http://www.cambridge.org/9780521651677

ii

This page intentionally left blank

the cambridge companion to

R A W L S

Each volume of this series of companions to major philoso-phers contains specially commissioned essays by an inter-national team of scholars, together with a substantial bibli-ography, and will serve as a reference work for students andnonspecialists. One aim of the series is to dispel the intimi-dation such readers often feel when faced with the work of adifficult and challenging thinker.

John Rawls is the most significant and influential polit-ical and moral philosopher of the twentieth-century. Hiswork has profoundly shaped contemporary discussions ofsocial, political, and economic justice in philosophy, law,political science, economics, and other social disciplines. Inthis exciting collection of new essays, many of the world’sleading political and moral theorists discuss the full rangeof Rawls’s contribution to the concepts of political and eco-nomic justice, democracy, liberalism, constitutionalism, andinternational justice. There are also assessments of Rawls’scontroversial relationships with feminism, utilitarianism,and communitarianism.

New readers will find this the most convenient and acces-sible guide to Rawls currently available. Advanced studentsand specialists will find a conspectus of recent developmentsin the interpretation of Rawls.

Samuel Freeman is Professor of Philosophy and Law at theUniversity of Pennsylvania.

i

ii

other volumes in the series of cambridge companions:

AQUINAS Edited by norman kretzmann andeleonore stump

HANNAH ARENDT Edited by dana villaARISTOTLE Edited by jonathan barnesAUGUSTINE Edited by eleonore stump and

norman kretzmannBACON Edited by markku peltonenDESCARTES Edited by john cottinghamDUNS SCOTUS Edited by thomas williamsEARLY GREEK PHILOSOPHY Edited by a. a. longFEMINISM IN PHILOSOPHY Edited by miranda

fricker and jennifer hornsbyFOUCAULT Edited by gary guttingFREUD Edited by jerome neuGADAMER Edited by robert dostalGALILEO Edited by peter machamerGERMAN IDEALISM Edited by karl ameriksHABERMAS Edited by stephen k. whiteHEGEL Edited by frederick beiserHEIDEGGER Edited by charles guignonHOBBES Edited by tom sorellHUME Edited by david fate nortonHUSSERL Edited by barry smith and

david woodruff smithWILLIAM JAMES Edited by ruth anna putnamKANT Edited by paul guyerKIERKEGAARD Edited by alastair hannay and

gordon marinoLEIBNIZ Edited by nicholas jolleyLOCKE Edited by vere chappellMALEBRANCHE Edited by stephen nadlerMARX Edited by terrell carverMILL Edited by john skorupskiNEWTON Edited by i . bernard cohen and

george e. smithNIETZSCHE Edited by bernd magnus and

kathleen higginsOCKHAM Edited by paul vincent spadePLATO Edited by richard krautPLOTINUS Edited by lloyd p. gersonROUSSEAU Edited by patrick rileySARTRE Edited by christina howellsSCHOPENHAUER Edited by christopher janawaySPINOZA Edited by don garrettWITTGENSTEIN Edited by hans sluga and

david stern

iii

iv

The Cambridge Companion to

RAWLS

Edited by

Samuel FreemanUniversity of Pennsylvania

v

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University PressThe Edinburgh Building, Cambridge , United Kingdom

First published in print format

ISBN-13 978-0-521-65167-7 hardback

ISBN-13 978-0-521-65706-8 paperback

ISBN-13 978-0-511-06637-5 eBook (NetLibrary)

© Cambridge University Press 2003

2002

Information on this title: www.cambridge.org/9780521651677

This book is in copyright. Subject to statutory exception and to the provision ofrelevant collective licensing agreements, no reproduction of any part may take placewithout the written permission of Cambridge University Press.

ISBN-10 0-511-06637-6 eBook (NetLibrary)

ISBN-10 0-521-65167-0 hardback

ISBN-10 0-521-65706-7 paperback

Cambridge University Press has no responsibility for the persistence or accuracy ofs for external or third-party internet websites referred to in this book, and does notguarantee that any content on such websites is, or will remain, accurate or appropriate.

Published in the United States by Cambridge University Press, New York

www.cambridge.org

http://www.cambridge.orghttp://www.cambridge.org/9780521651677

contents

Contributors page ix

Introduction: John Rawls – An Overview 1samuel freeman

1 Rawls and Liberalism 62thomas nagel

2 For a Democratic Society 86joshua cohen

3 Rawls on Justification 139t.m. scanlon

4 Rawls on the Relationship between Liberalismand Democracy 168amy gutmann

5 Difference Principles 200philippe van parijs

6 Democratic Equality: Rawls’sComplex Egalitarianism 241norman daniels

7 Congruence and the Good of Justice 277samuel freeman

8 On Rawls and Political Liberalism 316burton dreben

9 Constructivism in Rawls and Kant 347onora o’neill

vii

viii contents

10 Public Reason 368charles larmore

11 Rawls on Constitutionalismand Constitutional Law 394frank i. michelman

12 Rawls and Utilitarianism 426samuel scheffler

13 Rawls and Communitarianism 460stephen mulhall and adam swift

14 Rawls and Feminism 488martha c. nussbaum

Bibliography 521Index 557

contributors

joshua cohen is Professor of Philosophy and Political Science andGoldberg Professor of the Humanities at MIT. He writes on demo-cratic theory, is coauthor of On Democracy and Associations andDemocracy, and has written numerous articles on the theory andpractice of deliberative democracy. Cohen is editor-in-chief of BostonReview and associate editor of Philosophy and Public Affairs.

norman daniels is Professor of Ethics and Population Health,School of Public Health, Harvard University. His main work in ethicsand political philosophy includes Reading Rawls (ed., 1975), JustHealth Care (1985), Am I My Parents’ Keeper? (1988), Seeking FairTreatment (1995), Justice and Justification (1996), Benchmarks ofFairness for Health Care Reform (with Don Light and Ronald Caplan,1996), From Chance to Choice (with Allen Buchanan, Dan Brock, andDan Wikler, 2000), Is Inequality Bad for Our Health? (with BruceKennedy and Ichiro Kawachi, 2000), and Setting Limits Fairly (withJames Sabin, 2002).

burton dreben was Professor of Philosophy at Harvard Univer-sity and at Boston University until his death in 1999. He is the au-thor of several essays on the history of logic and analytic philosophy,as well as (with Warren Goldfarb) The Decision Problem: SolvableClasses of Unsolvable Formulas.

samuel freeman is Steven F. Goldstone Term Professor of Phi-losophy and Law at the University of Pennsylvania. He has writ-ten numerous articles in political, moral, and legal philosophy onsuch subjects as liberalism and libertarianism, deliberative democ-racy, utilitarianism and deontology, constitutional interpretation,

ix

x contributors

contractualism, and Rawls and social contract theory. He edited JohnRawls’s Collected Papers (1999).

amy gutmann is Provost and Laurance S. Rockefeller UniversityProfessor of Politics at Princeton University. She is the author ofDemocratic Education and Liberal Equality and of the forthcom-ing Identity Groups in Democracy: A Humanist View; coauthor ofDemocracy and Disagreement and Color Conscious: The PoliticalMorality of Race; and editor of eight books, including Democracyand the Welfare State, Multiculturalism, and Freedom of Associa-tion.

charles larmore is Chester D. Tripp Professor in the Humani-ties at the University of Chicago. His books include Patterns of MoralComplexity (1987), The Morals of Modernity (1996), and The Roman-tic Legacy (1996). He is currently completing a book (in French) onthe nature of self, entitled Les Pratiques du Moi.

frank i. michelman is Robert Walmsley University Professor,Harvard University, where he has taught since 1963. He is the authorof Brennan and Democracy (1999) and has published widely in thefields of constitutional law and theory, property law and theory, localgovernment law, and jurisprudence.

stephen mulhall is Fellow and Tutor in Philosophy at New Col-lege, Oxford. His research interests include political philosophy, phi-losophy of religion, Wittgenstein, Heidegger, and Kierkegaard. Hismost recent books are Inheritance and Originality (OUP: 2001) andOn Film (Routledge: 2001).

thomas nagel is Professor of Philosophy and Fiorello LaGuardiaProfessor of Law at New York University and author of Equalityand Partiality, The Last Word, The View from Nowhere, and ThePossibility of Altruism.

martha c. nussbaum is Ernst Freund Distinguished Service Pro-fessor of Law and Ethics at the University of Chicago with appoint-ments in the Law School, Philosophy Department, and the DivinitySchool. She is on the Board of the Center for Gender Studies. Hermost recent books are Women and Human Development (2000) andUpheavals of Thought: The Intelligence of Emotions (2001).

contributors xi

onora o ’neill is Principal of Newnham College, Cambridge. Shehas written widely on Kant’s practical philosophy, on internationaljustice, and on ethics. Her recent books include Bounds of Justice(CUP 2000) and Autonomy and Trust in Bioethics (CUP 2002).

