THE PLACE OF AUTONOMY IN A CAPABILITY THEORY OF JUSTICE

Transcription

1 THE PLACE OF AUTONOMY IN A CAPABILITY THEORY OF JUSTICE Rutger Claassen, Leiden University Draft text, for presentation at Politicologenetmaal, Please do not quote. Introduction The role of freedom or autonomy in the capability approach is very much in dispute. This has to do, amongst other things, with the fact that the approach tries to combine and balance concerns for well-being (broadly conceived, as including human flourishing, human development, etc.) with concerns for freedom (broadly conceived, as including autonomy or agency). In this paper I will look at the place that has been assigned to the concept of autonomy by authors who have worked on the capability approach so far, and then argue that the capability approach should be tied to autonomy in a different and more fundamental way. In the traditional picture a concern for autonomy motivates the move from functioning to capability. Autonomy is exercised because people are allowed to choose whether or not to realize a specific functioning that they have a capability to. I will refer to this as autonomy s functionings-choosing role. I will argue this role for autonomy is important, but still insufficient. In addition we should select our list of basic capabilities on the basis of each capability s contribution to the realization of personal autonomy. Thus, I will argue for a capability theory which is not geared towards the realization of opportunities for a good life (or human flourishing, or well-being), but towards the realization of opportunities for an autonomous life. The conceptual place for autonomy which I propose is capabilityselecting: it determines which capabilities are to count as basic capabilities in the first place. A caveat is in order. A distinction needs to be made between the capability approach as a general approach to evaluation and more specific capability theories, which can be seen as applications of the general approach. In this paper I will be concerned only with capability theories of justice, not with other applications of the capability approach. It is a matter of some dispute whether this is a substantial restriction (given the often all-encompassing nature of theories of justice), but it nonetheless helps to define our focus here. The hall-mark of the application of the capability approach to justice is that it (i) uses capabilities as the relevant metric of justice, (ii) identifies some list of basic capabilities which a political entity (most often but not necessarily a state) must guarantee for its citizens as a matter of rights or 1

2 entitlements 1, and (iii) specifies which distributive rule it uses for distributing these basic capabilities (Anderson 2010, 82). Along these lines several more or less worked-out capability theories of justice have been proposed, such as those by Martha Nussbaum, Elizabeth Anderson and Ingrid Robeyns (Nussbaum 2000; Anderson 1999; Nussbaum 2006; Robeyns 2003). The capability theory whose embryonic core is sketched in this paper should be understood as a competing theory. Since Nussbaum s theory is the most worked-out example, I will refer to it at many stages of my argument in order to clarify the contrast I have in mind with these other theories. 2 The paper is set up as follows. First I distinguish two interpretations of the capability notion: one which sees capabilities as constitutive parts of the good life, and another which sees capabilities as all-purpose means. The former ties the capability approach to a notion of well-being or human flourishing, while the latter is open to other possibilities. One of these is to treat capabilities as all-purpose means to the development of autonomy (section 1). Second I investigate more closely Nussbaum s explicit arguments for refusing to tie her capability theory to autonomy. I show how they fail to provide decisive arguments against the use of a concept of autonomy (section 2). From this point, the paper is directed to the positive case for the move to autonomy. First I discuss which conception of autonomy fits the capability approach best. I use a concept of autonomy inspired by Joseph Raz. His three conditions of autonomy (personal abilities, independence and an adequate variety of options) should be at the heart of an autonomy-developing capability theory of justice. This can be done by making a crucial distinction between a list of primary (which guarantees personal abilities and independence) and secondary capabilities (for an adequate variety of options). Each of these has a different status in a theory of justice (section 3). Finally, I discuss how such a theory has to deal with the problem of selecting the relevant primary capabilities. Here I will explore the idea that the primary capabilities can be modeled as analogous to the items on Rawls s list of primary goods. This leads me to endorse a list which includes the Rawlsian basic liberties and access to income and wealth (but also economic security in a broader sense) and adds separate capabilities for health and education (section 4). 2. The Functionings-Choosing Role for Autonomy The functionings-choosing role for autonomy is most often described in terms of freedom, not autonomy. The capability approach s commitment to freedom is the reason to endorse 1 This leaves open the question whether philosophers or democratic publics should be the ones to make such lists. The only thing required is that a full theory would include such a list to define a just society. 2 It can be a matter of dispute whether these other theories resemble Nussbaum s theory or my theory more closely. 2

