Jump to ContentJump to Main Navigation
Women, Culture, and DevelopmentA Study of Human Capabilities$

Martha C. Nussbaum and Jonathan Glover

Print publication date: 1995

Print ISBN-13: 9780198289647

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0198289642.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 16 June 2021

Justice, Capabilities, and Vulnerabilities

Justice, Capabilities, and Vulnerabilities

(p.140) Justice, Capabilities, and Vulnerabilities
Women, Culture, and Development

Onora O'Neill

Oxford University Press

Abstract and Keywords

O’Neill investigates the relationship between capability and vulnerability in forms of dependence and victimization, arguing that the satisfaction of preferences ought not to be taken as a normative criterion in political economy. Defending the capabilities approach by connecting it to a strand of Kantianism, O’Neill contends that Kant's categorical imperative continues to be a valuable test in social policy and supplies a powerful safeguard against the exploitation of vulnerability.

Keywords:   capabilities, dependence and victimization, independence and autonomy, justice, Kant, practical reasoning, preferences

Thinking about justice for those who endure acute poverty and vulnerability in distant parts of the world raises numerous difficulties, and suggests various possibilities. Jonathan Glover has charted both some possibilities and some of their difficulties, and in many respects I agree with his map. Putting the problem simply, it seems that cosmopolitan accounts of morality and justice, including many liberal accounts of justice, are often fellow‐travellers with old and ugly forms of imperialism, and sometimes rely on exaggerated, indeed false, assumptions about human rationality, independence, and self‐sufficiency. On the other hand, anti‐cosmopolitan (e.g. communitarian) accounts of morality and justice fail the distant poor and vulnerable in even sharper ways, because they have nothing much to say about action towards those who are distant or different, since they exclude ‘outsiders’ from the domain of justice.

Both approaches, it seems, underplay or discount the predicaments and vulnerabilities of many people in the Third World, and in particular pay little attention to the special vulnerability of poor women in poor societies.1 Yet in the contemporary world, action at a social distance, indeed action whose effects are felt across the world, is ubiquitous, and we can hardly expect to act justly or to keep our hands clean just by opting out. In our world we need an account of cosmopolitan justice that neither fails to appreciate the vulnerabilities of the vulnerable nor connives with the powers of the powerful.

1 Justice and Preferences

Many contemporary accounts of liberal justice have a cosmopolitan surface but are fundamentally statist:2 they tend to depict justice as justice within a State. A more deeply cosmopolitan account of justice needs a wider view of the domain of justice. Questions about the scope of principles of justice are being worked on by numerous authors in several disciplines, and I shall (p.141) leave them aside here. In my view, although I shall not argue for it in this paper, a genuinely cosmopolitan conception of the scope of justice is possible, and an irredeemably communitarian position is a non‐starter for thinking seriously about justice to the most vulnerable.

However, even if questions of scope can be resolved, an adequate cosmopolitan account of justice is not easily provided, for several reasons. One reason is that the broadly empiricist starting‐points relied on by most contemporay accounts of cosmopolitan justice (even if freed from statist thinking) are ill‐adapted to take the vulnerabilities of the poor, and in particular of poor women with dependants in markedly patriarchal societies, seriously.

Jonathan Glover points out that the cosmopolitan positions advocated by contemporary theorists of justice often disagree about the relative weight to be given to liberties and to equalities. For example, this weighting is in dispute between ‘social justice’ liberals such as John Rawls and libertarian liberals such as Robert Nozick. However, the deeper difficulties with these would‐be cosmopolitan accounts of justice may lie not in these areas of disagreement, but in the premisses by which conclusions such as these are reached. The supposedly strong contrast between Rawls and Nozick, and their respective followers, is much less evident in their starting‐points. Like most contemporary work on justice, including Utilitarian approaches, both start from up‐to‐date versions of broadly empiricist views of action, so accord with economic theory in depicting action as motivated by preferences. Correspondingly, both see practical reasoning as used to choose means to satisfy preferences. As Hume put the point, reason by itself is ‘inert’; it has no ends of its own, and cannot yield us an account of good or of justice, but can work out means to given ends.