t.m. scanlon is Professor of Philosophy at Harvard University.His writings in moral and political philosophy include many articlesand a book, What We Owe to Each Other.

samuel scheffler is Professor of Philosophy and Law at the Uni-versity of California, Berkeley. He is the author of The Rejectionof Consequentialism, Human Morality, and Boundaries and Alle-giances.

adam swift is Fellow in Politics and Sociology at Balliol Col-lege, Oxford. He coauthored Liberals and Communitarians (2nd ed.,Blackwell 1996) and Against the Odds? Social Class and Social Jus-tice in Industrial Societies (OUP 1997) but wrote Political Philoso-phy: A Beginner’s Guide for Students and Politicians (Polity 2001)on his own.

philippe van parijs directs the Hoover Chair of economic and so-cial ethics at the Catholic University of Louvain (Belgium). His booksin English include Evolutionary Explanation in the Social Sciences(1981), Arguing for Basic Income (1992), Marxism Recycled (1993),Real Freedom for All (1995), and What’s Wrong with a Free Lunch?(2001). He is currently working on the relationships between linguis-tic diversity, justice, and democracy.

xii

samuel freeman

IntroductionJohn Rawls – An Overview

I. preliminaries

John Rawls’s published works extend over fifty years from the mid-dle of the twentieth century to the present.1 During this period hiswritings have come to define a substantial portion of the agendafor Anglo–American political philosophy, and they increasingly in-fluence political philosophy in the rest of the world. His primarywork, A Theory of Justice (TJ), has been translated into twenty-sevenlanguages.2 Only ten years after Theory was published, a bibliographyof articles on Rawls listed more than 2,500 entries.3 This extensivecommentary indicates the widespread influence of Rawls’s ideas aswell as the intellectual controversy his ideas stimulate.

From the outset Rawls’s work has been guided by the question,“What is the most appropriate moral conception of justice for a demo-cratic society?” (TJ, p. viii/xiii rev.).4 In Theory he pursued this ques-tion as part of a more general inquiry into the nature of social justiceand its compatibility with human nature and a person’s good. HereRawls aimed to redress the predominance of utilitarianism in mod-ern moral philosophy. As an alternative to utilitarianism, Rawls,drawing on the social contract tradition, developed a conception ofjustice “that is highly Kantian in nature” (TJ, p. viii/xviii rev.). Ac-cording to this conception, justice generally requires that basic socialgoods – liberty and opportunity, income and wealth, and the basesof self-respect – be equally distributed, unless an unequal distribu-tion is to everyone’s advantage (TJ, p. 62/54 rev.). But under favorablesocial conditions a special conception, “justice as fairness,” applies;it requires giving priority to certain liberties and opportunities viathe institutions of a liberal constitutional democracy. Rawls’s two

1

2 samuel freeman

principles of justice require that certain important liberties be pro-vided equally for all, that these “basic” liberties have priority overaggregate social welfare and perfectionist values, that “fair” (not just“formal”) opportunities be provided equally for all citizens, and thatdifferences in income and wealth and in social positions be struc-tured so as to maximally benefit the worst-off members of society.

Theory depicts justice as fairness as a universal moral ideal tobe aspired to by all societies. Over two decades after Theory Rawlspublished Political Liberalism (PL). Here, because of the demands ofliberalism itself, Rawls revises the argument for justice as fairness tolimit its applicability. No longer does Rawls take issue directly withutilitarianism, perfectionism, or other general moral conceptions.Political liberalism instead addresses the culture of a constitutionaldemocracy. Its guiding question is, What is the most just and fea-sible arrangement of basic social institutions that realizes the coredemocratic values of freedom and equality for all citizens?

To appreciate the development of Rawls’s views it is essential tounderstand that all along he has sought to work out a realistic idealof justice (a “realistic utopia”5). His conception is ideal insofar asit is designed for the ideal conditions of a “well-ordered society,”where reasonable persons who are free and equal all accept the sameconception of justice. Rawls’s account of justice is realistic since itis designed to apply neither to moral saints or perfect altruists onthe one hand, nor to natural sinners or rational egoists on the other,but to what humans at their best are capable of, given their nature,under normal conditions of social life.6 To situate Rawls’s realisticideal in terms of the predecessors that most influenced him: Akinto Kant, Rawls seeks to discover the fundamental moral principlesthat regulate reasoning and judgments about justice. These principleshe presumes to be deeply implicit in ordinary moral awareness andare evidenced by our most considered moral judgments. But Rawlsrejects Kant’s dualisms;7 he does not suppose principles of justiceare a priori or based in “pure practical reason” alone. Human na-ture and the fixed empirical conditions within which practical rea-son is normally exercised are relevant to discovering and justifyingprinciples of justice. Rawls here moves some way toward the more“sentimental” and “naturalistic” accounts suggested by Rousseauand Hume. He conditions the justification of principles of justice oncertain psychological tendencies of human nature and our capacities

Introduction 3

for sociability.8 This explains Rawls’s emphasis on the “stability” orfeasibility of a moral conception of justice. A conception of justiceis stable when its realization would foster in people a steadfast willto do justice and a disposition to uphold just institutions (as thatconception defines them). It is because of his concern for the sta-bility of justice as fairness that Rawls is led eventually to make themodifications that lead to political liberalism.

In what follows I discuss some of the main features of justiceas fairness, as presented both in Theory (Section II) and in PoliticalLiberalism (Section III). In Section IV, I briefly discuss Rawls’s recentaccount of international justice in The Law of Peoples. I aspire not toa comprehensive overview but to emphasize certain crucial ideas toaid the reader in understanding Rawls and the contributions to thisvolume.

II. a theory of justice: justice as fairness

A. The Principles of Justice

Rawls describes Theory as an attempt “to generalize and carry to ahigher order of abstraction the traditional theory of the social con-tract as represented by Locke, Rousseau, and Kant” (TJ, p. viii/xviiirev.). This tradition’s main idea is that the political constitution andthe laws are just when they could be agreed to by free rational per-sons from a position of equal right and equal political jurisdiction.Rawls applies the idea of a hypothetical social agreement to argue forprinciples of justice. These principles apply in the first instance todecide the justice of the institutions that constitute the basic struc-ture of society. Individuals and their actions are just insofar as theyconform to the demands of just institutions. The basic structure isthe interconnected system of rules and practices that define thepolitical constitution, legal procedures and the system of trials, theinstitution of property, the laws and conventions which regulatemarkets and economic production and exchange, and the institutionof the family (which is primarily responsible for the reproductionof society and the care and education of its new members). Theseinstitutions can be individually organized and jointly combined inseveral different ways. How they are specified and integrated into asocial system deeply affects people’s characters, desires and plans,

4 samuel freeman

and their future prospects, as well as the kinds of persons they aspireto be. Because of the profound effects of these institutions on thekinds of persons we are, Rawls says the basic structure of society is“the primary subject of justice” (TJ, p. 7/6 rev.).

The significance of the basic structure comes out especially inRawls’s treatment of economic rights of property and freedom ofcontract. Rawls takes a “holistic” approach to these rights and to dis-tributive justice more generally.9 This means that we cannot decidewhat economic rights and duties people have without first determin-ing the effects of various systems of economic rights and practiceson others – particularly on others’ capacities to exercise their ba-sic rights and liberties. Rawls’s principle of distributive justice isthen closely aligned with his principle of equal basic liberties (as ex-plained at the end of this Section A). We will begin then with the firstprinciple of justice.

Rawls’s first principle, the principle of equal basic liberties, paral-lels J.S. Mill’s principle of liberty in that it is conceived as definingconstitutional limits on democratic government. Rawls sees certainliberties as “basic.” These include liberty of conscience and freedomof thought, freedom of association, and the rights and liberties thatdefine the freedom and integrity of the person (including freedomof movement, occupation, and choice of careers, and a right to per-sonal property); also included for Rawls are equal political rights ofparticipation and the rights and liberties that maintain the rule oflaw.10 To call these liberties “basic” means (in part) that they aremore important than others. Most people would readily admit thatit is far more important that people be free to speak their minds,practice their faiths or lack thereof, choose their careers, and marry,befriend, or associate with people they choose than that they be freeto harass others, drive recklessly and as fast as they please, or walkthe streets naked and relieve themselves in public view. Few wouldcall laws restricting these latter actions restrictions on a person’sfreedom at all. Most any purported liberal would agree with theserestrictions and with the greater significance of Rawls’s basic lib-erties. But what makes Rawls’s list of basic liberties more impor-tant than other normally permissible liberties many people arguefor, such as freedom to enter contracts of all kinds, to own weapons,or to accumulate, use, and dispose of productive resources as onepleases?