3 capabilities, not functionings. For example, Amartya Sen makes a distinction between wellbeing and agency and a further distinction in each of these between freedom and achievement. This generates four categories: wellbeing-freedom, well-being achievement, agency freedom and agency achievement. According to Sen the theoretical apparatus of the capability approach only deals with well-being, both well-being freedom (capability) and well-being achievement (functionings)(sen 1985, , 1992, 56, 1993, 35). In this construction the term freedom is reserved to mark the idea that we are free to choose which forms of wellbeing (which functionings) to achieve. I think that what Sen refers to as freedom here can well be described as an exercise in autonomy. 3. That it is autonomy which is at stake in the ability to choose functionings is perhaps brought out most clearly by Serena Olsaretti: My main claim is that capability approach can plausibly be seen to adopt a model of well-being which views the endorsement of valuable functionings as constitutive of well-being ( ). Freedom to achieve valuable functionings, as well as the freedom to forgo them, is then an appropriate standard of individual advantage (Olsaretti 2005, 98). In this view, the overall focus of the capability approach on well-being derived from achieving functionings is maintained. A crucial component of deriving well-being from your achieved functionings, however, is that you realize the functionings you voluntarily endorse. The idea of voluntary endorsement (which Olsaretti explicitly relates to the work of Raz, Dworkin and Scanlon) is nothing else than the exercise of an autonomous choice. When we endorse a functioning, we actively put our weight behind it it is not something that we passively accept as happening to us. While we need not agree with the details of the endorsement model that Olsaretti subsequently develops, her basic idea of voluntary endorsement describes very well what we also find in Nussbaum, Sen and others when they talk about the need for the state to realize capabilities, and leave the choice for specific functionings to individuals. 4 This role gives autonomy an important role in capability theories of justice and I will not remove or transform it in what follows. I will however argue that it is insufficient, for the 3 Sometimes Sen himself refers to positive freedom. E.g. (Sen 2002, 586). 4 It is a separate question whether such a restriction to capabilities is always tenable. For example, Nussbaum has repeatedly argued that in some situations the state will have to make sure that citizens actually function in certain ways, instead of being satisfied to guarantee capabilities for such functioning (Nussbaum 2000, 2006). I think Nussbaum is right that such exceptions to the capability approach s basic line, where direct coercion is asked for, will need to be granted. However, I also agree with her that when this is suitably dealt with, these exceptions do not undermine the basic commitment to restrict the theory to the realization of capabilities. I have treated this question extensively in [omitted for refereeing]. 3

4 functionings-choosing role leaves intact the capability approach s strong commitment to a theory of the objective human good in deciding which capabilities to function are to count as basic. Capabilities are constitutive parts of the good life. I will call this the objective good interpretation of functionings (and thus also of capabilities to these functionings). This interpretation pervades Nussbaum s capability theory, especially in her earlier work. In one typical remark, she states that the capabilities are not just instrumental to further pursuits: they are held to have value in themselves, in making the life that includes them fully human (Nussbaum 2000, 74). To bring out this idea, we can best look at an earlier article, in which Nussbaum argued for a broadly Aristotelian strategy to answer the challenge of relativism. This strategy involves three steps: What he [Aristotle, author] does, in each case, is to isolate a sphere of human experience that figures in more or less any human life, and in which more or less any human being will have to make some choices rather than others, and act in some way rather than some other. ( ) Aristotle then asks, what is it to choose and respond well within that sphere? And what is it to choose defectively? The thin account of each virtue is that it is whatever being stably disposed to act appropriately in that sphere consists in. There may be, and usually are, various competing specifications of what acting well, in each case, in fact comes to. Aristotle goes on to defend in each case some specification, producing, at the end, a full or thick definition of the virtue. (Nussbaum 1993, 245) The identification of different spheres of existence is the key to this strategy. The idea of good (virtuous) action is to be defined relative to each sphere of existence. Examples of Aristotelian spheres that Nussbaum mentions are fear of important damages, especially death (appropriate virtue for this sphere: courage), bodily appetites and their pleasures (virtue: moderation), distribution of limited resources (virtue: justice), etc. In the remainder of her article, Nussbaum defends this procedure against the relativist and makes it clear that she has a similar strategy in mind for her capability-oriented (rather than virtue-oriented) theory: it puts forwards conceptions of what it means to be well-functioning in the most important spheres of human existence. If we follow this argumentative strategy, the basic capabilities are clearly constitutive parts of the good life each capability with respect to another sphere of existence. The good life, for Nussbaum, is defined as a fully human life her criterion of selecting basic capabilities is to ask which capabilities we cannot do without, on pain of not recognizing our life as a fully human life (Nussbaum 1990, 1995). The length and definiteness of Nussbaum s list of basic capabilities can be seen as making explicit the consequences of such an appeal to a criterion of human flourishing. 4