Those who see human action in this way offer a rather convincing account of motivation (which does not concern us here) but have some difficulty in providing a convincing account of the good or of justice. Most settle for a subjective account of the good, and identify it with the satisfaction of preferences. Utilitarians generalize this strategy by offering a subjective account of the general good—maximal utility—which they use as the basis for their account of justice. Other contemporary work on justice is more cautious. It too takes a subjective view of individual good: individual preferences provide reasons for action for individuals, and satisfaction of those preferences is a good for individuals, but it asserts that there is no way to aggregate preferences or preference satisfactions to provide an account of the general good, on which an account of justice could be based. Justice must then be seen as a set of constraints on preference satisfaction, rather than as the optimum of preference satisfaction for all. This is how liberties and equalities come to be seen as the components of justice in contemporary non‐utilitarian liberal writing. Both social justice and libertarian accounts of justice depict human beings as pursuing their individual (p.142) ‘good’ in doing what they prefer, and doing so justly when their action stays within constraints defined by certain liberties and required for certain equalities.

No doubt liberties and equalities, or at least some liberties and equalities, are important for justice. But the disputes about the merits of trying to build an account of justice out of these very abstract elements seems to me to obscure what is most fundamental to most recent liberal approaches of justice, which is the thought that rational action is efficient pursuit of individuals' subjective goods—or preferences. This claim is often seen as uncontroversial background; I believe that it is controversial and moreover that it is a reason why cosmopolitan accounts of justice can be derailed and even become hostage to questionable ideals and attitudes, which undercut justice towards distant and vulnerable others. However, it can be hard to trace exactly how these underlying assumptions create problems.

In part this obscurity arises because contemporary work on justice takes varying views of preferences. Some hold realist views of preferences; others see preferences as revealed in action; yet others slide between these two incompatible views of preferences. On the realist view, preferences are real states of agents and cause their acts; agents are viewed as rational if they pursue what they most prefer, and it is an empirical truth that human beings are more or less rational, so aim to satisfy their preferences to the greatest extent possible. On the revealed view, preferences are not seen as real states of agents but are ascribed to them on the basis of what they do, and it is a conceptual truth that rational agents always pursue what they prefer (because what they pursue will be taken as revealing what they prefer).

One difficulty, which Jonathan Glover identifies, arises from the possibility of false consciousness. If preferences, understood realistically, can reflect false views of one's own advantage and good, preference satisfaction may not provide suitable materials for constructing a theory of justice. In effect, the very data for building an account of justice may be systematically corrupted. It is well established that the most deprived often acquiesce in their lot—this is the traditional problem of the ‘happy slave’. More specifically, it is well known that preferences adapt to realities, even to grim realities. So too do beliefs. Poor women, for example, have been shown to underestimate their own health problems, and to acquiesce in harsh subordination. The fact that preferences and beliefs ‘adapt’ to deprivation seems to me to put great strain on any project of working out what is ethically important or just within an account of action as governed by preferences. Where people accept that small mercies are all they deserve, when they see their oppressors as benevolent father‐figures, these distorted preferences will be given undeserved weight within any fundamentally subjectivist approach to ethics and justice. Absent preferences which would, we may (p.143) think, reflect a more accurate view of the situation, will have no part in ethical reasoning. Moreover, it is hard to see how ‘false preferences’ could be discounted without a powerful theory of real needs or interests, which is not available to those who start with a preference‐based conception of action.

But if one takes a revealed view of preferences, matters are probably worse: for action is then seen as the evidence for preferences, and so the compliant actions of the intimidated and vulnerable will be interpreted as showing that they prefer to comply, so suggesting that they are getting what they prefer. The problem in this case is not that ‘false preferences’ will be given weight they may not deserve, but that the very notion of a ‘false preference’ may lack sense.

