition clearly presupposes, however, an
account of what properly belongs to a person, and of what is due to him. Now such
entitlements are, I believe, very often derivedfrom social institutions and the legitimate
expectations to which they give rise. There is no reason to think that Aristotle
would disagree with this, and certainly he has a conception of social j ustice

notion [emphasis added] ( 1 0- I I ) .

to account for these claims . . . . There is no conflict with the tradi tional

I n comparing j ustice as fairness with traditional conceptions,

Rawls confirms its novelty rather than denies it. What he presents as
an incidental qualification to j ustice as classically conceived turns
out on inspection to signal a striking departure. As Rawls suggests,
traditional notions freely refer to 'what properly belongs to a person' ,
institutions, presumably, aside; they presuppose thickly-constituted
persons with a fixity of character, certain features of which are taken
to be essential, 'all the way down ' . On Rawls' conception, however,
none of these concepts is available. In so far as a theory of j ustice
'presupposes an account of what properly belongs to a person' (in the
strong sense of 'belongs' ) , Rawls effectively acknowledges that he
has none. Nor, he seems to imply, given the precedence of plurality,
the priority of right, and the theory of the person they require, is it
reasonable to think that such a theory ofj ustice could be true. We are
not essentially thick enough selves to bear rights and deserts antece­
dent to the institutions that define them . Given these constraints, the
only alternative is to opt for a theory ofj ustice based on entitlements
to legitimate expectations , ruling out desert altogether. Rawls
hedges this claim at first, saying only that 'such entitlements are, I
believe, very often derived from social institutions and the legitimate
expectations to which they give rise' [emphasis added] ( 1 0) . But as
86
The basis of desert

the full consequences of Rawls' view emerge, 'very often' becomes

' always ' , for it becomes clear that 'such entitlements' can arise in no
other way. While Aristotle might not disagree that entitlements can
arise in this way, it seems far from his view that they can arise in no
other way. In denying that j ustice has to do with giving people what
they deserve, j ustice as fairness departs decisively from the tradition­
al notion after all .
Rawls' apparent view that no one can properly be said to deserve
anything, and the connection of this view with the notion of the self as
'essentially unencumbered' , emerges more fully in his discussion of
legitimate expectations and moral desert. He begins by acknowledg­
ing that j ustice as fairness, in rejecting desert, runs counter to
common sense.

There is a tendency for common sense to suppose that income and wealth ,
and the good things in life generally, should be distributed according to
moral desert. Justice is happiness according to virtue. While it is recognized
that this ideal can never be fully carried out, it is the appropriate conception
of distributive justice, at least as a prima facie principle, and society should
try to realize it as circumstances permit. Now j ustice as fairness rej ects this
conception. S uch a principle would not be chosen in the original position.
There seems to be no way of defining the requisite criterion in that situation
(3 1 0- 1 1 ) .

There seems to be no way of defining the requisite criterion of a

person's virtue or moral worth in the original position because no
substantive theory of the person antecedent to social institutions
exists . For moral desert to provide an independent criterion of
j ustice, there must be some substantive theory of the person, or of the
worth of persons , to get it going. But for Rawls, the worth of persons
is subsequent to institutions, not independent of them . And so a
person's moral claims must await their arrival .
This leads to the distinction between moral desert and legitimate
expectations. Once a person does the various things established
institutions encourage him to do, he acquires certain rights, but not
before . He is entitled that institutions honor the claims they
announce they will reward , but he is not entitled that they undertake
to reward any particular kind of claim in the first place.
A j ust scheme, then, answers to what men are entitled to; it satisfies their
legitimate expectations as founded upon social institutions. But what they
Possession, desert, and distributive justice

are entitled to is not proportional to nor dependent upon their intrinsic

worth . The principles ofj ustice that regulate the basic structure and specify
the duties and obligations of individuals do not mention moral desert, and
there is no tendency for distributive shares to correspond to it ( 3 1 1 ) .

The principles of j ustice d o not mention moral desert because,

strictly speaking, no one can be said to deserve anything. Similarly,
the reason people's entitlements are not proportional to nor depen­
dent upon their intrinsic worth is that, on Rawls' view, people have no
intrinsic worth, no worth that is intrinsic in the sense that it is theirs
prior to or independent of or apart from what j ust institutions
attribute to them.
The essential point is that the concept of moral worth does not provide a
first principle of distributive justice. This is because it cannot be introduced
until after the principles ofj ustice and of natural duty and obligation have
been acknowledged . . . . [T] he concept of moral worth is secondary to those
of right and j ustice, and it plays no role in the substantive definition of
distributive shares (3 1 2- 1 3 ) .

Rawls could agree with Feinberg that 'desert i s a moral concept in

the sense that it is logically prior to and independent of public
institutions and their rules ' , but would deny that there is any ' antece­
dent standard for its definition' ( 3 1 3) , and so disagree with Feinberg
that 'one of the aims of [a system of public bestowals] is to give
people what they deserve' ( 1 970: 86) . For Rawls, the principles of
j ustice aim neither at rewarding virtue nor at giving people what
they deserve, but instead at calling forth the resources and talents
necessary to serve the common interest.
None of the precepts of j ustice aims at rewarding virtue. The premiums
earned by scarce natural talents, for example, are to cover the costs of
training and to encourage the efforts of learning, as well as to direct ability
to where it best furthers the common interest. The distributive shares that
result do not correlate with moral worth (3 1 1 ) .

To illustrate the priority ofj ust institutions with respect to virtue

and moral worth, Rawls suggests an analogy to the relation between
the rules of property and the law of robbery and theft.

These offenses and the demerits they entail presuppose the institution of
property which is established for prior and independent social ends. For a
society to organize itself with the aim of rewarding moral desert as a first

88
The basis of desert

principle would be like having the institution of property in order to punish

thieves. The criterion to each according to his virtue would not, then, be
chosen in the original position (3 I 3 ) .

The analogy i s intriguing, but one wonders whether it works

entirely to Rawls' advantage. While it is apparent that the institution
of property has a certain priority with respect to its correlative
offenses, it is less clear why the dependence must run only in one
direction, especially given Rawls' own commitment to the method of
reflective equilibrium. For example, is our belief in the validity of the
institution of property in no way enhanced by a conviction that
robbery and theft are wrong? Would our confidence in the institution
of property in no way be diminished if it turned out that those it
defined as robbers and thieves were invariably good and virtuous
men? And what of more extreme cases? While the norms and rules
protecting human life can no doubt be defended on a variety of
grounds, such as keeping people alive, avoiding suffering, and so on,
is it logically mistaken to think that one j ustification of prohibitions
against murder could be to punish murderers?
Rawls' position here appears especially perplexing in the light of a
contrast he draws between distributive j ustice and retributive j us­
tice, suggesting that in the second case, some notion of moral desert
may be appropriate after all. The view that distributive shares
should match moral worth to the extent possible, writes Rawls, 'may
arise from thinking of distributive j ustice as somehow the opposite of
retributive j ustice'. But the analogy is mistaken. In a reasonably
well-ordered society, 'Those who are punished for violatingjust laws
have normally done something wrong. This is because the purpose of
the criminal law is to uphold basic natural duties . . . and punish­
ments are to serve this end ' .
They are not simply a scheme o f taxes and burdens designed t o put a price
on certain forms of conduct and in this way to guide men's conduct for
mutual advantage. I t would be far better if the acts prescribed by penal
statutes were never done. Thus a propensity to commit such acts is a mark of bad
character, and in a j ust society legal punishments will only fall upon those
who display these faults.
It is clear that the distribution of economic and social advantages is
entirely different. These arrangements are not the converse, so to speak, of
the criminal law so that just as the one punishes certain offenses, the other
rewards moral worth. The function of unequal distributive shares is to

8g
Possession, desert, and distributive justice

cover the costs of training and education, to attract individuals to places

and associations where they are most needed from a social point of view,
and so on . . . . To think of distributive and retributive justice as converses of one
another is completely misleading and suggests a moral basis of distributive shares where
none exists [emphasis added) ( 3 1 4 - 1 5 ) .
Unlike the benefits that flow from distributive arrangements , the
punishments and prohibitions associated with the criminal law are
not simply a non- moral system of incen tives and deterrents designed
to encourage some forms of behavior and discourage others . For
Rawls, the pre-institutional moral notions excluded in distributive
j ustice somehow find meaning for retributive purposes , and there is a
tendency for punishment to correspond to them .
The immediate puzzle is how this account can possibly fit with the
analogy of property and theft . If retributive j ustice differs from
distributive justice precisely in virtue of its prior moral basis , it is
difficult to see how the example of property and theft could dem­
onstrate the priority of social institutions with respect to virtue and
moral worth, if this priority holds for distributive j ustice alone. This
relatively minor confusion aside, the more basic question is how
Rawls can admit desert in retributive j ustice without contradicting
the theory of the self and related assumptions that ruled it out for
purposes of distributive j ustice . If such notions as pre-institutional
moral claims and intrinsic moral worth are excluded from a theory of
distributive j ustice in virtue of an essentially unencumbered self too
slender to support them , it is difficult to see how retri butive justice
could differ in any relevant way . 3
D o not the same arguments from arbitrariness exclude desert as a
basis for punishment as for distributive shares? Is the propensity to
commit crimes, any less than the propensity to do good, the result of
factors arbitrary from a moral point of view? And if not, why would the
parties to the original position not agree to share one another's fate for
the purpose of criminal liability as well as distributive arrangements?
Since under the veil of ignorance, none can know whether he shall have
the misfortune to be born into the unfavorable social and family
3 In a footnote, Rawls ( 3 1 5 ) cites Feinberg in apparent support of this claim, but Feinberg
allows a role for desert in both distributive and retributive j ustice. Feinberg's point is that
retributive j ustice involves what he calls polar desert (where one either deserves good or
deserves ill ) , whereas distributive j ustice involves nonpolar desert (where, as with a prize,
some deserve and others do not) . But both cases involve desert in the moral, pre­
institutional sense ( Feinberg 1 970: 62).

go
The basis of desert
circumstances that lead to a life of crime, why would the parties not
adopt a kind of difference principle for punishments as well as distribu­
tive shares, and agree, in effect, to regard the distribution of natural and
social liabilities as a common burden?
Rawls holds that ' those who are punished for violating j ust laws
have normally done something wrong' , and so deserve their punish­
ment ( 3 1 4) . But suppose, by an act of vandalism, I deprive the
community of a certain measure of well-being, say by throwing a
brick through a window. I s there any reason why I deserve to bear
the full costs of my destructiveness any more than the person who
produced the window deserves to enjoy the full benefits of his produc­
tiveness? Rawls may reply that my 'propensity to commit such acts is
a mark of bad character' . But if the worker's industriousness in
making the window is not a mark of good character ( in the moral,
pre-institu tional sense) , why is my maliciousness in breaking the
window a mark of bad character (in the moral, pre-institu tional
sense) ? To be sure ( following Rawls, p. 1 03 ) , given a j ust system of
criminal law, those who have done what the system announces it will
punish are properly dealt with accordingly and in this sense are
'deserving' of their penalty . 'But this sense of desert presupposes the
existence of the [retributive) scheme; it is irrelevant to the question
whether in the first place the scheme is to be designed in accordance
with the difference principle or some other criterion ' ( 1 03 ) .
Some may think that the criminal deserves his punishment in the
strong moral sense because he deserves the low character his crimi­
nality reflects . Perhaps this is what Rawls has in mind when he
writes that ' propensity to commit such acts is a mark of bad
character' , and punishments properly fall on those who display these
faults . Because the transgressor is less worthy in this sense, he
deserves the misfortune that befalls him. But again (following Rawls,
p . 1 04) , this view is surely incorrect. I t seems to be one of the fixed
points of our considered j udgments that no one deserves his place in
the distribution of native endowments or liabilities, any more than
one deserves one's initial starting place in society . The assertion that
a man deserves the inferior character that prevents him from over­

coming his liabilities is equally problematic; for his character

depends in large part upon unfortunate family and social circum­
stances for which he cannot be blamed . The notion of desert seems
not to apply to these cases . None of which is to say that, generally
Possession, desert, and distributive justice

speaking, a non-moral theory of distributive j ustice is incompatible

with a moral, or desert-based theory of punishment, only that given
Rawls' reasons for rejecting desert-based distributive arrangements,
he seems clearly committed to rejecting desert-based retributive
ones as well.
The apparent inconsistency between Rawls' retributive and dis­
tributive theories need not do serious damage to the theory as a
whole. Given the method of reflective equilibrium, 'justification is a

fitting together into one coherent view' ( 2 1 ) . From the standpoint of

matter of the mutual support of many considerations, of everything

the overall theory, little hangs on Rawls' retributive theory, apart

from the measure of plausibility it lends j ustice as fairness for those
committed to a strong, desert-based notion of punishment. IfRawls'
distinction succeeds, they need not choose between their retributive
intuitions and the difference principle; if it does not, one or the other
of those convictions must give way. If, on reflection, a non-moral
theory of punishment appears unacceptable, even in the light of the
arbitrariness of criminal characteristics and dispositions , then the
difference principle - rejecting as it does the notion of desert - would
be called into serious question. If, on the other hand, our intuition
that criminals deserve punishment proves no more indispensable
than our intuition that virtue deserves reward ( an intuition of
common sense Rawls explicitly rejects ) , then we may adj ust our
intuitions in a direction that affirms the difference principle rather
than opposes it. Desert would be rejected as the basis for both
distributive and retributive arrangements, and so the inconsistency
resolved .
But such a resolution returns us to the larger difficulties of a theory
of j ustice without desert and a notion of the self as essentially
dispossessed, or barren of constituent traits . Nozick argues against
Rawls that the foundations underlying desert need not themselves be
deserved, all the way down. But as we have seen, Rawls' denial of
desert does not depend on the thesis Nozick refutes, but instead on
the notion ofthe self as a pure, unadulterated , 'essentially unencum­
bered ' subj ect of possession . Rawls is not committed to the view that
a person can only deserve a thing he produces if he deserves every­
thing he used in producing it, but rather to the view that no one
possesses anything in the strong, constitutive sense necessary to a
desert base. No one can be said to deserve anything (in the strong,
The basis of desert

pre-institutional sense) , because no one can be said to possess any­

thing (in the strong, constitutive sense) . This is the philosophical
force of the argument from arbitrariness .

That the argument from arbitrariness works in this way can be seen
by viewing the moves from natural liberty to fair opportunity to the
democratic conception, as traced by Rawls, as stages in the dispos­
session of the person. With each transition, a substantive self, thick
with particular traits, is progressively shorn of characteristics once
taken to be essential to its identity; as more of its features are seen to
be arbitrarily given, they are relegated from presumed constituents
to mere attributes of the self. More becomes mine, and less remains
me, to recall our earlier formulation, until ultimately the self is
purged of empirical constituents altogether, and transformed into a
condition of agency standing beyond the objects of its possession .
The logic of Rawls' argument might be reconstructed as follows:
At the far end of the spectrum, even before natural liberty appears,
are aristocratic and caste societies; in such societies, a person's life
prospects are tied to a hierarchy into which he is born and from
which his person is inseparable. Here, the self is most fully ascribed,
merged almost indistinguishably with its condition, embedded in its
situation. The system of natural liberty removes fixed status of birth
as an assumed constituent of the person, and regards each as free,
given his capacities and resources, to compete in the marketplace as
best he can, and to reap his reward . By shifting the basis of expecta­
tions from status to contract, the system of natural liberty repairs the
arbitrariness of hierarchical societies by taking the person more
narrowly, so to speak, as distinct and separable from his surround­
ings . Still, some arbitrariness remains, most notably in the form of
social and cultural contingencies. In the regime of natural liberty, a
person's life prospects are gflverned by factors no more ascribable to
the person (in the strong, constitutive sense) than his inherited
status . Having relieved the person of his hierarchical baggage, the
principle of natural liberty still conceives a thickly-constituted self,
burdened by the accidents of social and cultural contingency. And so
the move to fair opportunity, which strips the self of social and
cultural accidents as well as inherited status. In a 'fair meritocracy',
the effects of class status and cultural disadvantage are understood to
reflect more on the society and less on the person. Those with

93
Possession, desert, and distributive justice

comparable talents and 'the same willingness to use them, should

have the same prospects of success regardless of their initial place in
the social system, that is, irrespective of the income class into which
they are born' ( 73) . In this way, the meritocratic conception extends
the logic of natural liberty by ascribing less to the self and more to its
situation.
But even the principle offair opportunity, in rewarding individual
effort, conceives the province of the self too expansively. For even
'the effort a person is willing to make is influenced by his natural
abilities and skills and the alternatives open to him. The better
endowed are more likely, other things equal, to strive conscientious­
ly, and there seems to be no way to discount for their greater good
fortune' ( 3 1 2 ) . The self is still over-ascribed . Given its arbitrariness,
even the character that determines a person's motivation cannot
properly be regarded as an essential constituent of his identity. And
so finally the move to the democratic conception, in which the self,
shorn of all contingently-given attributes, assumes a kind of supra­
empirical status, essentially unencumbered, bounded in advance
and given prior to its ends, a pure subject of agency and possession,
ultimately thin. Not only my character but even my values and
deepest convictions are relegated to the contingent, as features of my
condition rather than as constituents of my person . 'That we have
one conception of the good rather than another is not relevant from a
moral standpoint. In acquiring it we are influenced by the same sort
of contingencies that lead us to rule out a knowledge of our sex and
class' ( Rawls 1 975: 53 7) . Only in this way is it possible to install the
self as invulnerable, to assure its sovereignty once and for all in a
world threatening always to engulf it. Only if the fate of the self is
thus detached from the fate of its attributes and aims, subject as they
are to the vagaries of circumstance, can its priority be preserved and
its agency guaranteed .
This is the vision of the person that Nozick and Bell, as defenders
of natural liberty and meritocracy, respectively, emphatically rej ect,
even if they do not spell out in any detail the conception of the self
they rely on instead . Both object that the argument from arbitrari­
ness, consistently applied , leads ineluctably to the dissolution of the
person, and the abnegation of individual responsibility and moral
choice. 'This line of argument can succeed in blocking the introdu c­
tion of a person's autonomous choices and activities ( and their

94
The basis of desert

results) only by attributing everything noteworthy about the person

completely to certain sorts of "external" factors' , writes Nozick.
Echoing his argument against the notion of common assets, Nozick
questions whether, on Rawls' account, any coherent conception of
the person remains, and if so, whether it is any longer the kind of
person worth the moral fuss deontological liberalism makes on its
behalf.
So denigrating a person's autonomy and prime responsibility for his actions
i s a risky line to take for a theory that otherwise wishes to buttress the
dignity and self-respect of autonomous beings; especially for a theory that
founds so much (including a theory of the good) upon a person's choices .
One doubts that the unexalted picture of human beings Rawls' theory
presupposes and rests upon can be made to fit together with the view of
human dignity it is designed to lead to and embody ( 1 974: 2 1 4) .

Bell summarizes the objection in an epigram : 'The person has

disappeared . Only attributes remain' ( 1 97 3 : 4 1 9) . Where Rawls
seeks to assure the autonomy of the self by disengaging it from the
world, his critics say he ends by dissolving the self in order to
preserve it.
To recapitulate our reconstructed version of the argument be­
tween Rawls and Nozick on the issue of desert: Nozick first argues
that the arbitrariness of assets does not undermine desert, because
desert may depend not only on things I deserve, but also on things I
j ust have, not illegitimately. Rawls' response is to invoke the distinc­
tion between the self and its possessions in the strongest version of
that distinction, and so to claim that, strictly speaking, there is
nothing that ' I ' , qua pure subject of possession, have - nothing that is
attached, rather than related, to me - nothing at least in the strong,
constitutive sense of possession necessary to a desert base. Nozick's
rej oinder is that this defense cannot succeed for long, for it has the
consequence of leaving us with a subject so shorn of empirically­
identifiable characteristics as to resemble once more the Kantian
transcendent or disembodied subj ect Rawls resolved to avoid . I t
makes the individual inviolable only by making him invisible, and
calls into question the dignity and autonomy this liberalism seeks
above all to secure.

95
Possession, desert, and distributive justice
INDIVIDUAL A N D S O C I A L C LAIM S : WHO OWNS WHAT ?
But Nozick has a further objection, independent of the first, that
goes something like this: even if Rawls is right that arbitrariness
undermines individual possession and hence individual desert, the
difference principle is not the inevitable result; something like the
principle of natural liberty - and here Nozick prefers to speak in
terms of his own version, which he calls the 'entitlement theory' -
could still be true. For even if no one deserved any of his natural
assets , it might still be that people were entitled to them, and to what
flows from them.
At best, this argument maintains, Rawls' case for the difference
principle is underdetermined by the argument from arbitrariness.
To show that individuals, as individuals, do not deserve or possess
'their' assets is not necessarily to show that society as a whole does
deserve or possess them. Simply because the attributes accidentally
located in me are not my assets, why must it follow, as Rawls seems to
think, that they are common assets, rather than nobody's assets? If
they cannot properly be said to belong to me, why assume automati­
cally that they belong to the community? Is their location in the
communiry 's province any less accidental, any less arbitrary from a
moral point of view? And if not, why not regard them as free-floating
assets, unattached in advance to a'V' subj ect of possession, whether
individual or social?
Here it is necessary to be more precise about the terms of relation
between the person and the endowments he bears . Three descrip­
tions seem to cover the possible cases; depending on the sense of
possession intended, I may be described as the owner, the guardian,
or the repository of the endowments I bear. In its strongest version,
the notion of ownership implies that I have absolute, unqualified,
exclusive rights with respect to my endowments, in its more moder­
ate versions that I have certain privileged claims with respect to
them, a bundle of rights, while not unlimited , at least more extensive
with respect to my assets than any bundle of rights anyone else may
have with respect to them . This is the sense of individual possession
against which Rawls' argument from arbitrariness is addressed and
which, if successful, it undermines .
If Rawls is right, and I cannot properly be described as the owner
of my assets and attributes , two alternate accounts remain. One is
the notion of guardianship, which denies individual ownership in

g6
Individual and social claims

favor of a more ultimate owner or subject of possession of which the

individual person is the agent. To say that I am the guardian of the
endowments I bear is to imply that they are owned by some other
subject, on whose behalf, or in whose name, or by whose grace I
cultivate and exercise them . This is a notion of possession reminis­
cent of the early Christian notion of property, in which man had
what he had as the guardian of assets belonging truly to God , and it is
a notion that fits with various communitarian notions of property as

well .•
In the third account, I am neither the owner nor the guardian but
rather the repository of the assets and attributes accidentally located
in my person . The notion of the individual as repository of endow­
ments does not presuppose some other subject of possession whose
endowments they ultimately are, but does without possession
altogether. In so far as I am the repository of natural assets, there
need be no further question to whom these assets ultimately belong;
their residing in me has no consequence for claims I , or others, may
have with respect to them.
Now in terms of these distinctions, Nozick claims in effect that
Rawls' argument from arbitrariness, even ifit succeeds in undermin­
ing individual possession and hence desert, warrants only the third
description ( that individuals are repositories of assets) and not the
second . But if this is all the argument from arbitrariness establishes,
then it does not lead to the difference principle. For the difference
principle must presuppose the second description - that I am the
guardian of assets to which the community as a whole has some prior
title or claim. If all arbitrariness means is that I am the repository of
assets which belong to no one in particular, then it cannot be
assumed that the community owns them any more than I do. I t
4 The Rev . Vernon Bartlet ( 1 9 1 5 : 97 -8) writes that 'the essential Christian attitude' held the
property rights of any individual to be 'purely relative. not only as compared with God ' s
absol ute rights as Prod ucer and Owner both of all things and of all persons. but also as
compared with the paramount h uman or derivative rights of Society as representing the
com mon weal . Of this. the individ ual's weal is only a dependent part, and should be
lim ited by the rights of all others to the conditions of personal wel l-being . . . . The resulting
practical principles, viz. the stewardship of property on behalf both of God and Society.
and the moral duty of fidelity in this relation as the condition of any correla tive rights of
private personal enjoyment, is too deeply em bedded in C hrist 's teaching. notably in the
parables, to need demonstration . ' Bartlet q uotes St. Pau l . I Corinthians iv. 7 1 : ' For who
maketh thee to excel? And what hast thou that thou hast not received ? But if thou didst
receive it , why dost thou glory as if thou hadst not received it?'

97
Possession, desert, and distributive justice

would be as though assets, and the benefits that flow from them, fell
like manna from heaven ( the image is Nozick's) , unmarked by prior
claims and unattached to any subject of possession whether indi­
vidual or social.
How then should they be distributed? On what basis should
competing claims to such spontaneously-generated bounty be asses­
sed? From the standpoint of desert, there would seem to be no
grounds on which to choose between letting the assets lie where they
fall, and trying to distribute them in some other way. Unless dis­
tribution under such assumptions is to be regarded as a stand-off, a
matter of moral indifference, people's entitlements must depend on
considerations other than notions of possession or desert. On this,
Rawls and Nozick seem prepared to agree. But what could these
considerations be? On this, they part ways. Each thinks he can
adduce considerations unrelated to desert to overcome the stand-off
in favor of his own conception .
For Nozick, the absence of desert creates a presumption in favor of
letting assets lie where they fall, at least once it is accepted that things
do not come into the world like manna from heaven but come into
being already held , attached to particular persons.

Since things come into being already held (or with agreements already
made about how they are to be held) , there is no need to search for s.o me
pattern for unheld holdings to fit . . . . The situation is not an appropriate
one for wondering, 'After all, what is to become of these things; what are we
to do with them. ' In the non-manna-from-heaven world in which things
have to be made or produced or transformed by people, there is no separate
process of distribution for a theory of distribution to be a theory of ( I 974:
2 1 9) .

Nozick goes on to argue that if a person has an asset to which no

one else is entitled , then, although he may not deserve the asset, he is
nonetheless entitled to it, and to whatever flows from it by a process
that does not violate anyone else's entitlements . 'Whether or no t
people's natural assets are arbitrary from a moral point of view, they
are entitled to them, and to what flows from them' ( 1 974: 2 26) .
For Rawls, on the other hand , the absence of individual desert
creates a pres urn ption in favor of regarding the distribution of talents
as a common asset. The lack of desert or a pre-institutional concept
of virtue means that institutions are unconstrained by antecedent

g8
Individual and social claims

moral claims in their pursuit of the primary virtue of social j ustice. In

this sense, the analogy of manna from heaven is apt. The array of
assets dealt by fortune is neither j ust nor unjust. 'These are simply
natural facts . What is j ust and unj ust is the way that institutions deal
with these facts' ( 1 02 ) . There is no reason to let assets and the
benefits that flow from them lie where they fall. This would be simply
to incorporate and affirm the arbitrariness of nature. The discovery
that virtue and entitlements await social institutions rather than
constrain them is a reason to pursue j ustice all the more insistently,
not a reason to freeze arbitrariness in place.
What, then, are we to make of these attempts to overcome the
apparent moral stand-off created by the presumed absence of desert?
Nozick is prepared to accept that people may not deserve their
natural assets, but claims they are entitled to them nonetheless . 'If
people have X , and their having X (whether or not they deserve to
have it) does not violate anyone else's ( Lockean) right or entitlement
to X , and Y flows from (arises out of, and so on) X by a process that
does not itself violate anyone's ( Lockean) rights or entitlements,
then the person is entitled to Y' ( 1 974: 225) . But he does not show
why this is so, nor is he clear on what precisely the difference between
desert and entitlement consists in. Rawls and Feinberg agree that
'desert is a moral concept in the sense that it is logically prior to and
independent of public institutions and their rules' ( Feinberg 1 970:
8 7 ) . Entitlements, by contrast, are claims that can only arise under
the rules or qualifying conditions of institutions already established ,
what Rawls describes as legitimate expectations founded on social
institutions . It is a consequence of this view that, for the purpose of
designing or assessing social institutions, people's entitlements,
being derivative from institutions, are without moral or critical force.
Assessing the j ustice of an institution in the light of what people were
entitled to would be like j udging the validity of a rule in the light of
claims arising under the rule; to recall our earlier discussion, it would
be appealing to a standard of appraisal thoroughly implicated in the
object of appraisal. It is for this reason that the concept of entitle­
ment cannot provide a first principle ofj ustice. As Rawls explains, it
'presupposes the existence of the co-operative scheme, and so is
irrelevant to the question whether in the first place the scheme is to
be designed in accordance with the difference principle or some other
criterion' ( 1 03 ) .