Introduction 5

Rawls calls the liberties of the first principle “basic” since theyare morally more significant to the freedom of democratic citizensthan are the “nonbasic” liberties just mentioned. This means, first,that the basic liberties are necessary for pursuing a wide range of con-ceptions of the good. Second, the basic liberties are essential to theexercise and development of the two moral powers that define theconception of the person implicit in Rawls’s constructivist view.11

The two moral powers are the capacity for a sense of justice (to un-derstand, apply, and act on and for the sake of principles of justice)and the capacity for a conception of the good (to form, revise, andrationally pursue a rational plan of life). In Theory Rawls sees themoral powers in Kantian terms; as the powers of practical reasoningin matters of justice, they are the essential capacities for moral andrational agency. By virtue of these capacities we see ourselves andeach other as free and responsible agents. As such the moral powersare the grounds for full autonomy. Subsequently, in Political Liber-alism, the moral powers are characterized in less ambitious terms;they are the capacities that anyone needs if he or she is to occupythe role of citizen and engage in, benefit from, and comply with thedemands of social cooperation in a democratic society.

Because of their role in defining the conception of moral personsthat underlies Rawls’s view, justice as fairness assigns the basic liber-ties strict priority over other social goods. This means basic libertiescan be limited only for the sake of maintaining other basic liberties.They cannot be compromised to promote greater aggregate happinessin society, to increase national wealth, or to promote perfectionistvalues of culture. The basic liberties cannot be limited even for thesake of better realizing the purposes of Rawls’s difference principle.That the worst off may be willing to give up some of their basicliberties (such as their right to vote) in exchange for added incomesupplements is of no political consequence. For the first priority ofjustice for Rawls is to maintain equal freedom and respect for per-sons in their capacity as democratic citizens. This indicates the wayjustice as fairness is grounded in an ideal of persons as free and equalcitizens who exercise their capacities for justice and rationality (thetwo moral powers) as they jointly govern public matters and freelypursue their conceptions of a good life.12 The political liberties, be-sides being necessary to a person’s sense of self-respect, are also es-sential to the full development of the capacity for a sense of justice

6 samuel freeman

that partly defines this ideal of citizens. It is because the basic lib-erties are essential to the exercise of the moral powers that they areinalienable: there is no right to give up or trade away the libertiesneeded to define a citizen’s status as a free and equal person. Thisis one of several ways justice as fairness differs from libertarianism.The unrestricted freedoms of contract and transfer that are definingfeatures of libertarianism are not basic or even protected libertiesaccording to Rawls’s liberal view.

Now let us turn to Rawls’s second principle and particularlythe question of distributive justice. For Rawls economic rights ofproperty and contract are institutional but not conventional. To sayproperty is an institution means in part that it consists in a sys-tem of social (mainly legal) rules and practices that specify exclusiverights and duties with regard to the use and control of things. Tosay property is conventional means that the institutional rights ofproperty people have are specified exclusively by existing legal rulesand institutions and that these rules are valid only so long as theyare effective and enforced. In the conventional view then, peoplehave no claim to property independent of existing legal rules and in-stitutional arrangements. Justice in distribution is simply enforcingcurrent property conventions, thereby giving each person his or herdue. Hobbes and Hume held such a view. Rawls does not.

Natural rights theory was designed to combat the conventionalview. The idea of a state of nature emphasizes that certain rights arenot conventional but are moral and apply to persons whatever theirsocial circumstances. So Locke contends that governments have noauthority to prohibit freedom of religious association, for this is aninalienable right people have independent of political society. Sim-ilarly, if the Crown confiscates people’s property, an injustice hasbeen done since they have arbitrarily been denied their livelihoodand means of independence. Now Rawls says that because of thefirst principle, “justice as fairness has the characteristic marks ofa natural rights theory” (TJ, p. 506n/443n). But he denies the ac-count of “natural” or presocial property argued for by libertariansand Lockeans.

The idea of natural, presocial property effectively deals with theproblem of oppressive confiscations by governments. But the ideaof natural property is inadequate when used to address questions ofthe kinds of property rights and distributions that ought to exist in

Introduction 7

modern industrialized and democratic societies. In the isolated stateof nature, where natural property claims hypothetically originate,people need not be so concerned with the effects on others of pos-sessing and exercising their rights, for few are around to be adverselyaffected. When this conception of presocial property is applied to so-cial conditions, it implies that people may accumulate, use, transfer,and exchange their possessions as if they were in an isolated stateof nature, and no matter what the effects or how badly off othersmight be made as a result of a system of institutionalized naturalproperty.13 Surely there must be some other way to argue that peo-ple can be morally entitled to their possessions without relying on apresocial state of nature.

Rawls, like natural property advocates, distinguishes betweenconventional property and the property rights people ought to have.But Rawls does not derive the property rights people ought to havefrom a preinstitutional state of nature. He refers instead to an idealof social cooperation where institutions are designed to benefiteveryone on a basis of reciprocity. They benefit everyone, not inthe weak sense that all are made better off than in an apolitical stateof nature, but in the strong sense that all are made better off thanthey would be in a state of equality and where no one benefits at theexpense of the poorest. The role of Rawls’s difference principle is todefine this ideal of reciprocity.14 The institution of property is justlyordered when it is part of a social and economic system that spec-ifies property relations so as to make the worst-off class better offthan they could be under the institutions of any feasible alternativeeconomic system (subject to the conditions that equal basic liber-ties and fair equality of opportunities are always maintained). It isthe responsibility of political institutions to structure economic andproperty relations so that, over a lifetime, the economic prospects ofthe worst-off class (which might be defined as the average wage ofunskilled workers, or in some other way) are as favorable as they canbe.15

Here it should be emphasized that “worst off” is defined in termsof certain resources that Rawls calls “primary social goods” withspecial focus on income and wealth. From the standpoint of justicethe worst off are the poorest among us. They are not necessarilythe unhappiest (as in welfarist views) or the most disabled physi-cally or mentally (as in Sen’s capability approach).16 Rawls’s reasons

8 samuel freeman

for eschewing happiness or welfare as a basis for interpersonal com-parisons is connected with his (Kantian) emphasis on freedom andresponsibility. Agency requires that people see themselves as actingfreely and as responsible for their ends. Because of our capacities toreflect critically on our desires and rationally structure our ends intoa coherent plan of life, we normally do not see ourselves as saddledwith desires and ends we can do nothing about. To encourage thisself-conception and the development and exercise of these capacitiesfor rational (and moral) agency, Rawls contends that a conception ofjustice should not simply take as given whatever desires people hap-pen to have and distribute welfare as if people’s ends were imposedon them. Instead, people should be held responsible for their endsand expected to adjust their desires to the fair share of resourcesthey can legitimately expect. What individuals may fairly and legiti-mately expect is specified by the difference principle, which is itselfgeared towards providing resources adequate for realizing everyone’scapacities for free and responsible agency.

It is because of his (Kantian) conception of agency that Rawls treatssevere mental and physical handicaps as a special case. He abstractsfrom such handicaps in the initial argument for principles of justice,leaving special principles to be worked out for them to the legisla-tive stage of his “four-stage sequence” (Restatement, pp. 171–76).This does not mean that such problems are unimportant or that thedisabled are not due special consideration because of their handi-caps. But it does imply that for Rawls justice is not primarily aboutredressing inequalities imposed by nature or misfortune. Rather jus-tice is primarily about providing each person with resources that aresufficient to their realizing their “moral powers” of free, responsible,and rational agency. As a result, Rawls (unlike Sen) does not give thenaturally handicapped absolute priority in decisions of justice.17 Hetreats their situation similar to problems of partial compliance. Prin-ciples of justice are initially chosen for the ideal case of a well-orderedsociety, where it is assumed all have the capacities for cooperationand that there will be “strict compliance.” Just as the parties in theoriginal position assume that the members of a well-ordered societyhave an effective sense of justice and normally will not violate justlaws, they assume that members are normal cooperating membersof society over a complete life who have the capacities needed for so-cial cooperation (the moral powers). These are idealizations (like the

Introduction 9

assumption of perfect competition in price theory). Rawls says theseidealizations present a more tractable problem of choice and providea basis for dealing with less than ideal circumstances, such as par-tial compliance or the special problems of the disabled.18 But whatprimarily underlies these assumptions is a view about the bases ofsocial justice. It is an ideal of a society of free and equal citizens whotake responsibility for their ends and cooperate with one another ona basis of reciprocity and mutual respect. It is this ideal, not the idealof redressing undeserved inequalities of welfare, resources, or luck,that is at the foundation of Rawls’s view.