This approach may fail to recognize major injustices. If preferences reflect action, and preference satisfaction is taken as the source of ethical legitimation, all sorts of oppression and domination may be made to seem legitimate. More specifically, once preferences are taken as the basis for action, and their satisfaction (subject to constraints) as just, it will be hard to criticize the range of social phenomena we call patriarchy. For the victims of patriarchy have preferences that are highly adapted to their circumstances: yet on ordinary, pre‐theoretical views such adaptation seems to many to exacerbate rather than obliterate or mitigate injustice.

If we want a better account of justice to the most vulnerable, it will surely be important not to set out from a perspective that so readily legitimates or minimizes vulnerability and oppression. Those who view action as governed by preferences, and justice as a set of constraints on preference satisfaction, face double trouble here. If they take a realist view of preferences they must acknowledge that the vulnerable will adapt their preferences to the realities; if they take a revealed view of preferences they will see all compliance and subordination as chosen. Consideration of the realities of the vulnerable lives of poor women in the Third World destroys the plausibility of thinking that preferences provide a guide to each individual's good. At most they provide a guide to each individual's current subjective perception of good, but one that can often be deeply misleading. The search for an account of justice cannot be seen merely as a search for constraints on the pursuit of preferences, once preferences are seen in this light.

2 From Preferences to Capabilities

For these and other reasons I believe, as many others now do, that there is great promise in Amartya Sen's project of taking capabilities rather than preference satisfaction as the basis for approaching questions of development (p.144) and justice. Concentration on securing capabilities, and in the process eliminating vulnerabilities, leads to a view in which both empowerment and liberty can be taken seriously, and in which apathy, false consciousness and adapted preferences cannot be invoked to legitimate action that harms the vulnerable.

However, if capabilities are to be the concern of justice, we need to consider how they are to be individuated, counted, and compared. Which capabilities and sets of capabilities are important for justice? How are they to be compared and judged better and worse? It won't be enough to say that one bundle of capabilities is better than another when it includes all that the other does and more besides. Pareto optimality will provide too weak an ordering of bundles of capabilities to be useful. The mere fact that we cannot this evening have the cake—or the bread—that we ate this morning highlights the fact that the most elementary decisions require us to weigh alternatives of which neither is Pareto optimal. In the real world we constantly face real costs as well as real opportunities.

Individuating and ranking (sets of) capabilities is not easy. As Jonathan Glover points out, their individuation will depend on the descriptions that are chosen,3 and these have no intrinsic metric. This suggests that the capability approach is going to need in the background some other considerations—a theory of basic needs, an Aristotelian account of human flourishing, or of course both in tandem, or some further possibility. This background theory will have to provide the considerations by which better and worse, more and less important, capabilities and sets of capabilities are distinguished. Of course, one need not assume that such a theory would be able to recapture the whole metric structure which preference‐based conceptions of action and the accounts of justice that build on them purportedly provide. However, it may be that very much less metric structure would be enough to make many judgements about which capabilities are more and which less important for public policy and personal life.

A highly‐sympathetic way of providing a background theory, which would enable us to distinguish better from worse (sets of) capabilities, is the Aristotelian route that Martha Nussbaum has explored.4 It would provide an account of objectivity in ethics, and as Jonathan Glover points out, could still be less difficult to establish than a more theoretical, Platonist form of moral realism. I do not doubt that if we could establish a convincing Aristotelianism it would provide a way of comparing and ranking bundles of capabilities. However, I am less sure how or whether this can be done.