99
Possession, desert, and distributive justice

Nozick never comes to terms with this difficulty. In making his

argument, he explicitly adopts the language of entitlement rather
than desert, but does not acknowledge its lesser moral force. In
Nozick's usage, the concept of entitlement does the same work as
desert, but without its pre-institutional credentials ever being estab­
lished . He begins with the premise that 'people are entitled to their
natural assets ' , and proceeds to argue that they are entitled to the
benefits that flow from them ( 1 974: 22 5-6) . But he never says why
people are entitled to their assets in any sense of entitlement strong
enough to get the argument going.
At one point, Nozick seems to claim that people are entitled to the
things they 'j ust may have, not illegitimately' , their natural assets
presumably being among such things . But having something, not
illegitimately, is not the same as being entitled to it; it is simply
having it, in some unspecified sense of possession . While my intelli­
gence, or physical strength, or good health may be among the things
I have, not illegitimately, it does not follow that I am entitled to these
things, for entitlements depend, as we have seen, on some scheme of
co-operation already being in effect.
At other points, Nozick seems to argue instead that people are
entitled to their natural assets and the benefits that flow from them in
some sense of entitlement antecedent to social institutions, which is
to say in some sense of entitlement equivalent to desert. ( ' I t needn't
be that the foundations underlying desert are themselves deserved,
all the wqy down.') This would require that I possess the relevant
assets in the strong sense of possession adequate to a desert base, and
this in turn requires a theory of the person on which I possess some
things, at least, as constituents and not merely as attributes of the
self. But Nozick's theory of the person is not easy to discern. He
complains that the 'purified' self implicit in Rawls' theory seems
radically at odds with our more familiar notion of ourselves as beings
'thick with particular traits' ( 1 974: 2 28) , and further obj ects that so
'purified' a self, even if coherent, seems to threaten not only indi­
vidual desert, but also such indispensable notions as individual
autonomy, responsibility, and the very human dignity and self­
respect deontological liberalism sets out to affirm ( 1 974: 2 1 4) . These,
to be sure, are powerful obj ections. But they do little to articulate a
substantive theory of the person that manages to leave the person
'thick with particular traits' while at the same time avoiding the

1 00
Individual and social claims

moral and epistemological difficulties Rawls identifies with the self

laden with contingencies . It is one thing simply to assert what is in
some sense undeniable, that we are 'thick with particular traits ' , and
quite another to show how this can be true in a way not subj ect to the
rival incoherences associated with a rad ically situated self, indefi­
nitely conditioned by its surroundings and constantly subject to
transformation by experience. For this reason , Nozick's proposed
solution to the stand-off can be of critical interes t alone .
What, then, of Rawls' attempt? If Nozick fails to show that the
absence of individual desert leads to a presumption in favor ofletting
natural assets lie where they fall, with what success does Rawls show
a presum ption in favor of a general social claim to such assets ?
According to Nozick, Rawls' view is that 'everyone has some entitle­
ment or claim on the totality of natural assets (viewed as a pool ) ,
with no one having differential claims' ( 1 974 2 2 8 ) , and indeed the
notion of common assets seems to imply some such view . The
question is what kind of 'entitlement or claim' is involved, and how
it might be established . Two possibilities suggest themselves; the
general social claim on the totality of natural assets might be a claim
of desert, or it might be a claim to an entitlement, or legitimate
expectation. If it is meant to be a claim of desert, then that claim is at
best underdetermined by Rawls' theory, for as we have seen, the
argument from arbitrariness works only to undermine individual
desert, not necessarily to install a social one. Moreover, for the
community as a whole to deserve the natural assets in its province
and the benefits that flow from them, it is necessary to assume that
society has some pre-institutional status that individuals lack, for
only in this way could the community be said to possess its assets in
the strong, constitutive sense of possession necessary to a desert base.
But such a view would run counter to Rawls ' individ ualistic assump­
tions , and in particular to his view that society is not 'an organic
whole with a life of its own distinct from and superior to that of all its
mem bers in their relations with one another' ( 2 64) .
The alternative would be to view society's claim on the distribu­
tion of assets as an entitlement given by legitimate expectations
founded upon social institutions 'established for prior and indepen­
dent social ends' ( 3 1 3) , prior and independent, that is, to the entitle­
ment itself. Rawls might argue along these lines that the difference
principle does not assume that society has some pre-institutional

101
Possession, desert, and distributive justice

status denied to individuals, only that the parties to the original

position would agree to regard the distribution of natural talents as a
common asset and to share in the benefits of this distribution . The
notion of society as the owner of natural assets for which individuals
are the guardians would then be seen as the result of the original
position and not its premise.
But this still leaves us wondering abQut the 'prior and independent
social ends' , and the source of their priority and independence.
Somehow the society's claim on the distribution of assets must make
a difference before institutions get established . Some social claim on
the distribution of assets would seem a necessary presupposition of
the agreement in the original position, otherwise the parties would
be deliberating about how to allocate shares that were not (yet)
rightfully theirs to allocate. Nozick argues, along similar lines,

Do the people in the original position ever wonder whether they have the
right to decide how everything is to be divided up? Perhaps they reason that
since they are deciding this question, they must assume they are entitled to
do so; and so particular people can't have particular entitlements to hold­
ings (for then they wouldn' t have the right to decide together on how all
holdings are to be divided ) ( 1 974: 1 ggn) .

To this Rawls might reply that no antecedent social claim is

involved , since the parties to the original position are not faced,
strictly speaking, with the moral question of how they ought to
allocate distributive shares (which would indeed imply their having
an antecedent 'right' to decide) , but only with the prudential ques­
tion of how, given the relevant constraints on information, etc., they
would, from the standpoint of individual self-interest, prefer that
individual shares be allocated . The original position, Rawls might
remind us, is not an actual site of allocation, but rather a way of
thinking.
Still, we want to know why this way of thinking is appropriate to
questions of distributive j ustice, and whether the appropriateness of
this way of thinking does not itself depend on the parties to the
original position having some antecedent claim on the totality of
natural assets . For even if the parties to the original position would,
as a matter of rational prudence, reason in the way Rawls says they
would , it is not immediately clear, absent some prior social claim,
why their collective choice should determine the j ust distribution of

1 02
Individual and social claims

these assets . This leads us to the issue of j ustification and the

question of how the argument from the original position serves to
j ustify the principles that result. If society's claim to the distribution
of natural assets can be shown to be a product of the original
agreement rather than its premise, then Rawls will have resolved the
stand-off in favor of the difference principle without having to
attribute an antecedent claim of desert to society as a whole . If, on
the other hand , the notion of common assets should turn out to be a
presupposition of the original agreement, then Rawls will have
overcome the stand-off only by implicit reliance on a claim of social
desert, and hence reliance on a wider subj ect of possession, presum­
ably the community , held to own the assets we individually bear.
Once again an intersubj ective dimension would intrude on Rawls'
individualistic proj ect. Although such a result would do consider­
able damage to the deontological ethic Rawls seeks to defend , I shall
try to show that his version of con tract theory cannot avoid it.

1 03
3

Contract Theory and Justification

In assessing the validity of the difference principle, we are led

ultimately to the question of justification, and in particular to the
question why the original position provides an appropriate way of
thinking about j ustice at all . Even assuming, for the sake of argu­
ment, that the parties to the original position really would choose the
principles Rawls says they would, why does this give us reason to
believe that these principles are j ust? Rawls writes that the principles
of j ustice are those principles 'that free and rational persons con­
cerned to further their own interests would accept in an initial
position of equality' ( 1 1 ) , and that the original position is ' the
appropriate initial status quo which insures that the fundamental
agreements reached in it are fair' ( 1 7 ) . 'Understood in this way the
question of j ustification is settled by working out a problem of
deliberation : we have to ascertain which principles it would be

theory of j ustice with the theory of rational choice' ( q ) .

rational to adopt given the contractual situation . This connects the

But it is not immediately clear how the original position confers

moral status on the results of an exercise in rational choice, not
obvious what the j ustificatory force of the argument from the original
position consists in. The question of j ustification is complicated by
the fact that Rawls seems simultaneously to rely on two different
sorts of j ustification , one appealing to the method of reflective
equilibrium, the other to the tradition of the social contract, and
sorting out their respective roles poses certain difficulties . ( Lyons,
Reading Rawls : 1 4 1 -68) .
For the moment, however, I shall put these difficulties aside in
order to focus attention on the contractarian aspect of j ustice as
fairness. In so far as the principles of j ustice depend for their
j ustification on a contractarian appeal , what does the moral force of
its appeal consist in? Exploring this question should provide us a
further test of the internal coherence ofRawls' conception, and of the fit
between the theory ofjustice and its correlative theory of the person . I
shall try to show that the argument from the original position can
The moraliry of contract

be seen to j ustify its results only at considerable cost to certain

voluntarist and individualist assumptions central to the deontologic­
al proj ect.

T H E M O RALITY O F C ONTRACT

Rawls locates his theory ofj ustice in the tradition of social contract
theory going back to Locke, Rousseau, and Kant. The 'guiding idea'
is that the principles of j ustice are the obj ect of an original agree­
ment. 'Thus we are to imagine that those who engage in social
co-operation choose together, in one joint act, the principles which
are to assign basic rights and duties and to determine the division of
social benefits' ( 1 1 ) . In designing the social contract, 'a group of
persons must decide once and for all what is to count among them as
just and unjust', and the principles they choose are ' to regulate all
subsequent criticism and reform of institutions' ( 1 2- 1 3 ) . In this
respect, the original contract would seem a kind of ordinary contract
writ large.
But for Rawls, as for some of his contractarian predecessors , the
original agreement is not an actual historical contract, only a
hypothetical one ( 1 2 ) . I ts validity does not depend on its terms
actually having been agreed to, but rather on the idea that they would
have been agreed to under the requisite hypothetical conditions . I n
fact, Rawls' hypothetical social contract i s even more imaginary
than most. Not only did his contract never really happen; it is
imagined to take place among the sorts of beings who never really
existed , that is , beings struck with the kind of complicated amnesia
necessary to the veil of ignorance. In this sense, Rawls' theory is
doubly hypothetical·. It imagines an event that never really hap­
pened, involving the sorts of beings who never really existed .
But this would seem to undermine the moral analogy that gives
contract theory much of its intuitive appeal . Once the social contract
turns hypothetical, the original agreement is no longer a contract
writ large, only a contract that might have been writ large but never
was . And as Ronald Dworkin has written, 'A hypothetical contract is
not simply a pale form of an actual contract; it is no contract at all'
( 1 9 7 7a : 1 7- 1 8) . How, then, can it serve to j ustify the principles it
yields , to certify their status as principles ofjustice?
To answer this question, we might begin by taking up a simpler

1 05
Contract theory and justification

question, and inquire into the moral force of contracts and agree­
men ts generally. Once we can say something abou t how justification
works with actual contracts, we may see more clearly how it works
with hypothetical ones .

When two people make an agreement we may typically assess its

j ustice from two points of view. We may ask about the conditions
under which the agreement was made, whether the parties were free
or coerced, or we may ask about the terms of the agreement, whether
each party received a fair share. While these two considerations may
well be related, they are by no means identical, and barring some
special philosophical assumptions to be considered later, cannot
normally be reduced the one to the other. Practically speaking, a
contract freely agreed to may be more likely than others to yield
terms that are fair, and a (substantively) fair exchange may well
suggest a free contract rather than a coerced one, but there is no
necessary connection either way.
Of any contractual agreement, however free, it is always intelli­
gible and often reasonable to ask the further question , 'But is it fair,
what they have agreed to?' , where this question cannot be translated
to the merely vacuous question, 'But is it what they have agreed to,
what they have agreed to?' What makes it fair is not just that it was
agreed to, but is a further question .
Similarly, any transaction or arrangement, however fair; is open
in principle to the further question, 'But was it freely arrived at, this
fair arrangement?' , where this question cannot be reduced to the
trivial question , 'But is it a fair arrangement, this fair arrangement?'
What makes a transaction free is not that it ended fairly; being treated
fairly neither makes us free nor entails that we are free. This , too, is a
further question .
The distinction between these two sorts of questions suggests that
we may think of the morality of contract as consisting of two related
yet distinguishable ideals. One is the ideal of autonomy, which sees a
contract as an act of will , whose morality consists in the voluntary
character of the transaction . The other is the ideal of reciprocity,
which sees a contract as an instrument of mutual benefit, whose
morality depends on the underlying fairness of the exchange . '
1 Valuable discussions of the moral foundations of con tract law, stressing the ideals of
be found in Fri ed ( 1 98 1 ) and A tiyah ( 1 9 79 ) .
a u tonomy and reci p roci ty respectively, can

1 06
The morality of contract

Each ideal suggests a different basis for contractual obligation .

From the s tandpoint of au tonomy, a contract's moral force derives
from the fact of its voluntary agreement; when I enter freely into an
agreement, I am bound by its terms, whatever they may be. Whether
its provisions are fair or inequitable, favorable or harsh, I have
' brought them on myself' , and the fact that they are self-imposed
provides one reason at least why I am obligated to fulfill them .
The ideal of reciprocity, on the other hand , derives contractual
obligation from the mutual benefits of co-operative arrangements .
Where autonomy points to the contract itself as the source of obliga­
tion, reciprocity points through the contract to an antecedent moral
requirement to abide by fair arrangements, and thus implies an
independent moral principle by which the fairness of an exchange
may be assessed . With reciprocity, the emphasis is less on the fact of
my agreement than on the benefits · I enjoy; contracts bind not
because they are willingly incurred but because (or in so far as) they
tend to produce results that are fair.
In its account of obligation , each ideal can be seen to highlight the
moral incompleteness of the other. From the standpoint of auto­
nomy, my obligations are limited to those I voluntarily incur, but
these may include provisions onerous and harsh. From the stand­
point of reciprocity, hard deals bind less , but on the other hand, the
need for consent fades , and I may be obligated in virtue of benefits I
do not want or dependencies beyond my control. I n the first I may be
bound to terms that are unfair; in the second I may be bound in ways
I did not choose.
Finally, each account of contractual obligation relates contract to
justification in a different way. On the ideal of autonomy, the con­
tract imparts the j ustification; in so far as it is free, the process itself
serves to j ustify the outcome, 'whatever it happens to be' . On the
ideal of reciprocity, by contrast, the contract approximates justice
rather than confers it; the process is instrumental to, rather than
definitive of, a j ust result. On the first, a fair result is defined as the
outcome of a process that is free; on the second, a free process is
simply one means of arriving at a result which is (independently)
fair.
Unlike obligations voluntarity incurred , obligations arising under
the ideal of reciprocity must presu ppose some criterion of fairness
independen t of con tract, some way in which the obj ective fairness of

1 07
Contract theory and justification

an exchange may be assessed . Such obligations are thus not contrac­

tual in the strict sense that the contract creates the obligation , but
rather in the limited epistemic or heuristic sense that the contract
helps to identify or clarify an obligation that is already there (Atiyah
1 9 79 : 1 43-6) . One consequence of this feature of benefit-based
obligations is that the carrying out of a contract is not essential to the
existence of the obligation . In principle at least, there may be ways of
identifying such obligations without recourse to contract.
Those obligations arising on the ideal of autonomy, however,
presume no quality ofj ustice intrinsic to certain results which could,
even in principle, be identified apart from or antecedent to the
process that produced them. With obligations of this sort, no result
can be identified as j ust without reference to a procedure actually
having been carried out. What is j ust cannot be known directly
because it is, by definition , the product of a process of a certain kind;
it cannot be known directly because it must be created , and until it is
created , it cannot be known .
The contrasting ways in which the two ideals relate contract to
j u stification may be illuminated by Rawls' distinction between pure
procedural justice and perfect (or imperfect) procedural j ustice. In
pure procedural justice, ' there is no independent criterion for the
right result: instead there is a correct or fair procedure such that the
outcome is likewise correct or fair, whatever it is, provided that the
procedure has been properly followed ' ( 86) . In perfect and imperfect
procedural j ustice, on the other hand , 'there is an independent
standard for deciding which outcome is j ust' , and the question is
simply whether or not a procedure can be found which is guaranteed
to lead to it (85-6) .
Now in so far as a contract realizes the ideal of autonomy, it
approaches the case of pure procedural j ustice, in which the outcome
is j ust, whatever it is, in virtue of the contract that produced it. On
the ideal of reciprocity, a contract is a case of imperfect procedural
justice, seeking as it does to approximate a standard of justice
independently defined . As Rawls points out, 'A distinctive feature of
pure procedural justice is that the procedure for determining the j ust
result must actually be carried out; for in these cases there is no
independent criterion by reference to which a definite outcome can
be known to be just . . . . A fair procedure translates itsfairness to the
outcome only when it is actually carried out' [emphasis added] ( 86) .

1 08
Contracts versus contractarian arguments

The answer, then , to our preliminary question how actual con­

tracts j ustify seems ' incompletely ' . As the non-trivial coherence of
the 'further question' attests ( 'But is it fair , what they have agreed
to?' ) , actual contracts are not self-sufficient moral instruments but
presuppose a background morality in the light of which the obliga­
tions arising from them may be qualified and assessed . While it may
be j ust, under certain circumstances, to hold a person to the terms of
his prior agreement, it does not follow from his agreement that the
terms themselves are j ust. Common sense suggests various reasons
why, in practice, actual contracts may turn out unfairly; one or both
of the parties may be coerced or otherwise disadvantaged by an
unfavorable bargaining position, or deceived or otherwise mistaken
about the value of the things being exchanged , or unclear about their
own needs and interests. But even where an agreement turns out
fairly ( as when the effects of such factors are countervailing, for
example) , and where the fairness of the agreement provides a reason
for its enforcement, it cannot be assumed that what makes it just is the
fact that i t was agreed to . Actual contracts are typically cases of
imperfect procedural j ustice; pure procedural j ustice rarely, if ever,
appears in the world .

C O NTRACTS V E R S U S C O N T R A C T ARIAN A R G U M ENTS

Rawls would not likely disagree with this formulation . Notwith­

standing the contractarian basis of his theory, he does not suppose
that the mere fact of an agreement is the test of its fairness, or that
actual contracts are self-sufficient moral instruments that j ustify
their own results , or that obligations voluntarily incurred are im­
mune from criticism in the light of pre-existing principles ofj ustice.
Understanding the sense in which Rawls' theory does not rely on the
notion of contract as an instrument of j ustification is essential to
understanding the sense in which it does .
The first point to emphasize in this connection is that the agree­
ment in the original position gives rise not to obligations ( at least not
directly) , but to principles ofj ustice. The principles ofjustice include
principles of two sorts - ' principles for institutions' , which apply to
the basic structure of society, and ' principles for individuals' , which
set out the duties and obligations of persons with respect to institu­
tions and each other. The first define what makes an institution or a

1 09
Contract theory and justification

social practice just, while the second specify the terms on which
individuals are bound to abide by them .
The principles for individuals specify two different ways in which
persons may be bound - as a matter of natural duty or of obligation .
Natural duties are those moral claims that apply to persons irrespec­
tive of their consent, such as the duties to help others in distress, not
be cruel , to do j ustice, and so on . Such duties are 'natural' in the

arrangements but are owed to persons generally ( I I 4- I 5) .

sense that they are not tied to any particular institutions or social

Obligations, by contrast, describe those moral ties we voluntarily

incur, whether by contract or promise or other expression of consent.
The obligations of public office voluntarily sought are one such
example. But even with obligations, consent is not sufficient to create
the tie. A further condition is that the institution or practice agreed to
be j ust (or nearly j ust) , in accordance that is, with the two principles
of j ustice. Rawls emphasizes that, notwithstanding their voluntary
dimension, our actual obligations are never born of consent alone
but inevitably presuppose an antecedent background morality, inde­
pendently derived , in the light of which it is always possible to ask
whether one ought to have consented or not.
Obligations arise only if certain background conditions are satisfied . Ac­
quiescence in, or even consent to, clearly unjust institutions does not give
rise to obligations . It is generally agreed that extorted promises are void ab
initio. But similarly, unj ust social arrangements are themselves a kind of
extortion , even violence, and consent to them does not bind ( 343 ) .
I n particular, it i s not possible t o have a n obligation t o au tocratic and
arbitrary forms of government. The necessary background does not exist for
obligations to arise from consensual or other acts , however expressed.

of the circumstances ( 1 1 2 ) .
Obligatory ties presuppose j ust institutions, or ones reasonably j ust in view

Even promises cannot alone give rise to obligations . Rawls distin­

guishes here between the rule of promising and the principle of
fidelity, and argues that the obligation to keep a promise is not a
conseq uence of the promise, but of a moral principle antecedent to
the promise, deriving from a theory of j ustice. ' I t is essential to
distinguish between the rule of promising and the principle of
fidelity. The rule is simply a constitutive convention, whereas the principle
of fidelity is a moral principle, a consequence of the pri nci ple of
1 10
Contracts versus contractarian arguments

fairness . . . . The obligation to keep a promise is a consequence of the

p rinciple of fairness ' (346) . As a constitutive practice or convention ,
the rule of promising is analogous to legal rules , or the rules of a
game; whether they are just or not is always a further question which
cannot be answered without recourse to a moral standard indepen­
dent of the practice . 'There are many variations of promising j ust as
there are of the law of contract. Whether the particular practice . . . is
j ust remains to be determined by the principles of j ustice' ( 345-6 ) .
Strictly speaking, then, it i s not promises that bind , but the principle
of fidelity that binds us to (certain of ) our promises, and this
p rinciple derives from the original position .
Even the rule of promising does not give rise to a moral obligation by itself.
To account for fiduciary obligations we must take the principle offairness as
a premise. Thus along with most other ethical theories, j ustice as fairness
holds that natural duties and obligations arise only in virtue of ethical
principles . These principles are those that would be chosen in the original
position ( 348) .

But if Rawls does not take actual contracts or promises to be

binding, at least not in themselves , in what sense is his theory

in real life from its role in the original position . While consent is
contractarian? Here it is important to distinguish the role of consent

decisive in the original position, it plays less central a role in our

actual duties and obligations . Notwithstanding their contractarian
derivation, the natural duties apply without reference to our volun­
tary acts, and the consent obligations require is in any case distinct
from the consent involved in the original position .
Even though the principles of natural duty are derived from a contractarian
point of view, they do not presuppose an act of consent, express or tacit, or
indeed any volun tary act, in order to apply . The principles that hold for
individuals, just as the principles for institutions, are those that would be
acknowledged in the original position . These principles are understood as
the outcome of a hypothetical agreement. If their formulation shows that no
binding action, consensual or otherwise, is a presupposition of their
application, then they apply unconditionally . The reason why obligations
depend upon voluntary acts is given by the second part of the principle of

tual nature of justice as fairness ( I I 5- I 6) .

fairness which states this condition . It has nothing to do with the contrac­

Real contracts issue in exchanges or arrangements whose j ustifica-

I I I
Contract theory and justification

tion must await a principle ofj ustice; the hypothetical contract issues
in principles ofj ustice capable of assessing those arrangements an d
defining the moral consequences of contracts generally . As things
turn out, the role the parties decide to assign to ( actual) agreemen ts
is different from the role their ( hypothetical) agreement plays in
j ustification.
Nozick obj ects that it is somehow inconsistent for a contract
theory to produce principles ofj ustice that do not give full j ustifica­
tory force to voluntary exchanges . If contracts are binding, Nozick
suggests, then Rawls' theory is wrong for yielding principles which
would deny their force in many cases , and if contracts cannot bind,
then Rawls ' theory is undermined since it is founded on a contract.
Contract arguments embody the assumption that anything that emerges
from a certain process is just. V pon the force of this fundamental assump­
tion rests the force of a contract argument. Surely then no contract argu­
ment should be structured so as to preclude process principles being the
fundamen tal principles of distributive justice by which to j udge the institu­
tions of a society; no contract argument should be structured so as to make it
impossible that its results be of the same sort as the assumptions upon
which it rests. If processes are good enough to found a theory upon, they are
good enough to be the possible result of the theory. One can' t have i t both
ways ( 1 974: 208-g ) .
But this obj ection overlooks the distinction between the imperfect
procedural j ustice that typically describes our actual agreements,
and the pure procedural j ustice that obtains, or at least is meant to
obtain , in the original position . Or to put the point another way, the
obj ection confuses contracts with contract arguments . As Rawls
observes, real contracts are not arguments but social facts, whose
moral consequences depend on some moral theory, contractarian or
otherwise. Some such distinction, between the fact of an agreement
and the grounds ofits j ustification, is essential if we are to account for
the coherence of the 'further question' ( ' But is it fair, what they have
agreed to?' ) , or to make sense of the related question whether one
ought to have consented or not, or allow for the rival but correlative
claims of autonomy and reciprocity in arguments about j ustice.
Seen in this light, there is no contradiction in a contract argumen t
producing principles that limit the justificatory role of contracts . I n
fact, there appears a n important sense i n which the results o f a
contract argument cannot be 'of the same sort' as the assumptions on

I I2
Liberalism and the priority ofprocedure

which it rests . For if, as Nozick argues, ' contract arguments embody
the assumption that anything that emerges from a certain process is
j ust' [emphasis added] ( I 9 74: 208) , it seems unlikely to suppose that
j ust any agreement, arrived at under j ust any conditions, could
produce results guaranteed to be j ust. And once the circumstances of
an agreement are seen to be relevant to its justification, it cannot be
claimed that the agreement itself does all the j ustifying. To acknow­
ledge the relevance of circumstance is already to acknowledge a
moral sanction independent of the agreement by which the morally
necessary features of the situation are identified .
As Rawls shows in his analysis of promises, the source of this
sanction cannot be a further promise or agreement ( such as a
promise to keep promises) , since the credentials of this background
promise would be equally open to question. The back-up of a
promise (or a contract) must be more than j ust another promise (or
contract) . I t must be a premise of a different sort. This premise,
which on contract theory 'looks much like an agreement to keep
agreements and yet which, strictly speaking, cannot be one' ( 34 9 ) , is
what the hypothetical agreement in the original position seeks to
provide. Rawls believes this device can supply such a premise in a
way that preserves the voluntarist appeal of contract theory without
lapsing into the question-begging regress associated with a mere
agreement to agree. Before assessing Rawls' solution, it may be
helpful to summarize the problem ofjustification it seeks to address
and briefly to consider two alternate solutions which Rawls rej ects .
I n this way i t may be possible to establish some connections between
Rawls' account of justification and certain central features of his
deontological proj ect.