This raises again the question of the relationship between the dif-ference principle and the equal basic liberties. Rawls believes thetwo principles of justice cannot be appreciated or justified in isola-tion from one another. To be a liberal conception it is not enoughto recognize basic liberties and assign them priority. A liberal con-ception of justice also recognizes a social minimum, a basic socialentitlement to enabling resources, particularly income and wealth.For without a social minimum, the basic liberties are merely formalprotections and are worth little to people who are impoverished andwithout the means to take advantage of their liberties. So, Rawlscontends, any liberal view provides some kind of social minimum toguarantee the worth of the basic liberties (PL, p. 6, 156f.). What dis-tinguishes justice as fairness is its egalitarianism: it defines the socialminimum in terms of the difference principle.19 Now the differenceprinciple has a distinct relationship to the principle of equal basicliberties. It permits inequalities in income and wealth in order tomaximally promote the effective exercise of the equal basic libertiesby the worst off:

Taking the two principles together, the basic structure is to be arranged tomaximize the worth to the least advantaged of the complete scheme of equalliberty shared by all. This defines the end of social justice. (TJ, p. 205/179rev., emphases added)

The “end of social justice” is not simply that everyone’s equalfreedoms be formally protected but that the basic liberties be effec-tively exercisable by all to the degree that the worth of freedom to theworst off is maximal. Its guarantee of the maximal worth of equalliberties provides one of the more compelling reasons for Rawls’sdifference principle.20 In every other economic system, the value of

10 samuel freeman

liberty to the least advantaged is less than in justice as fairness. ForRawls this means that the effective freedom of the least fortunate isbeing compromised for the sake of those better off. Only the differ-ence principle achieves reciprocity in the sense that gains to thosebetter off are not achieved at the expense of the poorest members ofsociety (Restatement, pp. 123–24).

B. The Argument from the Original Position

Appeal to a hypothetical agreement – of what people could or wouldagree to under certain conditions – is characteristic of social contractdoctrine. None of the major historical proponents of contractarian-ism (Hobbes, Locke, Rousseau, and Kant) saw the existing statusquo as the appropriate perspective from which to achieve agreementon laws and social institutions. For even supposing agreement wereachievable under current conditions, it would presuppose the va-lidity of existing distributions of rights, bargaining advantages, andthe very laws and institutions whose justice is to be decided by thesocial contract. In order to abstract from the influence of existingconditions, Hobbes and Locke assume that general agreement takesplace in the prepolitical (and for Hobbes, presocial) circumstances ofa hypothetical state of nature. Now a state of nature is historical inthe following sense: Its inhabitants have knowledge of their circum-stances and interests; they know everything about themselves thatany historically situated individual might know about prevailing cir-cumstances. So like any other contract, a social contract in a state ofnature would be affected by its parties’ access to information aboutthemselves and others’ situations.

One respect in which awareness of one’s historical situationaffects the resulting distribution of political power is evident fromLocke’s justification of passive citizenship. For Locke a political con-stitution is just only if free persons could agree to it via a series ofagreements starting from a state of nature wherein each person hasequal political jurisdiction. But while Locke’s contracting partiesbegin with equal political rights, their knowledge of their circum-stances in the state of nature leads to the peculiar consequence thatthe majority of free persons could agree to alienate their equal polit-ical status in exchange for other benefits. For Locke political rightsare alienable in a way that freedom of conscience and the “right of

Introduction 11

private judgment” are not. So Locke does not envision that women,or even the majority of men (those who do not satisfy property qual-ifications) retain their political rights under the constitution.

This is a peculiar result for a view that assumes equal politicalrights are a defining feature of our natural condition. If not in Locke’sday, it conflicts now with our considered convictions of justice.21

For why should gender or property ownership affect one’s having theright to vote or hold office, or one’s status as an equal citizen? Itis only because the occupants of Locke’s state of nature know theirparticular characteristics and social circumstances that the class ofaffluent men are in a position to take advantage of their privilegedposition and persuade others to give up the equal political statusall originally have in exchange for enjoying the benefits of politicalsociety. But a person’s gender or wealth, even if relevant to certainkinds of contracts, is not morally relevant to agreement on princi-ples of justice for the basic structure of society. So Rawls imposes aveil of ignorance on the parties, depriving them of knowledge of thisinformation, as well as any other information that might advantageor disadvantage the parties in their discussions and agreement. De-cision on principles of justice is then rendered ahistorical to makethe decision strictly impartial with respect to peoples’ social status,natural characteristics and abilities, and even their conceptions ofthe good.

The veil of ignorance rules out information that, Rawls contends,is morally irrelevant to decision on principles of justice.22 Philoso-phers of course disagree about the moral relevance of information todeciding principles of justice. Libertarians will say that knowledgeof one’s property rights and bargaining position is relevant to anycontract so long as one is entitled to them according to libertarianprinciples. They would then reject the idea of an ahistorical hypo-thetical agreement on principles of justice.23 So do utilitarians andcontemporary Hobbesians, but for different reasons.

According to many utilitarians, the appropriate solution to prob-lems of partiality and people’s taking advantage of their position inmaking moral decisions is to impose a “thin” veil of ignorance: al-low the parties to an original position full historical information,including knowledge of everyone’s desires and interests, and sim-ply deprive them of knowledge of their identity in society. Peopleare then not in a position to take advantage of knowledge of their

12 samuel freeman

particular situations and conceptions of the good.24 Under this moremodest impartiality condition, parties still have knowledge of exist-ing desires and interests. Surely, utilitarians argue, this informationis relevant, for what could the end of justice be if not promotingexisting human interests?

One problem with this proposal is that it still does not en-tirely remedy the problem of partiality towards dominant interests.Though parties may not know their situation in society, a thinlyveiled initial position still leads them to play the odds in hopes thatthey are among those who endorse the dominant majority positionand values. Moreover, the initial condition of knowing particular so-cial facts while ignorant of one’s social position still does not addressthe crucial question of the justice of existing background conditionsthat led to the status quo. People’s desires are consequently taken asgiven no matter what the conditions of their origin or how unreason-able they are; these desires are then allowed to influence and evendetermine final agreement on principles of justice. For example, toimpose only a thin veil of ignorance on a racially segregated societyor (to emphasize the point) on a slave society still leaves people opento assume that there is nothing unjust about racism and apartheid,or even about owning slaves, so long as it is rational to take a chanceon entering society as a free member of the dominant class. But whyshould we take such clearly unjust circumstances and the unreason-able desires they generate as a relevant touchstone for deciding thejustice of social institutions and the reasonableness of expectations?To remedy this problem, it will not suffice to rule out such “anti-social” desires as the desire to own slaves or the desire to live apartfrom racial minorities.25 This proposed solution, even if it works inthis instance, only addresses the more extreme examples of a generalproblem. Should we impose only a thin veil over the inhabitants ofa traditional hierarchical society, where the overwhelming majorityare satisfied with the status quo and would gladly deny themselvesand others freedom of religion and speech or the right to vote? Somecommunitarians and utilitarians may think so, for on the basis oftheir views information about existing conditions and conceptionsof the good is relevant to deciding basic principles of justice.

In ruling out knowledge of all historical information, Rawls’s“thick” veil of ignorance ensures that principles of justice are notcontoured to the conditions of any particular social situation or

Introduction 13

designed to promote or especially favor any particular conceptionof the good. This does not mean the original position is “neutral”among conceptions of the good, for it is not. The principles chosenin the original position clearly render certain conceptions imper-missible (e.g., racist and other intolerant doctrines) and others diffi-cult to achieve under conditions of a well-ordered society (e.g., tra-ditionalist religions requiring conditions of minimal education andwidespread social conformity for their success). But it is not as ifneutrality among conceptions of the good were a desirable goal tobegin with – what’s the point, after all, of being neutral towardsracist or other intolerant doctrines which deny the values of fairnessor equal freedom? Instead of aiming to be neutral among desires orfair to conceptions of the good, Rawls’s original position seeks fair-ness to persons, conceived as equals with the capacities to criticallyreflect on, adopt, and pursue their conceptions of the good.26 In thisregard the original position assigns importance, not to existing endsbut to people’s freedom and moral powers to shape and amend theirends. Consequently, the original position’s conditions encourage adiversity of conceptions of the good. In this regard, the original posi-tion necessarily results in distinctly liberal principles that guaranteeequality of basic liberties.

There are many objections to Rawls’s original position and to hisargument from this perspective for the principles of justice. Here Ican only consider one of the more prominent objections. This shouldprovide a better idea how Rawls sees the original position’s connec-tion with the principles of justice and differentiate his account fromostensibly similar views.