It is striking that Aristotelians today are one thing in the English‐speaking world and another in the German‐speaking world. All Aristotelians hope to discern what human flourishing entails, but whereas many recent (p.145) Anglophone Aristotelians emphasize that human flourishing is highly variable and sensitive to context, many German‐speaking Aristotelians tend to think that human flourishing is far more socially determinate, indeed that it is not that different from the Thomist vision of flourishing accepted in Catholic social thought. At a guess, it is the latter, socially conservative Aristotelian thinking that has the deeper and wider political significance today, via its influence on Christian Democratic politics. This makes a lot of difference: Christian Democratic Aristotelianism tends to endorse very different social practices from those recommended by Social Democratic Aristotelianism. Patriarchal social relations seem to many Christian Democratic Aristotelians the proper structure for human flourishing. Not so to Social Democrats. Of course, a great thinker will have many readings, and I am not suggesting that either Aristotle is the right one—but it does point out that if an Aristotelian account of human flourishing is to enable us to choose between bundles of capabilities, we will need to go one way or the other. Like Jonathan Glover I therefore want to understand where the Aristotelian research programme will lead. Will it, for example, show that the capabilities women need for poverty and patriarchy are better or worse than the capabilities women need for poverty and isolation?5

Martha Nussbaum has suggested that Social Democratic Aristotelians look to a thick vague conception of the good.6 I have no difficulty with the vagueness. Any moral reflection at this level is going to be relatively indeterminate. Nor indeed do I have trouble with the thickness, that is with the thought that there are a lot of different matters that must be taken into account. But I am not sure whether what we will be looking at when we get this far is best thought of as a conception of the good, or of the good life, or rather as a set of constraints within which many good lives, and perhaps also some less good, yet not unjust, lives may lie. The metaphors are perhaps revealing. Martha Nussbaum uses the traditional Aristotelian metaphor of the target. But if we speak in terms of this metaphor, we will also be tending to the thought that there is a centre to the target, an optimal way of achieving the human good, an ideal of life. This picture puts heavy demands on any account of human flourishing, that were ostensibly met by many traditional readings of Aristotle. If it can provide this account of flourishing, Aristotelian Social Democracy will have to be rich not only in its conclusions, but also in its premisses. Yet thinking about justice in a way that takes account of human vulnerability and powerlessness should perhaps consider starting from more modest assumptions.

(p.146) 3 Poverty and Practical Reasoning

The poverty of philosophy is usually taken to be a criticism. It is taken to mean that philosophy leads to meagre conclusions. But there could be another way of taking the metaphor, which perhaps would be particularly appropriate in a discussion of capabilities, poverty, and development. This would stress the meagreness of the initial materials, but also the possibility of seeking conclusions which were rich in relation to those starting‐points. Poverty is then seen as a matter of having rather little material to start out with, and dealing well with poverty as a matter of making good use of that little. This image of how to proceed might be useful for seeing how we might work out which capabilities are important without needing to show what constitutes a flourishing life, nor which are the most flourishing lives. We might simply concentrate on establishing constraints that must be observed for any flourishing life.

In terms of a metaphor that Martha Nussbaum also draws from Aristotle, we would then be thinking about the outline or limits rather than the target. The shift in metaphor may help, for it suggests that what is important is that acts and policies not lie beyond a certain limit, rather that they cluster as close as possible to some central ideal.

There thoughts lead me in roughly the same direction as Jonathan Glover. Like him I shall try to trace a route that starts out thinking directly about action, about functionings and capabilities, but uses minimal and general considerations in trying to see which capabilities are the most important for human lives. The considerations which Jonathan Glover proposes as providing a method for ethics are clarification and consistency, which, he suggests, will lead us close to Rawls's Reflective Equilibrium.7 I share his view that we should avoid strong and controversial assumptions, even if it means that we have to forgo some strong conclusions that might be reached by a full‐blown Aristotelian position or by certain preference‐based approaches to justice.