LIBERALISM AND T H E P R I O RITY O F P R O C E D U R E

To justify an exchange or institutional arrangement, it is not enough

to show that it arose from a voluntary agreement between the parties
involved, for at least two different sorts of reasons - one moral, the
other epistemological . Although Rawls does not distinguish these
arguments explicitly, both are implicit in his account, and each
serves to reinforce the other. We might call the first the argument
from contingency and the second the argument from conventional­
ism . The first recalls the argument from arbitrariness deployed in

I I3
Contract theory and justification

support of the difference principle against meritocratic, aristocratic,

and other rival conceptions of equality falling short of Rawls'
'democratic conception' . I n the case of j ustification , it begins with
the observation that in practice, agreements turn out unfairly for a
variety of reasons , as already suggested ; one or the other party may
be coerced or otherwise disadvantaged in his bargaining position , or
misled or otherwise misinformed about the value of the objects being
exchanged, or confused or mistaken about his own needs and
interests, or, where uncertain future returns are involved , a bad
j udge of risk, and so on. I n some of these cases, notably those
involving outright coercion or deception, we may be tempted to say
that the exchange was not truly a voluntary one, or that the
'contract' is invalid , and so attribute the unfairness of the result to an
inadequacy of consent. Libertarians and others who argue that
voluntary agreements are wholly self-j ustifying are anxious to rule
out such cases by invoking distinctions between coercive and non­
coercive influences , legitimate and illegitimate bargaining tactics,
threats and inducements, and so on (Nozick I 9 7 2 ; Kronman I g8o) .
But Rawls would deny that any such distinction could succeed in
marking out a range of self-j u stifying agreements as long as some
morally arbitrary influences were allowed to remain. However
strictly one defines the requ irements of a voluntary agreement, the
fact that different persons are situated differently will assure that
some differences of power and knowledge persist, allowing agree­
ments, even 'voluntary' ones, to be influenced by factors arbitrary
from a moral point of view. 'Somehow we must nullify the effects of
specific contingencies which put men at odds and tempt them to
exploit social and natural circumstances to their own advantage'
( I 36) .
Even voluntary agreements are likely to fall short of the ideal of
autonomy, in which the obligations incurred are self-imposed in the
strict sense of 'self defined as prior to its attributes and ends and thus
free from heteronomous determinations. Only this sense of self, and
the notion of autonomy it permits, rule out arbitrary influences
completely. Ruling out coercion alone cannot justify a contract any
more than ruling out, say, class privileges alone can j ustify a
meritocracy . I n both cases, too much is left subject to contingencies
arbitrary from a moral point of view. Once we are bothered by the
most conspicuous obstacles to individual autonomy, we are bound

I I4
Liberalism and the priority of procedure

on reflection to reject heteronomous influences wherever they

appear!
Beyond the moral difficulty with the notion that contracts are
self-j ustifying lies an epistemological difficulty . This concerns the
status of contracts as 'constitutive conventions ' , in Rawls' phrase,
and recalls the problem that arose in connection with the Archime­
dean point, the problem of distinguishing a standard of appraisal
from the thing being assessed . In the case of contracts, the parallel
distinction is between a moral principle on the one hand and a rule or
a practice or convention on the other. Notwithstanding the norma­
tive import of, say, the practice of promising or the rules of a game or
the law of contracts, practices and rules and laws as such cannot
j ustify anything on their own, but must depend for their moral
consequences on some principle independent of them . 'The contract
doctrine holds that no moral requirements follow from the existence
ofinstitutions alone' ( 348 ) . Given its status as a constitutive conven­
tion, an exchange can no more be justified by showing that it was
voluntarily agreed to than a law can be justified by showing that it
was duly enacted . The fact that a transaction is agreed to or a law
enacted m ay be sufficient, given the relevant background norms, to
establish a legal or institutional requirement to abide by it, but
'whether these requirements are connected with moral duties and
obligations is a separate question' ( 349) . We still need to know
whether the parties ought to have given their consent or whether the
legi slators ought to have voted the way they did.
I f actual contracts must presuppose an antecedent principle to
justify their results, the question naturally arises how such a princi­
ple might be derived . We might be tempted to seek such a principle
in a prior, more general agreement, setting out the terms on which
particular agreements are j ust. But this solution is quickly under­
mined once it is seen that the problems of contingency and conven­
tionalism would simply be deferred. Once the problems of contingency
and conventionalism are acknowledged , there is no reason to think
that second order contracts can be made self-justifying in a way that
particular contracts cannot. A contract can no more be sanctioned
by a prior agreement to keep agreements than a law can be j ustified
by a law about legislation . In each case, the 'further question' is not
dissolved but postponed . A convention about conventions
2 Compare Rawls on the difference principle ( 1 97 1 : 7 4-5) .

1 15
Contract theory and justification

does not make a moral principle, only a further social fact.

I f no actual contract, however general , can j ustify contracts,
owing to its implication in the practices and conventions of some
particular society, the alternative would seem recourse to a principle
of j ustice somehow prior to particular practices and conventions.
This, in fact, is the solution sought by traditional contract theorists,
who backed up the Social Contract by an appeal to Natural Law
(Barker 1 948: x-xi ) . Thus , for Locke, it is 'the law ofGod and Nature'
that sanctions the original compact and sets bounds on the powers of
the commonwealth the compact brings forth .

Thus the law of nature stands as an eternal rule to all men, legislators as
well as others . The rules that they make for other men's actions must, as
well as their own, and other men's actions, be conformable to the law of
nature, i . e . , to the will of God , of which that is a declaration , and the

sanction can be good or valid against it ( 1 6go: go ) .

fundamental law of nature being the preservation of mankind, no human

But this traditional solution is unavailable to Rawls for at least two

reasons . The most obvious is that reliance on ' the law of God and
Nature' involves a more substantial theological and metaphysical
commitment than Rawls is prepared to assume. So controversial an
assumption would clash with his determination to argue from
'generally shared and preferably weak conditions' (20) , and ' to
insure that the principles of j ustice do not depend upon strong
assumptions . At the basis of the theory , one tries to assume as little as
possible' ( 1 29) .
For Rawls' purpose, a further difficulty with the idea of founding
j ustice on a premise of natural law such as Locke invokes is that it
runs counter to the core assumptions of the deontological proj ect . As
we have seen, a central aspiration of deontological liberalism is to
derive a set of regulative principles that do not presuppose any
particular conception of the good, nor depend on any particular
theory of human motivation. Connected with this aim are the views
that conceptions of the good are diverse, that there is no single,
dominant human end , that man is a being whose ends are chosen
rather than given, and that the well-ordered society is therefore one
in which people are free to pursue their various aims , whatever they
may be, on terms that are j ust. ' Li berty in adopting a conception of
the good is limited only by principles that are deduced from a

1 16
Liberalism and the priority of procedure

doctrine which imposes no prior constraints on those conceptions'

( 2 5 3 ) . To emphasize the voluntarist conception of human agency
that underlies those principles, Rawls assumes that the parties to the
original position are unbound by prior moral ties , and ' think of
themselves as beings who can and do choose their final ends ( always
plural in number) ' (563 ) .
By appealing to the law of nature as a premise of the original
compact, Locke founds j ustice on certain claims about human ends
and motivations which a deontological ethic would be reluctant to
admit. When he assumes that men join in society with others 'in
order to unite for the mutual preservation of their lives , liberties , and
estates, which I call by the general name, property' , he attributes to
human beings a certain dominant end and makes this end a premise
of the principles that result. 'The great and chief end , therefore, of
men's uniting into commonwealths, and putting themselves under
government, is the preservation of their property; to which in the
state of nature there are many things wanting' ( 1 6go: go) .
But for Rawls, to found principles ofj ustice on ends or desires said
to be given by nature, whether they be the pursuit ofhappiness or the
preservation of property or of life itself, is to place the good prior to
the right, to deny the essential plurality of human ends and to posit
instead a single dominant end, to base j ustice on certain natural
contingencies, and to reverse the priority of self and ends by conceiv­
ing man as a subj ect of ends given in advance rather than a willing
subject of ends he has chosen .
Once a dominant end given by nature is admitted as a premise of
the original compact, the parties can no longer ' think of themselves
as beings who can and do choose their final ends', but instead are
bound in advance . The voluntarist dimension of their enterprise
fades, and the terms of the contract are no longer a matter of choice
but are determined in advance, given by the requirements of the
natural law that is their prerequisite ( Pitkin 1 9 65: ggo-g) .
That a deontological liberal would rej ect a Lockean solution on
these grounds is strengthened by the fact that Kant mounts a similar
obj ection to traditional contract views . Though Kant's critique is
directed against Hobbes , his objections would seem equally applic­
able in this respect to Locke.
Kant distinguishes between those social agreements among men
designed to further some common end they may happen to share,

1 17
Contract theory and justification

and those dedicated to the one end in itself which all ought to share,
namely the principle of right. Only the second sort of contract
constitutes a civil state, in which the freedom of each is made to
harmonize with the freedom of everyone else under terms governed
by right. But such an arrangement cannot be based on any particular
view of human . nature nor motivated by any merely contingent
human ends.

The whole concept o f an external right is derived entirely from t h e concept

offreedom in the mutual external relationships of human beings , and has
nothing to do with the end which all men have by nature ( i . e . , the aim of
achieving happiness) or with the recognized means of attaining this end .

the laws governing external right ( I 793 : 73) .

And thus the latter end must on no account interfere as a determinant with

For Kant, the principle that serves as sanction to the original

contract is not ' the great and chief end of preserving property' , or
pursuing happiness, but the duty in itself which is ' the highest formal
condition of all other external duties , the right of men under coercive

secured against attack from any others' ( I 793 : 7 3 ) . And this duty
public laws by which each can be given what is due to him and

comes not from nature but 'is the requirement of pure reason, which
legislates a priori , regardless of all empirical ends (which can all be
summed up under the general heading of happiness ) ' . Since men
have different views on the empirical end of happiness and what it
consists of, happiness could not bring their will under any external
law harmonizing with the freedom of everyone. The civil state can
thus only be established in accordance with 'pure rational principles
of external human right' given a priori . It must be based on a priori
principles, for neither nature nor experience can provide knowledge
of what is right ( Kant I 793 : 7 3- 4 , 86) . The result is a liberalism that
departs significantly from Locke's, and contains the essentials of the
deontological ethic taken up by Rawls . As Kant writes,
We are not concerned here with any happiness which the subj ect might
expect to derive from the institutions or adminis tration of the common­
wealth, but primarily with the rights which would thereby be secured for
everyone . A nd this is the highest principle from which all maxims relating to the
commonwealth must begin, and which cannot be qualified by arry other principles. No
generally valid principle of legislation can be based on happiness. For both
the current circumstances and the highly conflicting and variable illusions

1 18
Liberalism and the priority of procedure

as to what happiness is ( and no-one can prescri be to others how they should
attain it) make all fixed principles impossible, so that happiness alone can
never be a suitable principle of legislation [emphasis added) ( 1 7 9 3 : 8o) .
Where Locke backs up the original contract with the law of God
and Nature, Kant backs it up with a principle of right given not by
nature but by pure reason . Of the two solutions, Kant's is the more
congenial to Rawls' conception in that it avoids deriving the right
from the good and so preserves deontological assumptions . But as we
have seen , Rawls is resistant to Kant's solu tion in so far as it seems to
depend on metaphysical assumptions he finds obj ectionable. He is
dubious of the idealist metaphysics by which pure reason does its
work, and is trou bled by what seems to be the arbitrariness of the a
priori derivation of the Kantian moral law. And so rather than adopt
Kant's solution to the problem ofjustification directly, Rawls seeks
instead to reformulate Kant's deontological teachings , ' to detach the
underlying structure of Kant's doctrine from its metaphysical sur­
roundings so that it can be seen more clearly and presented relatively
free from obj ection' ( 264) . Here we return to the mission of the
original position in its bid to provide an Archimedean point: to find a
middle way between conventionalism and arbitrariness , to seek a
standard of appraisal neither compromised by its implication in the
world nor dissociated and so disqualified by detachment.
With contract theory, the challenge posed by the Archimedean
point takes more determinate form . Clearly, j ustification involves
some sort of interplay between contracts and principles . Actual
contracts presuppose principles ofj ustice, which derive in turn from
a hypothetical original contract. But how does j ustification work
there? Is recourse to yet a further layer of antecedent principles
required? Or is contract at that stage morally self-sufficient, and fully
self-j ustifying? At times the search for the ultimate sanction appears
an infinitely elusive dance of procedure and principle, each receding
in turn behind the other. For given the assumptions of contract
theory, neither seems to offer a stable resting point on which to found
the other. If the parties to the original contract choose the principles of
j ustice, what is to say that they have chosen rightly? And if they
choose in the light of principles antecedently given, in what sense can
it be said that they have chosen at all? The question of justification
thus becomes a question of priority ; which comes first - really,
ultimately first - the contract or the principle?
I l9
Contract theory and justification

With Kant, it is unclear whether the principle of right is the

product of the original agreement or its premise, and his reliance on
'pure reason, which legislates a priori' , ( I 793 : 7 3 ) , seems to suggest
the latter. This , in any case, is the point at which Rawls seeks to
reformulate the Kantian position, to assert the priority of contract
and so to emphasize the connection of j ustice with the theory of
rational choice. 'The merit of the contract terminology is that it
conveys the idea that principles of j ustice may be conceived as
principles that would be chosen by rational persons, and that in this
way conceptions of j ustice may be explained and justified' ( I 6) .
By founding j ustice on an original contract, Rawls seeks to express
what he takes to be the central Kantian insight, ' the idea that moral
principles are the obj ect of rational choice' ( 25 I ) . He describes the
original position as 'a procedural interpretation of Kant's conception
of autonomy and the categorical imperative . . . a natural procedural
rendering of Kant's conception of the kingdom of ends, and of the
notions of autonomy and the categorical imperative' [emphasis
added] ( 2 56, 2 64) .
Why a procedural rendering? Why is it necessary to amend Kant
in order to give these principles an explicitly procedural or contrac­
tual derivation? Rawls' answer must be that by casting the moral law
as the outcome of a certain process of rational choice, however
hypothetical, it is possible to establish its claim on human experience
in a way that might not otherwise be apparent. 'No longer are these
notions purely transcendent and lacking explicable connections with
human conduct, for the procedural conception of the original posi­
tion allows us to make these ties ' ( 2 56) .
The priority of procedure in Rawls' account ofj ustification recalls
the parallel priorities of the right over the good , and of the selfover its
ends. I t thus connects the account ofj ustification with the theory of
the person which j ustice as fairness was seen to entail, and suggests
the importance of contract theory to the deontological proj ect gener­
ally. As the self is prior to the ends it affirms , so the contract is prior to
the principles it generates . Of course, not j ust any contract is prior to
the principles of justice; as we have seen, actual contracts cannot
j ustify precisely because they are typically situated in the practices
and conventions which j ustice must assess . Similarly, real persons,
ordinarily conceived as ' thick with particular traits ' , are not strictly
prior with respect to their ends, but are embedded in and con-

1 20
Liberalism and the prioriry of procedure

ditioned by the values and interests and desires from among which
the 'sovereign ' self, qua subject of possession, would take its pur­
poses . To assert the priority of the self whose sovereign agency is
assured , it was necessary to identify an 'essentially unencumbered'
self, conceived as a pure subj ect of possession, distinct from its
contingent aims and attributes, standing always behind them .
I n the case of contract, the priority of procedure depends on
distinguishing the special case of pure procedural justice, a sort of
purified , pre-situated version of ordinary procedure in which no
independent criteria offairness are available . Only the purified self is
guaranteed to be a sovereign agent, and only pure procedural j ustice
is guaranteed to produce results that are fair. It is this notion of
procedu re that Rawls invokes in the original position .
The idea of the original position is to set up a fair procedure so that any

procedu ral j ustice as a basis of theory ( 1 36) .

principles agreed to will be j ust. The aim is to use the notion of pure

Pure procedural j ustice obtains when there is no independent criterion for

the right result: instead there is a correct or fair procedure such that the
outcome is likewise correct or fair, whatever it is, provided that the proce­
dure has been properly followed (86) .

Two further parallels link Rawls' account ofj ustification with his
theory of the person . One highlights the role of choice in deontologi­
cal ethics ; the other emphasizes the assumption of plurality . In our
discussion of the self, we considered two accounts of agency by which
the self might come by its ends , a voluntarist account which related
self to ends as willing subject to objects of choice, and a cognitive
account which related self to ends as knowing subj ect to objects of
understanding. The priority of the self over its ends was seen to
require the voluntarist account.
Once we imagine the parties to the original position seeking
principles ofj ustice, we can similarly conceive two possible accounts
of justification , a voluntarist account in which the parties arrive at
the principles through an act of choice or agreement, and a cognitive
account in which the parties arrive at the principles through an act of
discovery or collective insight. As with agency, so with justification :
for contract to be prior to principle, the parties must choose the
principles of j ustice rather than find them . Both the priority of the
self and the priority of procedure req uire the voluntarist notions of

121
Contract theory and justification

agency and j ustification respectively. For the self to be prior, its aims
must be chosen rather than given; for contract to be prior, the
principles of justice must be products of agreement rather than
obj ects of discovery .
I n addition to the emphasis on choice, the assumption of plurality
is common to both the theory of the person and the account of
j us tification on the contractarian view. Just as ' the pl urali ty of
distinct persons with separate systems of ends' ( 2 9 ) is essential to
Rawls' notion of the subj ect, so the antecedent plurality of the parties
to the original position is essential to the notion of their hypothetical
agreement. For a contract to be a contract requires a plurality of
persons . I cannot make a contract or an agreement with myself
except in a metaphorical sense in which part of the metaphor in­
volves speaking as though ' I ' were two persons rather than one, a
plurality of selves within a single human being. As Rawls notes ,
principles ofj ustice ' apply to the relations among several persons or
groups. The word "contract" suggests this plurality' ( I 6) .
Having reconstructed the problem ofj ustification Rawls seeks to
address and considered the form of solution he undertakes , it re­
mains to be shown whether this solution succeeds in providing the
foundation the deontological ethic requires . And so we turn at last to
the hypothetical contract in the original position, in order to see what
exactly goes on there, and how it j ustifies , if that is what it do�s . At
the risk of belaboring the familiar, we must explore the text with
some detail if the phenomenology of the original agreement is to
become clear.

WHAT R E A L LY GOES O N B E H I N D THE VEIL O F I G N O R A N C E

What goes on in the original position is first of all a choice, or more

precisely, a choosing together, an agreement among parties . What
the parties agree to are the principles ofj ustice. Unlike most actual
contracts , which cannot j ustify, the hypothetical contract the parties
agree to does j ustify; the principles they choose are j ust in virtue of
their choosing them . As the voluntarist account of j ustification
would suggest, the prin ciples of j ustice are the prod ucts of choice.

The guiding idea is that the principles of j ustice for the basic structure of
society are the obj ect of the original agreement [emphasis added] ( I I ) .

I 22
Behind the veil of ignorance

Thus we are to imagine that those who engage in social cooperation choose
together, in one joint act, the principles which are to assign basic rights and
duties and to determine the division of social benefits . Men are to decide in
advance how they are to regulate their claims against one another [ emph­
asis added) ( I I ) .

Just as each person must decide by rational reflection what constitutes his
good , that is, the system of ends which it is rational for him to pursue, so a
group ofpersons must decide once and for all what is to count among them as j ust
and unj ust. The choice which rational men would make in this hypothetical
situation of equal liberty, assuming for the present that this choice problem
has a solution, determines the principles of j ustice [emphasis added]
( I I - I 2) .

Since all are similarly situated and no one is able to design principles to
favor his particular condition, the principles ofj ustice are the result of a fair
agreement or bargain [emphasis added] ( I 2 ) .

Justice a s fairness begins, as I have said , with one of the most general of all
choices which persons might make together, namely with the choice of the first
principles of a conception of j ustice [emphasis added] ( I 3) .
The principles of j ustice are those which would be chosen in the original
position. They are the outcome of a certain choice situation [emphasis
added] (4 I -2 ) .

Justice as fairness the principles ofj ustice are not thought of as self-evident,
but have their justification in the fact that they would be chosen [emphasis added]
(42 ) .

On a contract doctrine the moral facts are determined by the principles

which would be chosen in the original position . . . [I] t is up to the persons in the
.

original position to choose these principles [emphasis added] (45) .

Justice as fairness differs from traditional contract theories in that

' the relevant agreement is not to enter a given society or to adopt a
given form of government, but to accept certain moral principles'
( 1 6) . The result of the agreement is not a set of obligations applying
to individuals, at least not directly, but principles ofjustice applying
to the basic structure of society. Still, the voluntarist aspect of
j ustification corresponds in some sense to the notion of society as a
voluntary agreement. Rawls writes that living in a society governed
by principles ofj ustice derived from a voluntary account ofj ustifica-

1 23
Contract theory and justification

tion is, in effect, the next best thing to living in a society we have
actually chosen.
No society can, of course, be a scheme of co-operation which men enter
voluntarily in a literal sense; each person finds himself placed at birth in
some particular position in some particular society , and the nature of this
position materially affects his life prospects. Yet a society satisfying the
principles of j ustice as fairness comes as close as a society can to being a
voluntary scheme, for it meets the principles which free and equal persons
would assent to under circums tances which are fair. I n this sense i ts mem bers

added] ( r 3 ) .
are autonomous and the obligations they recognize self-imposed [emphasis

A s our reconstruction suggests, the voluntarist nature of Rawls'

contract view is bound up with the essential plurality of human
subjects and the need to resolve conflicting claims. Wi thout plural­
ity, contracts, and for that matter principles of j ustice, would be
neither possible nor necessary . ' Principles of justice deal with con­
flicting claims upon the advantages won by social co-operation; they
apply to the relations among several persons or groups. The word
"contract" suggests this plurality as well as the condition that the

principles acceptable to all parties' ( 1 6) .

appropriate division of advantages must be in accordance with

As previously seen , j ustice as fairness differs from utilitarianism in

its emphasis on the plurality and distinctiveness of individuals , and
this difference is embodied in the role contract plays in j ustification .
Whereas the utilitarian extends to society the principles of choice for one
man, j ustice as fairness, being a contract view, assumes that the principles of
social choice, and so the principles ofjustice, are themselves the object of an
original agreement [emphasis added] ( 28) .
From the standpoint of contract theory one cannot arrive at a principle of
social choice merely by extending the principle of rational prudence to the
system of desires constructed by the impartial spectator. To do this is not to
take seriously the plurality and distinctness ofindividuals, nor to recognize as the
basis of justice that to which men would consent [emphasis added] ( 29) .

I n basing the principles ofjustice on an agreement among parties,

Rawls emphasizes two characteristics that the hypothetical contract
shares with actual ones, namely choice and plurality . But we have
already seen that the ingredients of choice and plurality are not
sufficient to make justice; actual contracts , which include both,

1 24
Behind the veil of ignorance

cannot j ustify. This is due to the problems we have described as

contingency and conventionalism . Actual agreements often turn out
unfairly because of the various ( coercive and non-coercive) contin­
gencies associated with the inevitable differences of power and
knowledge among persons differently situated . But in the original
position, such contingencies are cured . Due to the veil of ignorance
and other conditions of equality, all are similarly situated , and so
none can take advantage, even inadvertently, of a more favorable
bargaining position .
The original position is designed to overcome the problem of
conventionalism as well . Where actual contracts are inescapably
embedded in the practices and conventions of some particular
society , the agreement in the original position is not implicated in the
same way. It is not an actual contract, only a hypothetical one. Since
it is imagined to occur before the principles of j ustice arrive on the
scene, it may be thought of as 'pre-situated' in the relevant sense, a
status quo antecedent to the arrival of justice such that no prior
moral principles are available by which its results might be im­
pugned . In this way it is able to realize the ideal of pure procedural
justice. ( I ronically, where the hypothetical nature of the original
agreement at first appeared to weaken its j ustificatory force, it now
appears as a positive, perhaps indispensable, advantage . Where
Rawls emphasizes that 'nothing resembling [the original agreement]
need ever have taken place' ( 1 20) , it might be the case that no such
agreement ever could take place and still overcome the problem of
conventionalism . )
Since all are similarry situated and n o one i s able to design principles to favor
his particular condition, the principles of j ustice are the result of a fair
agreement or bargain [emphasis added] ( I 2 ) .
The original position is, one might say, the appropriate initial status quo, and
thus the fundamental agreements reached in it are fair. This explains the
propriety of the name 'j ustice as fairness ' : i t conveys the idea that the
principles ofj ustice are agreed to in an initial situation that is fair [emphasis
added] ( 1 2 ) .

I t is a state of affairs in which the parties are equally represented as moral

persons and the outcome is not conditioned by arbitrary contingencies or the relative
balance of social forces . Thus j ustice as fairness is able to use the idea of pure
procedural j ustice from the beginning [emphasis added] ( 1 20) .

1 25
Contract theory and justification

By imposing the veil of ignorance it is possible to ' nullify the effects

of specific contingencies which put men at odds and tempt them to
exploit social and natural circumstances to their own advantage'
( I 36) .
If a knowledge of particulars is allowed , then the outcome is biased by
arbitrary contingencies. As already observed , to each according to his threat
advantage is not a principle of j ustice. If the original position is to yield
agreements that are j ust, the parties must be fairly situated and treated
equally as moral persons . The arbitrariness of the world must be corrected
for by adj usting the circumstances of the initial contractual situation
[ emphasis added] ( I 4 I ) .

Once the parties to an agreement are assumed to be similarly

situated in all relevant respects , differences of power and knowledge
disappear, and the possible sources of unfairness are thus eradicated .
Since no one is able to choose on the basis of contingently-given
attributes , the ideal of autonomy, implicit but imperfect in actual
con tracts , is fulfilled , the ideal of reciprocity is realized as a matter of
course, and the vulnerability of contract to the 'further question'
( 'But is it fair?' ) is eliminated . 'The veil of ignorance deprives the
persons in the original position of the knowledge that would enable
them to choose heteronomous principles . The parties arrive at their
choice together as free and equal rational persons knowing only that
those circumstances obtain which give rise to the need for principles
of justice' ( 2 5 2 ) .
Once the 'further question' offairness loses its independent moral
force, owing to the fact that the parties are situated in such a way that
no unfairness conceivably could result, any agreement reached be­
comes a case of pure procedural j ustice; its outcome is fair, 'whatever
it is', in virtue of its agreement alone. Under such conditions, a
contract ceases to be a constitutive convention and becomes instead
an instrument of j ustification .
The aim is to characterize this situation so that the principles that would be
chosen, whatever they tum out to be, are acceptable from a moral point of view.
The original position is defined in such a way that it is a status quo in which
any agreements reached are fair [emphasis added] ( I 20) .