5 The centrality of such a theory of the objective good creates obvious problems of perfectionism. The ambition to prescribe, in each sphere of experience, what the relevant capability should be and to make the ensuing list the cornerstone of citizen entitlements to government action, has been taken by many as a sign that the capability approach is objectionably perfectionist. 5 For example, Eric Nelson recently argued that the celibate, the misanthrope, the Christian Scientist and others could reasonably complain that Nussbaum s capability list includes items that are offensive to them (capabilities to sexual satisfaction, affiliation and use of certain medicines). The standard move that they do not have to act upon their capabilities doesn t help, Nelson claims, since these people are nonetheless required to fund a state that guarantees the actual ability to exercise the functionings to all: even if I myself consider sex to be sinful, I am required to fund somebody else s Viagra (Nelson 2008, 100). This criticism seems to me convincing. The objective good interpretation, in prescribing a specific account of the good life, does overreach the boundaries of what a political community should be expected to do for its citizens, even if it only provides for capabilities. The fact that it provides certain capabilities and not others makes for substantive commitments that some citizens would rightly reject. It is important, though, to be clear about where actually the problem lies. The problem starts when a political community shows a commitment to certain moral ideals even when it only promotes certain capabilities to realize specific functionings. Such a commitment is evident from the fact that it chooses to promote these capabilities and not others. Any list is in that sense non-neutral. It is telling that Nussbaum never seriously discussed what it would mean for a state to actually give people capabilities to many of the things on her list. Why should the state make sure people are able to laugh (part of her capability to play), able to live with concern for and in relation to animals, plants, and the world of nature (her capability for other species ), or to love those who love and care for us (part of her capability for emotions )? Does Nussbaum think that modern states already are preoccupied with these kinds of things? (so that her list codifies actual practices). Many of these things may be important for a lot of citizens (though not all: some serious types have little interest in laughing, some metropolitan types have no interest in nature, etc.). But even if the moral value of these capabilities is widely recognized, that does not mean that there is a political obligation to realize them. What is lacking is a separate argument for the political importance of these capabilities. In many cases, such an argument would be hard to make. Practically, the state may not be best placed to realize them. Principally, the problem is in the 5 Although not always expressed in these terms, see (Arneson 2000) (Nelson 2008) (Jaggar 2006) (Okin 2003) 5

6 thick account of well-being that many items on Nussbaum s list express. They remain controversial for at least part of the population. If the objective good interpretation would be the only interpretation available, this criticism would be fatal to the capability approach as a whole. This is what Richard Arneson has claimed. He argues that for a full theory of justice it needs to be possible to rank the relevant capabilities objectively, according to the well-being value of each functioning (Arneson 2010, 108). As far as I know, it is agreed by almost all commentators that the capability approach presupposes a list of objectively valuable functionings. Sen developed the capability approach, after all, as an alternative to subjective measures (such as utility or preference). Now the trick is that Arneson equates such an objective ranking with a ranking of objective good or objective well-being (or again, an objective standard for assessing quality of life, 2010, 115). Other options do not occur to him. It seems to be impossible for him as for others to detach the idea of well-being from the capability approach. Functionings represent individuals well-being, that s their whole point. But since it is the theory of objective well-being that makes the capability approach so controversial as a normative political theory, it is important to see if other interpretations are possible. We can save the capability approach if we give the concepts of capabilities and functionings a different interpretation; not as objective goods but as primary goods. Such a primary good interpretation of capabilities has indeed been hinted at in Nussbaum s more recent writings, where she has claimed that the basic capabilities are very close to primary goods, and that it is rational to want them whatever else one wants (Nussbaum 2000, 88). It seems that Nussbaum thinks these interpretations can simply be juxtaposed. I think, however, that this cannot be right. 6 For Rawls primary goods are means to ensure that each person can realize the conception of the good of her own choosing. 7 Nussbaum, however, cannot give the same answer, at least not as long as she also wants to maintain the objective good interpretation of the capabilities notion. For in the latter interpretation the substantive question of the good life is not relegated to individuals choices, but it consists of a well-specified list of basic capabilities (where the only choice left is whether or not to convert these capabilities into functionings). Thus, the objective good interpretation and the primary good interpretation of capabilities are incompatible. Since Nussbaum doesn t acknowledge this incompatibility, she hasn t faced the question what a consistent primary good interpretation of the capabilities would mean for the other parts of her or any other capability theory. In the final sections I 6 I here use Rawls original formulation and ignore Rawls s later reformulation of primary goods as things citizens need as free and equal persons, to fit his turn to political liberalism. See (Rawls 2005 [1993], 180) For the problem with the latter formulation see (Nelson 2008, 114) 7 This is his thin theory of the good. See (Rawls 1999 [1971]) 6