Consider first Rawls’s account of the rationality of the parties inthe original position (TJ, Sec. 25). They are rational in the thin senseused in the social sciences insofar as they choose principles solely topromote their interests. The parties are assigned no moral motiva-tions, and with respect to one another’s welfare and position they areindifferent under the special circumstances of the original position.Now it would be a mistake to say, as many do, that Rawls’s partiesare purely self-interested, like rational egoists. They are not egoistsany more than chess players who play to win or buyers who shop forthe lowest price are egoists. Just as chess players and ordinary con-sumers usually have all sorts of moral convictions and motives aswell as benevolent affections for others, so the parties in the original

14 samuel freeman

position are assumed to have them too. Indeed, their moral inter-ests and benevolent concerns are among the interests they aim toprotect in their choice among principles of justice. Moreover, theparties are assumed to be distinctly nonegoistical since they have acapacity for an effective sense of justice – a desire to act not just ac-cording to, but also for the sake of, justice. This moral motivation inthe end proves crucial to Rawls’s argument for the principles of jus-tice and their stability. But for purposes of this particular decision –namely, deciding on principles of justice for the basic structure ofsociety – the parties do not act from their benevolent affections andmoral sentiments. They “take no interest in one another’s interest”as contracting agents but are concerned only with promoting theirown interests.27 The idea of the original position is to devise a choicesituation where rational decision is subject to reasonable (moral)constraints imposed by the conditions on choice in the original po-sition. Rawls believes this way of proceeding promotes greater clarityand that the alternative method of assigning moral motivations tothe parties would not result in the definite choice of a conception ofjustice (TJ, pp. 148–9/128–9 rev.; 584/512 rev.).

Describing the parties as strictly rational agents whose choiceis subject to moral constraints allows Rawls to invoke the theoryof rational choice and decision under conditions of uncertainty. Indecision theory it is standard practice under conditions of uncer-tainty to assign probability estimates to alternatives based on thelimited knowledge that one has. According to Bayesian decision the-ory, where there is no such limited knowledge upon which to basejudgments of probability, rational choosers would assign an equalprobability to each outcome they confront. Now assume that a ra-tional chooser seeks to maximize individual utility. The economistJohn Harsanyi has argued that, confronted with a thin veil of igno-rance which allows complete knowledge of society and everyone’spreferences in it but ignorance of his or her own identity, a rationalchooser would assign an equal likelihood to being each person insociety. To maximize individual utility, the rational chooser wouldsympathetically identify with the preferences of each person, andask, How much utility would I experience if I had the preferences ofthis individual i under conditions C? Multiplying the utility that therational chooser would experience under C if he or she were i1, i2,i3 . . . in, by 1/n (according to the equiprobability assumption), and

Introduction 15

adding up the results, the rational chooser achieves a measure of theexpected utility if outcome X or Y under C is chosen. That is:

Urational chooser = 1/nUi1 + 1/nUi2 + 1/nUi3. . . . + 1/nUin.

In maximizing his or her utility behind a thin veil of ignorance itis as if the rational chooser were applying the principle of averageutility.

What now about the maximin rule of choice that Rawls has theparties apply in his original position? Harsanyi contends that theabsurdity of maximin is easy to demonstrate. For example, supposeyou have the opportunity to choose between two bets, (1) and (2),each of which has two outcomes, A and B. You have no knowledgeof the likelihood of outcomes A and B, but you do know the bets’respective payoffs:

A B(1) $1 $2(2) $ .10 $1,000,000

It is clear by looking at these examples that a rational person wouldnot hesitate to choose alternative (2). Following the equiprobabil-ity rule suggests this too.28 The maximin strategy, however, sinceit requires that we focus on the worst-off position, requires choiceof (1) (since if circumstances A prevail, you avoid the worst out-come, namely ten cents.) But by choosing (1) you give up the chanceof enormous gains for the sake of assuring yourself an additional90 cents. This seems clearly irrational. So Harsanyi says, “Rawlsmakes the technical mistake of basing his analysis on a highly ir-rational decision rule, the maximin principle, which [has] absurdpractical implications.”29

No doubt maximin is an irrational strategy under most circum-stances of uncertainty. Rawls admits this.30 But simply because it ismost often irrational does not mean that it is never rational to followmaximin. Suppose, in the preceding example, you need $1 immedi-ately to save your child’s life? Or suppose a tycoon as rich as Bill Gatesderives perverse pleasure from giving people the opportunity to playRussian roulette in hopes of gaining his enormous wealth. You knownothing about the gun(s) he uses. Whether or not you are inclined toplay, it would not be rational to base your decision on the assump-tion of an equal probability of becoming the world’s richest person or

16 samuel freeman

being shot, or on any assumption about the makeup of the weaponused. The tycoon’s gun(s) may have only two chambers and one bul-let, or, if you are lucky, six chambers and one bullet. They may alsohave six chambers and five bullets; or he may play with 10,000 gunsand pick at random from 9,999 that are fully loaded and one six-shooter that has five bullets. You have no basis for knowing anythingabout these facts. Under these dire circumstances, one does not haveto be risk averse to make no probability assumptions whatsoever andprudently refuse the tycoon’s wager.

Now consider a decision problem more like the one envisionedby Rawls (though still involving factual information his original po-sition does not allow). Consider the following game of chance. Youare confronted with the following choices and schedule of payoffs:

Sum expected utilityStates of World assuming equal probabilityA B

(1) −100 500 = 200Choices

(2) 100 200 = 150,

where (1) and (2) are two potential choices, and A and B representtwo states of the world that may occur. The numbers representpotential payoffs, what you receive if you choose alternative (1) or(2) and state of the world A or B transpires. These payoffs may bewinnings (in dollars) from bets, or the average utilities experienced(over a lifetime) by individuals in different social classes A and B insocieties (1) and (2), respectively.

Now assume that you have no knowledge whatsoever about thelikelihood of either alternative, A or B. Bayesians say that, ignorantof probabilities of A or B we are to apply the principle of insufficientreason: We are to calculate the expected utilities of choices (1) and(2) as if the probability of A and B were each .5 and then pick thecourse of action whose expected utility is maximal. Therefore, weshould choose (1) over (2) (since the expected utility of (1) [−100(.5) +500(.5) = 200] exceeds the expected utility of (2) [100(.5) + 200(.5) =150].

But now assume that, unbeknownst to you the chooser, 90% ofchoosers end up in Class A, and 10% in Class B. Given accurate in-formation we can say then that the real likelihood of A is .9 and

Introduction 17

of B is .1. Then the expected utility of choosing (1) is −40 [−100(.9)+ 500(.1) = −40], while the expected utility of choosing (2) is110 [100(.9) + 200(.1) = 110]. That is,

States of World Sum expected utilityA B given accurate information

(1) −100 500 = −40(2) 100 200 = 110

Depending on what is at stake, forming expectations according tothe principle of insufficient reason can result in a minor setback orcomplete ruin. If all that is involved in choosing between (1) and (2)is a choice between bets on horses or voting on some minor piece oflegislation, insufficient reason might be the rational strategy if youare ignorant of relevant information, for you will have the opportu-nity to play again and recoup your losses. If, however, a person’s lifeis at stake, or all future prospects, then it seems a different matterentirely. Suppose you are choosing which society to enter: (1) is lib-ertarian laissez-faire capitalism, while (2) is Social Democracy thatprovides a social minimum. There are two classes in each society,A and B, which represent the unfortunates (A) and the well off (B)in each society. Given that the risk of being an unfortunate is .9, itwould be most unfortunate indeed to assume otherwise, choose ac-cording to the principle of insufficient reason, and end up in A, theworst-off class in a libertarian society.

Strict Bayesians contend that the principle of insufficient reasonis the rational strategy to adopt under complete uncertainty no mat-ter what is at stake. But in what sense can it be rational to risk one’sfuture, or life itself, on a strategy (namely insufficient reason) unsup-ported by evidence and which one otherwise has no reason to adopt?The principle of insufficient reason says that if there is no reasonfor assigning one set of probabilities rather than another, then weought to assign equal probability to each outcome.31 But if there isinsufficient reason for one set of probabilities rather than another,it should follow that there is no reason to assign equal probabilityeither. If every probability assignment is groundless, then perhapsthe rational thing to do is to assign no probability at all,32 at leastunder life and death circumstances or when one’s life prospects areat stake and there is an acceptable alternative.

This roughly represents the reasoning behind Rawls’s argumentagainst adopting the principle of insufficient reason in the original

18 samuel freeman

position.33 But by itself the rejection of the equiprobability assump-tion under conditions of complete uncertainty does not necessarilyspeak in favor of adopting the maximin strategy. Two other condi-tions need to hold to make maximin rational. These conditions are,first, that the choice singled out by following a maximin strategy isan acceptable alternative we can live with. There is little point in fol-lowing maximin if it results in an unacceptable worst outcome thatis only marginally better than some other possible worst outcome.When this condition is satisfied, then every position is acceptableno matter what the outcome. Second, all the other alternatives have(worst) outcomes that we could not live with and accept (so if someother alternative has a worst outcome you could live with and accept,then maximin may not be an appropriate strategy).34

Rawls contends these three conditions for following the maximinstrategy are satisfied in the original position when choice is madebetween the principles of justice and different versions of the prin-ciple of utility. Because all one’s future prospects are at stake in theoriginal position, and there is no hope of renegotiating the outcome,a rational person would agree to the principles of justice instead ofthe principle of utility (average or aggregate). For the principles ofjustice imply that no matter what position you occupy in society,you will have the rights and resources needed to enable you to ex-ercise your rational capacities to pursue a wide range of conceptionsof the good. With the principle of utility there is no such guarantee.Of course, by choosing the principles of justice, there is no guaran-tee that everyone will be in a position to pursue his or her currentconception of the good either. A rich plutocrat, for example, mayno longer enjoy his privileged position. But neither the principle ofutility nor any other traditional moral conception would guaranteethe rights and resources the plutocrat currently enjoys. Almost allmoral theories agree that justice is not simply a Pareto improve-ment on existing conditions, for the status quo itself may be unjust.At least with justice as fairness all persons have the necessary free-doms, opportunities, and resources to freely pursue a wide range ofconceptions of the good. This is not guaranteed under the principle ofutility.