The idea of the original position is to set up a fair procedure so that any
principles agreed to will be just. The aim is to use the notion of pure procedural
justice as a basis of theory [emphasis added] ( I 36) .
Behind the veil of ignorance

But at this point a crucial ambiguity arises, for it is not clear what
exactly it means ' to use the notion of pure procedural j ustice as a
basis of theory ' . Rawls claims that once the situation is appropriately
characterized , then the principles chosen, whatever they tum out to be, are
acceptable from a moral point of view; once the original position is
p roperly defined , then arry agreem ents reached in it are fair; once a fair
p rocedure is established , then any principles agreed to will be j ust.
What is unclear is how generous these provisions are to the
choosers . On one reading, the terms seem generous indeed , the very
embodiment of the voluntarist provisions suggested above . Once the
p arties find themselves in a fair situation, anything goes; the scope
for their choice is unlimited . The results of their deliberations will be
morally acceptable 'whatever they turn out to be' . No matter what
p rinciples they choose, those principles will count as j us t .
But there is another, less expansive reading o f their situation ,
which gives considerably less scope to their enterprise. On this
interpretation, what it means to say that the principles chosen will be
j ust 'whatever they turn out to be' is simply that, given their situa­
tion, the parties are guaranteed to choose the right principles . While
it may be true that, strictly speaking, they can choose any principles
they wish, their situation is designed in such a way that they are
guaranteed to 'wish' to choose only certain principles . On this view,
'any agreements reached ' in the original position are fair, not be­
cause the procedure sanctifies j ust any outcome, but because the
situation guarantees a particular outcome. But if the principles
agreed to are j ust because only ( the) j ust principles can be agreed to,
the voluntarist aspect of the enterprise is not as spacious as would
first appear. The distinction between pure and perfect procedural
j ustice fades , and it becomes unclear whether the proced ure ' trans­
lates its fairness to the outcome' , or whether the fairness of the
procedure is given by the fact that it necessarily leads to the right
result.
Rawls confirms the less voluntarist reading when he writes , 'The
acceptance of these principles is not conj ectured as a psychological
law or probability . Ideally anyway, I should like to show that their
acknowledgement is the only choice [sic] consis ten t with the full

be strictly deductive' ( I 2 I ) . The notion that the full description of

description of the original position . The argument aims eventually to

the original position determines a single 'choice' which the parties

1 27
Contract theory and justification

cannot but acknowledge seems to introduce a cognitive element to

j ustification after all and to call into question the priority of proce­
dure over principle which the contract view - and the deontological
proj ect generally - seemed to require. But a more immediate con­
sequence of this reading is that it complicates our account of wha t
goes on in the original position. Rawls maintains that what happens
behind the veil of ignorance is that a plurality of persons come to a
unanimous agreement on a particular conception of j ustice. I t is
worth examining his description closely.
To begin with, it is clear that since the differences among the parties are
unknown to them, and everyone is equally rational and similarly situated,
each is convinced by the same arguments. Therefore, we can view the choice in the
original position from the standpoint of one person selected at random. if

unanimous agreement can be reached. [emphasis added] ( 1 39) .

a�one after due reflection prefers a conception ojjustice to another, then they all do, and a

Rawls suggests that, to make the circumstances more vivid , we

might imagine that the parties communicate with each other
through a referee, who transmits proposed alternatives , informs the
parties when they have come to an understanding, and so on. 'But
such a referee is actually superfluous, assuming that the delibera­
tions of the parties must be similar' ( 1 3 9 ) .
Thus there follows the very important consequence that the parties have no
basis for bargaining in the usual sense. No one knows his situation in society

principles to his advantage ( 1 39) .

nor his natural assets , and therefore no one is in a position to tailor

The veil of ignorance makes possible a unanimous choice of a particular

ing problem of the original position would be hopelessly complicated ( 1 40) .

conception ofjustice. Without these limitations on knowledge the bargain­

Since the parties are 'similarly situated ' , they are guaranteed to
reason in the same way, and have no basis for bargaining 'in the
usual sense' . This would seem to imply that they have a basis for

result of a fair agreement or bargain . ' ( 1 2 ) ) , but it is difficult to

barga ining in some other sense ( 'The principles of j ustice are the

imagine what this sense could be. Bargaining in aTry sense requires
some difference in the interests or preferences or power or knowledge
of the bargainers , but in the original position, there are none . Under
such conditions , it is difficult to imagine how any bargain, in any
sense, could ever get going.
Behind the veil of ignorance

If no bargaining could take place, the question also arises whether

any discussion could take place. Rawls suggests that various alterna­
tives might be proposed before the final agreement is reached . But if
the parties are assumed to reason in the same way and to be con­
vinced by the same arguments , it seems unlikely that a certain idea
could occur to one but not to another. Discussion , like bargaining,
p res upposes some differences in the perceptions or in teres ts or know­
ledge or concerns of the discussants, but in the original position ,
there are no such differences . We must therefore assume that the
'deliberations ' of the parties proceed in silence and issue in a single
conception which is unanimously agreed to.
But this makes the account of the agreement in the original
p osition more puzzling still. For if there is no basis for bargaining or
discussion, it is doubtful that there can be any basis for agreement,
even a unanimous agreement, either. Rawls states , ' I f anyone after
due reflection prefers a conception ofj ustice to another, then they all
do, and a unanimous agreement can be reached ' [emphasis added]
( 1 39 ) . But why 'and ' ? What does the agreement add once the
discovery has been made? Suppose that everyone, after due reflec­
tion, found that he preferred a particular conception ofjustice, and
suppose further that everyone knew that all preferred the same one.
Would they then go on to agree to this conception ? What would it
mean for them to m ake this discovery first, and then go on to make an
agreement about it? Even if we could imagine what it would mean to
go on to make an agreement under such circumstances, what would
the agreement add to the d iscovery that all preferred the same
conception? Would the conception be j us tified after they 'went on to
make the agreement' in a way that it was not j ustified when they saw
that all preferred the conception but before they ' made the agree­
ment'?
At this point it is important to distinguish two different s e nses of
' agreement' . The first involves agreement with a person (or persons )
with respect t o a proposition, the second agreement t o a proposition .
The first sort of agreement is a kind of ' choosing together' , and
requires a plurality of persons. ( One will not do, except in the
metaphorical sense in which I make an agreement with myself. ) It is
this sort of agreement that is typically engaged in making a contract,
where part of the agreement involves forming an intention. Although
we may speak of two persons agreeing to a contract, what we mean is

1 29
Contract theory and just�fication

that two persons agree with each other to abide by certain terms. The
agreement and the terms , taken together, constitute the contract.
Since agreement in this sense requires an intentional act, or an
exercise of will, we might describe it as agreement in the voluntarist
sense.
The second sort of agreement, an agreement to a proposition, does
not require more than a single person, and does not involve an
exercise of will. I n this sense of agreement, to agree to a proposition
amounts to acknowledging its validity, and this requires neither that
others be involved nor that I take the validity of the proposition to be
a matter of choice. It may be enough that I see it to be valid, as when I
agree to (or accept, or acknowledge) the proposition that 2 + 2 = 4·
To agree in this sense is to grasp something already there. Although I
may say I have 'decided' that the answer to this difficult problem in
mathematics is 'x' , it is not a decision that decides anything except
whether I have got it right. Since agreement in this second sense is a
question of knowing rather than willing, we might describe it as
agreement in the cognitive sense.
Once this distinction is borne in mind, Rawls' account of the
original agreement appears in a new light . Passages that first seemed
to describe an agreement in the voluntarist sense can now be seen to
admit a cognitive interpretation as well. Where at first Rawls writes
as though ' the choice . . . determines the principles ofj ustice' [emphasis
added] ( 1 2 ) , in other places he writes as though the parties have
merely to acknowledge principles already there .
The relevant agreement is not to enter a given society or to adopt a give n
form of government, but to accept certain moral principles [emphasis added]
( I6).

I argue that the two principles :would be acknowledged [in the original
position] [emphasis added] ( I I 8) .

They are the principles that free and rational perspns concerned to further
their own interests would accept in an initial position of equality [emphasis
added] ( I I ) .

[The members of a well-ordered society] could all view their arrangements

as meeting the stipulations which they would acknowledge in an initial
situation that embodies widely accepted and reasonable constraints on the
choice of principles [emphasis added] ( I 3) .
Behind the veil of ignorance
Thus men exhibit their freedom, their independence from the contingencies
of nature and society, by acting in ways they would acknowledge in the
original position [emphasis added] ( 2 56 ) .

I ronically, the Kantian interpretation o fj ustice a s fairness high­

lights the shift from the voluntarist interpretation to the cognitive
one. Although some reference to choice remains, the parties are
described less as willing agents than as subjects who perceive the
world in a certain way .
My suggestion is that we think of the original position as the point of view
from which noumenal selves see the world. The parties qua noumenal selves
have complete freedom to choose whatever principles they wish; but they
also have a desire to express their nature as rational and equal members of
the intelligible realm with precisely this liberty to choose, that is , as beings
who can look at the world in this way and express this perspective in their life as
members of society [emphasis added] ( 255) .

I n both Rawls' theory of the person and his account of j ustifica­

tion, the assumptions of choice and of plurality have stood together
as central features of the conception . As the voluntarist interpreta­
tion of the original position gives way to a cognitive one, the
assumption of plurality is called into question as well . Rawls speaks
throughout of the parties to the original position and in the Kantian
interpretation even speaks of noumenal selves. But since the veil of
ignorance has the effect of depriving the parties , qua parties to the
original position, of all distinguishing characteristics , it becomes
difficult to see what their plurality could possibly consist in.
Rawls acknowledges this condition in part when he writes that ' all
are similarly situated ' ( 1 2 ) , and that in this way a unanimous
agreement is guaranteed . But once all individuating characteristics
are excluded , the parties are not merely similarly situated (as persons
in real life with similar life circumstances and certain overlapping
interests ) , but identically situated . And as we have seen, Rawls' own
theory of the person acknowledges that no two subjects could ever be
regarded as identically situated and still count as distinguishable
persons . The notion that not persons but only a single subject is to be
found behind the veil of ignorance would explain why no bargaining
or discussion can take place there. It would also explain why there
can be no contract or agreement in the voluntarist sense. For
contracts, like discussions, require a plurality of persons, and when
Contract theory and justification

the veil of ignorance descends , this plurality dissolves .

At the beginning of the book, and again at the end , Rawls asks
why, if the original position is merely hypothetical , we should take
any interest in it, moral or otherwise. His answer in each case is that
' the conditions embodied in the description of the original position
are ones that we do in fact accept . Or if we do not, then perhaps we
can be persuaded to do so by philosophical refiecton' ( 2 1 , also 58 7 ) .
The philosophical considerations by which Rawls would persuade
us set out from the contractarian tradition . The well-ordered society

scheme' ( 1 3 ) . But what begins as an ethic of choice and consent ends,

he recommends ' comes as close as a society can to being a voluntary

however unwi t tingly, as an ethic of insigh t and self-understanding.

I n the final passage of the book, the language of choosing and willing
is displaced by the language of seeing and perceiving, as the volun­
tarist image of Kant gives way to the cognitive image of Spinoza.

from the required point of view . . . . Thus to see our place in society from the
Once we grasp this conception , we can at any time look at the social world

perspective of this position is to see it sub specie aeternitatis: it is to regard the

human situation not only from all social but also from all temporal points of
view. The perspective of eterniry is not a perspective from a certain place beyond
the world, nor the point of view of a transcendent being; rather it is a certain
form of thought and feeling that rational persons can adopt within the world . .
. . Purity of heart, if one could attain it, would be to see clearly and to act with
grace and self-command from this point of view [emphasis added] ( 58 7 ) .

The secret to the original position - and the key to its j ustificatory
force - lies not in what they do there but rather in what they
apprehend there . What matters is not what they choose but what they
see, not what they decide but what they discover. What goes on in the
original position is not a contract after all, but the coming to
self-awareness of an intersubj ective being.
4

Justice and the Good

We set out to assess Rawls' claim for the primacy ofjustice and found
that it required a certain conception of the moral subject. We sought
then to examine this conception in the light ofRawls' moral theory as
a whole to check for its consistency with that theory and for its
plausibility generally . We hoped eventually in this way to assess first
Rawls' theory of the subject and finally the claim for the primacy of
j ustice it must support.
Thus far we have considered Rawls' theory of the subject primari­
ly in relation to his theory of j ustice, or conception of right . But as
Rawls points out, a full moral theory must give some account of the
good as well as the right, and the final third of his book seeks to
provide one. Indeed the primacy of j ustice is itself a claim not only
about j ustice but about the relation ofj ustice to those virtues falling
under the concept of the good . So before we can assess this ultimate
claim, we must consider Rawls ' theory of the subj ect in relation to
this theory of the good as well.

T H E UNITY O F T H E S E L F

We might begin by recalling the main points of correspondence

between Rawls' moral theory on the one hand and his theory of the
subj ect on the other. Where the morality of right corresponds to the
bounds of the self and speaks to that which distinguishes us, the
morality of good corresponds to the unity of persons and speaks to
that which connects us. On a deontological ethic, where the right is
prior to the good, this means that what separates us is in some
important sense prior to what connects us - epistemologically prior
as well as morally prior. We are distinct individuals first, and then we
form relationships and engage in co-operative arrangements with
others; hence the priority of plurality over unity. We are barren
subj ects of possession first, and then we choose the ends we would
possess; hence the priority of the self over its ends.
These in brief are the interlocking claims of moral theory and
1 33
justice and the good

philosophical a n t h ropolgy on which deontological l i bera l i s m has

been seen to depend . In consideri n g Rawls ' theory of the s u bj ect
from the standpoint of the righ t , we have focused on the d i s tinctness
of the self and t h e con s t i t u tion of i t s bound s . In con sid ering Rawls '
th eory of the s u bj ect now from the standpoint of the goo d , o u r foc u s
wi ll s h i ft to the u n i ty of the s e l f and the q u estion how i t s bounds m ay
be negotiated or traversed . This q u estion comes in two parts . The
fi rs t concerns Raw l s ' th eory of com m u n i ty and i ts acco u n t of how
an tecedently individu ated persons come to j oi n together i n social
u n i o n . The second concerns h i s th eory of age n cy and it s acco u n t of
how s u bj ects of possession bounded i n advance come to acq u i re their
p u rposes and end s .
W e have spoken of t h e two central features o f t h e self - i t s
d i s tinctness and i ts u n ity - as though e a c h w e r e in s o m e s e n s e
s e l f- s u fficien t , as t h o u g h e a c h c o u l d be d escribed independently o f
t h e other. B u t i t i s d i ffi c u l t i n practice to observe t h e d i s tinction
between these two fea t u res of the s e l f without also remarking their
i n ternal connecti o n . Even as we have focused on the b ound s of the
s e l f and the con ception of righ t , we h ave already had occasion to seek
a principle of u n i ty and to a n t i cipate a theory of com m u n i ty that
m i g ht prov i d e it. Even from the s tandpoint of righ t the bounds of the
s e l f as posi ted by Rawls have been seen to give way .
I n so far as the d i fference principle has been seen to req uire a wider
s u bj ect of pos s e s s i o n , the pri n ci p l e s of j u stice h ave s p i l led over the
bounds of the an tecedently individu ated s u bj ec t , so to speak, and
relied i n advance on a form of u n i ty reserved by Rawls to the
p rovi nce of t h e good . Where Rawls would fix the identities of persons
i n d ependently of their com mo n a l i ty and define the righ t without
reference to the ( fu l l th eory of t h e ) good , the notion that the
d i fference principle re l ie s on a theory of com m u n ity ' from the s tart'
would deny these priorities i n i m portant ways . Where Rawls sees a
th eory of t h e good as a com p l e m e n t rather than a prereq u i s i t e of
j u s ti c e , d esigned to s how its s t a b i l i ty and its tendency to generate i t s
own s u pport, t h e w i d e r n o t i o n of possession i m p l icit i n t h e d i fference
principle would req uire certai n theories of co m m u n i ty and agen cy a t
t h e fou n d ation of j u s tice and n o t only at its peri m e t e r .
We n e e d t h e refore a s s e s s Rawl s ' th eory of the good , and i n
particular h i s a c c o u n t s of co m m u n i ty a n d agency , not only for t h e i r
plausibility gen erally , but also for their ability to provid e the sort o f

1 34
The case of affirmative action

account the theory ofj us tice requires for its completion . I shall try to
show that Rawls' conception fails in both respects , and for similar
reasons . But before taking up his theory of the good directly, it may
be helpful to consider a concrete illustration of what exactly is at
stake for j ustice in a theory of community, and what goes wrong with
the liberal position when it tries to do without one . For this purpose I
propose to consider an argument by Ronald Dworkin in favor of
affirmative action, or preferential treatment of minorities in univer­
sity admissions. Although Dworkin's argument is not identical to the
one Rawls might make, it has much in common with Rawls' general
view of merit, desert, and the nature of the moral subj ect, and serves
to highlight the deontological assumptions with which we are
concerned .

T H E CASE OF AFFIRMATIVE ACTION

Dworkin defends affirmative-action admissions policies for profes­

sional schools such as medicine and law on the grounds that they are
an effective, or at least possibly effective means to a desirable social
goal , namely to increase the presence of blacks and other minorities
in these socially strategic professions, and so eventually ' to reduce
the degree to which American society is over-all a racially conscious
society' ( 1 9 7 7 b : 1 1 ) . H is basic argument is an argument of social
utility. Affirmative action is j ustified , not because those who are
given preference are entitled to an advantage, whether in compensa­
tion for past discrimination or for any other reason , but simply
because ' helping them is now an effective way of attacking a national
problem ' ( 1 9 7 7 b: 1 2 ) .
But Dworkin, like Rawls , believes that no social policy can be
j ustified , however well it serves the general welfare, if it violates
individual rights . He therefore considers the argument that affirma­
tive action violates the rights of those whites it puts at a disadvantage
and in some cases excludes . He concludes in the negative : the idea
that preferential treatment 'presents a conflict between a desirable
social goal and important individual rights is a piece of intellectual
confusion' ( 1 9 7 7 b: 1 2 ) .
One version of the argument Dworkin considers is a claim that
taking race into account is unfair because it fixes on a quality beyond
a person ' s control . Dworkin answers that this does not distinguish
1 35
justice and the good

race as a criterion but applies equally to most standards typically

used in college and university admissions, including intelligence.
While it is true that persons do not choose their race,
it is also true that those who score low in aptitude or admissions tests do not
choose their levels of intelligence. Nor do those denied admission because
they are too old , or because they do not come from a part of the country
underrepresented in the school, or because they cannot play basketball
well, choose not to have the qualities that made the difference ( 1 9 77b: 1 5 ) .

Race may seem a different factor because exclusions based on race

have historically expressed prej udice or contempt for the excluded
race as such. But whites excluded as a result of affirmative action are
excluded not out of contempt but only on the same sort of in­
strumental calculation that justifies the more familiar criteria. While
it is true that a white with marginal test scores would have been
accepted if he were black, 'it is also true, and in exactly the same
sense, that he would have been accepted if he had been more
intelligent, or made a better impression in his interview . . . . Race is
not, in his case, a different matter from these other factors equally
beyond his control' ( 1 97 7b: 1 5 ) .
Another version of the argument Dworkin considers is the claim
that by admitting blacks with lower test scores than those achieved
by some whites who are excluded, affirmative action violates the
right of applicants to be j udged on the basis of merit. Dworkin
responds that what counts as merit cannot be determined in the
abstract but depends on those qualities deemed relevant to the social
purpose the institution serves . I n the case of medical and law
schools, intelligence as measured by standardized tests may well be
among the relevant characteristics, but it is by no means the only
appropriate consideration, as the long-standing practice of admis­
sions committees attests . Other attributes of person and background
are typically weighed in assessing the likely ability of the applicant to
perform the needed function, and where being black is relevant to the
social purpose at hand, being black must count as merit as well .

There is no combination of abilities and skills and traits that constitutes

'merit' in the abstract; if quick hands count as 'merit' in the case of a
prospective surgeon, this is because quick hands will enable him to serve the
public better and for no other reason. If a black skin will, as a matter of
regrettable fact, enable another doctor to do a differen t medical job better,
The case of affirmative action

then that black skin is by the same token 'merit' as well ( I 97 7 b: I 3 ) .

Dworkin acknowledges that some may find dangerous the argument

counting race as a form of merit, but 'only because they confuse its
conclusion - that black skin may be a socially useful trait in
particular circumstances - with the very different and despicable
idea that one race may be inherently more worthy than another'
( 1 9 7 7b: 1 2 ) .
Implicit in much of Dworkin's argument is the idea that no one
can j ustly claim his rights are violated by affirmative action prog­
rams , because no one, white or black, deserves to go to medical school
or law school to begin with; no one has an antecedent right to be
admitted . To be sure, those who meet to the fullest extent the
conditions established for admission are entitled to be admitted , and it
would be wrong to exclude them . But it cannot be said that they or
any others deserve to be admitted, for at least two reasons . First, their
having the relevant characteristics is in most cases no doing of theirs;
their native intelligence, family environment, social and cultural
opportunites and so on are for the most part factors beyond their
control, a matter of good fortune. And in any case, no one is entitled
that medical schools or law schools undertake to reward any particu­
lar kind of qualifications in the first place. What counts as a
qualification for any particular task depends on the qualities that
task happens to require, nothing more. The benefits associated with
the professions are thus not rewards for superior attainment but
incentives to attract the relevant qualities . There can therefore be no
antecedent right to be j udged according to any particular set of
criteria.
From this it seems clear that Dworkin's arguments coincide with
Rawls' theory in several respects . The notion that traditional criteria
of admission, as well as race, are no doing of the applicant recalls
Rawls' argument that the advantages of the fortunate are arbitrary
from a moral point of view. Dworkin's argument that there is no such
thing as 'merit' in the abstract, without reference to the purposes
institutions may define and pursue, parallels Rawls' argument
against meritocracy that the concepts of merit and virtue and moral
worth have no antecedent or pre-institutional moral status and so
cannot provide an independent standpoint from which otherwise
j ust institutions could be criticized . A nd the general implication of
I 37
justice and the good

Dworkin's argument, that no one, black or white, deserves to go to

medical or law school, that no one has an antecedent right to be
admitted , corresponds to Rawls' distinction between moral desert
and legitimate expectations.
Rawls' and Dworkin's positions are similar in a more general
sense as well . Both are rights-based theories, defined in explicit
opposition to utilitarian conceptions , and seek to defend certain
individual claims against the calculus of social interests . But not­
withstanding their individualist aspirations, both rely on a theory of
the subj ect that has the paradoxical effect of confirming the ultimate
frailty, perhaps even incoherence, of the individual whose rights they
seek above all to secure. We have already seen how on Rawls'
conception the self threatens at different points in the argument
either to dissolve into a radically disembodied subject or to collapse
into a radically situated subject. As we shall now see, Dworkin's
argument for affirmative action illustrates how these perplexities,
identified first in the abstract, find consequence in practice.
Central to any case for affirmative action is the ability to disting­
uish discrimination against blacks and other minorities, as in histor­
ic color bars and anti-Jewish quotas, from discrimination in favor of
blacks and other minorities of the kind involved in affirmative action
programs . Dworkin argues that j ustification for the first sort of
discrimination typically depends in part on 'the despicable idea that
one race may be inherently more worthy than another' , while
j ustification for the second depends instead on the utilitarian notion
that society as a whole would gain by having more widely repre­
sentative medical and legal professions .
With respect to the first j ustification, Rawls like Dworkin would
clearly reject the idea that one race may be inherently more worthy
than another. What is striking to recall is why, on Rawls' theory of
the subj ect at least, this despicable idea must be wrong. For Rawls,
the fallacy with the claim that whites are inherently more worthy
than blacks is not that it denies the intrinsic worth of blacks but that
it falsely attributes an intrinsic worth to whites, and so attributes to
them an unfounded claim of desert. The reason is that for Rawls, the
concept of moral worth, like the concept of the good , is 'secondary to
those of right and j ustice, and it plays no role in the substantive
definition of distributive shares' ( 3 1 2- 1 3 ) . Persons can no more have
an intrinsic worth than they can have intrinsic merit or desert, that
The case of affirmative action

is, a worth or merit or desert that is theirs prior to or independent of

what j u s t institutions may attribute to them . And as we have seen, no
one can strictly speaking be said to deserve anything because no one
can be said to have anything, at least not in the undistanced ,
constitutive sense of possession necessary to a desert base. On Rawls'
theory of the subj ect, no person or race can be inherently more
worthy or deserving than another, not because all are intrinsically
worthy and deserving - and eq ually so - but because none is
intrinsically worthy or deserving, and so all claims must equally
await the arrival of j ust institutions .
Some will obj ect to Dworkin's argument for affirmative action -
and to Rawls' theory of j ustice in so far as it supports it - on the
standard meritocratic grounds that the individ ual possesses his
attributes in some unproblematic sense and therefore deserves the
benefits that flow from them , and that part of what it means for an
institu tion or distributive scheme to be j ust is that it rewards
individuals antecedently worthy of reward . But Rawls and Dworkin
present powerful arguments against these assumptions which defen­
ders of meritocracy would be hard-pressed to meet . The difficulty
with Dworkin 's argument, it seems to me, lies elsewhere; it concerns
the possible alternative visions of the subj ect that remai!l once the
meritocratic conception of the individ ual is rej ected . And this re­
turns u s to the problem of the bounds of the self.
We h ave already considered the difficulties associated with the
notion of a person essentially dispossessed , barren of constituent
features , without intrinsic worth or desert, and wholly dependent for
his life prospects on the rights and opportunities institutions of
justice may dispense. We have remarked as well the irony that a
person so morally disempowered should be the prod uct of a liberal
ethic designed to establish the rights of the individual as inviolable.
But if the denial ofindividual desert and the insistence on the bounds
between the self and its attributes lead in the direction of a radically
disembodied subj ect, the notion of common assets poses a different
threat to the integrity of the self in its implication that the bounds
between the self and the other must somehow be relaxed . For unless
some principle of individuation other than a merely empirical one
can be found , the danger here is the drift into a radically situated
subj ect.
On Dworkin's argument for affirmative action , this perplexity
1 39
justice and the good

takes the following form : Once admission or exclusion cannot plaus­

ibly be seen to depend on a notion of 'merit' in the abstract or on an
antecedent individual claim , the alternative is to assume that the
collective ends of the society as a whole should automatically prevail .
But the bounds o f t h e relevant society are never established , its
status as the appropriate subj ect of possession never confirmed .
Once the self, qua individual self, is dispossessed, the claims of the
individual fade to betray an underlying utilitarianism which is never
j ustified . And as Rawls implies early on, utilitarianism is in a sense
the ethic of the unbounded subj ect, the ethic that fails to take
seriously the distinction between persons.
For Dworkin, however, utilitarian considerations are precisely the
ones that distinguish the legitimate discrimination involved in
affirmative action from the unj ustifiable sort based on prej udice and
contempt. If it cannot be said that some are inherently more worthy
than others, it can at least be said that some are more valuable than
others with respect to the social purposes at hand, and discrimina­
tion on this basis is j ustifiable. So long as a policy of preferential
treatment uses people for the sake of worthy ends rather than judges
people as more or less worthy in themselves, it is permissible. So long
as an exclusion based on race is motivated not by prej udice but by an
'instrumental calculation ' , a ' rational calculation about the socially
most beneficial use of limited resources' , or an idea such as the one
that ' black skin may be a socially useful trait' ( I 9 7 7 b : I 2 ) , the
exclusion is consistent with utilitarian considerations and may be
j us tifiable. A person 's expectations, unless they are founded on
rights in Dworkin's special sense of the term , must always give way
in the face of a ' more general social concern' , as when a small
businessman must go under so that a new and superior road might
be built ( I 9 7 7b : I 5) . Although their disappointment is understand­
able, even worthy of our sympathy, rejected applicants can no more
stand in the way of the medical profession society needs than the
small businessman can stand in the way of the superhighway.
Although Dworkin's argument assumes that where no individual
rights are at s take, social policy is properly decided on utilitarian
grounds, he never says why this should be so. Apart from showing
why utilitarian arguments cannot defeat individual rights, his theory
does not offer an explicit defense of utilitarian ethics as such, and
says little about why utilitarianism should prevail when individual
The case of affirmative action

rights are not involved . Dworkin may not feel the need to j ustify his
underlying utilitarian assumptions because they seem on the surface
to have a certain self-evident appeal . If no individual has an antece­
dent claim to the benefits of his accidentally-given assets and
endowments , it might seem natural to suppose that the society as a
whole therefore does . But as we saw in the discussion of common
assets and the difference principle, this assumption is without war­
rant. The arbitrariness of an individual's assets argues only against
the proposition that the individual owns them or has a privileged
claim to their benefits , not in favor of the proposition that some
particular society owns them or has a privileged claim with respect to
them. And unless this second proposition can be established , there
would seem no grounds for favoring a utilitarian dispensation of such
assets and endowments rather than j ust letting them lie where they
fall.
Without some conception of a wider subject of possession , such as
Rawls' notion of common assets seems also to req uire, there would
seem no obvious reason why these assets should be made to serve
general social ends rather than individual ones . To the contrary; in
the absence of some wider subj ect of possession, to regard ' my'
abilities and endowments as mere instruments of a wider social
purpose is to use me as a means to others' ends, and thus to violate a
central Rawlsian and Kantian moral injunction .
The moral oddness of basing university admissions on the
assumption Rawls and Dworkin suggest, whether or not affirmative
action is involved , might be illustrated by imagining the following
letters of rej ection and acceptance written to convey the moral basis
of the policy they recommend :

Dear ( U n successful) Applicant,

We regret to inform you that your application for admission has been
rej ected . Please understand that we intend no offense by our decision . Your
rej ection indicates neither that we hold you in contempt nor even that we
regard you as less deserving of admission than those who were accepted .
I t is not your fault that when you came along society happened not to
need the qualities you had to offer. Those admitted instead of you were not
themselves deserving of a place, nor worthy of praise for the factors that led
to their admission. We are in any case only using them - and you - as
instruments of a wider social purpose.
You will likely find this news disappointing in the sense that your hopes of
Justice and the good

reaping the benefits given those whose qualities do coincide with society 's
needs at any given momen t will not be realized . But this sort of disappoint­
ment occurs whenever an individual's preferences must give way to socie­
ty 's preferences, and should not be exaggera ted by the though t that you r
rej ection reflects in any way on your intrinsic moral worth; please be
assured that those who were admitted are intrinsically as worthless as you .
You have our symp� thy in the sense that it is too bad you did not happen
to have the qualities society happened to want when you applied . Better
luck next time. Sincerely yours . . .

Dear ( Successful) Applicant,

We are pleased to inform you that your application for admission has been
accepted . Through no doing of your own , it turns out that you happen to
have the traits that society needs at the momen t, so we propose to exploit
your assets for society's advantage by admitting you to the study of
medicine/law.
No praise is intended or to be inferred from this decision , as your having
the relevant qualities is arbitrary from a moral point of view . You are to be
congratulated , not in the sense that you deserve credit for having the
qualities that led to your admission - you do not - but only in the sense that
the winner of a lottery is to be congratulated . You are lucky to have come
along with the right traits at the right moment, and if you choose to accept
our offer you will ultimately be entitled to the benefits that attach to being
used in this way. For this, you may properly celebrate.
You, or more likely your parents, may be tempted to celebrate in the
further sense that you take this admission to reflect favorably, if not on your
native endowments, at least on the conscientious effort you have made to
cultivate your abilities and overcome the obstacles to your achievements .
But the assumption that you deserve even the su perior character necessary
to your effort is equally problematic, for your character also depends on
fortunate circumstances of various kinds for which you can claim no credit.
The notion of desert seems not to apply to your case.
We look forward nonetheless to seeing you in the fall. Sincerely yours . . .