However, if we do not insert considerations of clarification and consistency into an Aristotelian account of flourishing, or a preference‐based account of action, where should they be deployed? An alternative would be to apply them to the principles and norms already established in our lives and institutions or proposed for their reform. This is a plausible way of proceeding, because formal standards of clarity and consistency can apply directly only to elements that have propositional structure and content. Considerations of consistency and clarity cannot apply directly to lives and (p.147) institutions, but they can be applied to the (system of) principles, norms or maxims by which we structure lives and institutions, and to the categories by which we describe, individuate, and enumerate both capabilities and vulnerabilities. This shift of starting‐point brings us close to Kant's own account of practical reasoning, but distances us from contemporary ‘Kantian’ work on justice, which for the most part takes a broadly empiricist view of action, and see preferences and their satisfaction rather than principles of action and their embodiment in human lives as the locus of reasoning about justice.8

Yet when this conception of action is combined with the spare, formal considerations to which Jonathan Glover draws our attention, it has surprising implications. For the combination of a conception of action that is not preference‐based and a conception of critical clarification and consistency leads in the direction not of contemporary Kantian theories of justice but of Kant's own account of ethics, as oriented by the requirement of acting only on principles that can be acted on by all. Principles that do not meet this standard cannot coherently be recommended as principles for all.

Attempts to domesticate versions of Kant's criterion within a preference‐based theory of action have been numerous, but also have proved endlessly paradoxical and highly destructive of the point of the criterion. Instead of the thought that we should act only on principles on which we hold that others too can act, we are led in the direction of conceptions of justice which demand institutions and policies that receive either actual consent from all affected or the hypothetical consent of beings with enhanced, idealized rationality or knowledge. It is obscure why either criterion should be thought of as defining justice. Kantian considerations cannot easily be exported into an empiricist framework in which actions are determined by preferences.

Yet the simple thought that justice demands principles on which others too could act has powerful implications for the range of capabilities that might be compatible with justice. For example, norms or principles of injuring others plainly cannot be regarded as norms or principles for all, because those who are injured will have their capabilities damaged, so will be hindered from injuring in their turn. Of course, sporadic injury can be reciprocal—think of vendettas—but systematic injury by all is impossible. A commitment to injury—by violence, by coercion, by intimidation, by deception, by poverty or by patriarchy—will always be a commitment that is possible for perpetrators but not for victims. It cannot be enacted by all, so is unjust. The capabilities that all can have do not therefore include capabilities (p.148) for injury, oppression, manipulation, coercion, deceit and the like. If none of these capabilities can be justly enjoyed, then institutions which make some vulnerable to others will be unjust. In particular, extremes of poverty, dependence, social isolation, overwork, and patriarchy, which burden so many women in poor, and even in less poor, economies are thereby shown unjust.

More concretely, since poverty is an enormous source of vulnerability and dependence of many sorts, it cannot be right to leave in place the institutional structures which produce and perpetuate poverty. Equally, since patriarchal social forms consist of unsafe structures of dependence that institutionalize vulnerabilities, these social forms cannot justly be left in place. Of course, if somebody suggests that poverty and patriarchy need only be mitigated and not abolished for justice to be instituted, this will be a matter for discussion—but the burden of proof will always be on those who suggest that there are ways of securing everyone's capabilities to act on the same deep principles while retaining these ancient structures of deprivation and subordination.

In some ways this Kantian conception of just social relations and their institutionalization is not far from the one that Martha Nussbaum sketches in ‘Aristotelian Social Democracy’ and elsewhere under the (rather Kantian) heading ‘Architectonic Functionings’. There she suggests that the two key human functionings, which shape all others, are those of reasoning practically and of affiliation. She characterizes practical reasoning as planning and organizing all other functionings and affiliation as acting with and to others. These two ideas are part and parcel of the Kantian conception of practical reason and of justice as requiring both consistency and connection to others, which is articulated in the thought that right action conforms to principles all can share.9

4 Autonomy and Independence

These conclusions may meet a sceptical response because it is generally supposed that Kantian approaches to the predicaments of the poor and vulnerable are flawed, in that they assume an excessive, idealized conception of human beings and of human autonomy. Kantian thinking is said to be too concerned with autonomy, which it depicts as an ideal or virtue, indeed sometimes as the cardinal virtue of a liberal political order. Communitarians, virtue ethicists, and certain feminist writers have criticized (p.149) the view that self‐sufficiency and independence are ideals of human life, and accused those who see these and other elements of autonomy as ideals of neglecting the realities of human vulnerability and powerlessness, and the importance of intimacy, affiliation, and other forms of relationship.