7 will defend a specific version of the primary good interpretation, according to which the capabilities are all-purpose means to the development of personal autonomy. However, first I will discuss two distinct reasons Nussbaum has given for not tying her theory more closely to the concept of autonomy. I will argue that we need not accept these reasons, but I will also show that Nussbaum does raise two concerns that an autonomy-based capability theory needs to address. Thus, this discussion provides us with an agenda of challenges for an autonomybased capability theory. 2. Nussbaum s Criticisms of Autonomy In earlier work Nussbaum argued that the basic capabilities on her list are generated through an interpretation of the concept of human nature ; by asking what it is for a life to be fully human. 8 She presented this notion of humanness explicitly as an alternative to a liberal notion of autonomy. Here is how she counters the objection that her Aristotelian conception is neglectful of autonomy: the Aristotelian insists that choice is not pure spontaneity, flourishing independently of material and social conditions. If one cares about autonomy, then one must care about the rest of the form of life that supports it and the material conditions that enable one to live that form of life. Thus the Aristotelian claims that her own comprehensive concern with flourishing across all areas of life is a better way of promoting choice than is the liberal s narrower concern with spontaneity alone, which sometimes tolerates situations in which individuals are in other ways cut off from the fully human use of their faculties. (Nussbaum 1992, ) In this passage, it seems to me, Nussbaum presents a false opposition between her Aristotelianism and liberalism. It is not the case that one must either resort to a liberal theory which is neglectful of the material and social conditions of autonomy or endorse an Aristotelian theory which promotes such conditions through a vision of flourishing in all areas 8 I leave out of consideration her more recent assimilation of her theory to Rawlsian political liberalism, by which the list of capabilities is thought to represent part of an overlapping consensus. The strategy of political liberalism, geared as it is towards the elucidation of the political values inherent in actually existing democratic societies, seems to me to be obviously at odds with he capability approach s universalist aspirations. Moreover, Nussbaum nowhere makes clear how her conversion to political liberalism affects her capability theory; it is apparently to be simply attached to the robust Aristotelian framework without necessitating any changes in the framework. This seems to me unconvincing. Similarly, for a rejection of the view that Nussbaum succeeds in being a political liberal (which I share) and an argument that Nussbaum already is a comprehensive liberal (which I would deny), see (Barclay 2003) 7

8 of life. By contrast, a liberal can very well be concerned with the material and social conditions of autonomy without resorting to this kind of Aristotelianism. She can do so by listing conditions which enhance the capacity for choice, to be used in all areas of life, without defining flourishing in all these areas. The crucial distinction which allows the liberal this third way between the formal type of liberalism that Nussbaum sets up only to reject and her own thick Aristotelianism, is the distinction between the development of autonomy and its exercise. For Nussbaum autonomy is apparently only realized if one has the opportunity for flourishing in all areas of life: in the passage above this is treated as a necessary means of acquiring the material and social conditions mentioned. However, this overstates what is necessary for building up autonomy, potentially assigning to all these spheres an autonomy-developing function. In many actions in many spheres of a life we are not so much developing our capacity for autonomy, but rather exercising it in this or that way. If I am right about the importance of this distinction, then the liberal has a strong argument when he complains to the Aristotelian that he wants to prescribe how persons should flourish in all areas of life instead of leaving this to their own choices. Instead, the capability theory should be geared towards the realization of those capabilities which necessarily contribute to the development of personal autonomy. When taking this route, the liberal has a perfectionist goal: that of giving people the opportunity to develop their autonomy. In other words, he does recognize a conception of the good life, i.e. the good life is a life led autonomously. Compared to other perfectionist conceptions, this conception of the good life is the most liberal one available, since it leaves more room for people to choose their ends in life than any other conception. This does raise the question how an autonomy-based capability theory would distinguish between capabilities which develop autonomy and capabilities in which autonomy is merely exercised; and whether the latter have a role to play in such a capability theory at all. This is a first item on our agenda, which I will take up below (see section 3). In a more recent exchange with Linda Barclay, Nussbaum has given a second reason for rejecting a fundamental role for a concept of autonomy. Here, faithful to her political liberal turn, Nussbaum is anxious to refute Barclay s contention that she is a comprehensive liberal. 9 She points to the specific Western tradition of the idea of autonomy, and argues that it includes the idea that one denies that God is a necessary part of the justification of moral 9 Barclay had based this claim mainly on the fact that Nussbaum argues in favour of capabilities, not functionings. As I have argued earlier in this paper, I do not believe that this functionings-choosing role for autonomy is sufficiently strong. Whether or not it is sufficient to qualify for the label comprehensive liberal is a matter that depends on one s definition of comprehensive liberalism. However, I do find it strange that Barclay thinks Nussbaum s theory with its elaborate moral anthropology is just aimed at realizing the value of autonomous choice. 8