As these letters suggest, the policy Rawls and Dworkin defend can be
trou bling even for those who do not hold the meritocratic assump­
tions they effectively call into question . One can imagine, for exam­
ple, a response along the following lines .
I do not claim that I , as an individual, either possess ( in any
exclusive sense) the assets with which I am endowed , or that I have
any special moral claim on the fruits of their exercise . I acknowledge
that I am ind ebted in a complex variety of ways for the constitution
The case of affirmative action

of my identity - to parents, family, city, tri be, class , nation , culture,

historical epoch, possibly God , Nature, and maybe chance - and I
can therefore claim little or no credit ( or for that matter, blame) for
having turned out the way I have. Sorting out just who or what is
accountable for this or that part of me is a difficult if at times
indispensable moral activity which after a certain point may become
impossible to complete . But I agree in any case that I do not deserve
to be admitted to any particular opportunity in any antecedent
moral sense, first because I do not possess in my own right the
q ualities that would make me eligible, and second , beca use even if I
did, I would not be entitled that the rules in force reward any
particular set of attributes or qualifications rather than others .
From this it seems reasonable to su ppose that what at firs t glance
appear as ' my' assets are more properly described as common assets
in some sense; since others made me, and in various ways continue to
make me, the person I am, it seems appropriate to regard them, in so
far as I can identify them , as participants in 'my' achievements and
common beneficiaries of the rewards they bring. Where this sense of
participation in the achievements and endeavors of ( certain) others
engages the reflective self-understandings of the participants , we
may come to regard ourselves , over the range of our various activi­
ties, less as individuated subj ects with certain things in common, and
more as members of a wider ( but still determinate) subjectivity, less
as 'others ' and more as participan ts in a common identity, be it a
family or community or class or people or nation .
One consequence of an enlarged self-understanding such as this is
that when 'my' assets or life prospects are enlis ted in the service of a
common endeavor, I am likely to experience this less as a case of
being used for others' ends and more as a way of contributing to the
purposes of a community I regard as my own . The j ustifica tion of my
sacrifice, if it can be called a sacrifice , is not the abstract assurance
that unknown others will gain more than I will lose, but the rather
more compelling notion that by my efforts I con tribute to the
realization of a way of life in which I take pride and with which my
identity is bound . While it would of course remain true that I could not,
as an individual, claim credit for possessing the qualities relevant to the
common endeavor, I could none the less take pride in my fitness to
contribute in this way, and this fitness, perhaps even more than the
benefits I might glean, would be just cause for celebration.
14 3
justice and the good
This is not of course to say that a claim on 'my' resources from j ust
any quarter can be described in this way . The scope of community
ties , however expansive, is not without limit. Even an enlarged self,
conceived as a community, has its bounds , however provisional its
contours may be. The bounds between the self and { some) others are
thus relaxed on the intersubj ective account, but not so completely
relaxed as to give way to a radically situated subj ect. The bounds
that remain are not given by the physical, bodily differences between
individual human beings , but by the capaci ty of the self through
reflection to participate in the constitution of its identity, and where
circumstances permit, to arrive at an expansive self-understanding.
A further feature of the intersubj ective description of common
assets is that it renders the dispossession of the person as it appears
from the individualistic point of view less ultimately disempowering.
While the argument from arbitrariness systematically deprives the
subj ect, qua individual person , of its attributes and possessions,
leaving a self so shorn of empirically identifiable features as to
dissolve into abstraction ( 'The person has disappeared ; only attri­
butes remain . ' ) , the notion of a wider subj ect of possession goes some
way toward reconstituting the person and restoring its powers . If I
cannot be the owner I can at least be the guardian of the assets
located ' here ' , and what is more, a guardian for a community of
which I count myself a mem ber.
None of this is an argument against affirmative action as such . But
it does suggest a further moral issue that Dworkin must address
before his argument for affirmative action can be complete, and that
is the question of how to establish the relevant subject of possession,
or how to identify those among whom the assets I accidentally bear
are properly regarded as common. To put the point another way,
utilitarianism is an ethic of sharing. { I n this respect, it resembles the
difference principle . ) As such it must presuppose some antecedent
bond or tie among those whose satisfactions it would maximize and
whose efforts and expectations it would expend in the process .
Otherwise it is simply a formula for using some as means to others'
ends, a formula deontological liberals are committed to rej ect.
But Dworkin ' s position on this q uestion is ambiguous at best. At
times he speaks as tho � gh no account of a wider subj ect of possession
is req uired , as though it is enough for a utilitarian argument to
succeed that an individual expectation come up against 'some more
The case of affirmative action

general social concern' , where that expectation is not protected as a

matter of right. On this interpretation, I must share 'my' assets wth
'society as a whole' not because this particular society has made me
what I am and so is responsible for these assets and endowments in a
way that I , individually, am not, but rather on the dubious assump­
tion that 'society' is the residuary beneficiary of the free-floating
assets that remain once the individual is dispossesed . This assumes
without argument that 'society' in some indeterminate sense (all of
humankind ?) has a prior claim on whatever assets the individual
does not. But simply because I, as an individual, do not have a
privileged claim on the assets accidentally residing 'here' , it does not
follow that everyone in the world collectively does . For there is no
reason to think in advance that their location in 'society's' province
(or for that matter, within the province of humankind ) is any less
arbitrary from a moral point of view. And if their arbitrariness within
me makes them ineligible to serve my ends, there seems no obvious
reason why their arbitrariness within a particular society should not
make them ineligible to serve that society's ends as well.
Dworkin speaks at other times as though he does have a determin­
ate subject of possession in mind after all, and that it is the nation­
state. He writes, for example, that 'American sociery is currently a
racially conscious society', and that it is the goal of affirmative action
' to reduce the degree to which American sociery is over-all a racially
conscious society' . These programs are said to provide 'an effective

1 1 - 1 2 ) . B u t ifDworkin means t o claim that, for the sake of determin­

way of attacking a national problem' [emphasis added] ( 1 97 7 b :

ing university admissions and career prospects, the purposes of the

national community properly predominate, then he must say a good
deal more about why this should be so. And part of this argument
would have to include some evidence of the nation's responsibility
for having cultivated the qualities and endowments it would now
enlist, its capacity to engage the reflective self-understanding of its
members as the basis of their common identity, and its ability to
claim if not agreement at least allegiance to the purposes that would
arise from this identity. It would need to demonstrate, in short, that
of the various communities and forms of identity, the nation is the
one that is properly entitled to define the common purpose and to
deploy the common assets necessary to its pursuit, at least in so far as
university education and the choice of certain professional careers
1 45
justice and the good

are concerned . I t may or may not be the case that the American
nation today defines a community in the relevant sense; ' but in so far
as Dworkin means to invoke the nation as the relevant subj ect of
possession, it remains for him to show that this is so .
Despite Dworkin's passing references to the nation, both he and
Rawls seem generally to assume that once the rights of the individual
are dealt with, an unspecified social claim predominates without any
account of a determinate community or wider subject of possession
being required . Thus Dworkin speaks of the need to serve the ' more
general social concern ' , and to provide 'what the more general
society most needs' ( 1 9 7 7 b: 1 5) , and Rawls writes of the need to
arrange distributive schemes so as to further ' the common interest'
( 3 1 1 ) , and to serve 'prior and independent social ends ' ( 3 1 3 ) .
We might summarize the difficulties with this assumption as
follows : first, there is no such thing as ' the society as a whole', or 'the
more general society ' , taken in the abs tract, no single 'ultimate'
community whose pre-eminence just goes without argument or
further description. Each of us moves in an indefinite number of
communities, some more inclusive than others , each making diffe­
rent claims on our allegiance, and there is nc. saying in advance
which is the society or community whose purposes s hould govern the
disposition of any particular set of our attributes and endowments .
Second, if there is no such thing as 'the society as a whole', taken in
the abstract, then it would seem unlikely that any particular society,
arbitrarily identified , could have any greater claim to some particu­
lar set of endowments than the individ ual in whom they accidentally
reside, for surely their location within the province of such an
arbitrarily-identified community could be no less arbitrary from a
moral point of view. In particular there would be no obvious reason
why 'more general social concerns ' as such should in all cases defeat
more local or particular concerns merely in virtue of their generality .
I t is interesting t o note in this connection that utilitarianism in its
earlier, theological versions ( as in Tucker and Paley) did offer an
explicit account of the ultimate subject of possession - namely God -

1 An illuminating discussion of the nation as a community can be found in Beer ( 1 g66) . He

distinguishes between the centralization of government and the process of national
integration and points out that the two tendencies, however interdependent, are not
guaranteed to coincide. I n national integration, the nation is 'made more of a community',
and the sense in which its members share a common life deepens ( 1 966: So- 2 ) .
Three COT/Ceptions of communiry

whose purposes necessarily predominated over more local concerns

( Macintyre 1 9 67 : 462-6) . But once utilitarianism turns secular, the
relevant subject of possession is no longer a settled matter, and the
grounds for asserting the precedence of one range of concerns over
another must await some further description of the relevant subj ect
or community and the basis of its claims.
Finally, unless it is possible to identify the relevant community
across which 'my' assets are properly shared and to establish its
credentials, Dworkin's argument for affirmative action and Rawls'
notion of common assets have the effect either of contradicting the
central Kantian and Rawlsian inj unction against using some as
means to others' ends , or evading this contradiction by relaxing
altogether the bounds between the self and the other, thus lapsing
into a radically situated subject.
Having seen a practical illustration of how j ustice, on the deonto­
logical ethic, req uires a notion of community for its very coherence
and not simply to demonstrate its 'congruence and stability' , we
must now consider whether Rawls' conception can supply it. Rawls
writes that 'j ustice as fairness has a central place for the value of
community ' , and claims in this respect a resemblance to the idealist
side of Kant he otherwise rej ects ( 2 64) . The question is whether the
theory of community Rawls provides is capable of completing the
principles ofj ustice as required and also of accounting for the virtue
of community generally.

T H R E E C O N C E PTIONS OF C O M M U N ITY

In assessing Rawls' theory of community it may be helpful to recall

that the individ ualism ofhis conception describes the subject and not
the obj ect of motivations. The interests pursued by the parties to the
original position are not interests in the selfbut interests of a self, and
more specifically the interests of an antecedently individuated self.
By identifying the individualism ofhis theory with the subj ect rather
than the obj ect of desires, Rawls believes he can avoid relying on any
particular theory of human motivations and especially the assump­
tion, common to some traditional liberal theories , that man is by
nature selfish or egoistic. Deriving a theory of justice withou t
reference to any particular motivations or conceptions of the good is
essential to the deontological project and has the further consequ-

1 47
Justice and the good

ence, Rawls believes, of allowing a fuller theory of community than is

available on traditional individualistic assumptions . Where the
content of motivations is left open, it is possible to suppose that
individuals may pursue social or communitarian aims as well as
merely private ones , especially in a society governed by a scheme of
reciprocity that works to affirm their sense of self-esteem .

There is no reason why a well-ordered society should encourage primarily

individualistic values if this means ways of life that lead individuals to
pursue their own way and to have no concern for the interests of others
(although respecting their rights and liberties ) . Normally one would expect
most people to belong to one or more associations and to have at least some
collective ends in this sense ( Rawls 1 9 75: 550) .

I n his discussion of the ' idea of social union' (section 7 9 ) , Rawls

distinguishes two senses of the 'good of community ' . The first is
based on conventional individualist assumptions which take for
granted the self-interested motivations of the agents . This account
conceives community in wholly instrumental terms and evokes the
image of a 'private society ' , where individuals regard social arrange­
ments as a necessary burden and cooperate only for the sake of
pursuing their private ends. From this instrumental conception
Rawls distinguishes his own view of community in which the partici­
pants have certain 'shared final ends' and regard the scheme of
co-operation as a good in itself. Their interests are not uniformly
antagonistic but in some cases complementary and overlapping.
Since Rawls does not assume in advance that all are given to selfish
motivations alone, he does not foreclose the possibility that some
may take account of others ' welfare and seek to promote it. 'We need
not suppose . . . that persons never make substantial sacrifices for one
another, since moved by affection and ties of sentiment they often do.

structure of society' ( 1 78) . .

But such actions are not demanded as a matter ofjustice by the basic

Of the two accounts of community Rawls presents, both are

individualistic, although the way they are individualistic differs in
each case. The instrumental account is individualistic in that the
subj ects of co-operation are assumed to be governed by self­
interested motivations alone, and the good of community consists
solely in the advantages individuals derive from co-operating in
pursuit of their egoistic ends. Rawls' account is individualistic in the
Three conceptionr of communiry

sense of assuming the antecedent individuation of the subj ects of

co-operation, whose actual motivations may include benevolent
aims as well as selfish ones . As a result, the good of community for
Rawls consists not only in the direct benefits of social co-operation
but also in the quality of motivations and ties of sentiment that may
attend this co-operation and be enhanced in the process . Where
community on the first account is wholly external to the aims and
interests of the individuals who comprise it, community on Rawls'
view is partly internal to the subjects , in that i t reaches the feelings
and sentiments of those engaged in a co-operative scheme . In
contrast to the instrumental conception of a community, we might
therefore describe Rawls ' account as the sentimental conception.
But neither the instrumental nor the sentimental account seems
capable of generating the strong theory of community which Rawls'
and Dworkin's arguments seem to require - Rawls' to redeem the
notion of common assets involved in the difference principle, Dwor­
kin's to define the relevant community of sharing in his argument for
affirmative action. As we have seen , both arguments seem to require
for their completion a wider subj ect of possession capable of laying
legitimate claim to the assets necessary to its purposes without using
some as means to others' ends and without collapsing into a radically
situated subj ect. But neither the instrumental nor the sentimental
account of community, presupposing as they do the antecedent
individuation of the subj ect, can offer a way in which the bounds of
the subj ect might be redrawn; neither seems capable of relaxing the
bounds between the self and the other without producing a radically
situated subject.
For this, one would have to imagine a conception of community
that could penetrate the self more profoundly than even the sen­
timental view permits . For while Rawls allows that the good of
community can be internal to the extent of engaging the aims and
values of the self, it cannot be so thoroughgoing as to reach beyond
the motivations to the subj ect of motivations. The good of commu­
nity cannot reach that far, for to do so would be to violate the priority
of the self over its ends, to deny its antecedent individuation, to
reverse the priority of plurality over unity, and to allow the good a
hand in the constitution of the self, which on Rawls' view is reserved
to the concept of right. ( 'The essential unity of the self is already
given by the concept of right' ( 563 ) . )

1 49
justice and the good

A theory of community whose province extended to the subject as

well as the object of motivations would be individ ualistic in neither
the conventional sense nor in Rawls ' . It would resemble Rawls'
conception in that the sense of community would be manifest in the
aims and val ues of the participants - as fraternal sentiments and
fellow-feeling, for example - but would differ from Rawls ' conception
in that community would describe not j ust a feeling but a mode of
self-understanding partly constitutive of the agent's identity . On this
strong view, to say that the members of a society are bound by a sense
of community is not simply to say that a great many of them profess
communi tarian sentiments and pursue communitarian aims, but
rather that they conceive their identity - the subject and not just the
obj ect of their feelings and aspirations - as defined to some extent by
the community of which they are a part. For them , community
describes not j ust what they have as fellow citizens but also what they
are, not a relationship they choose ( as in a volun tary association) but
an attachment they discover, not merely an attribute but a consti­
tuent of their identity . In contrast to the instrumental and sen­
timental conceptions of community, we might describe this strong
view as the constitutive conception .
Despite Rawls' resistance to the constitutive conception of com­
munity and the theory of the subj ect it req uires , we have already seen
how his language seems at times to carry him beyond the sentimental
conception, as if implicitly to acknowledge what we have argued is
the case, that his theory of j ustice depends ultimately for its coher­
ence on precisely the intersubjective dimension he officially rejects .
I n the account of the difference principle we are told that the
distribution of natural talents is best described as a 'common asset' ,
and that inj us tice as fairness men agree to 'share one another' s fate' .
I n the account of social union, the bounds between empirical , bodily
persons seem more attenuated still. Human beings are said to have
'shared final end s ' , and to participate through community 'in the
total sum of the realized natural assets of the others ' . We are thus led
' to the notion of the community of humankind ' , whose boundaries
can be imagined to extend even across time and space, 'for those
widely separated by history and circumstance can nevertheless
co-operate in realizing their common nature' ( 52 7 ) . ' Only in social
union is the individual complete' , for it is here that 'we cease to be
mere fragments' ( 5 2 9 ) . The mem bers of community 'participate in
Three conceptions of communiry

one another's nature ' , and ' the self is realized in the activities of
many selves' (565 ) .
I t is difficult to know how seriously to take these 'intersubjective­
sounding' passages , for much is couched in metaphor, and often the
metaphor is mixed . lntersubjective and individ ualistic images
appear in uneasy, sometime infelicitous combination, as if to betray
the incompatible commitments contending within. Assets described
as 'common' in one passage turn ' collective' in another. A concep­
tion in which men ' share one another's fate' is later re-descri bed as a
principle of ' reciprocity' and ' mutual benefit' . Those who ' partici­
pate in one another's nature' at one point are said elsewhere more
distantly to engage in 'associative activities ' . And those who at one
moment can overcome their partiality and realize their nature only
in community later find their communitarian imperative red uced to
the mere likelihood that they will j oin one or more associations and
'have at least some collective ends in this sense ' . In perhaps the most
conspicuously unsettled imagery of all, community is said at one
point to consist in the fact that 'different persons with similar or
complementary capacities may co-operate so to speak in realizing their
common or matching nature' [emphasis added] ( 5 2 3 ) .
But as the distinction between the sentimental and constitu tive
conceptions of community suggests, the moral vocabulary of com­
munity in the strong sense cannot in all cases be captured by a
conception that 'in its theoretical basis is individ ualistic' . Thus a
'community' cannot always be translated without loss to an 'associa­
tion ' , nor an 'attachment' to a 'relationship ' , nor 'sharing' to 'recip­
rocating' , nor 'participation' to 'co-operation ' , nor what is 'common'
to what is 'collective' . Though Rawls' argument for the priority of
plurality over unity may normally apply to the second of each of
these pairs , it does not necessarily hold for the first. Where 'collec­
tive' assets imply endowments once separately held , now ceded to
society as a whole, 'common ' assets do not necessarily; they need not
logically presuppose a prior individuation . And while 'reciprocity'
implies a principle of exchange and hence a plurality of agents, the
notion of 'sharing' may suggest a solidarity such that no exchange
need be involved , as in sharing a joke, or an aspiration , or an
understanding. And while 'association ' and 'co-operation ' typically
presuppose the antecedent plurali ty of those who join together to
associate or co-operate, 'community' and 'participation' may de-
justice and the good

scribe a form oflife in which the members find themselves commonly

situated 'to begin with ' , their commonality consisting less in rela­
tionships they have entered than in attachments they have found .

So it would appear that community in the strong, constitu tive sense

req uired by both Rawls and Dworkin cannot be accounted for by a
conception that is individualistic even in Rawls' special sense of the
term . For the individualistic account takes the bounds of the subject
as antecedently given and finally fixed , but Rawls and Dworkin
require a conception capable of marking out a wider subj ect of
possession, a conception in which the subj ect is empowered to
participate in the constitution ofits identity. That such a conception
is unavailable on deontological assumptions can be seen as follows .
For a subject to play a role in shaping the contours of its identity
requires a certain faculty of reflection . Will alone is not enough .
What is req uired is a certain capacity for self-knowledge, a capacity
for what we have called agency in the cognitive sense . This can be
seen by recalling the two accounts of agency and possession we
considered in our initial reconstruction of Rawls' theory of the
subj ect. The first account, corresponding to Rawls' conception , took
the bounds of the self as given and related self to ends by agency in its
voluntarist sense, as willing subj ect to objects of choice . This sort of
agency depended on the faculty of will , for it is the will that allows the
self to reach beyond itself, to transcend the bounds that are fixed in
advance, to grasp the ends it would possess and hold them as it
always must, external to itself.
The second account, by contrast, took the bounds of the self as
open and conceived the identity of the subject as the product rather
than the premise of its agency. The relevant agency here was not
voluntarist but cognitive; the self came by its ends not by choice but
by reflection , as knowing (or inquiring) subj ect to object of ( self- )
understanding. The problem here was not the distance o f the self
from its ends, but rather the fact that the self, being unbounded in
advance, was awash with possible purposes and ends, all impinging
indiscriminately on its identity, threatening always to engulf it. The
challenge to the agent was to sort out the limits or the boundaries of
the self, to distinguish the subj ect from its situation , and so to forge
its identity .
For the subj ect whose identity is constitu ted in the light of ends
Three conceptions of communi�

already before it, agency consists less in summoning the will than in
seeking self-understanding. Unlike the capacity for choice, which
enables the self to reach beyond itself, the capacity for reflection
enables the self to turn its lights inward upon itself, to inquire into its
constituent nature, to survey its various attachments and acknow­
ledge their respective claims, to sort out the bounds - now expansive,
now constrained - between the self and the other, to arrive at a
self-understanding less opaque if never perfectly transparent, a
subjectivity less fluid if never finally fixed, and so gradually ,
throughout a lifetime, to participate in the constitution of its identity.
Now the capacity for reflection suggested by the cognitive account
would seem precisely the feature Rawls' 'wider subject of possession'
req uires if it is not to dissolve into a radically situated subj ect, for this
capacity holds out the possibility of arriving at the bounds of the self
without taking them to be given in advance. Indeed , once the
presumed antecedent individuation of the subject is called into
question, the predicament of the self would seem to approach the
dispossession described on the cognitive account, in which the greater
threat to agency is not the distance of the self from its purposes and
ends but rather the surfeit of seemingly indispensable aims which
only sober self-examination can hope to sort out.
But on Rawls' moral epistemology, the scope for reflection would
appear seriously limited . Self-knowledge seems not to be a possibility
in the relevant sense, for the bounds it would define are taken as
given in advance, unreflectively , once and for all , by a principle of
antecedent individuation . But once these bounds are seen to fall
away, there is nothing to take their place. For a subj ect such as
Rawls' the paradigmatic moral question is not 'Who am I ? ' , for the
answer to this question is regarded as self-evident, but rather 'What
ends shall I choose?' , and this is a question addressed to the will.
Rawls' subj ect would thus appear epistemologically impoverished
where the self is concerned , conceptually ill-equipped to engage in
the sort of self-reflection capable of going beyond an attention to its
preferences and desires to contemplate, and so to re-describe, the
subj ect that contains them .
I t seems clear at least that the question of community leads
naturally to the question of reflection, and that in order to assess the
role of reflection in Rawls ' scheme, we need to examine in greater
detail Rawls ' theory of agency, his account of how the self arrives at

1 53
justice and the good

its ends. We have seen that for Rawls the self comes by its ends by
choosing them , or more elaborately, that the self is related to its ends
as willing subj ect to objects of choice, and we have described this
ability to choose as agency in the voluntarist sense. But what exactly
goes on in this moment of choice, and what role, if any, does
reflection play in arriving at it?

A G E N C Y AND THE ROLE O F REFLECTION

For Rawls, the account of agency and ends falls under the conception
of good . Like the right, the good is conceived voluntaristically; it is
founded in choice . As the principles of right are the product of a
collective choice in the original position , conceptions of the good are
the products of individual choices in the real world .
But here there arises an important difference. For while both the
right and the good are founded in choice, the special ( i . e . hypotheti­
cal ) conditions under which the right is 'chosen' mean that actual
persons do not have a hand in it. What counts as right or j ust is not
something we are free to choose, because the principles ofj ustice are
in force from the moment the veil of ignorance disappears, that is,
before any actual choosing can begin. The principles ofj ustice, being
antecedently derived , are not subj ect to our agency; they apply
whether we like them or not.
With the good it is different. Here, each person is free to choose for
himself, free to adopt whatever conception of the good he desires .
Different things are good for different people, and subject only to the
constraints ofjustice, each is 'free to plan his life as he pleases' ( 447 ) .
While there is assumed to be a single 'correct conception of j ustice
from a philosophical point of view' (446) , which everyone must
adhere to, there is no comparably correct conception of the good
from a philosophical point of view, and so each is free to launch out
on his own .
Here then is further illustration of the priority of the right over the
good, in both its moral and epistemological dimensions. The moral
priority consists in the fact that the principles ofjustice limit the concep­
tions of the good individuals may choose to pursue; where a person's
values clash with justice, it is justice that prevails. As Rawls repeatedly
acknowledges, the principles of justice are not compatible with all
conceivable plans oflife, and plans that do not conform must be rejected.
1 54
Ageng and t he role of reflection
I n j ustice as fairness the concept of right is prior to that of the good . I n
contras t with teleological theories , something i s good only i f it fits into ways
of life consistent with the principles of right already on hand ( 3 96) .

I ndeed , even rational plans oflife which determine what things are good for
human beings, the values of human life so to speak, are themselves
constrained by the principles of j ustice ( 398) .

Our way of life, whatever our particular circumstances, must always

conform to the principles ofj ustice that are arrived at independently (449) .

These principles are then given absolute precedence so that they regulate
social insti tutions without question and each frames his plan in conformity
with them. Plans that happen to be out of line must be revised ( 565 ) .

[ Men's] desires and aspirations are restricted from the outset by the
principles ofjustice which specify the boundaries that men's systems of ends
must respect . . . . The priority ofj ustice is accoun ted for, in part, by holding
that the interests requiring the violation ofj ustice have no value. Having no
merit i n the first place, they cannot override its claims ( 3 1 ) .

The priority of right over good provides a meta-ethical back­

ground to the familiar liberal notion that the preferences and
convictions of the majority, however intensely held , cannot defeat a
legitimate claim of individual rights . On a deontological ethic, the
convictions of the majority merely reflect particular conceptions of
the good . As such, they cannot claim to be 'correct from a philo­
sophical point of view', only preferred from a majoritarian point of
view, and no mere preference can override the requirements of
j ustice.

The intense convictions of the majority, if they are indeed mere preferences
without any foundation in the principles ofjustice antecedently established ,
have no weight to begin with. The satisfaction of these feelings has no value
that can be put in the scales against the claims of equal liberty . . . . Against
these principles neither the intensity of feeling nor its being shared by the
majority counts for anything (450) .

Another way of viewing the contrast between the right and the
good is to recall that the good , whether individual or collective,
includes as ingredients various contingencies which are arbitrary
from a moral point of view, while the right is free of such arbitrari-
1 55
Justice and the good

ness. The need , stressed by Rawls, to regulate the distribution of

opportunities and benefits in a way that is not arbitrary from a moral
point of view suggests one reason at least why the right must be prior.
Once the precedence of right is secured , however, there is no
obj ection to the pursuit of conceptions of the good tinged with
contingency and arbitrariness. The principles of justice effectively
domesticate such conceptions and keep them safely within bounds.

There is no obj ection t o fitting rational plans t o these con tingencies, since
the principles ofj ustice have already been chosen and constrain the con tent
of these plans, the ends that they encourage and the means they use ( 449 ) .

Th u s the arbitrary features of plans of life d o not affect these principles, or

how the basic structure is to be arranged . The indeterminacy in the notion
of rationa i ity does not translate itself into legitimate claims that men can
impose on one another . . . . Since men's claims on one another are not
affected , the indeterminacy is relatively innocuous ( 449, 564) .