Controversies surrounding the rationality of maximin choice inthe original position have obscured three other arguments Rawls re-lies on to support justice as fairness (all in Section 29 of Theory). Each

Introduction 19

of these depends on the concept of a “well-ordered society.” Sincetwo of them also rely on the idea of the stability of a well-ordered so-ciety, they will be discussed in the next section. The first of Rawls’sthree arguments proceeds from the idea that choice in the originalposition is an agreement and as such involves certain “strains ofcommitment.”35 As background to this argument, consider the fre-quent objection that there is no genuine agreement among personsin the original position because of the veil of ignorance. The partiesin effect are described in the same way since they know nothing dis-tinctive about themselves, and so there is no basis for bargaining tobegin with. But if not, there is no point in describing their decisionas a contract or agreement. The original position really involves therational choice of one person.36

The objection assumes that all contracts are like economic con-tracts in that they must involve bargaining based on conflictinginterests. But none of the major historical contractarians – Locke,Rousseau, Kant, or even Hobbes – conceived of their social contractson the order of an economic bargain.37 Underlying their appeals (andRawls’s) to a social contract is the idea that every member of societyshould be able to accept the same terms of cooperation because theyachieve certain interests everyone has. The mutual acknowledg-ment of principles involved here warrants the term “agreement,” andthe mutual precommitment involved might just as well be called a“contract.” For each agrees only on condition others do too, and all tiethemselves into social and political relations permanently to achievecertain common purposes as well as their individual interests.38 Mar-riage contracts are just one example of a mutual precommitment thatis not a bargain or based in a conflict of interests but which assumescommon purposes.

To understand the original position in this way indicates why itmay be said to involve a contract or agreement. By mutual accord andon condition that others do too, all the parties commit themselvesin advance to certain principles in perpetuity. Their precommitmentis reflected in the fact that once these principles become embodiedin institutions, no means would allow one or more persons to departfrom the terms of their commitment. The result is that the partieshave to take seriously the legal obligations and social pressures theywill incur by their agreement, for there is no going back to the initialsituation. So if they do not sincerely believe that they can accept the

20 samuel freeman

requirements of a conception of justice and conform their actionsand life plans to it, then these are strong reasons to avoid choosingthose principles. It would not be rational for a party to take risks,falsely assuming that if it turns out badly, he or she can violate theterms of agreement and later recoup the situation.

Rawls gives special poignancy to this precommitment require-ment by making it a condition that the parties cannot choose andagree to principles in bad faith; they have to be able, not just to livewith, but also to endorse the principles of justice once in society.Essential to Rawls’s argument is the assumption of willing compli-ance with requirements of justice. This is what he means by “thestrains of commitment.” The parties are choosing principles for awell-ordered society (discussed in the next section), where everyoneis assumed to have a sense of justice: they accept the principles ofjustice and want to act as these principles demand. Given this re-striction, the parties can only choose principles they believe theywill be able to accept and comply with come what may. A party,then, cannot take risks with principles he or she knows it will bedifficult to comply with. That would be making an agreement inbad faith, and this is ruled out by the conditions of the originalposition.

Rawls contends that these “strains of commitment” created bythe agreement strongly favor the principles of justice over the princi-ples of utility and other consequentialist views. For it is much moredifficult for those who end up worse off in a utilitarian society towillingly accept their situation. Given what we know about humannature, the person is rare who can freely and without resentmentsacrifice his or her life prospects so that those who are better offcan have even greater comforts, honors, and enjoyments. This is toomuch to demand of our capacities for human benevolence. Besides,why should we want to encourage such subservient dispositions andthe accompanying lack of self-respect? The principles of justice, bycontrast, conform better with our desire for self-respect and naturalmoral capabilities to reciprocally recognize and respect others’ le-gitimate interests while freely promoting our own good. The strainsof commitment incurred by agreement in the original position pro-vide strong reasons for the parties to choose the principles of justiceand reject the risks involved in choosing the principles of average oraggregate utility.

Introduction 21

C. The Good of Justice and the Stabilityof a Well-Ordered Society

Rawls’s strains of commitment argument indicates a rarely notedfeature of his account: it involves in effect two social contracts.First, hypothetical agents situated equally in the original positionunanimously agree to principles of justice. This agreement has at-tracted the most attention from Rawls’s critics, especially the argu-ment from the maximin criterion. But hypothetical agreement in theoriginal position itself is patterned on the general acceptability andfeasibility of a conception of justice among the members of society.39

There must be a genuine possibility that real persons, given humannature, can agree and act on principles of justice. Rawls’s argumentfor stability has attracted little commentary. But it is crucial to un-derstanding Rawls’s argument for justice as fairness.

Rawls expresses this second contractarian requirement via thecondition that principles of justice are to be agreed to in the orig-inal position only if they can be generally acknowledged within andremain stable under conditions of a “well-ordered society.” This is acentral idea in Rawls’s theory.40 It provides a way for testing whethermoral conceptions of justice are consistent with psychological andsocial theory. In a well-ordered society, (1) everyone accepts the samepublic conception of justice, and their general acceptance is publicknowledge; (2) society consistently realizes the generally acceptedconception in its institutions; and (3) everyone has an effective senseof justice, which leads them to want to do what justice requires ofthem. A well-ordered society is an ideal social world.41 For it is de-sirable that people know and freely accept the principles of justiceregulating their basic social institutions.

The stability requirement says that the parties in the original posi-tion are to choose principles that will be feasible and enduring withina well-ordered society. It bears emphasizing that Rawls is not con-cerned with Hobbesian stability or peace and tranquillity for theirown sake. The stability of a grossly unjust society is worth littleor nothing by itself, particularly if its destabilization will result ina more just situation without great loss of life. Rawls’s concern iswith the stability of a presumptively just (or “well-ordered”) soci-ety, which depends on its members having certain moral motives.A conception of justice is stable when “those taking part in [just]

22 samuel freeman

arrangements acquire the corresponding sense of justice and desireto do their part in maintaining them” (TJ, p. 454/398 rev.). One con-ception of justice is relatively more stable than another when peopleare more willing to observe its requirements under conditions of awell-ordered society. The stability question raised in Theory is,Which conception of justice is more likely to engage our moral sensi-bilities and our sense of justice? This requires an inquiry into moralpsychology and the human good.

Rawls initially appeals to stability in the two remaining argu-ments (in TJ, Sec. 29) from the original position to argue against utili-tarianism and perfectionism. Briefly, Rawls first contends that theseand other “teleological” conceptions are not likely to be practicallyfeasible when made public under the conditions of a well-orderedsociety.42 A feature of a well-ordered society is that its regulativeprinciples of justice are publicly known and regularly are appealedto as a basis for deciding and justifying laws and basic institutions.Rawls contends that under this publicity condition justice as fairnessremains more stable than utilitarianism.43 For public knowledge thatreasons of maximum average (or aggregate) utility determine the dis-tribution of benefits and burdens understandably would lead thoseworse off to resent their situation. After all, their well-being andinterests are being sacrificed for the greater good of those more for-tunate, and it is too much to expect of human nature that peopleshould freely acquiesce in and embrace such terms of cooperation.44

The principles of justice, by contrast, are designed to advance recip-rocally everyone’s position in society, and those who are better off donot achieve their gains at the expense of the less advantaged. “Sinceeveryone’s good is affirmed, all acquire inclinations to uphold thescheme” (TJ, p. 177/155 rev.).