9 claims (Nussbaum 2003, 41). Since this notion of autonomy is unacceptable to some (reasonable) religious comprehensive doctrines, Nussbaum rejects it. Instead and partly to compensate for this, she emphasizes that her list does include the capability for practical reason: everybody should have the ability to form a conception of the good and to engage in critical reflection about the planning of one s life (Nussbaum 2003, 39). In reply, we should start by recognizing that the historical notion of autonomy that Nussbaum has in mind, however important, is highly specific. The contemporary literature on autonomy contains an overwhelming amount of different notions of autonomy. None of them a priori has a less legitimate claim to be the notion of autonomy than the one that Nussbaum associates autonomy with. This throws doubts on a strategy which takes a highly specific interpretation of autonomy and then concludes from one of its alleged unsatisfactory features (i.e. a certain dismissiveness towards religion) that autonomy as such is to be rejected. Maybe, however, her rejection of this specific notion of autonomy is not of so much importance if we can put the emphasis on the other half of her strategy, that autonomy is already included in the form of the capability for practical reason. This may be especially so since she has argued that this somehow is a privileged capability (together with the capability for sociability) in that it suffuses all the other capabilities (Nussbaum 2000, 82). Doesn t that guarantee a sufficiently strong place for autonomy in the capability theory? The problem with this strategy is that the inclusion of many of the other capabilities besides practical reason remains contentious (for the reasons given in section 1). These other capabilities contribute to well-being, but it is not clear why states should, as a matter of justice, make sure that people have opportunities for choosing these welfare improvements. Of course we might imagine a capability theory which would have removed all these other capabilities from the list, so that we now have a one-item list, with only the capability for practical reason on it. Don t we have what we want then? In a sense, we do. Such a capability theory would restrict itself to the development of autonomy. However, such a one-item list would be uninformative. It is not enough to state at a higher level of abstraction that the capability for practical reason is to be promoted; what we want to know is how this abstract capability for practical reason is to be promoted; i.e. what more specific capabilities are needed to realize it (and some capabilities on Nussbaum s list may be necessary to that end). In other words, we are looking for a list of autonomy-developing capabilities, with autonomy (or practical reason) itself as the relevant background criterion (replacing Nussbaum s elusive criterion of human flourishing). What we need to start with, then, is a conception of autonomy itself, which is going to inform our list. This is the other item on our agenda, which I will now start with. 3. A Capability Conception of Autonomy 9

10 In a broad sense the concept of personal autonomy refers to the idea that people should make their own lives ( ) The ideal of personal autonomy is the vision of people controlling, to some degree, their own destiny, fashioning it through successive decisions throughout their lives (Raz 1986, 369). This idea can be developed in several directions and there are many details that would have to be defended for a full theory of autonomy. In this section I can only offer some main clarifications to show what kind of conception of autonomy would be suitable as the guiding criterion in a capability theory of justice. This will inevitably leave room for further differentiations and disagreements, but the main aim is to present a conception of autonomy which is sufficiently determinate to show the attractions of an autonomy-based capability theory. A good starting-point is Joel Feinberg s distinction between four different meanings of the word autonomy: as the capacity to govern oneself, as the actual condition of selfgovernment, as an ideal of character, and as the sovereign right to govern oneself (Feinberg 1986, 28). An autonomy-based capability theory would combine Feinberg s first and fourth senses of autonomy. Its final goal, as a normative political theory, is to ensure people s right to self-government. It ensures this right by giving them the capacity to govern themselves. If people so disposed act upon their capacity they will find themselves in an actual condition of self-government (Feinberg s second sense of autonomy). Whether or not this is what they do, however, is outside of the capability theory s concern, strictly speaking. The capability theory only guarantees the right to the capacity for self-government. 10 The next question is what that capacity entails. To answer that question, it is useful to distinguish two components to autonomy (here and hereafter I use the word in Feinberg s first sense: as the capacity for self-government). An often-used distinction is that autonomy involves two crucial abilities: the capacity to form and revise one s plan of life and the capacity to realize this plan in one s actions. The first part has to do with the agents preferences, beliefs, and reasons for them and their internal coherence in a plan of life. It asks for the agents control in the process of forming and revising this plan. The second part asks for control at the moment when in our actual course of life we try to realize this plan. Both parts are crucial and belong together. There is a lack of autonomy both when we are the slave to our own conception of the good and when we are unable to realize this conception. The next question is which conditions for autonomy have to be present for an individual to possess these two components of autonomy. 10 Such a restricted ambition also means that the capability theory does not treat autonomy as a moral ideal (Feinberg s third sense). It only treats the opportunity to live autonomously as a moral ideal. 10