The priority of right might finally be viewed in terms of its

antecedent derivation , and the need for some ultimately ' unchosen'
background as a precondition of choice in conceptions of the good . If
the principles of justice were themselves up for grabs , then ' the
freedom of choice that j ustice as fairness assures to individuals and
groups within the framework of j ustice' (447) would no longer be
assured . Something must remain beyond choice ( and so constrain it)
if choice itself is to be secured . This is the epistemological priority
that deontological ethics carries over into a moral priority ! Thus the
morality of right, which assures freedom of choice within its bounds,
cannot itself be vulnerable to any choice that would challenge or
restrict it, for the moral force of such a challenge would in all cases
fall short of the moral force of the framework within which it was
conceived . For as we saw in our discussion of contract theory, the
sanction of this framework, unlike the sanction for the aims and
values arising within it, is not simply a choice or even a contract but a
hypothetical agreement conceived under special conditions whose
fairness is independently established .
The epistemological aspect of the priority of right recalls the
2 Compare Kant's argument in the transcendental deduction that concepts of objects in
general must underlie all empirical knowledge as its a priori conditions. 'The objective
validi ty of the categories as a priori rests, therefore, on the fact that, so far as the form of
thought is concerned, through them alone does experience become possible' ( 1 78 1 : 1 26) .
Age119' and the role of reflection

parallel priority of the self over its ends. In both cases, an 'unchosen'
framework, antecedently given, is seen as a prerequisite of choice. As
the principles of j ustice must be antecedently given ( and hence
beyond choice) in order to assure the possibility of choice in concep­
tions of the good , so the bounds of the self must be antecedently given
( and hence beyond choice) in order to assure the agency of the
subj ect, its capacity to choose its ends. While the bounds imposed by
j ustice may seem an undue restriction on choice in that persons
cannot participate in their constitution but can only choose within
them, in fact these bounds secure the equal liberty of each to choose
his ends for himself against the vagaries of a public opinion that
might one day prefer otherwise. Similarly, while the bounds of the
self may seem an undue restriction on agency in that the self cannot
participate in their constitution , they are in fact a prerequisite of
agency. For these are the bounds that hold the world off, so to speak,
and provide the subj ect the detachment it needs to choose for itself.
They secure for the selfthe capacity to choose against the vagaries of
circumstances that would otherwise engulf it. The notion that the
principles of j ustice, like the bounds of the self, provide a basis for
choice that is not itself chosen is not a contradiction but a necessary
presupposition of a subj ect capable of choice. In this it can be
compared with parallel notions running throughout the deontologi­
cal conception, including a subj ect of possession not itself possessed ,
a basis of d esert not itself deserved, and a foundation of contract not
itself contractual.
We have seen that the choice of ends is constrained from the start
by the principles of j ustice antecedently defined . But in order to
describe Rawls' account of choice and to assess the role of reflection ,
if any, we need to know in greater detail how the constraint ofj ustice
makes itself felt, how exactly it enters the deliberation of the agent.
Are the constraints of right somehow built into the activity of
deliberation such that only j ust desires or conceptions of the good
can arise in the first 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 j us tice?
At times Rawls writes as though the principles of j ustice shape a
person's conception of the good from the start, even as the concep­
tion is formulated . ' I n drawing up plans and in deciding on aspira­
tions men are to take these constraints into account . . . [T] heir

1 57
justice and the good

desires and aspirations are restricted from the outset by the princi­

ends must respect' ( 3 I ) . At other times Rawls seems to favor the

ples of justice which specify the boundaries that men's systems of

second account, as when he writes that in j ustice as fairness, persons

'implicitly agree . . . . to conform their conceptions of the good to

directly violate them' [emphasis added] ( 3 I ) .

what the principles ofj ustice require, or at least not to press claims which

If it is unclear whether j ustice intervenes at the point I choose my

life plans or only later, at the point I would pursue them, it seems
clear at least that neither account introduces an element of self­
reflection of the sort we are concerned with; both allow that the right
does not wholly determine my good , that my conception of the good ,
however constrained within a certain range, still remains for me to
choose. At this point Rawls introduces another set of considerations
to narrow our choice, certain ' counting principles ' , as he describes
them, which amount roughly to the basic tenets of instrumental
rationality . These recommend , for example, that I choose more
rather than less effective means to given ends, a more inclusive plan
over a less inclusive one, the plan offering a greater rather than a
lesser probability of success , and so on . Rawls acknowledges , howev­
er, that the counting principles , even when supplemented with
various other principles of rational choice, 'do not suffice to order
plans' ( 4 I 6) , and that the constraints of right and of instrumental
rationality taken together are not enough to lead us to a single
determinate choice . Once these principles run out, we must simply
choose. 'We may narrow the scope of purely preferential choice, but
we cannot eliminate it altogether . . . . We eventually reach a point

further guidance from principle' ( 5 5 2 , 55 I ) .

where we j ust have to d ecide which plan we most prefer without

It is at this point, according to Rawls, that our reflection must be

engaged in order to determine as best we can what things we want and
how much we want them, and to ascertain in the light of all the relevant
facts the plan most likely to realize these desires most completely.

I shall suppose that while rational principles can focus our j udgments and
set up guidelines for reflection, we must finally choose for ourselves in the
sense that the choice often rests on our direct self-knowledge not only of
what things we want but also of how much we want them . Sometimes there
is no way to avoid having to assess the relative intensity of our desires (4 1 6) .
Agerl9' and the role of reflection
We can say that the rational plan for a person is the one ( among those
consistent with the counting principles and other principles of rational
choice once these are established) which he would choose with delibera­
tive rationality. It is the plan that would be decided upon as the outcome
of careful reflection in which the agent reviewed , in the light of all the
relevant facts, what it would be like to carry out these plans and thereby
ascertained the course of action that would bes t realize his more
fundamental desires ( 4 1 7 ) .

Rawls' account of how w e choose would seem to confirm the

limited scope for reflection on his conception . While the plan of life
or conception of the good most appropriate to a particular person is

obj ects of this reflection are restricted to ( I ) the various alternative

said to be ' the outcome of careful reflection' , it is clear that the

plans and their likely consequences for the realization of the agent's
desires , and (2) the agent's wants and desires themselves, and their
relative intensities . In neither case does reflection take as its object
the self qua subj ect of desires . The reflection involved in ( I ) , sizing
up the alternatives and estimating their likely consequences, is
scarcely a form of self-reflection at all; it looks outward rather than
inward , and amounts to a kind of prudential reasoning that could
in principle be carried out with equal or greater success by an
outside expert who knew relatively li ttle about the agen t but a good
deal about the alternatives involved and the sorts of interests and
desires they typically satisfy.
The reflection involved in ( 2 ) , assessing the relative intensity of
desires, looks inward in a sense but not all the way in. It takes as its
obj ects the contingent wants and desires and preferences of the self,
but not the self itself I t does not extend its lights to the self standing
behind the wants and desires it surveys; it cannot reach the self qua
subject of desires . Since for Rawls the faculty of self-reflection is
limited to weighing the relative intensity of existing wants and
desires , the deliberation it entails cannot inquire into the identity of
the agent, ( 'Who am I , really?' ) only into the feelings and
sentiments of the agent ( 'What do I really fee/ like or most prefer?' ) .
Because this sort of deliberation is restricted to assessing the desires
of a subject whose identity is given ( unreflectively) in advance, it
cannot lead to self-understanding in the strong sense which enables
the agent to participate in the constitution of its identity .
Although Rawls does speak briefly o f 'our direct self-knowledge'

r sg
justice and the good

of what things we want and how much we want them , 'self­

knowledge' in this sense seems little more than an awareness of our
immediate wants and desires . And in so far as this self-knowledge is
'direct' in the strict sense that it is given transparently to our
awareness, it is difficult to imagine how anything resembling
reflection or deli beration in the ordinary sense could ever take
place, since we would likely know all we would need to know 'in an
instant', before anything recognizably deliberative could begin . But
even if 'our direct self-knowledge' permits some uncertainty for
reflection to sort out, the self that is known once the uncertainty is
resolved is not really the self in the strict sense distinguished
throughout but merely the contingent accidents and attributes of
the self.
The difference between the sort of reflection that attends to the
desires of the agent alone and the sort that extends to the subj ect of
desires and explores its identity corresponds in part to a distinction
made by Charles Taylor in his account of human agency between
the 'simple weigher' and the 'strong evaluator' . Central to both
distinctions are the images of superficiality and depth . For Taylor,
the 'simple weigher' is reflective in the minimal sense that he is
capable of evaluating courses of action and acting out of his
evaluations. But the reflection of the simple weigher lacks depth in
that his evaluations are limited to the inarticulate 'feel ' of the
alternatives .

Whereas for the simple weigher what is at s take is the desirability of

different consummations, those defined by his de facto desires, for the
strong evaluator reflection also examines the different possible modes of
being of the agent. Whereas a reflection about what we feel like more,
which is all a simple weigher can do in assessing motivations , keeps us as
it were at the periphery; a reflection on the kind of beings we are takes us
to the center of our exis tence as agents . . . . It is in this sense deeper ( I 977:
1 1 4- 1 5) .

For Rawls, reflection 'on the kind of beings we are' rather than on
the kind of desires we have is not a possibility, first because the kind
of beings we are is antecedently given and not subject to revision in
the light of reflection or any other form of agency, and second ,
because Rawls' self is conceived as barren of constituent traits,
possessed only of contingent attributes held always at a certain

1 60
Ageng and the role of choice

distance, and so there is nothing in the self for reflection to survey or

apprehend . For Rawls, the identity of the subject can never be at
stake in moments of choice or deliberation (although its future aims
and attributes may of course be affected ) , for the bounds that define
it are beyond the reach of the agency - whether voluntarist or
cognitive - that would contribute to its transformation.
The distinction between a notion of reflection such as Rawls' ,
which i s limited to the objects o f desire, and one such a s Taylor' s,
which penetrates further to reach the subject of desire, corresponds
to the distinction between the sentimental and the constitutive
conceptions of community we identified earlier. For on the
sentimental conception, the good of community was limited to the
communitarian aims and sentiments of antecedently individuated
subj ects, while on the constitutive conception, the good of comm­
unity was seen to penetrate the person more profoundly so as to
describe not j ust his feeling but a mode of self-understanding partly
constitutive of his identity, partly defininitive of who he was .
W e have seen that Rawls' theory o f justice requires for its
coherence a conception of community in the constitutive sense,
which requires in turn a notion of agency in the cognitive sense,
and we have found that Rawls' theory of the good can allow for
neither. This calls into question the theory of j ustice, or the theory
of the good, or both. But beyond the difficulties they raise for
Rawls' conception as a whole, Rawls' limited accounts of agency
and reflection are implausible in themselves, incapable of making
sense of what choice and deliberation could possibly consist in. Or
so at least I shall try to show, by considering the picture of choice
that remains once reflection in the strong, thoroughgoing sense has
been ruled out.

AGENCY AND THE ROLE OF C H O ICE

As we have seen, Rawls' theory of the good is voluntaristic; our

fundamental aims, values, and conceptions of the good are for us to
choose, and in choosing them, we exercise our agency. As Rawls
describes it, once the principles of rational (i.e. instrumental) choice
run out, ' We must Jinal!Y choose for ourselves in the sense that the choice
often rests on our direct self-knowledge not only of what things we
want but also of how much we want them . . . . It is clearly left to the

161
justice and the good

agent himself to decide what it is that he most wants ' [emphasis

added] ( 4 1 6) . Since the principles of rational choice do not specify a
single best plan of life, 'a great deal remains to be decided . . . . We
eventually reach a point where we just have to decide which plan we
most prefer without further guidance from principle . . . . [W] e may
narrow the scope ofpurely preferential choice, but we cannot eliminate it
altogether . . . . The person himself must make this decision, taking into
account the full range of his inclinations and desires, present and
future' [emphasis added] (44 9, 55 1 , 552, 55 7 ) .
If it is clear that Rawls would describe my values and conceptions
of the good as the products of choice or decision, it remains to be seen
what exactly this choice consists in and how I come to make it.
According to Rawls, we 'choose for ourselves in the sense that the
choice often rests on our direct self-knowledge' of what we want and
how much we want it. But a choice that is a choice 'in the sense that'
it 'often rests on' (is determined by?) my existing wants and desires is
a choice only in a peculiar sense of the word . For assuming with
Rawls that the wants and desires on which my choice 'rests' are not
themselves chosen but are the prod ucts of circums tance, ( 'We do not
choose now what to desire now' (4 1 5 ) ) , such a ' choice' would involve
less a voluntary act than a factual accounting of what these wants
and desires really are . And once I succeed in ascertaining, by 'direct
self-knowledge' , this piece of psychological information, there would
seem nothing left for me to choose. I would have still to match my wants
and desires, thus ascertained, to the best available means of satisfying
them, but this is a prudential question which involves no volition or
exercise of will.
When Rawls writes that it is 'left to the agen t himself to decide what
it is he most wants' (4 1 6) , and that 'we j ust have to decide which plans
we most prefer' (55 1 ) , the 'decision' the agent must make amounts to
nothing more than an estimate or psychic inventory of the wants and
preferences he already has, not a choice of the values he would
profess or the aims he would pursue. As with the collective 'choice' or
'agreement' in the original position, such a 'decision' decides no­
thing except how accurately the agent has perceived something
already there, in this case the shape and intensity of his pre-existing
desires . But if this is so, then the voluntarist aspect of agency would
seem to fade altogether.
To arrive at a plan of life or a conception of the good simply by

1 62
Agen91 and the role of choice

heeding my existing wants and desires is to choose neither the plan

nor the desires ; it is simply to match the ends I already have with
the best available means of satisfying them . Under such a
description, my aims, values , and conceptions of the good are not
the products of choice but the objects of a certain superficial
introspection, j ust 'inward ' enough to survey uncritically the
motives and desires with which the accidents of my circumstance
have left me; I simply know them as I feel them and seek my way as
best I can to their consummation .
It might be suggested that Rawls could escape the apparent
collapse of this account of agency and choice in one of two ways .
The first would be to introduce the idea that persons are capable of
reflecting on their desires not only in the sense of assessing their
intensity but also in the sense of assessing their desirability;
capable, that is, of forming second-order desires , desires whose
obj ects are certain first-order desires ( Frankfurt 1 9 7 1 ) . I may thus
want to have certain desires and not others, or regard certain sorts
of desires as desirable and others less so. The fact that something
was desired (and not unj ust} would no longer be enough to make it
good , for this would depend on the further question whether it was
a desirable sort of desire or not. Once I ascertained what I ( really)
wanted as a matter of first-order desire, it would remain for me to
assess the desirability of my desire and in this sense to affirm or
rej ect it.
I ndeed , Rawls seems vaguely to admit such a possibility when he
writes that although 'we do not choose now what to desire now ' , we
can at least 'choose now which desires we shall have at a later time
. . . . We can certainly decide now to do something that we know
will affect the desires we shall have in the future . . . . Thus we
choose between future desires in the light of our existing desires'
(4 1 5 ) .
But even if a Rawlsian agent were capable of forming desires for
certain other desires , his agency would not in any meaningful sense
be restored . For he would have no grounds , apart from the mere
fact of his second-order desire, on which to justify or defend the
desirability of one sort of desire over another. He would still have
only the psychological fact of his ( now, second-order) preference to
appeal to and only its relative intensity to assess. Neither the
intrinsic worth of a desire nor its essential connection with the
Justice and the good

identity of the agent could provide a basis for affirming it, since on
Rawls' account, the worth of a desire only appears in the light of a
person's good , and the identity of the agent is barren of constituent
traits so that no aim or desire can be essential to it. The affirmation
or rej ection of desires suggested by the formation of second-order
desires would on Rawls' assumptions introduce no further element
of reflection or volition, for such an assessmen t could only reflect a
slightly more complicated estimate of the relative intensity of
pre-existing desires , first- and second-order desires included . The
resulting conception of the good could no more be said to be chosen
than one arising from first-order desires alone.
A second possi ble attempt to restore the coherence of choice on
Rawls' conception might be to imagine a case in which the various
desires of the agent, properly weighed for their respective intensi­
ties , led to a tie, and where deliberation had already taken account
of all relevant preferences such that no further preferences could be
introduced to break the tie. In such a case, this account might
continue, the agent would have no alternative but to plump, just
arbitrarily, one way or the other, without relying on any preference or
desire at all. It might be suggested that a 'choice' thus independent
from the influence of pre-existing wants and desires - a 'radically
free choice' , as it is sometimes described - would allow for the
voluntarist aspect of agency seemingly unavailable when the agent
is bound to 'choose' in conformity with his pre-existing wants and
desires.
But Rawls rej ects a wholly arbitrary form of agency that would
escape the influence of pre-existing wants and desires altogether.
'The notion of radical choice . . . finds no place in j ustice as fairness'
(Rawls I g8o: s68) . Unlike the principles of right, which express the
autonomy of the agent and must be free from contingencies,
conceptions of the good are understood to be heteronomous
throughout. Where incompatible aims arise, Rawls speaks not of
radically free or arbitrary choice, but instead of 'purely preferential
choice', suggesting the form of (non-) agency we first considered . In
any case, the notion of a purely arbitrary 'choice' governed by no
considerations at all is hardly more plausible an account of
voluntarist agency than a 'choice' governed wholly by pre­
determined preferences and desires . Neither 'purely preferential
choice' nor ' purely arbitrary choice' can redeem Rawls' notion of
The status of the good

agency in the voluntarist sense; the first confuses choice with

necessity, the second with caprice. Together they reflect the limited
scope for reflection on Rawls ' account, and the implausible account
of human agency that results .

THE STATUS OF THE GOOD

The difficulty with Rawls ' theory of the good is epistemological as

well as moral, and in this it recalls a problem that arose in
connection with the concept of right - that of distinguishing a
standard of assessment from the thing being assessed . If my
fundamental values and final ends are to enable me, as surely they
must, to evaluate and regulate my immediate wants and desires,
these values and ends must have a sanction independent of the
mere fact that I happen to hold them with a certain intensity . But if
my conception of the good is simply the product of my immediate
wants and desires, there is no reason to suppose that the critical
standpoint it provides is any more worthy or valid than the desires
it seeks to assess; as the product of those desires , it would be
governed by the same contingencies.
Rawls responds to this difficulty in the case of the right by seeking in
justice as fairness an Archimedean point that 'is not at the mercy, so to
speak, of existing wants and interests' (26 1 ) . But as we have seen,
Rawls' concept of right does not extend to private morality,
nor does any other instrument of detachment save the good from
thoroughgoing implication in the agent's existing wan ts and
desires . ' Purely preferential choice' is thoroughly heteronomous
choice, and no person' s values or conception of the good can
possibly reach beyond it. As Rawls strikingly concedes, 'That we
have one conception of the good rather than another is not relevant
from a moral standpoint. In acquiring it we are influenced by the
same sort of contingencies that lead us to rule out a knowledge of
our sex and class' ( 1 97 5 : 5 3 7 ) .
The limited scope for reflection on Rawls' account, and the
problematic, even impoverished theory of the good that results
reveal the extent to which deontological liberalism accepts an
essentially utilitarian account of the good , however its theory of
right may differ. This utilitarian background first appeared in our
discussion of Dworkin' s defense of affirmative action; once no

r 6s
justice and the good

individ ual rights were seen to be at stake, utilitarian considerations

automatically prevailed . Although Dworkin defends what he calls an
'anti-utili tarian concept of right' , the scope of this right is strictly ( if
elusively) circumscribed , such that 'the vas t bulk of the laws that
diminish my li berty are justified on utili tarian grounds as being in
the general interest or for the general welfare' ( 1 9 7 7a: 26g) .'
The utilitarian background to Rawls' conception most clearly
appears in his references to individual moral life. Where justice as
fairness rejects u tilitaranism as the basis of social, or public morality,
it has no apparent argu ment with utilitarianism as the basis of
individ ual, or private morality, the Kantian notion of 'duty to
oneself' to the contrary. Rawls describes the utilitarian account of
private morality, without discernible objection, as follows :

A person quite properly acts , at least when others are not affected , to
achieve his own greatest good , to advance his rational ends as far as possible
. . . . [T] he principle for an individ ual is to advance as far as possi ble his own
welfare, his own system of desires ( 2 3 ) .

To be sure, there is one formal principle that seems to provide a general

answer [to an individual's choice of life plan] . This is the principle to
adopt that plan which maximizes the expected net balance of satisfaction
(4 1 6 ) .

For Rawls , utilitarianism goes wrong not i n conceiving the good

as the satisfaction of arbitrarily-given desires undifferentiated as to
worth - for j ustice as fairness shares in this - but only in being
indifferen t to the way these consummations are spread across
individuals . I ts mistake as he sees it is to adopt 'for so.:: i ety as a
whole the principle of rational choice for one man ' , to com bine ' the
desires of all persons into one coherent system of desire ' , and to
seek its overall satisfaction ( 26- 7 ) . In so doing, it 'fuses' or
'conflates ' all persons into one, it reduces social choice to
'essentially a question of efficient administration' ( as , presumably,
individual choice can properly be reduced ) , and so fails to take
seriously the distinction between persons ( 2 7 , 3 3 ) .
Justice as fairness seeks to remedy these shortcomings by
emphasizing the distinction between persons and by insisting on
the separateness of those diverse 'systems of desires' that utilitar-
3 For a compelling critiq ue of Dworkin's view in this respect, see H . L.A. Hart ( 1 979: 86-9) .

1 66
The status of the good
1amsm conflates . But the grounds for Rawls ' departure from
utilitarianism in this respect are not immediately apparent.
Although he seems firm in his view that to each individual human
being there corresponds exactly one 'system of desires ' , he never
says why this must be so, or what exactly a 'system of desires'
consists in, or why it is wrong to conflate them . I s a 'system of
desires' a set of desires ordered in a certain way, arranged in a
hierarchy of relative worth or essential connection with the identity
of the agent, or is it simply a concatenation of desires arbitrarily
arrayed , distinguishable only by their relative intensity and
accidental location? If it is the second, if a system of desires means
nothing more than an arbitrary collection of desires accidentally
embodied in some particular human being, then it is unclear why
the integrity of such a ' system' should be taken so morally and
metaphysically seriously. If desires can properly be conflated
within persons , why not between persons as well?
I f, on the other hand , what makes a system of desires is a
hierarchical ordering of qualitatively distinguishable desires , then
it would be no more j ustifiable to 'conflate' desires within a person
than between persons, and what is wrong with utilitarianism would
also be wrong, in this respect at least, with j ustice as fairness . The
tendency to conflate desires , whether within persons or between
them, would reflect the failure to order them , or to acknowledge the
qualitative distinctions between them . But this failure cuts across
the distinction between individual and social choice, for there is no
reason to suppose that a 'system of desires ' in this sense corresponds
in all cases to the empirically-individuated person . Communities of
various sorts could count as distinct 'systems of desires ' in this
sense, so long as they were identifiable in part by an order or
structure of shared values partly constitutive of a common identity
or form of life. From this point of view, the utilitarian failure to take
seriously the distinction between persons would appear a mere
symptom of its larger failure to take seriously the qualitative
distinctions of worth between different orders of desires , a failure
rooted in an impoverished account of the good which j ustice as
fairness has been seen to share .
For a deontological doctrine such as Rawls' it might be thought
that viewing the good as wholly mired in contingency, despite its
implausibility generally, would have at least the redeeming
justice and the good

advantage of making the primacy of right all the more compelling.

If the good is nothing more than the indiscriminate satisfaction of
arbitrarily-given preferences , regardless of worth, it is not difficult
to imagine that the right (and for that matter a good many other
sorts of claims) must outweigh it. But in fact the morally
diminished status of the good must inevitably call into question the
status of j ustice as well. For once it is conceded that our
conceptions of the good are morally arbitrary, it becomes difficult
to see why the highest of all ( social) virtues should be the one that
enables us to pursue these arbitrary conceptions ' as fully as
circumstances permit' .

T H E M O R A L EPISTE M O LOGY OF J USTIC E

Our discussion of the good thus brings us back to the question of

justice and the claim for its priority, and with this we return to the
circumstances of justice in the original position . Here, the
distinctness or separateness of persons on which Rawls insists as a
corrective to utilitarianism is installed as the key assumption of
mutual disinterest, the notion that individuals take no interest in
one another's interests ( 2 1 8) . When first we surveyed the condi­
tions in the original position , this assumption in particular and the
empiricist rendering of the circumstances of j ustice in general
seemed to undermine the primacy ofj ustice in various ways : Where
j ustice depended for its virtue on the existence of certain empirical
pre-conditions , the virtue ofjustice was no longer absolute, as truth
to theories, but only cond itional , as physical courage to a war zone;
it presupposed a rival virtue or set of virtues of at least correlative
status; it assumed in certain circumstances a remedial dimension;
finally, where inappropriately displayed , j ustice appeared as a vice
rather than a virtue. In sum, a Humean account of the circumst­
ances of j ustice - such as Rawls explicitly adopts - seemed
incompatible with the privileged status ofj ustice required by Rawls
and defended by Kant only by recourse to a moral metaphysic
Rawls found unacceptable.
Hume's own view ofjustice confirms its partiality, at least in so far as
it is derived from premises which Hume and Rawls seem to share. For
Hume, the circumstances of justice describe certain unfortunate if
unavoidable material and motivational conditions of actual

1 68
The moral epistemology ofjustice

human societies, most notably moderate scarcity and 'limited generosity' .

Toge ther, t h e s e circu mstances demons trate t h e s e n s e in
which the arrival of justice signifies the absence of certain nobler but
rarer virtues.
' I f every man had a tender regard for another, or if nature supplied
abundantly all our wants and desires . . . the jealousy of interest,
which j ustice supposes, could no longer have place'; nor, says Hume,
would there be any occasion for distinctions of property and posses­
sion. 'En crease to a sufficient degree the benevolence of men, or the
bounty of nature, and you render j ustice useless, by supplying its
place with much nobler virtues, and more valuable blessings ' . If
material scarcity were replaced with abundance, 'or if everyone had
the same affection and tender regard for everyone as for himself;
j ustice and inj ustice would be equally unknown among mankind' .
And so, Hume concludes, ' ' tis only from the scanty provision nature
has made for his wants, that j ustice derives its origin' ( 1 73 9 : 494-5) .
For Hume, j ustice cannot be the first virtue of social institutions
(at least not in any categorical sense) , and in some cases is doubtfully
a virtue at all . I n the institution of the family, for example, affections
may be enlarged to such an extent that j ustice is scarcely engaged ,
much less as ' the first virtue ' . And even in the wider society, where
generosity is more limited and justice more extensively engaged , its
virtue can only be accounted for against a background of higher or
nobler virtues whose absence calls justice into being. In so far as
mutual benevolence and enlarged affections could be cultivated
more widely, the need for ' the cautious , jealous virtue of j ustice'
would diminish in proportion, and mankind would be the better for
it. Were scarcity or selfishness overcome altogether, then 'j ustice,
being totally useless . . . could never possibly have place in the
catalogue of virtue' ( 1 7 7 7 : 1 6) , much less the first place to which
Rawls would assign it.
But despite the parallel Rawls himself invites between Hume's
account and his own, the assumption of mutual disinterest has a
different meaning for Rawls. I t does not imply that human beings are
typically governed by 'selfishness and confined generosity ' ; indeed it
is not meant as a claim about human motivations at all . It is rather a
claim about the subj ect of motivations. I t assumes interests of a self,
not necessarily in a self, a subject of possession individuated in
advance and given prior to its ends.

1 69
justice and the good
From this there follow important consequences for the status of
justice. No longer is benevolence prior to j ustice and in some cases
able to supplant it. Since for Rawls , the virtue of j ustice does not
presuppose egoistic motivations to begin with, it need not await the
fading of benevolence to find its occasion, and even the full
flowering of 'enlarged affections ' cannot displace it. Justice ceases
to be merely remedial with respect to the 'nobler virtues' , for its
virtue no longer depends on their absence. To the contrary , where
persons are individuated in Rawls' sense, j ustice not only wins its
independence from prevailing sentiments and motivations , but
comes to stand above them as primary. For given the nature of the
subj ect as Rawls conceives it, j ustice is not merely a sentiment or a
feeling like other, lesser virtues, but above all a framework that
constrains these virtues and is 'regulative' with respect to them .
Therefore in order to realize our nature we have no alternative but to plan
to preserve our sense of justice as governing our other aims . This
sentiment cannot be fulfilled if it is compromised and balanced against
other ends as but one desire among the rest . . . . To the contrary , how far
we succeed in expressing our nature depends on how consistently we act
from our sense of j ustice as finally regulative. What we cannot do is
express our nature by following a plan that views the sense of j ustice as
but one desire to be weighed against others . For this sentiment reveals
what the person is, and to compromise it is not to achieve for the self free

(5 74-5) .
reign but to give way to the contingencies and accidents of the world

We have seen how the priority of j ustice, like the priority of the
self, derives in large part from its freedom from the contingencies
and accidents of the world. This much emerged in our discussion of
right and the bounds of the self. In the light of our discussion of the
good , we can now also see why on Rawls' theory of the subject,
such virtues as benevolence and even love are not self-sufficient
moral ideals but must await j ustice for their completion.
Given the limited role for reflection on Rawls' account, the
virtues of benevolence and love, as features of the good , are forms of
sentiment rather than insight, ways of feeling rather than knowing.
Unlike personal or first-order sentiments and feelings , whose
obj ects are given more or less directly to my awareness, benevo­
lence and love are desires whose obj ect is the good of another. But
given the separateness of persons and the in tractability of the

1 70
T he moral epistemology ofjustice

bounds between them , the content of this good ( that is, the good I
wish another) must be largely opaq ue to me. On Rawls' view, love
is blind , not for its intensity but rather for the opacity of the good
that is the object of its concern . 'The reason why the si tuation
remains obscure is that love and benevolence are second-order
notions: they seek to further the good of beloved individuals that is
already given' ( r g r ) .
I f arriving at one's own good is primarily a matter of surveying
existing preferences and assessing their relative intensities , it is not
the sort of inquiry in which another, even an intimate other, can
readily participate. Only the person himself can 'know' what he
really wants or 'decide' what he most prefers . ' Even when we take
up another's point of view and attempt to estimate what would be
to his advantage, we do so as an adviser, so to speak' (448) , and
given the limited cognitive access Rawls' conception allows , a
rather unprivileged adviser at that .
Although we may at times overcome the difficulty of knowing the
good of a beloved individual whose interests we would advance, the
problem becomes hopelessly compounded when we would extend
our love or benevolence to a plurality of persons whose interests
may conflict. For we could not hope to know their respective goods
well enough to sort them out and assess their relative claims. Even
if benevolence could be as widely cultivated as Hume in his
hypothetical vision suggests, its virtue would still not be self­
sufficient, for it would remain unclear, without more, what the love
of mankind would enj oin . ' I t is quite pointless to say that one is to
j udge the situation as benevolence dictates . This ass umes that we
are wrongly swayed by self-concern . Our problem lies elsewhere.
Benevolence is at sea as long as its many loves are in opposition in
the persons of its many obj ects' ( r go) . Not surprisingly, the anchor
this benevolence requires is s upplied by the virtue of justice;
benevolence, even at its most expansive, depends on justice for its
completion . 'A love of mankind that wishes to preserve the
distinction of persons, to recognize the separateness of life and

aims when the many goods it cherishes are in opposition' ( I g I ) .

experience, will use the two principles of j ustice to determine its

Even in the face of so noble a virtue as the love of mankind , the

primacy of justice prevails, although the love that remains is of an
oddly j udicial spirit.