The publicity condition is also crucial to Rawls’s second stabil-ity argument for the principles of justice (TJ, pp. 178–82/155–9 rev.).These principles, when publicly known, give greater support to citi-zens’ sense of self-respect than do utilitarian and perfectionist views.Of self-respect Rawls says it is “perhaps the most important pri-mary good” (TJ, p. 440/386 rev.), since few things seem worth doingif people have little sense of their own value or no confidence intheir abilities. The parties in the original position will then aim tochoose principles that best secure their sense of self-respect. Nowjustice as fairness, by protecting the priority of equal basic liberties,

Introduction 23

secures the status of each as equal citizens; there are no “passivecitizens” who must depend on others to protect their interests polit-ically. Moreover, the second principle secures fair opportunities andadequate resources for all to make everyone’s equal basic libertiesworthwhile. The aim is to make all citizens socially and economi-cally independent so that no one need be subservient to the will ofanother. Citizens then can respect one another as equals and not asmasters or subordinates. Equal basic liberties and political and eco-nomic independence are primary among the bases of self-respect ina democratic society. For in the absence of a generally accepted reli-gion or other shared system of beliefs, people confront one anothermainly as citizens. The parties in the original position should thenchoose the principles of justice over utilitarianism and other tele-ological views both to guarantee their equal status as citizens andsecure their sense of self-respect, and to procure the same for others,thereby guaranteeing greater overall stability.45

Rawls relies substantially on the publicity condition to argueagainst utilitarianism and perfectionism. He says the publicitycondition “arises naturally from a contractarian standpoint” (TJ,p. 133/115 rev.). His point ultimately is that giving people knowl-edge of the moral bases of coercive laws is a condition of fully ac-knowledging and respecting them as responsible agents. People thenhave knowledge of the reasons for their social and political relationsand the formative influences of the basic structure of society ontheir characters, plans, and prospects.46 Moreover, when made pub-lic, principles of justice can serve agents as bases for their practicalreasoning and for political argument and justification. These consid-erations underlie Rawls’s contention that having knowledge of theprinciples that determine the bases of social relations is a precondi-tion of individuals’ freedom.47

Now we turn to Part III of Theory to examine in more detailRawls’s argument for stability and the good of justice. The stabil-ity problem for Rawls requires showing how a conception of justiceis realistically possible given human nature and certain fixed con-ditions of social life. To do so, Rawls assumes the ideal case of awell-ordered society. If a conception of justice is not workable there,then it is not feasible under less than ideal conditions. In this re-gard a conception is utopian and thus it is not rational to seek toachieve it.

24 samuel freeman

There are two parts to the stability argument in Part III. In the firstRawls undertakes to show how citizens in a well-ordered society ofjustice as fairness can come to acquire a sense of justice, a disposi-tion to act not simply according to, but also for the sake of justice,as defined by the principles of justice and the legal and social normsthat satisfy them. The relevant question here is whether justice asfairness is compatible with human nature. The moral psychology ofTheory, Chapter VIII, assumes that moral motives are real and arenot grounded ultimately in self-interests. Rawls contends that byvirtue of our nature we are inclined to appreciate and develop sin-cere attachments towards persons and institutions that affirm ourgood. He offers three “reciprocity principles” that develop this basicidea. The third says that it is a psychological law that, once attitudesof love and trust, and of friendly feeling and mutual confidence, havebeen generated in the course of moral development, “then the recog-nition that we and those whom we care for are the beneficiaries of anestablished and enduring just institution tends to engender in us thecorresponding sense of justice” (TJ, pp. 473–4/415 rev.). If so, then itshould be a normal part of citizens’ development in a well-orderedsociety of justice as fairness that they will acquire a desire to supportand maintain its laws and institutions. For justice as fairness itselfis built upon an ideal of reciprocity: basic institutions are to be de-signed to promote reciprocally everyone’s good as measured from abaseline of equality.

The second stage of Rawls’s stability argument in Theory builds onthe first, for it assumes that citizens already have an effective senseof justice and want to do what is right and just for its own sake.The question now is, Is exercise and development of the sense ofjustice compatible with the human good, or is having this dispositioncontrary to our interests and something which undermines our good?The primary purpose of Chapter IX of Theory is to show how exerciseof a sense of justice is not only compatible with, but can also promote,and realize the human good.

For Rawls the concept of the ‘good’ is formally characterized interms of certain rational principles of choice and rationally formeddesires. A person’s good is what is rational for that person to want,assuming that he or she has full and accurate information and hascritically reflected on his or her ends, made them consistent, and de-cided on effective means for realizing them. Rawls defines the good

Introduction 25

in terms of the plan of life that a person would choose under suchideal conditions of deliberative rationality (TJ, Sec. 64). The idea hereis that everyone has primary ends and commitments which (if theyare not already apparent) are discoverable upon careful thought underconditions conducive to reflective deliberation. These aims and com-mitments can be arranged in terms of importance and consistentlyscheduled according to rational principles of choice (enumerated inTJ, Sec. 63) to yield a plan of life suitable for each person. There maybe several desirable plans each of us can imagine. The plan of lifethat a person would choose under ideal deliberative conditions offull information, imaginative appreciation of the options, and so on,is most rational for that person and specifies his or her objective good(TJ, Sec. 64).

To provide content to this formal account of a person’s good, Rawlsappeals to another psychological law, the Aristotelian principle (TJ,Sec. 65). “Other things equal, human beings enjoy the exercise oftheir realized capacities (their innate or trained abilities), and thisenjoyment increases the more the capacity is realized, or the greaterits complexity” (TJ, p. 426/374 rev.). This principle implies that weare not just moved by the desire for pleasurable sensations and thepressures to satisfy bodily needs and achieve its comforts. Instead, wedesire to engage in more complex and demanding activities so longas they are within our reach. If so, then it will generally be rationalfor people (when given the opportunity) to realize and train theirmature capacities and incorporate certain “higher activities” (J.S.Mill’s terms) into their plans of life (TJ, p. 428/376 rev.). Otherwise,life becomes monotonous and boring.

The question raised by the congruence argument (Theory,Chap. IX) for stability is as follows: Is it rational in a well-orderedsociety to exercise and develop the sense of justice, as defined byjustice as fairness, and incorporate this virtue into one’s conceptionof the good? That is, from the point of view of deliberative rational-ity, is it rational to choose a plan of life in which one is steadfastlypredisposed to take up the impartial standpoint of justice and acton the principles that would be chosen from this (original) position?If so, then choice from these two ideal perspectives coincide, and theright and the good are congruent.

Rawls advances two main arguments to show that justice as fair-ness is congruent with the human good. The first invokes the good

26 samuel freeman

of participation in community and develops the idea of a just so-ciety as a “social union of social unions” (TJ, Sec. 79).48 This ar-gument and the second congruence argument (TJ, Sec. 86) both in-voke the Aristotelian principle and also (particularly in the second)the Kantian Interpretation of Justice as Fairness (TJ, Sec. 40). In theKantian interpretation Rawls seeks to give specific content to Kant’snotion of autonomy, understood as reason giving a law to itself. “Theoriginal position may be viewed . . . as a procedural interpretation ofKant’s conception of autonomy and the categorical imperative” (TJ,p. 256/cf. 226 rev.). Rawls’s thought here roughly is this: We are ca-pable of autonomy by virtue of “our nature as free and equal rationalbeings” (TJ, p. 252/222 rev.). Our nature as free rational moral agentsconsists in the two moral powers mentioned earlier, the capacity fora sense of justice and the capacity for a conception of the good. Theseare, in effect, the capacities for practical reasoning as applied to jus-tice. They are essential to our being free self-governing agents whohave a conception of our good and who can take responsibility forour actions and ends and participate in social life. Now the Kantianinterpretation says that the conditions of the original position can beconstrued as a “procedural interpretation” of the conception of freeand equal rational beings. The original position “models” this con-ception of the person (in ways Rawls explains in detail in subsequentworks).49 When the original position is construed as “modeling” themoral powers, then the principles chosen there can be interpretedas principles that we give to ourselves out of our nature as free andequal rational beings. If so, then to act on and from these principlesis to act autonomously in Kant’s sense: it is to act for the sake ofprinciples that express our nature as free and equal rational beings.50

Against this background Rawls argues that it is rational (in the“thin” sense discussed above and in TJ, Sec. 64) and thus part oftheir good, for the members of a well-ordered society to realize theirnature as free and equal rational beings. To argue for the rationality ofjustice Rawls relies on the Aristotelian principle – the psychologicallaw that, given appropriate conditions, we prefer ways of life thatengage our higher faculties – to contend that the development andexercise of the moral powers is a good for people generally. Withoutgoing here into the details of this argument for congruence,51 themain point is that, if it is indeed rational for people generally todevelop and exercise their sense of justice in a well-ordered society,

Introduction 27

then there should be no concerns regarding its stability. For thenit is normally in people’s interests to do what justice as fairnessrequires of them. Rawls’s congruence argument ambitiously aims toshow that justice is rational in the strongest way. It is not simplyinstrumental (as Hobbes tried to show) to each person’s good to dowhat just laws require; instead, acting on justice is an intrinsic andsupremely regulative good, for it enables each to realize his or hernature as a morally and rationally autonomous being.