However, this line of criticism confuses two distinct problems. Cosmopolitan thinking about justice can take a Kantian approach to the valuing of capabilities without falling into the trap of neglecting vulnerability and powerlessness. The basic feature of a genuinely Kantian approach to justice is a certain attitude to justification, which is not to be derived either from actual preferences or from determinate ideals of human flourishing. As Kant himself makes quite explicit, neither empiricist nor perfectionist strategies are part of or compatible with Kantian ethics. The basic Kantian strategy is rather minimal, and best put as a modal claim: ethical principles, including principles of justice, must be principles that all can act on. In this respect Kantian strategies of justification fit well with a capabilities approach. For to secure capabilities is to ensure that certain ways of acting can be chosen.

Kantian justifications pick out certain principles as basic to just institutions and lives. The embodiment of these principles in actual institutions and lives secures certain capabilities, but equally eliminates other capabilities. The capabilities which can justly be secured are capabilities that will reduce and tend to eliminate the possibility of violence, coercion, deception, oppression and the like, since these latter capabilities embody principles that could not be followed by all. For example, the injustice of a ‘head’ of family's capability to prevent his dependants from undertaking activity outside the home, is evident from the fact that this capability cannot be extended to all.

Any set of capabilities which eliminates injustice must both exclude capabilities that cannot be held by all and secure capabilities that buttress the vulnerable and empower them, so that others do not enjoy unjust capabilities. This will demand not merely that capabilities for injuring and oppressing be limited, but that capabilities for nourishing and supporting self and others and for resisting personal and institutional demands for subservience be strengthened. Limiting vulnerability is the other face of limiting capabilities whose ‘ideal’ development would damage or undercut others' capabilities.

One reason why the capabilities approach fits well with a Kantian Strategy of justification is that both positions take liberty seriously. Capabilities are for the sake of functionings, i.e. for the sake of acting and responding; but the particular uses of capabilities on a given occasion are up to those whose capabilities they are. In this approach to justice, human beings are not seen just as bearers of preferences to be satisfied by receiving certain resources, but as having capabilities which they can, but need not, use in a variety of functionings. A concentration on action and empowerment for (p.150) action insists that the poor are to be thought of and respected as agents, rather than as mere recipients of development, that liberties are important for the vulnerable. This emphasis fits well with a conception of practical reasoning that focuses directly on principles of action, rather than invoking an ideal of flourishing as the fulcrum by which principles of action and their just embodiments are to be identified.

However, the importance of a focus on action has, I believe, been much distorted in some recent so‐called Kantian ethics in two ways. The first oddity—of which I'll say nothing, although it has deep implications for development ethics—is that much recent Kantian ethics has stressed rights rather than obligations, recipience rather than agency, vulnerabilities rather than capabilities. One defect of this approach, but not the only one, is that ethical concerns that cannot be claimed by recipients tend to fade from view.

The second oddity of recent so‐called Kantian ethics is that, contrary to its critics' claims, it has made a poor job of giving an adequate account of human autonomy. It is true that proponents of liberal accounts of justice lay much stress on the value of autonomy, and assert that other positions fail to do so. But their reliance on preference‐based conceptions of action and rationality makes it extraordinarily hard for them to offer any adequate account of autonomy. The best‐known approach is Frankfort's attempt to explain what it is to be an (autonomous) person in terms of being guided by second‐order preferences, and many others have tried to make this the basis of an account of autonomy. In my opinion all these attempts fail: since preferences are endlessly cross‐referring, second‐orderedness is just too commonplace to pick out anything of much importance.10