I7I
Justice and the good

This love is guided by what individuals themselves would consent to in a

fair initial situation which gives them equal representation as moral
persons ( 1 9 1 ) .

Thus we see that the assumption of the mutual disinterestedness of the

parties does not prevent a reasonable interpretation of benevolence and of
the love of mankind within the framework of justice as fairness [emphasis
added) ( 1 92 ) .

For Rawls, the consequences of taking seriously the distinction

between persons are not directly moral but more decisively
epistemological . What the bounds between persons confine is less
the reach of our sentiments - this they do not prej udge - than the
reach of our understanding, of our cognitive access to others . And it
is this epistemic deficit (which derives from the nature of the subj ect)
more than any shortage of benevolence (which is in any case
variable and contingent) that requires j ustice for its remedy and so
accounts for its pre-eminence . Where for Hume, we need j ustice
because we do not love each other well enough , for Rawls we need
j ustice because we cannot know each other well enough for even love
to serve alone.
But as our discussion of agency and reflection suggests, we are
neither as transparen t to ourselves nor as opaq ue to others as
Rawls' moral epistemology requires . If our agency is to consist in
something more than the exercise in 'efficient administration'
which Rawls' account implies, we must be capable of a deeper
introspection than a 'direct self-knowledge' of our immediate wants
and desires allows . But to be capable of a more thoroughgoing
reflection, we cannot be wholly unencumbered subj ects of posses­
sion, individuated in advance and given prior to our ends, but must
be subjects constituted in part by our central aspirations and
attachments, always open , indeed vulnerable, to growth and
transformation in the light of revised self-understandings . And in so
far as our constitutive self-understandings comprehend a wider
subject than the individual alone, whether a family or tribe or city
or class or nation or people, to this extent they define a community
in the constitutive sense. And what marks such a community is not
merely a spirit of benevolence, or the prevalence of communitarian
values, or even certain 'shared final ends' alone, but a common
vocabulary of discourse and a background of implicit practices and

1 72
justice and communiry

understandings within which the opacity of the participants is

reduced if never finally dissolved . In so far as j ustice depends for its
pre-eminence on the separateness or boundedness of persons in the
cognitive sense, its priority would diminish as that opacity faded
and this community deepened .

J USTICE AND C O M M U N ITY

Of any society it can always be asked to what extent It Is JUSt, or

'well-ordered ' in Rawls' sense, and to what extent it is a
community, and the answer can in neither case fully be given by
reference to the sentiments and desires of the participants alone. As
Rawls observes, to ask whether a particular society is j ust is not
simply to ask whether a large number of its members happen to
have among their various desires the desire to act j ustly - although
this may be one feature of a j ust society - but whether the society is
itself a society of a certain kind , ordered in a certain way, such that
justice describes its ' basic structure' and not merely the disposi­
tions of persons within the structure . Thus Rawls writes that
although we call the attitudes and dispositions of persons j ust and
unj ust, for j ustice as fairness the 'primary subj ect of j ustice is the
basic s tructure of society' ( 7 ) . For a society to be j ust in this strong
sense, j ustice must be constitutive of its framework and not simply
an attribute of certain of the participants' plans of life.
Similarly, to ask whether a particular society is a community is
not simply to ask whether a large number of its members happen to
have among their various desires the desire to associate with others
or to promote communitarian aims - although this may be one
feature of a community - but whether the society is itself a society
of a certain kind , ordered in a certain way, such that community
describes its basic structure and not merely the dispositions of
persons within the structure. For a society to be a community in
this strong sense, community must be constitutive of the shared
self-understandings of the participants and embodied in their
institutional arrangements , not simply an attribute of certain of the
participants ' plans of life.
Rawls migh t obj ect that a cons titu tive conception of community
such as this should be rej ected 'for reasons of clarity among others ' ,
o r o n the grounds that it supposes society t o b e ' a n organic whole
1 73
Justice and the good

with a life of its own distinct from and superior to that of all its
members in their relations with one another' ( 2 64) . But a
constitutive conception of community is no more metaphysically
problematic than a constitutive conception ofjustice such as Rawls
defends . For if this notion of community describes a framework of
self-understandings that is distinguishable from and in some sense
prior to the sentiments and dispositions of individuals within the
framework, it is only in the same sense that justice as fairness
describes a ' basic structure' or framework that is likewise dis ting­
uishable from and prior to the sentiments and dispositions of
individuals within it.
If utilitarianism fails to take seriously our distinctness , justice as
fairness fails to take seriously our commonality . I n regarding the
bounds of the self as prior, fixed once and for all, it relegates our
commonality to an aspect of the good , and relegates the good to a
mere contingency, a product of indiscriminate wants and desires
'not relevant from a moral standpoint' . Given a conception of the
good that is diminished in this way , the priority of right would seem
an unexceptionable claim indeed . But utilitarianism gave the good
a bad name, and in adopting it uncritically, j ustice as fairness wins
for deontology a false victory .

1 74
Conclusion
Liberalism and the Limits of J ustice
For j ustice to be the first virtue, certain things must be true of us. We
must be creatures of a certain kind , related to human circumstance
in a certain way . We must stand at a certain distance from our
circumstance, whether as transcendental subj ect in the case of Kant,
or as essentially unencum bered subject of possession in the case of
Rawls . Either way , we must regard ourselves as independent: inde­
pendent from the interests and attachments we may have at any
moment, never identified by our aims but always capable of standing
back to survey and assess and possibly to revise them (Rawls 1 9 79: 7 ;
1 9 8o: 544-s) .

'
DE O N T O L O G Y S LIBERATING P R OJ E C T

Bound up with the notion of an independent self is a vision of the

moral universe this self must inhabit. Unlike classical Greek and
medieval Christian conceptions, the universe of the deontological
ethic is a place devoid ofinheren t meaning, a world 'disenchanted ' in
Max Weber' s phrase, a world without an objective moral order.
Only in a universe empty of telos, such as seventeenth-century
science and philosophy affirmed , ' is it possible to conceive a subject
apart from and prior to its purposes and ends. Only a world
ungoverned by a purposive order leaves principles ofj ustice open to
human construction and conceptions of the good to individual
choice . In this the depth of opposition between deontological liberal­
ism and teleological world views mos t fully appears .
Where neither nature nor cosmos supplies a meaningful order to
be grasped or apprehended , it falls to human subjects to constitute
meaning on their own . This would explain the prominence of
con tract theory from Hobbes onward , and the corresponding
emphasis on voluntarist as against cognitive ethics culminating in
1 For discussion of the moral, political, and epistemological consequences of the seven­
teenth-century scientific revolution and world-view, see Strauss 1 953; Arendt 1 958 :
248-325; Wolin 1 960: 239-85; and Taylor 1 975: 3-50.

1 75
Conclusion: Liberalism and the limits ofjustice

Kant. What can no longer be found remains somehow to be created . •

Rawls describes his own view in this connection as a version of
Kantian 'constructivism ' .
The parties t o the original position d o not agree o n what the moral facts
are, as if there were already such facts . It is not that, being situated
impartially, they have a clear and undistorted view of a prior and
independent moral order. Rather (for constructivism) , there is no such order,
and therefore no such facts apart from the procedure as a whole [emphasis
added] ( 1 g8o: 568) .

Similarly for Kant, the moral law is not a discovery of theoretical

reason but a deliverance of practical reason, the product of pure
will . 'The elementary practical concepts have as their foundation
the form of a pure will given in reason' , and what makes this will
au thori tative is that it legislates in a world where meaning has yet
to arrive. Practical reason finds its advantage over theoretical
reason precisely in this voluntarist faculty, in its capacity to
generate practical precepts directly, without recourse to cognition .
'Since in all precepts of the pure will it is only a question of the
determination of will, ' there is no need for these precepts ' to wait
upon intuitions in order to acquire a meaning. This occurs for the
noteworthy reason that they themselves produce the reality of that to which
they refer' [emphasis added] ( 1 788 : 67-8) .
I t is important to recall that, on the deontological view, the
notion of a self barren of essential aims and attachments does not
imply that we are beings wholly without purpose or incapable of
moral ties, but rather that the values and relations we have are the
products of choice, the possessions of a self given prior to its ends. I t
i s similar with deontology's universe. Though it rej ects the
possibility of an obj ective moral order, this liberalism does not hold
that j u s t anything goes . It affirms j u s tice, not nihilism . The notion
of a universe empty of intrinsic meaning does not, on the
deontological view, imply a world wholly ungoverned by regulative
principles, but rather a moral universe inhabited by subjects
capable of constituting meaning on their own - as agents of
2 As one li beral writer boldly asserts, 'The hard truth is this: There is no moral meaning
hidden in the bowels of the universe . . . . Yet there is no need to be overwhelmed by the
void . We may create our own meanings, you and I' (Ackerman 1 980: 368) . Oddly enough,
he insists nonetheless that liberalism is committed to no particular metaphysic or
epistemology, nor any 'Big Questions of a highly controversial character' ( 356-7 , 36 1 ) .
Deontology 's liberating project

construction in case of the right, as agents of choice in the case of the

good . Qua noumenal selves, or parties to the original position, we
arrive at principles of j ustice; qua actual, individual selves , we
arrive at conceptions of the good . And the principles we construct
as noumenal selves constrain ( but do not determine) the purposes
we choose as individual selves . This reflects the priority of the right
over the good .
The deontological universe and the independent self that moves
within it, taken together, hold out a liberating vision . Freed from
the dictates of nature and the sanction of social roles, the
deontological subj ect is installed as sovereign, cast as the author of
the only moral meanings there are . As inhabitants of a world
without telos, we are free to construct principles of j ustice
unconstrained by an order of value antecedently given . Although
the principles ofj ustice are not strictly speaking a matter of choice,
the society they define 'comes as close as a society can to being a
voluntary scheme' ( 1 3 ) , for they arise from a pure will or act of
cons truction not answerable to a prior moral order. And as
independent selves, we are free to choose our purposes and ends
unconstrained by such an order, or by cus tom or tradition or
inherited s tatus . So long as they are not unj ust, our conceptions of
the good carry weight, whatever they are, simply in virtue of our
having chosen them. We are 'self-originating sources of valid
claims' ( Rawls 1 g8o: 543 ) .
Now j ustice is the virtue that embodies deontology's liberating
vision and allows it to unfold . It embodies this vision by describing
those principles the sovereign subj ect is said to construct while
situated prior to the constitution of all value. It allows the vision to
unfold in that, equipped with these principles, the j ust society
regulates each person's choice of ends in a way compatible with a
similar liberty for all . Citizens governed by j ustice are thus enabled
to realize deontology's liberating project - to exercise their capacity
as 'self-originating sources of valid claims' - as fully as circumst­
ances permit. So the primacy of j ustice at once expresses and
advances the liberating aspirations of the deontological world view
and conception of the self.
But the deontological vision is flawed , both within its own terms
and more generally as an account of our moral experience . Within
its own terms, the deontological self, stripped of all possible
Conclusion: Liberalism and the limits ofjustice

constitutive attachments, is less liberated than disempowered . As

we have seen, neither the right nor the good admits of the
voluntarist derivation deontology requires . As agents of construc­
tion we do not really construct ( chapter 3 ) , and as agents of choice
we do not really choose ( chapter 4) . What goes on behind the veil of
ignorance is not a contract or an agreement but if anything a kind
of discovery; and what goes on in ' purely preferential choice' is less
a choosing of ends than a matching of pre-existing desires,
undifferentiated as to worth, with the best available means of
satisfying them . For the parties to the original position, as for the
parties to ordinary deliberative rationality, the liberating moment
fades before it arrives; the sovereign subject is left at sea in the
circumstances it was thought to command .
The moral frailty of the deontological self also appears at the
level of first-order principles . Here we found that the independent
self, being essentially dispossessed, was too thin to be capable of
desert in the ordinary sense ( chapter 2 ) . For claims of desert
presuppose thickly-constituted selves, beings capable of possession
in the constitutive sense, but the deontological self is wholly
without possessions of this kind . Acknowledging this lack, Rawls
would found entitlements on legitimate expectations instead . If we
are incapable of desert, at least we are entitled that institutions
honor the expectations to which they give rise.
But the difference principle requires more. I t begins with the
thought, congenial to the deontological view, that the assets I have
are only accidentally mine. But it ends by assuming that these
assets are therefore common assets and that society has a prior
claim on the fruits of their exercise. This either disempowers the
deontological self or denies its independence. Either my prospects
are left at the mercy of institutions established for ' prior and
independent social ends' ( 3 1 3 ) , ends which may or may not
coincide with my own , or I must count myself a member of a
community defined in part by those ends, in which case I cease to
be unencumbered by constitutive attachments . Either way, the
difference principle contradicts the liberating aspiration of the
deontological proj ect. We cannot be persons for whom j ustice is
primary and also be persons for whom the difference principle is a
principle of j ustice.
Character, self-knowledge, andfriendship

CHARACTER, S E L F - K N O W L E DG E , AND FRI E N D S H I P

I f the deontological ethic fails to redeem its own liberating promise,

it also fails plausibly to account for certain indispensable aspects of
our moral experience. For deontology insists that we view ourselves
as independent selves, independent in the sense that our identity is
never tied to our aims and attachments . Given our 'moral power to
form , to revise, and rationally to pursue a conception of the good '
( Rawls I g8o : 544) , the continuity of our identity is unproblemati­
cally assured . No transformation of my aims and attachments could
call into question the person I am, for no such allegiances, however
deeply held, could possibly engage my identity to begin with.
But we cannot regard ourselves as independent in this way
without great cost to those loyalties and convictions whose moral
force consists partly in the fact that living by them is inseparable
from understanding ourselves as the particular persons we are - as
members of this family or community or nation or people, as
bearers of this history, as sons and daughters of that revolution, as
citizens of this republic. Allegiances such as these are more than
values I happen to have or aims I 'espou se at any given time' . They
go beyond the obligations I voluntarily incur and the ' natural
duties ' I owe to human beings as such. They allow that to some I
owe more than j ustice requires or even permits, not by reason of
agreements I have made but instead in virtue of those more or less
enduring attachments and commitments which taken together
partly define the person I am.
To imagine a person incapable of constitutive attachments such
as these is not to conceive an ideally free and rational agent, but to
imagine a person wholly without character, without moral depth .
For to have character is to know that I move in a history I neither
summon nor command , which carries consequences none the less
for my choices and conduct. I t draws me closer to some and more
distant from others ; it makes some aims more appropriate, others
less so. As a self-interpreting being, I am able to reflect on my
history and in this sense to distance myself from it, but the distance
is always precarious and provisional, the point of reflection never
finally secured outside the his tory i tself. A person wi th character
thus knows that he is implicated in various ways even as he reflects,
and feels the moral weight of what he knows .

I 79
Conclusion: Liberalism and the limits ofjustice

This makes a difference for agency and self-knowledge. For, as

we have seen , the deontological self, being wholly without
character, is incapable of self-knowledge in any morally serious
sense. Where the self is unencumbered and essentially disposses­
sed , no person is left for self-reflection to reflect upon . This is why,
on the deontological view, deliberation about ends can only be an
exercise in arbitrariness. In the absence of constitutive attach­
ments , deliberation issues in 'purely preferential choice' , which
means the ends we seek, being mired in contingency, 'are not
relevant from a moral standpoint' ( Rawls 1 97 5 : 5 3 7 ) .
When I act out of more or less enduring qualities of character, by
contrast, my choice of ends is not arbitrary in the same way . I n
consulting m y preferences, I have not only to weigh their intensity
but also to assess their suitability to the person I (already) am . I
ask, as I deliberate, not only what I really want but who I really
am, and this last question takes me beyond an attention to my
desires alone to reflect on my identity itself. While the contours of
my identity will in some ways be open and subj ect to revision, they
are not wholly without shape. And the fact that they are not
enables me to discriminate among my more immediate wants and
desires ; some now appear essential, others merely incidental to my
defining proj ects and commitments. Although there may be a
certain ultimate contingency in my having wound up the person I
am - only theology can say for sure - it makes a moral difference
none the Jess that, being the person I am, I affirm these ends rather
than those, turn this way rather than that. While the notion of
constitutive attachments may at first seem an obstacle to agency -
the self, now encumbered , is no longer strictly prior - some relative
fixity of character appears essential to prevent the lapse into
arbitrariness which the deontologicai self is unable to avoid .
The possibility of character in the constitutive sense is also
indispensable to a certain kind of friendship, a friendship marked
by mutual insight as well as sentiment. By any account, friendship
is bound up with certain feelings . We like our friends ; we have
affection for them , and wish them well. We hope that their desires
find satisfaction , that their plans meet with success, and we commit
ourselves in various ways to advancing their end s .
But for persons presumed incapable o f constitutive attachments,
acts of friendship such as these face a powerful constraint. However

1 80
Character, self-krwwledge, andfriendship

much I might hope for the good of a friend and stand ready to
advance it, only the friend himself can know what that good is. This
restricted access to the good of others follows from the limited scope
for self-reflection, which betrays in turn the thinness of the
deontological self to begin with . Where deliberating about my good
means no more than attending to wants and desires given directly
to my awareness, I must do it on my own; it neither requires nor
admits the participation of others . Every act of friendship thus
becomes parasitic on a good identifiable in advance. 'Benevolence

beloved individuals that is already given' ( 1 9 1 ) . Even the friend­

and love are second-order notions: they seek to further the good of

liest sentiments must await a moment of introspection itself

inaccessible to friendship. To expect more of any friend , or to offer
more, can only be a presumption against the ultimate privacy of
self-knowledge.
For persons encumbered in part by a history they share with
others, by contrast, knowing oneself is a more complicated thing. It
is also a less strictly private thing. Where seeking my good is bound
up with exploring my identity and interpreting my life history , the
knowledge I seek is less transparent to me and less opaque to
others . Friendship becomes a way of knowing as well as liking.
Uncertain which path to take, I consult a friend who knows me
well, and together we deliberate, offering and assessing by turns
competing descriptions of the person I am, and of the alternatives I
face as they bear on my identity . To take seriously such
deliberation is to allow that my friend may grasp something I have
missed , may offer a more adequate account of the way my identity
is engaged in the alternatives before me. To adopt this new
description is to see myself in a new way; my old self-image now
seems partial or occluded , and I may say in retrospect that my
friend knew me better than I knew myself. To deliberate with
friends is to admit this possibility, which presupposes in turn a
more richly-constituted self than deontology allows . While there
will of course remain times when friendship requires deference to
the self-image of a friend , however flawed, this too requires insight;
here the need to defer implies the ability to know.
So to see ourselves as deontology would see us is to deprive us of
those qualities of character, reflectiveness, and friendship that
depend on the possibility of constitutive projects and attachments .

181
Conclusion: Liberalism and the limits ofjustice

And to see ourselves as given to commitments such as these is to

admit a deeper commonality than benevolence describes , a
commonality of shared self-understanding as well as 'enlarged
affections ' . As the independent self finds its limits in those aims and
attachments from which it cannot stand apart, so justice finds its
limits in those forms of community that engage the identity as well
as the interests of the participants .
To all of this, deontology might finally reply with a concession
and a distinction: it is one thing to allow that 'citizens in their
personal affairs . . . have attachments and loves that they believe
they would not, or could not, stand apart from ' , that they ' regard it
as unthinkable . . . to view themselves without certain religious and
philosophical convictions and commitments' ( Rawls 1 g8o : 545 ) .
But with public life it i s different. There, no loyalty o r allegiance
could be similarly essential to our sense of who we are. Unlike our
ties to family and friends, no devotion to city or nation, to party or
cause, could possibly run deep enough to be defining. By contrast
with our private identity, our 'public identity' as moral persons 'is
not affected by changes over time' in our conceptions of the good
( Rawls 1 g8o : 544-5) . While we may be thickly-constituted selves in
private, we must be wholly unencumbered selves in public, and it is
there that the primacy of j us tice prevails.
But once we recall the special status of the deontological claim, it
is unclear what the grounds fo r this distinction could be. I t might
seem at first glance a psychological distinction ; detachment comes
more easily in public life, where the ties we have are typically less
compelling; I can more easily step back from, say , my partisan
allegiances than certain personal loyalties and affections. But as we
have seen from the start, deontology's claim for the independence
of the self must be more than a claim of psychology or sociology.
Otherwise, the primacy of j ustice would hang on the degree of
benevolence and fellow-feeling any particular society managed to
inspire . The independence of the self does not mean that I can , as a
psychological matter, summon in this or that circumstance the
detachment required to stand outside my values and ends, rather
that I must regard myself as the bearer of a self distinct from my
values and ends, whatever they may be. It is above all an
epistemological claim, and has little to do with the relative intensity
of feeling associated with public or private relations .
Character, self-knowledge, andfriendship

Understood as an epistemological claim, however, the deontolo­

gical conception of the self cannot admit the distinction required .
Allowing constitutive possibilities where 'private' ends are at stake
would seem unavoidably to allow at least the possibility that
'public' ends could be constitutive as well. Once the bounds of the
self are no longer fixed , individuated in advance and given prior to
experience, there is no saying in principle what sorts of experiences
could shape or reshape them, no guarantee that only ' private' and
never 'public' events could conceivably be decisive.

Not egoists but strangers , sometimes benevolent, make for citizens

of the deontological republic; j ustice finds its occasion because we
cannot know each other, or our ends, well enough to govern by the
common good alone. This condition is not likely to fade altogether,
and so long as it does not, j ustice will be necessary . But neither is it
guaranteed always to predominate, and in so far as it does not,
community will be possible, and an unsettling presence for j ustice.
Liberalism teaches respect for the distance of self and ends, and
when this distance is lost, we are submerged in a circumstance that
ceases to be ours . But by seeking to secure this distance too
completely, liberalism undermines its own insight. By putting the
self beyond the reach of politics , it makes human agency an article
of faith rather than an object of continuing attention and concern, a
premise of politics rather than its precarious achievement. This
misses the pathos of politics and also its most inspiring possibilities .
I t overlooks the danger that when politics goes badly, not only
disappointments but also dislocations are likely to result. And it
forgets the possibility that when politics goes well, we can know a
good in common that we cannot know alone.
Bibliography

Listed are all references indicated in the text as well as some works whose
influence is present in the text although not referred to directly.

Ackerman, B.A. 1 980. Social justice in the Liberal State, New Haven
Arendt, H. 1 958. The Human Condition, Chicago
Aristotle. Nicomachean Ethics, translated by D. Ross, 1 92 5 . London
Politics, ed . E . Barker, 1 946. London
Atiyah, P.S. 1 9 79. The Rise and Fall of Freedom of Contract, Oxford
Barker, E . , ed . 1 948. Social Contract, New York
Bartlet, V. 1 9 1 5 . The Biblical and early Christian idea of property . I n
Property: Its Duties and Rights, e d . C . Gore, London
Beer, S. 1 966. Liberalism and the national idea, The Public Interest ( Fall)
No. 5, 70-82
Bell , D . 1 9 73. The Coming of Post-Industrial Society, Harmondsworth
Bickel, A . M . 1 97 5 . The Morality of Consent, New Haven
Dworkin, R. 1 9 77a. Taking Rights Seriousry, London
1 97 7 b . Why Bakke has no case, New York Review of Books ( November
1 0) , 1 1 - 1 5
1 978. Liberalism. I n Public and Private Morality, ed . S . Hampshire, pp.
1 1 3-43 . Cambridge
Feinberg, J. 1 9 70. Doing and Deserving, Princeton
Frankfurt, H. 1 9 7 1 . Freedom of the will and the concept of a person,
Journal of Philosophy 68, 5-20
Fried , C. 1 9 78. Right and Wrong, Cambridge
1 98 1 . Contract as Promise, Cam bridge
Hampshire, S. 1 959. Thought and Action, London
1 97 2 . Spinoza and the idea of freedom. In Freedom of Mind, pp. 1 83-209.
London
1 97 7 . Two Theories of Morality, Oxford
1 9 78. Morality and pessimism. In Public and Private Morality, ed . S .
Hampshire, pp . 1 -2 2 . Cambridge
Hart, H . L.A. 1 9 79. Between utility and rights. In The Idea of Freedom, ed.
A. Ryan, pp . 7 7-98 . Oxford
Hegel, G.W.F. 1 80 7 . Phenomenology of Spirit, translated by A.V. Miller,
1 9 7 7 . Oxford
Bibliography

1 8�2 1 . Philosophy of Right, translated by T . M . Knox, 1 95 2 . London

1 899. The Philosophy of History, translated by J. Sibree, 1 956. New York
Hume, D. 1 7 39. A Treatise of Human Nature, 2 nd edn , ed . L.A. Selby-Bigge,
1 978. Oxford
1 7 7 7 . An Enquiry Concerning the Principles of Morals, 1 966 edn , La Salle,
I llinois
Kant, I. 1 78 1 . Critique of Pure Reason, 1 st edn , translated by N. Kemp
Smith, 1 929. London
1 785. Groundwork of the Metaphysics of Morals, translated by H. J. Paton,
1 956. New York
1 78 7 . Critique of Pure Reason, 2nd edn , translated by N. Kemp Smith,
1 929. London
1 788. Critique of Practical Reason, translated by L. W . Beck, 1 956.
I ndianapolis
1 793. On the common saying: ' this may be true in theory, but it does
not apply in practice ' . In Kant 's Political Writings, ed. H. Reiss, 1 9 70,
pp. 6 1 -9 2 . Cambridge
1 79 7 . The Metaphysical Elements of justice, translated by J. Ladd, 1 965.
I ndianapolis
Kronman, A. 1 980. Contract law and distributive j ustice, Yale Law journal
89, 472-5 1 1
Locke, J . 1 690. Treatise of Civil Government, ed . C . L . Sherman, 1 93 7 . New
York
Lyons , D. n.d. Nature and soundness of the con tract and coherence
arguments . In Reading Rawls, ed . N. Daniels, pp. 1 4 1 -68 . New York
Macintyre, A. 1 96 7 . Egoism and altruism. In Encyclopedia of Philosophy, ed .
P. Edwards, vol . ii, pp. 462-6. New York
1 98 1 . After Virtue, Notre Dame
Marx, K. Early Writings, translated by R. Livingstone and G. Benton ,
1 9 75· Harmondsworth.
Mill, J . S . 1 849. On Liberty. I n The Utilitarians, 1 97 3 , pp. 47 3-600. Garden
City.
1 863. Utilitarianism. I n The Utilitarians, 1 973, pp. 399-47 2
Murdoch, I . 1 970. The idea o f perfection. I n The Sovereignty of Good, pp.
1 -45 . London
Nagel, T. 1 97 3 . Rawls on j ustice. In Reading Rawls, ed . N. Daniels , pp.
1 - 1 5. New York
Nozick, R. 1 9 7 2 . Coercion. In Philosophy, Politics, and Society, 4th series , eds
P. Laslett, W.G. Runciman and Q. Skinner, pp. I O I -35·
1 9 74. Anarchy, State, and Utopia, New York
Oakeshott, M. 1 962. Rationalism in Politics, London
Bibliography