11 To that end the three conditions of autonomy presented by Joseph Raz are useful (Raz 1986, ). First Raz states there have to be appropriate mental abilities. I would prefer to call these personal abilities, a term encompassing the necessary cognitive but also emotional and volitional abilities for forming, revising and executing a life plan, with all the perseverance and strategic rationality that may be necessary. 11 Second, there is what Raz calls independence, which is violated both by manipulation and by coercion. Manipulation refers to an interference with the formation of one s plan of life (or one of the beliefs, preferences etc. underlying that plan), while coercion interferes with the execution of that plan. Third, we need an adequate range of options. When we form a life plan we need to be aware of different possible life choices (e.g. several different career options), and when we execute the plan these options need to be actually open to us. Both their availability in reality and their availability in our imagination are to be included in this demand. As this brief overview hopefully makes clear, all three conditions of autonomy each help to realize both components of autonomy. With these general clarifications about autonomy in mind, we can now turn to the connection with capability theory. At the heart of a capability theory of justice is a list of basic capabilities which each citizen is entitled to. In the present theory these basic capabilities are those that develop the capacity for autonomy. This would mean that as a first step we need to make sure that together these basic capabilities cover the content of Raz s three conditions of autonomy. The conditions of autonomy will have to be translated into capabilities with the same content. I propose to treat the first two conditions are the proper subject of the list of basic capabilities. These basic capabilities ensure that people have the requisite personal abilities and independence. How to devise such a list probably is the biggest challenge to the development of an autonomy-based capability theory. This matter I will treat separately in the next section. For the moment presuppose that we succeed in establishing a satisfactory list of autonomy-developing capabilities. We then face the question of how to deal with the third condition, of an adequate range of options. Of course there is the pernicious question of how much variety of options is adequate (a problem for capability and non-capability theories alike). More pertinent to our problem of how to structure an autonomy-based capability theory of justice is whether the guarantee of such a range of options should be a political obligation in the same way that this is true for Raz s first two conditions. I do not think this is the case. This asymmetry is reason to treat the third condition separately. 11 I will leave it open what these precise abilities are and how they relate to the self. It is here that there is room for dispute, following the discussions in the literature on personal autonomy. According to Christman, the standard conception of autonomy involves both authentic conditions and competency conditions (Christman and Anderson 2005, 3). 11

12 Inclusion on the basic list would make this condition, like the other two, an entitlement for citizens to be guaranteed by the relevant political authority. Normally however these options are not created by political authorities, but by citizens themselves in the exercise of their autonomy. Just to mention one example, normally governments are not involved in the creation of any particular occupation, let alone a wide range of sufficiently differing occupations available on the labour market. What occupations are created and what vanish, depends on the spontaneous processes of economic life, where market demand, technological developments and other factors together determine the outlook of the division of labor at any point in time. If all works well, this spontaneous process will by itself create a sufficient variety of occupations to choose from. If so, it seems eccentric to ask the government to keep open any particular option, just because one citizen insists on this particular option. If an occupation vanishes, those who would like to hold it will simply have to deal with the consequences (the argument similarly applies to other areas of life: options for practicing different religions, different hobbies, different forms of family life, etc.). As a consequence, citizens in a just society will normally have two sets of capabilities. The basic or as I will call them henceforth primary capabilities are given by a fixed and finite list of capabilities corresponding to the autonomy conditions of personal ability and independence. These are the subject of entitlements to be safeguarded by political authority. Separate from that, there is an endless and ever-changing list of secondary capabilities to which citizens will have access because other citizens have created them through the free exercise of their capacity for autonomy. In modern societies this list will feature the capability to become an accountant and the capability to become a computer programmer, while in earlier times these items would not have been on the list, while others such as the capability to become a weaver or the capability to become an alchemist would (this is not to say that access to these occupations is, or ever was, unconditional). With such a split between primary and secondary capabilities, we can do justice to the distinction between autonomy development and autonomy exercise and give theoretical expression to the idea that the political responsibilities are different for each of them. This however doesn t mean that there are no political obligations with respect to the secondary capabilities at all. I have only stated that there is no direct political responsibility to guarantee entitlements to them, as for the primary capabilities. But there are two more indirect responsibilities which I would propose to acknowledge. 12 First, while the creation of an adequate range of secondary capabilities can be left to the spontaneous processes of civil 12 This indirect line of argument is a less perfectionist alternative to Raz insistence that government should directly promote valuable options. This feature of his theory has been criticized as overly perfectionist or even conservative. See (Brink 2000, 70-71), (Mulhall and Swift 1996, ), (Neal 1994, 47),(Waldron 1989, 1131). 12