Failure to provide an adequate account of autonomy has not, however, saved liberal theories of justice from the accusation that they have hinged everything on an excessive emphasis on autonomy. Critics of liberalism insist that what the liberal theories of justice support, under the heading of autonomy, is a fictional conception of the solitary and unaffiliated self, which is both empirically false and morally offensive. There is no time to enter into these debates here.11

(p.151) However, it is rather easy to show that much of the polemic against autonomy arises from widespread failure to distinguish two quite separate notions. The classical texts on autonomy in the modern period are Kant's, but Kant's view of autonomy is wholly different from the one targeted in all these criticisms. Kant reserved the term autonomy for the conception of principles as available (strictly: willable) for all. Autonomy in Kant's sense is applied to certain principles and not to others; it is to be contrasted with heteronomy, which applies to principles that are conditional on desires. Autonomy understood in this way is not a predicate of persons, and not to be contrasted with dependence or relationship or affiliation. Strictly speaking, autonomy in this traditional and technical Kantian sense cannot be fitted within a preference‐based theory of action at all.12

Contemporary concern with autonomy focuses on quite other considerations, and in particular on forms of independence or self‐sufficiency, and their absence. Whereas Kantian autonomy is a matter of our principles being willable for all, contemporary accounts of autonomy stress independence and thereby call into question relationships and dependence. As critics of contemporary Kantian work have pointed out, not every form of dependence and relationship is wrong or unjust, or a source of wrong or injustice. Some forms of dependence are essential and others commonplace or enjoyable; moreover total self‐sufficiency is mythical. We are all of us dependent in many ways.

These considerations cast a lot of doubt on positions that try to make independence itself an ideal of life. Autonomy as Kant conceived it may indeed be fundamental to ethics and ultimately to justice. However, that conception of autonomy, while critical of some ways of being dependent or making others dependent, is not invariably critical of human relations that secure and sustain others. On the contrary, a Kantian approach to autonomy is likely to be critical of forms of day‐to‐day autonomy that jeopardize capabilities for action, and in particular of those that jeopardize the meagre capabilities of the most vulnerable.

In our day‐to‐day discussions, the term autonomy is used as a relational predicate of persons and their characters, who are said to be more or less autonomous, autonomous of this or that but not of another agent or power. This everyday, gradated, relational sort of autonomy, is more perspicuously characterized as independence, which is clearly understood as gradated and relational. The structure of independence reveals why it is both valuable and threatening for the vulnerable. Independence is valuable if it consists of capabilities that limit how others can push one around; it is threatening (p.152) when it consists of capabilities that enable others to push one around. Independence is no unconditional ideal. It may be important to achieve specific configurations of independence, and to avoid others; but total independence is mythical, while over‐extended independence injures the most vulnerable.

The reasons why we are particularly interested in the acute forms of dependence and vulnerability described in Martha Chen's paper is that these are forms of dependence on others which can stifle, even threaten, lives. The reason why we have to set this interest in the context of an account of justice is that it is never simple to work out which capabilities should be strengthened to avert injustice. Many ways of strengthening one person's capabilities restrict another's. Although the distribution of capabilities is not a zero‐sum game, there are real conflicts. It is never possible to achieve more than a quite specific and limited configuration of capabilities—and of vulnerabilities. When we then ask which capabilities could in principle be secured for all, we discover that some capabilities cannot be achieved for all. These are capabilities that need and create victims; they are unjust capabilities. Other capabilities can be enjoyed by all, and when they are, will reduce vulnerability. To secure just rather than unjust configurations of capabilities, some sorts of dependence will no doubt have to be eliminated, but others may have to be secured. If justice is a matter of living by principles which could be followed by all it cannot be uniformly hostile to all forms of dependence and interdependence; it will be selectively hostile to unjust forms of dependence, unjust relationships—and unjust forms of independence.