P i t k i n , H . 1 96 5 . O b l igation and cons e n t - I , A merican Political Science Review

Rawls, J .
6 6 , 990-9
1 97 1 . A Theory ofJustice, Oxford
1 9 7 2 . Reply to Lyons and Teitelman , Jou m al of Philosoph_y 69, 556- 7
' 9 7 5 · Fairness to good n e s s , Philosophical Review 84, 536-54
' 9 7 7 · The basic s t r u c t u re as s u bj e c t , A merican Philosophical Quarter[y 1 4 ,

A well-ordered society .
1 5 9-65
1 979. In Philosophy, Politics, and Society, 5 t h series ,
eds P. Lasl e t t and J . F i s h k i n , p p . 6 - 2 0 . Oxford
1 980 . K a n tian co n s t r u c t i v i s m in moral t h eory , Journal of Philosoph] 77·

S I S- 7 2

Spi n o z a , B.
Strauss, L. ' 95 3 · Natural Right and Histo�y, C h i cago
1 670. A Theologico-Political Treatise, transla ted by R . H . M .
Elwes, 1 9 5 1 . N e w York
1 677. The Ethics, t ra n s l a ted by R . H . M . Elwes . In The Rationalists, 1 960 ,
pp. I 79-408 . Garden C i t y

is
Taylor, C . ' 9 7 5 · Hegel, C a m bridge
' 9 7 7 - What h u m a n agency? In The Self: Ps_vchological and Philosophical
Issues, ed . T . M i s c h e l , pp . 1 0 3 - 3 5 · Oxford
U n ger, R . ' 9 7 5 · Knowledge and Politics, New York
Wol i n , S. 1 96 0 . Politics and Vision, Boston

1 86
Index

Ackerman, B . , 1 0 , 1 76n intersubj ective description of, 79-82, 1 03 ,

affirmative action, 1 35-47, 1 49, 1 65 1 43-6, 1 5 1
agency, 2 2 , 56- g , 6s, 84, 8s, g 3, 94. 1 2 1 , 1 34, community, 1 4, 53, 6o-s, 8o-2, 96, 97-8, 97n,
1 52-64, 1 7 2 , 1 76, • n-8o, 1 83 1 0 1 , 1 02 , 1 03, 1 34-5, 1 43-54, 1 46n,
and choice, 1 6 1 -4 1 6 1 , 1 72-4, 1 78, 1 83
cognitive notion of, 2 2 , 58, 1 2 1 , 1 5 2-3, 1 6 1 constitutive conception of, 62, 63-4, 8o-2,
and possession, 56-9, 84, 93 , 1 2 1 , 1 44, 1 52-3 1 0 1 , 1 49-52 , 1 6 1 , J 72 , 1 7 3
Rawls' theory of, 1 34-5, 1 52-64, 1 7 2 good of, 53, 65, 1 47-9, 1 6 1
and reflection, 1 53-6 1 instrumental conception of, 1 48-5 1
volu ntarist notion of, 2 2 , 58- g , 1 1 7, 1 2 1 -2 , j ustice and, 1 72-4, 1 83
1 52-4, 1 62-4 Rawls' theory of, 6 1 -5, 1 34-5, 1 47-52
agreement, with a person/to a proposition, sentimental conception of, 1 48-5 1 , 1 6 1
1 29-3 1 consent, 1 1 0, 1 1 1 , 1 1 2 , 1 1 4, 1 1 5 , 1 24
altruism, 1 1 , 1 2 , 6o consequentialis m , 3
arbitrariness, 1 7 , 2 1 , 2 3 , 68-7 1 , n. 8 1 , 82-3, constructivism, 1 76-8
85, g o - 1 , 93- 1 0 1 , • • 3- 1 5 , • • g, 1 24-6, contingency, 1 7 , 23, 24, 26-8, 36-40, 68- 7 1 ,
1 3 7 , 1 4 1 -2 , 1 44, • 45, 1 46, • s 6, • 64, 74, 76, n-8, 8s, 93-4. 1 0 • · • • 3- • s.
J 66, 1 67 , 1 80 • • 7-20, 1 24-6, • s6, 1 64, 1 65 , • 66, 1 67 ,
Archimedean point, 1 7 , 2 1 , 2 3 , 24, 26, 40, 1 70, 1 78, 1 80
1 1 5, 1 1 9, 1 65 contract theory, 2 7 , 50, 53. 1 04-3 2 , 1 56-7 , 1 75
Arendt, H . , 1 75n contracts, 1 05-32, 1 56
Aristotle, 86, 8 7 actual, 1 05- 1 6, 1 1 8, 1 1 9, 1 20, 1 22 , 1 24-5,
Atiyah, P. S . , 1 06n, 1 08 1 26
autonomy, 6, g , 1 1 - 1 2 , 1 4, 1 9-20, 2 2 , 3 7 , 38, hypothetical, 1 05, 1 1 1 - 1 3 , 1 1 8, 1 1 9-20,
39. 94-5· 1 00, 1 06-7, 1 08, 1 1 2 , 1 1 4, 1 20, 1 22-32, • s 6-7
1 22 , 1 24, 1 26, 1 64, 1 76, 1 78, 1 82 ; Stt morality of, 1 05- 1 2
also freedom conventionalism, 1 1 3 , 1 1 5- 1 6, 1 1 9, 1 24-6
courage, 3 1 , 32, 33-4, 35
Barker, E., 1 1 6
deontology, 3 - 1 4, 1 5, 1 7- 1 9, 2 1 - 4, 30, 35-40,
Bartlet, V . , 97n
52, 53, sg, 64-s. 66, 1 03 , • • 7. 1 1 8, 1 2 1 ,
Beer, S . , 1 46n
1 22 , • 3 3 , • 47 · 1 52 , • ss. • s6, • s7 . • 74,
Bell, D., 73, 75, 94, 95
• 75-83
benevolence, 1 1 , 1 2 , 3 1 , 3 2 , 34-5, 44-6, 6o, 6 1 ,
1 49. 1 68- 7 2 , 1 8 1 , 1 82 , J 83
desert, 66, 7 1 - 7 , So, 82- 1 03 , g on, 1 3 7-43, 1 57 ,
Bickel, A . , 1 6n
1 78
difference principle, 66, 67, 68, 70-82, 85, 9 1 ,
categorical imperative, 1 4, 3 7 , 38, 39, 1 20 92, g6 , 97, 99, 1 0 1 -3 , 1 04, 1 1 4, • • sn ,
character, 7 1 , 74, 83, 84, 85, 8g, 9 1 , 94, 1 42 , • 34. • 4 • . • 44· • 49· • s o, • 78
1 79-8 1 dispossession, s6-8, 1 53
choice, 9, 1 0, 24-5, 45, 63n, 94 - 5 , 1 02-3, 1 04, distributive j ustice, 66- 1 03 , gon, 1 38, 1 39
1 1 9-20, 1 2 1 , 1 2 2-5, 1 2 7 , 1 28, 1 54. duties, see natural duties
• s 6-66, 1 67 , 1 76-8, 1 8o- • Dworkin, R., g- 1 0, 1 05 , 1 35-47, 1 49, 1 52 ,
Rawls ' theory of, 1 54, 1 56-66, 1 76-8, 1 8o- 1 • 6s, • 6sn
common assets, 70-2, 7 3 . n-82, 95. 98-9, 1 02 ,
1 03 , 1 39· 1 4 1 , 1 43 - 7 . 1 49· 1 50- 1 , 1 78 egalitarianism, 66- 7 7 ; see also equality
Index

t:goism, 1 2 , 53, 54, 6 1 , 1 47-9, 1 69 8 1 , 1 0 1 , 1 03, 1 38, 1 44. 1 47-52

entitlement, 7 1 -2 , 82, 85-8, 96, 98- 1 02 , 1 3 7, interests, in the self/of a self, 20- 1 , 54-5, 1 47,
1 78 1 59-6 1
epistemological priori ty, see priority, episte­
mological j ustice,
epistemology, see moral epistemology circumstances of, 1 4, 28-46, 52, 6o- 1 ;
equal liberty, 1 8- 1 9, 22, 25, 78, 1 2 3, 1 24, 1 55, empiricist interpretation. of, 28-43, 45, 60- 1 ,
1 57 65, 1 68
equality, 1 o, 67-77, 82-3, 93-4, 1 1 4 and community, 1 72-4, 1 83
democratic, 68, 69- 77, 93, 94, 1 1 4 distributive, 66- 1 03, 90n, 1 38, 1 39
liberal, 67-9, 70, 72-7, 82-3, 93-4, 1 1 4 entitlement theory of, 67-8, 96, 98
of opportunity, 68, 69, 70, 73, 93-4 and the good , 2 - 7 , 34, 6o, 1 33 , 1 34-5, 1 38,
of result, 69- 70 1 47-8, 1 54-7 . 1 65-7. 1 73-4
Hume on, 29-30, 3 1 , 32, 35-6, 1 68-9, 1 72
fairness, 1 , 1 6, 24, 25, 33, 75, 1 06, 1 0 7-9, increase in, 32-34
1 1 0- 1 1 , 1 2 1 , 1 23-7 . 1 56-7 limits of, 1 , 1 1 , 1 4, 1 8 1 -2
justice as, 1 5, 1 8, 2 1 , 27, 40, 52, 6 1 , 72-3, Bo- 1 , Mill on, 2, 3-5, 5 n
82, 86-7, 92, I I Q- 1 1 , 1 20, 1 22-6, 1 3 1 , 1 47, moral epistemology of, 1 68- 72
1 50, 1 55. 1 56, 1 58, 1 65, 1 66- 7, 1 7 1 , 1 73-4 primacy of, 1 - 7 , 1 0- 1 1 , 1 3, 1 4, 1 5- 2 3 , 28,
family, 3 1 , 33-4, 69, 1 69 30-5, 36. 40, 49. 55. 59. 64, 65, 66, 78,
Feinberg, J . , 83-4, 88, 90n, 99 88-90, 1 33 , 1 55, 1 68- 7 2 , 1 7 7 , 1 78, 1 82
Frankfurt, H . , 1 63 procedural, pure and perfect, 1 08-9, 1 1 2 ,
fraternity, 3 2 , 34-5 1 2 1 , 1 25, 1 26-7
freedom, Kant on, 5-7, 9, 24, 3 7-9, 1 1 8; see also Rawls ' derivation of principles of, 24-7,
autonomy 38-9, 42-3, 1 05, 1 1 6-32
Fried, C . , 1 0, w6n remedial aspect of, 3 1 -2 , 1 68, 1 69, 1 7 1 -2
friendship, 3 1 , 35, 1 80- 1 retributive,- 89-92 , 90n
good , 1 , 2, 3, 6, 7, g- w, 1 2, 1 3, 1 6, 1 7- 1 9, 5 1 , as a virtue, 2, 1 5 , 2 2 , 28, 29-35, 36, 49, 64,
54-5 . 6o , 62, 63-4 · 65, 67, 76, 94 . 1 1 6- 1 7, 87, 99. 1 68- 7 1 ' 1 75
1 1 9, 1 20, 1 23, 1 33· 1 34-5. 1 38, 1 47-9. see also fairness
1 5 4-7· 1 6 1 -7, 1 7 <>- 1 , 1 73-4 · 1 77, 1 79. j ustification, 2-3, 1 8, So, 1 04, 1 0 7-9, 1 1 1 -32
1 8 1 -2, 1 83 cognitive account of, 1 2 1 -2 , 1 2 7 , 1 32
of community, 53, 65, 1 47-9, 1 6 1 voluntarist account of, 1 2 1 -4, 1 2 7, 1 32

Kant, 1 . , 1 - 1 5, 23-4, 28, 35-40, 43, 49, 54,

j ustice and the, 2 - 7 , 34, 6o, 1 33 , 1 34-5, 1 38,
1 47-8, 1 54-7 . 1 65-7 . 1 73-4
Rawls' conception of, 94, 1 1 6- 1 7, 1 34-5, 57n, 6 3-4, 66, 78, 1 05, 1 1 7-20, 1 3 1 , 1 32 ,
1 38, 1 54-7 . 1 6 1 - 7, 1 70- 1 , 1 82 1 47 , 1 56n, 1 66, 1 68, 1 75-6
status of, 1 65-7 kingdom of ends, 1 4, 1 9, 3 7-8, 40, 1 20
thin theory of, 25-8, 39 Kronman, A . , 1 1 4
utilitarian view of, 5 1 , 1 38, 1 65-7
Grossman, A . , 56n law, primacy of, 1 6n
law of nature, 1 1 6- 1 7, 1 1 9
Hart, H . L.A. , 1 65n legitimate expectations, 7 1 -2 , 86-8, 99, 1 0 1 ,
hedonism, 2 1 1 38, q8
heteronomy, 7 , 9 , 36, 39, 1 1 4- 1 5, 1 26, 1 64, liberal equality, see equality, liberal
1 65 liberalism, 1 - 1 5, 2 1 -4, 1 47-8, 1 55, 1 83
Hobbes, T . , 1 1 7 , 1 75 deontological, 1 - 1 5, 1 8- 1 9, 2 1 -4, 28, 7 7-8,
Hulbert, M . , 1 4n 95. 1 00, 1 1 6- 1 7 , 1 1 8- 1 9, 1 34· 1 44. 1 48,
human agency, see agency 1 55· 1 65, q6
Humboldt, W. von, 8 1 foundations of, 1 -7, 1 3- 1 4, 2 1 -3, 1 1 4-22, 1 75-8
Hume, D. , 1 2- 1 4, 29, 30, 3 1 , 3 2 , 35-6, 40, Kant ' s, 1 , 4-6, 1 3- 1 4, 1 5, 28, 66- 7, 1 1 7-20,
1 68-9, 1 7 1 , 1 7 2 1 76
Locke's, 2, 1 1 6- 1 9
idealism, see transcendental idealism Mill's, 2-4
identity, see self, identity of Rawls', 1 3- 1 4, 1 5 , 1 8- 1 9, 2 1 -4, 28, 66-7, 95 •
individ ualism, 1 1 - 1 2 , 50, 5 1 , 53, 60-5, 66- 7, 1 1 6-20, 1 47-8, 1 76

1 88
Index
libertarianism, 66- 7 2 , 1 1 4; set also natural and agency, s6-g, 84, 93, 1 2 1 , 1 44, 1 52-3
liberty in constitutive sense, 62, So, 82, Bs-6, 92-3,
Locke, J . , 2, 66, 1 05, 1 1 6- J g 95. 1 00, 1 0 1 , 1 34. 1 39. 1 78
Lyons, D . , 1 04 and dispossession, s6-8, 1 53
in distancing sense, 20, 54-9, 62, 85, 1 52
Macintyre , A . , 1 47 as guardianship, 82, g6- 7, 1 02 , 1 44
merit, 7 2 , 1 35, 1 36-7 , 1 38-g, 1 40 as ownership, 82, g6, 1 02 , 1 03, 1 44
meritocracy, 67-9, 67n, 70, 72-7, 82-3, 93-4, as repository, 82, g6- 7
1 1 4, 1 36- 7 , 1 38-g, 1 42 subject of, see subject of possession
Mill, J. S . , 2-5, sn preferential treatment, 1 35-47, 1 49, 1 65
moral epistemology, 7-9, 1 2- 1 3, 1 6- 1 7, 20- 1 , primary goods, 25, 26, 2 7 , 39, 42
22-3, ·so, 53, 1 1 3 , 1 1 5- 1 6, 1 20-2, 1 53 , priority,
1 5 6- 7 , 1 68- 7 2 , 1 82-3 epistemological, 7-9, 1 2- 1 3, 1 6- 1 7 , 20- 1 ,
moral law, 2-3, 6- 7 , 24, 37, 38, 40, 1 1 8-2o, 1 76 2 2 - 3 , 5 3 . 64, 1 1 5- 1 6, 1 20-2, 1 56- 7,
moral priority, 2-7, 1 6, 1 7-20, 22-3, 64, 76- 7, 1 68-72, 1 82-3
88-go, 1 1 4- 1 5, 1 54-7 167 moral, 2-7, 16, 1 7-20, 22-3, 64, 76- 7 , 88-go,
moral subj ect, see person 1 1 4- 1 5, 1 54-7· 1 6 7
moral worth, 76- 7 , 87-9 1 , 1 3 7 , 1 38-g, 1 40, o f plurality, so-3, 86, 1 33 · 1 49. 1 5 1
1 42 o f procedure, 1 6, 38, 1 08-g, 1 1 2-22, 1 25,
mutual disinterest, 29, 42, 44-6, 48, 54-5, 1 26-7 , 1 76 see also right, priority of; self,
6o- 1 , 1 68, 1 69 priority of
procedural justice, 108-g, 1 1 2, 1 2 1 , 1 25, 1 26-7
Nagel, T., 6o procedure, 1 6, 38, 1 08-g, 1 1 2-22, 1 25, 1 26-7 ,
natural duties, 88, 8g, 1 09- 1 1 , 1 1 8, 1 79 1 76,
natural law, 1 1 6- 1 7 , 1 1 9 promises, I I 0- 1 1 , 1 1 3, 1 1 5
natural liberty, 67, 6g, n. 82-3, 93. 94. g6 property, 88-g,go, 97, 97n, 1 1 7 , 1 1 8
Nozick, R. , 66-7 , 7 7-85, 92, 94- 1 02 , 1 1 2- 1 3, punishment, 89-92
1 14
rational beings, 6, 1 7 , 24, 25, 36, 3 7 , 38-g, 59,
obligations, 1 0 7- 1 1 , 1 1 4, 1 79 1 04, 1 20, 1 23
origimll position, 1 4, 2 2 , 24-8, 29-30, 3 1 , 34, rational choice, 24-5, 45, 63n, 1 04, 1 1 9-20,
38-49, 5 4-5, 6o- 1 , 63-4, 66, 87, Bg, go, 1 22-5, 1 2 7, 1 28, 1 54, 1 56-66, 1 78, 1 8o;
1 02-3, 1 04-5. 1 09, I I I , 1 1 2 , 1 1 3, 1 1 7, see also choice
1 20, 1 2 1 -3 2 , 1 47 . 1 54· 1 68, 1 76-7 , 1 78 Rawls, J . , g, 1 0, 1 3- 1 4
motivational assumptions of, 25-7, 4 1 -6, o n agency, 1 34-5, 1 57-64, 1 7 7
48, 54-5, 6o- 1 , 1 68 on choice, 1 54, 1 56-66, 1 76-8, 1 8o- J
ownership, see possession, as ownership on community, 6 1 -5, 1 34-5, 1 47-52
conception of good, 94, J J 6- 1 7 , 1 34-5, 1 38,
Paley, W . , 1 46 1 54-7 · 1 6 1 -7 , 1 70- 1 , 1 82
person, deontological theory of the, 4- 1 4, 1 9-23, on j ustice: circumstances of, 1 4, 28-46, 52,
27-8, 1 79-83 6o- 1 ; derivation of principles of, 24-
Rawls ' , 1 9-23, 47-65, 66, 78-Bo, 85, 86, 87, 7, 38-g, 42-3, 1 05, 1 1 6-32; distributive,
9 2 , 9 3 , 94-s. 1 oo- 1 , 1 o4, 1 20-2, 1 3 1 , 66- 1 03 , 1 38, 1 39; procedural , I o8-g,
1 33 . 1 34· 1 38-g, 1 44, 1 60- 1 1 1 2 , 1 2 1 , 1 25, 1 26- 7; retribu tive and
persons, distributive, contrasted, 8g-g2 , gon; as
distinction between, 1 6, 5 1 -3, 66-7 , 78, So, first virtue, 1 5 , 2 2 , 29-35, 64, gg,
Bon, 1 2 2 , 1 24, 1 33 , 1 34, 1 40, 1 66- 7 , 1 68- 7 1
J 68, 1 70, 1 7 1 , 1 72 , 1 73.
plurali ty of, 1 6, so-53, 55, 63, 67, So, 86 , 1 2 1 -2,
on justification, So, 1 04, 1 o8-g, 1 1 1 -32
liberalism of, 1 3- 1 4, 1 5, 1 8- I g, 2 1 -4, 28,
1 24. 1 29, 1 3 1 , 1 33, 1 5 1 -2, 1 7 1 ; prior to
their unity, so-3, 86, 1 33, 1 49, 1 5 1
66- 7 , 95. 1 1 6-20, 1 47-8, 1 76
on reflection, 1 52-6 1 , 1 64, 1 65, 1 70, 1 72,
see also self; subject J 8o- J
Pitkin, H . , 1 1 7 on the idea ofsocial union, 8 1 -2 , 1 50- 1
plurality, see persons, plurality of see also arbitrariness; common assets; con­
possession, 54-9, 66, 7 2 , 7 7-86, 92-8, 97n, tract theory; difference principle; fair­
1 00, 1 34. 1 4 1 , 1 42-7 , 1 76 ness, j ustice as; legitimate expecta-

1 89
Index
tions; original posmon; person; per­ intersubj ective and intrasubj ective forms
sons; reflective equilibrium; veil of of, 62-3, 63n , • 43, 1 44, • 4s. • s o, 1 72 ,
ignorance • 73. 1 8 1
reciprocity, 28, 52-3, 7 7-8, 1 06-8 , 1 06n, 1 1 2 , self-realization, 8 1 - 2
1 26, 1 48, 1 5 1 , Sidgwick, A . , 3 7
reflection, Rawls' account of, 1 52-6 1 , 1 64, social con t ract, 1 04, 1 05, 1 1 6- I g ; see also
1 65, 1 70, 1 72 , 1 80- 1 contract theory; contracts
reflective equilibri u m , 4 1 , 43-4, 4 7-8, 59, 66, social union , the idea of, 8 1 -2 , 1 50- 1
8g, g 2 , 1 04 Spinoza, B, 1 32
regulative princip les, 1 , 1 2 , 38, 40, 5 1 , 1 1 6, Strauss, L., 1 75n
1 70, 1 76, subject,
retributive j ustice, 8 g- g 2 , 9 0n antecedently individuated, 6-7, 1 1 , 1 2 ,
right, 1 - 1 0, 1 2, 1 7-24, 63-4, 63n, 1 1 7, 1 1 8-20, 1 9· 2 1 , 53 · 9, 62-3, 64, 1 34· 1 47. 1 49.
1 34, 1 49•50, 1 54· 7 , 1 64, 1 65, 1 52 , 1 53 , 1 5 7 , 1 6 1 , 1 69 , 1 72 , 1 83
priority of, 1 - 1 0, 1 3 , 1 6-24, 26, 30, 36, 38, radically disembodied , 6- 1 4, 2 1 , 23, 26, 3 7 ,
76- 7 , 78, 86, 88-90, 1 20, 1 33, 1 38, 40, 5 4 . 79. 93 . 9 4 . 9 5 . 1 38, • 3 9 . 1 44
1 55-7, 1 67 , 1 68- 7 2 , 1 7 7 ; epistemologic­ radically situated, I J - 1 3, 20- 1 , 2 2 , 23, 26,
a l , 22-3 , 26, 64, 1 20-2, 1 56- 7 , 1 68-72 ; as 54· 7 9 . 93. 1 00- 1 , 1 38, • 3 9. 1 44, 1 47 ·
derived from freedom, 5- 1 0, 24, 37- g, 1 49. 1 53
1 1 8; moral, 2-3 , 1 6, 1 7- 1 8, 22-3, 64, transcendental, 6- 1 4, 1 7, 2 3 , 24, 3 7-9, 40,
76- 7 , 88-go, 1 54- 7 , 1 6 7 54· 79, 95 . 1 3 1 , 1 75 . 1 76-7
rights, 1 , 2 , 3-4, g- 1 0, 1 6, 1 8- 1 9 , 3 3, 6o, 6 6 , 67, see also person; persons; self
86-8, g 6, 9 7 n , 1 02 , 1 1 8, 1 35-8, 1 39, subj ect of possession, 20- 1 , 54- g , 62, 79-82,
1 4o- 1 , • ss , • 6 s 84, as, 9 2-8 1 03 , 1 1 7, 1 2 1 , 1 33 , 1 34,
and utilitarianism, 2, 1 35, 1 38, 1 40- 1 , 1 55, 1 40, 1 4 1 , 1 43 · 7 , 1 49· 1 52 , 1 53. 1 5 7.
1 65 1 59-6o, 1 6 9 , 1 72 , 1 75, 1 82
Rousseau, J .J . , 1 05 wider, 79-82, 1 03 , 1 34, 1 4 1 , 1 43-7, 1 49,
1 52-3, 1 72
self,
bounds of ' ss-9 . 6 • -s , 79-82, 1 33 · 4· 1 39- Taylor, C . , J 6o- J , 1 75n
4 7 . 1 49, • so- 1 , 1 5 2-3, 1 5 7, teleology, 3, 4, 7 , 1 0, 1 7- • g , 2 1 , so, 8 1 , 1 55,
1 7 1 , 1 73, 1 82-3 1 75· • 7 7
dispossessed , 84, 85, 86, 87, 92-4, 1 39, 1 40, transcendental argument, 7 - 1 1 , 24, 2 6 , 1 56n
1 44, 1 45. 1 53. 1 78, 1 7 9 -80 transcendental idealism, 7- 1 4, 23-4, 1 1 g , 1 20,
and its en ds, 6,7, 9, 1 1 , 1 2, 1 3 , 1 5-23, 54-9, 1 47
64, 78-g, 8o, 1 2 1 , 1 52-3, 1 54, 1 5 7 transcendental subj ect, see subj ect, transcen­
identity of, so, 55-59,62 , 62n, 64, 74. 79. dental
85, 93, 94, 1 34, 1 42-3, 1 44, • 45, • so, Tucker, J . , 1 46
1 52-3 , • s 7 . • sg- 6 • , 1 63-4, 1 66, • 7 9 ,
1 80-2
i n tersubjective conception of, 62, 63, 79- Utilitarianism, I , 3-5, 5n, 1 6, 1 7- 1 8, 22 , 36,
82, 1 03 , 1 3 2 , 1 43"4• 1 50- 1 s o- l , 63, 66 - 7, 78, 8on, 1 24, 1 35. 1 38,
i ntrasubj ective conception of, 62, 63, 63n 1 40- 1 , 1 44·5· 1 46-7, 1 65· 7, 1 68, 1 73-4
priority of, 6-7, 9 , 1 1 , 1 2 , 1 3, 1 5-23, : 9 , 64, conception of good, 5 1 , 1 38, 1 65-7
78- 9, 80, 94, 1 1 4 , 1 1 7 , 1 20 , 1 2 1 -2 , 1 33, and j ustice, 3-s. 5n
1 49· 1 5 7 . 1 70, 1 76 and rights, 2, 1 35, 1 38, 1 40- I , • ss. 1 65
as subj ect of possession,6, 7, g , 1 2 , 20- 1 ,
54-9, 62, Bo, 84, as, 9 2-3, 94, 95, • • 7, veil of ignorance, 24-5, 2 7-8, 38, 42, 44, go- 1 ,
1 2 1 , 1 33, 1 34, 1 5 7, 1 59-60, 1 69, 1 72 , 1 2 5 , 1 26, 1 28, 1 3 1 , 1 54· • 78
' 75 . J 82 virtue, 76- 7 , 88- 9, go, g :� , 9 8- g , 1 3 7
unity of, 8- 9 , 1 3 , 1 9 - 2 2 , 23, 24, 52-3, 64, j ustive as a, 2, 1 5, 2 2 , 28, 29-35, 36, 49 , 64 ,
1 33-s. 1 50 87, 9g , 1 68- 7 1 ' 1 75
see also person; persons; subject virtues, 2, 1 1 , 1 2 , 1 5, 3 • - s , 1 33 , 1 68- 72
self-knowledge, 8-9, 22, 24, so, ss. sB-9, 62-4,
• 43, • 44, 1 45 , • s o, 1 52-3, • s B-62, 1 72 , Weber, M., 1 75
1 73 . 1 79-8 1 Wolin, S . , 1 75n