The Kantian interpretation provides the subsequent basis forRawls’s Kantian constructivism.52 Constructivism in moral philoso-phy is an account of the validity and objectivity of moral statementsand principles. It denies both realism as well as the skeptical the-sis that moral statements do not possess a truth value or objectivevalidity. Constructivism says that, while moral statements are nottrue in the realist’s sense that they represent an independent orderof (moral) facts existing prior to moral reasoning, they nonethelesspossess a kind of validity that is appropriate to the distinct subjectmatter of moral theory. Moral statements are correct when they ac-cord with reasonable moral principles, and moral principles are rea-sonable when they are the product of a reasoning procedure that in-corporates all the relevant requirements of practical reason. Kantianconstructivism begins from a conception of the person and of prac-tical reason, the ideal of free and equal moral persons who are bothreasonable and rational. It “represents” or “models” this conceptionin a “procedure of construction,” (in Kant, the categorical impera-tive procedure, and in Rawls, the original position.)53 The principleschosen by the parties to this procedure are objective so long as allwho employ it reach the same or similar conclusions and the pro-cedure incorporates all relevant requirements of practical reason. Inthis regard, moral principles are said by Rawls to be “constructed”from a conception of the person and of practical reason.

Rawls offers Kantian constructivism as an alternative account ofthe objective validity of moral claims in contrast to “rational intu-itionism.” The traditional realist account of moral correctness (en-dorsed by Plato, the Rationalists, Sidgwick, G.E. Moore, and manyothers) says that moral principles are objective insofar as they aretrue, and they are true only insofar as they represent a prior andindependent order of (nonnatural) facts accessible to rational intu-ition. The problem with this account from a Kantian perspective is

28 samuel freeman

not simply that it posits a peculiar domain of nonnatural facts and aspecial faculty of reason with access to them. Rather, intuitionism, ifcorrect, undermines the Kantian idea of moral autonomy – that rea-son provides most basic moral principles out of its own resources.Kantian constructivism attempts to retain and make sense of Kant’sobscure idea of autonomy by depicting moral principles of justiceas the product of an objective procedure of deliberation that is de-signed to capture the main features of practical reasoning. Accordingto constructivism, objectivity of judgment (conceived as judgmentaccording to universal rules) precedes the notion of moral validity ortruth. Moral statements are sound or true, not in representing a priororder of moral facts but when they accord with principles that couldor would be accepted by fully rational persons in an objective proce-dure of practical reasoning.54 For Rawls, the procedure that bestowsobjective status on judgments of justice is the original position.

III. political liberalism

A. Problems with Rawls’s Argument for Stability

Political liberalism is seen by many as the result of Rawls’s com-ing to terms with his critics, especially communitarians.55 But thesecriticisms had little to do with Rawls’s development of his ideas.Instead, political liberalism results from certain tensions implicitwithin Rawls’s argument for stability in Theory, which becameonly more noticeable in Kantian constructivism (cf. PL, p. xvii/xixpb. ed.).

The Kantian conception of the person Rawls invokes in Theoryembodies a philosophical view of moral agency: We are free responsi-ble agents by virtue of the moral powers for practical reasoning. Uponthis basis Rawls develops accounts of autonomy as self-realization(TJ, Sec. 40), of the intrinsic good justice possesses by its relationto moral autonomy (TJ, Sec. 86), and of the epistemic objectivity ofjudgments of justice (TJ, Sec. 78). Rawls’s 1980 Dewey Lectures onKantian constructivism work out the details of these philosophicaltheses and mark a further stage in the development of justice as fair-ness as part of a general moral theory.

But Kantian constructivism proves to be a transitional stagein Rawls’s thought. Already in the Dewey Lectures Rawls says a

Introduction 29

conception of justice must have a “practical role” in a well-orderedsociety: It should provide a basis for “public justification” amongpersons who have different and conflicting conceptions of the good.“Because its principles are to serve as a shared point of view amongcitizens with opposing religious, philosophical and moral convic-tions, as well as diverse conceptions of the good, this point of viewneeds to be appropriately impartial among those differences.” So,the kinds of reasons that can be invoked to justify and apply the con-ception of justice must be limited to “part of the truth, and not thewhole.” “Justice as fairness tries to construct a conception of justicethat takes deep and unresolvable differences on matters of funda-mental significance as a permanent condition of human life.”56

But surely (one might argue) there will be deep and unresolvabledifferences in a well-ordered society of justice as fairness about thenature of morality and moral justification – even about the justifi-cation and truth of the shared public conception of justice govern-ing society. So, too, we might expect that there will be fundamen-tal differences regarding the good of autonomy. Imagine a liberalCatholic who accepts the principles and duties of justice as fair-ness but who sees them as natural laws, part of divine law willedby God to enable rational beings to achieve the summum bonum,the contemplation and glorification of God. According to the liberalThomist, principles of justice are divinely ordained and are know-able, some perhaps as self-evident truths, by the natural light ofreason. This denies Kantian constructivism’s position that moralprinciples originate in practical reason. Rejecting the constructivistview of justification and objectivity, the liberal Thomist also re-jects the Kantian conception of autonomy that underpins the con-gruence argument. Recall that the aim of the Kantian congruenceargument is to show that justice is a supremely rational good foreach in a well-ordered society. The argument depends on showingthat the sense of justice is the same as the desire to realize our na-ture as free and equal rational beings and thereby become morallyautonomous. But the liberal Thomist denies this identification ofdesires. The sense of justice is a desire to conform to God’s naturallaws, not a desire to express our nature as the author of these laws.Not only is autonomy not an intrinsic good; to think so is a con-ceit of human reason that comes from rejecting the divine source ofmorality.

30 samuel freeman

So we have a peculiar situation: There will be reasonable citizens57

in a well-ordered society of justice as fairness who can endorse theprinciples of justice and the institutions they authorize, but yet, be-cause of their religious views, cannot accept its public justification.Significantly, this problem is not peculiar to orthodox Catholics andother religious believers. It is typical of anyone who accepts lib-eral principles of justice on grounds at odds with Kantian construc-tivism. Nothing about a well-ordered society ensures against therebeing significant numbers who accept the public principles of jus-tice for utilitarian, perfectionist, or pluralist reasons; others may ac-cept these principles and yet endorse moral relativism or skepticismabout moral truth. Given basic liberties of freedom of thought, con-science, and association it is predictable that these and other moraland philosophical doctrines will gain adherents. This means that thepublic conception of justice Rawls envisions for a well-ordered soci-ety incorporates premises that starkly conflict with many citizens’conscientious beliefs. Consequently this conception, far from ensur-ing a shared basis for justification and stabilizing the social system,is likely to have destabilizing consequences. It can adversely affectthe sense of self-respect of those who hold non-Kantian philosophicaland ethical views and even undermine their allegiance and supportfor just institutions.

Do these problems mean that justice as fairness is unworkable?It may be unworkable as initially conceived. But Rawls believes jus-tice as fairness still can serve as a publicly shared conception of jus-tice that provides a common basis for justification if certain featuresare altered or relinquished. In Political Liberalism Rawls surrendersthe Kantian congruence argument for stability – the claim that moralautonomy and therewith justice are supremely rational goods in awell-ordered society. Given Rawls’s account of goodness in terms ofrational desire, this claim appears untrue for many reasonable per-sons – not just liberal Catholics and members of other faiths, butliberal utilitarians, perfectionists, pluralists, and others too.

Second, Rawls gives up the comprehensive constructivist positionthat says (a) that there is no moral order apart from the activity ofpractical reason, and (b) that moral objectivity and the validity ofjudgments are to be understood only in terms of correct judgmentfrom a suitably constructed social point of view.58 These philosophi-cal claims, even if true, cannot gain widespread agreement as part of a

Introduction 31

public conception of justice because they conflict with other not un-reasonable views about the nature of morality and moral objectivityand validity.

Third, the Kantian conception of the person (as free, equal, rea-sonable, and rational) is retained but not without modification. Itis no longer presented as part of a “comprehensive doctrine” thatgrounds morality in the conditions of moral agency and the powersof practical reason. Rather “the idea of the person as having moralpersonality with the full capacity of moral agency is transformed intothat of the citizen” (PL, pb. ed., p. xlv). This “political conceptionof the person” provides the basis for a “freestanding” conception ofjustice. As discussed in the next section, a political conception isnot to be committed to any particular philosophical doctrine but isto have its basis in ideas implicit in democratic thought and culture.

Before proceeding, something should be said about the idea of rea-sonableness that figures prominently in Rawls’s later works. WhileRawls relied on a distinction between reasonable and rational inTheory, the distinction is not made explicit. There the concept of ra-tionality was used to characterize the good and what is to a person’sadvantage.59 Reasonableness concerned the concept of the right, par-ticularly the ideas of fairness and willingness to moderate one’s de-mands out of respect for others and their legitimate claims.60 Be-ginning with “Kantian constructivism,” Rawls explicitly marks adistinction between “the Reasonable” and “the Rational.”61 But hestill does not ever try to define the concept of ‘reasonableness’. Thiscan be frustrating for the reader, since this concept is so crucial toRawls’s argument and is used by him in several different ways. Tosome it may seem that the idea is so obscure that it explains little,and its recurrent usage only masks the lack of a proper argument.

On Rawls’s behalf, it should be said that there should be no pre-sumption that a list of necessary and sufficient conditions can ex-haust the meaning