(1) For further discussion of the failure of contemporary work on justice to engage with the predicaments of women's lives, and above all of women's lives in the Third World, see Onora O'Neill, ‘Justice, Gender, and International Boundaries’, in M. Nussbaum and A. Sen (eds.), The Quality Of Life (Oxford: Clarendon Press, 1993).

(2) For discussions of the lack of genuinely cosmopolitan accounts of justice and some attempted remedies see for example the Ethics Symposium: ‘Duties Beyond Borders’ vol. 98, 1988; B. Barry and R. E. Goodin (eds.), Free Movement of Money and Peoples (Harvester and Pennsylvania State University, 1992); Chris Brown (ed.), Political Restructuring in Europe: Ethical Perspectives (Routledge, 1994).

(3) Glover, above p. 121.

(4) See above all her ‘Aristotelian Social Democracy’ in G. Mara and H. Richardson (eds.), Liberalism and the Good (Routledge, 1991).

(5) While subjective preferences cannot be decisive for reasons stated above, we should listen carefully to those Eastern European women who say that they long to resume traditional social roles in the newly expected affluence: they are putting capabilities for the dependence on husbands that patriarchy requires ahead of capabilities for economic independence.

(6) ‘Aristotelian Social Democracy’, above n. 4, 215.

(7) This specific direction seems to me unpromising. Although Rawls's approach is often labelled ‘Kantian’, by him and by others, it in fact relies on a fairly standard view of action as guided by preferences; and while he does not regard preference satisfaction as defining the aim for justice, he does see justice as a set of constraints on the pursuit of preference. It seems to me unlikely this approach will either preserve the most interesting features of a Kantian approach or escape the criticisms of preference‐based approaches to justice sketched above.

(8) A view of human action as informed by certain descriptions, principles, laws or policies should, moreover, be congenial not only to traditional Kantians but to many others, provided they do not imagine that principles, rules, etc., demand uniform enactment hence lead to rigourism in ethics. These others include communitarians and contemporary defenders of virtue ethics, as well as to those whose approach to social inquiry is basically legal or sociological rather than economic.

(9) Kant combines the notions of consistency and possible community into his very conception of practical reason, which is intrinsically dialogical. (This may surprise both his detractors and his alleged defenders—yet the textual basis for the claim is strong.) For Kant, to act or to think reasonably is to rely on basic principles held to be available to all. For textual evidence see Onora O'Neill, ‘Vindicating Reason’ in Paul Guyer (ed.), A Companion to Kant (Cambridge: Cambridge University Press, 1992).

(10) Harry Frankfort, ‘Freedom of the Will and the Concept of a Person’, Journal of Philosophy, 1970. I believe Frankfort was misled by concentrating on the special case of addiction. Second‐orderedness—endorsement of one preference by another—may be important in this case, but is far too common to provide evidence of ethically important characteristics. For criticisms of this and kindred approaches to autonomy see Onora O'Neill ‘Autonomy, Coherence and Independence’ in W. W. Miller and D. Milligan (eds.), Liberalism, Citizenship and Autonomy (Avebury, 1992), 202–29.

(11) For well‐known examples of this criticism see Iris Murdoch's eloquent attack on ‘Kantian man’ in The Sovereignty of the Good (London: Routledge & Kegan Paul, 1970) and Michael Sandel, Liberalism and the Limits of Justice, who accuses Kantian work (in this case Rawls and contemporary liberalism rather than Kant) of relying on the fiction of the ‘deontological self’ which lacks affiliation, connection, and relationships. Variants of the same criticism abound in feminist writing, see e.g. Carol Gilligan, In A Different Voice: Psychological Theory and Women's Development (Cambridge Mass.: Harvard University Press, 1982), and in certain critical theorists, see S. Benhabib and D. Cornell (eds.), Feminism as Critique (Oxford: Polity Press, 1986).

(12) See Thomas E. Hill Jr., ‘The Kantian Conception of Autonomy’, Dignity and Practical Reason in Kant's Moral Theory (Ithaca, NY: Cornell University Press, 1992), 76–96.