13 society, there may be a political task in making sure that each citizen is aware of these options. Some citizens may be situated in a social environments which restricts their horizon of experience, so that they are unaware of many important options open to them. In these cases political authorities may have to step in to increase their awareness of these options. This may alternatively be conceived as falling under the creation of independence (removing social barriers which manipulate their consciousness) so that it is already in the remit of the primary capabilities. Even if thus conceptualized it remains however important to note that the creation of independence explicitly has as of one of its subjects the awareness of the available secondary capabilities. A second indirect political responsibility arises when spontaneous processes for whatever reason fail to deliver an adequate range of secondary capabilities. One can imagine that in certain extreme cases civil society fails, whether due to natural disasters, dominating social groups or any other cause. If the range of available options becomes particularly sparse and monotonous, one can imagine that political authorities would have to play a stimulating role. A satisfactory account of this responsibility (which I cannot give here) here would show both the opportunities and limitations of this kind of political action. This is, in very rough sketch, what a capability-selecting role for autonomy could look like. In conclusion I would like to draw attention to the fact that this makes the functionings-choosing role for autonomy to appear in a different light as well. According to that role, individuals exercise autonomous choices to what extent to convert their primary and their secondary capabilities into achieved functioning. The conversion of the primary capabilities is especially interesting. Since these are now interpreted as autonomy-developing capabilities, an individual who makes full use of his primary capabilities comes to in life in a realized state of full autonomy (i.e. autonomy in Feinberg s second sense). An individual who makes only partial use of his primary capabilities will be in such a state to a lesser extent. He will have the capacity for autonomous living to the same extent (since that it guaranteed across the board), but will not seize upon that capacity at every turn. For example, he may decide to leave decisions in a certain area of his life to authorities to which he submits himself. The capacity for autonomy is a global capacity. Its possession is perfectly compatible with more local decisions not to exercise a specific personal ability for autonomy (Raz s first condition) or not to insist upon his independence (Raz s second condition) in a specific area of life. In this way, the capability theory s core distinction between capability and functioning neatly maps onto the distinction between a formal and a more substantive notion of autonomy (Dworkin 1988, 12). Every citizen is formally autonomous, and can decide for himself whether or not to lead a more substantively autonomous life. This allows the autonomy-based capability theory to immunize itself against the prominent criticism of 13

14 theories which adhere to a too substantive notion of autonomy as a life of self-realization. 13 This capability theory allows but does not require such a life. On the other hand, it becomes equally invulnerable to those who think that a formal notion is too meager for a theory of justice. Such a formal notion allows individuals a more substantively autonomous life. The capability theory proposed here has the theoretical resources to accommodate both the formal and the substantive notion of autonomy into its edifice, with the autonomous choice of individuals themselves as the bridge which holds both notions together. 4. The Selection of Primary Capabilities I now have to turn to the task that I have postponed so far, to discuss how to select autonomydeveloping capabilities. The suggestion that I want to explore in this section is that we should take the idea of capabilities as analogous to primary goods more literally than we have done so far. Up to this point the analogy consisted of the fact that capabilities, like primary goods, should be seen as all-purpose means (see section 1). In the literature on the relation between (basic) capabilities and primary goods as respective metrics of justice, the emphasis has been laid on the difference between the two. The main difference is that capabilities, unlike primary goods (or resources more generally) do justice to individual variations in the ability to convert these resources into a successful life. This difference however does not prevent us from constructing a theory in which the list of primary capabilities is analogous to the list of primary goods in a stricter sense: the same items that feature on the list of primary goods then feature on a list of capabilities. The reason why this suggestion is worth exploring is that Rawls s list of primary goods was originally meant to establish the same purpose as the capability theory proposed here: to enable people to live a life of their own choosing. Rawls s thin theory of the good runs parallel to the conception of autonomy specified in the previous section (even if not all the details match). So if Rawlsian primary goods serve to make people autonomous (in his terms), why not establish a list of primary capabilities which make people autonomous (in the sense defended here)? Rawls s list of primary goods contains five items (abbreviated presentation): (i) the basic rights and liberties 13 The autonomy notion to which Nussbaum objected (see section 2), is an example of this. But a similar criticism of a substantive autonomy notion inspired Rawls and Larmore s turn to political liberalism (Rawls 2005 [1993]) (Larmore 1987). The capability approach doesn t have to make such a turn. 14

15 (ii) freedom of movement and free choice of occupation (iii) powers and prerogatives of offices and positions of authority and responsibility (iv) income and wealth (v) the social bases of self-respect. 14 Could these primary goods be converted into primary capabilities? If so, the resulting capability theory would retain the advantage of catering to individual variability (compared to their primary good formulation), while it would avoid the criticism of perfectionism of a more extensive list (like Nussbaum s). We can discuss this question by distinguishing three categories of potential primary capabilities. First there are certain primary goods for which it may be easy to translate them in the language of capabilities, namely, those primary goods which are themselves described in terms of liberties (categories i, ii and iii above). Nussbaum had already made this step by including the political liberties on her list of basic capabilities (capability 10: being able to participate effectively in political choices that govern one s life; having the right of political participation, protections of free speech and association ). In a similar way we can take the civil and political liberties as items on an autonomy-based list of primary capabilities. After all, capabilities are freedoms, i.e. opportunities to enjoy freedom of movement, of association, of speech, etc. We might say the basic liberties are already capability-like items on Rawls s list of primary goods. They are included on our capability list mainly to satisfy Raz second condition of autonomy: independence. The basic liberties serve to make possible a sphere of free action for each individual, where he/she can follow his own life plan without interference by others. 15 In a second category we find items on Rawls s list of primary goods which are not so readily translatable into analogous capabilities. These are the primary goods of income and wealth and the social bases of self-respect (categories iv and v). I will leave the latter out of consideration here. 16 Income and wealth however are of special importance, since they are the most resourcist of all primary goods. Those working within the capability approach have denied that income and wealth are of direct importance with respect to measurements of 14 (Rawls 2001, 58-59) 15 [add discussion of what it means to have their fair value (Rawls) and whether individual variation in having these capabilities plays a role in the capability theory]. 16 The status of this primary good has always remained controversial in Rawls, probably because unlike the other primary goods it doesn t play a role in his two principles of justice. Rather, it underlies these others, especially the basis liberties (Rawls: the basis for self-respect in a just society is not then one s income share but the publicly affirmed distribution of fundamental rights and liberties (Rawls 1999 [1971], 477)). I propose therefore not to treat this as a separate primary capability. Just as for Rawls self-respect emerges in an otherwise just society which delivers the requisite primary goods, so it will emerge in a society which guarantees people s autonomy-based primary capabilities. 15

16 justice. The argument has always been that one cannot measure people s freedom (in terms of their capability sets) by measuring their incomes. If justice is to be measured in terms of people s capabilities, then, there seems to be no role for an analogous primary capability. Should we simply omit this item? It seems to me that that would underestimate the role income and wealth play in the development of people s autonomy. The condition of independence does not just require that one s basic liberties are secured. Such a situation still leaves all kinds of dependences in the private sphere. If there is to be independence in the full sense of the term, then, a certain level of income and wealth may very well be seen as a prerequisite. This guarantees the kind of social and economic standing that is necessary to speak up against these private others and risk conflicts with them (family members, powerful individuals within one s community or at work, etc.). If independence has such preconditions in the socio-economic sphere, the relevant primary capability should probably be broadened, not just to include income and wealth, but economic security in a wider sense. This term would then encompass a subset of specific capabilities which guarantee security. Just to mention one example, a sufficient level of income and wealth make it possible for one to quit one s job if one has a conflict with one s employer: it gives one an exit-option from any specific position and guarantees independence in this sense. But to use this exit-option is an often painful option of the last resort. Independence at one s work arguably should include more protections than the protection offered by one s bank account. For example, if one has been the victim of sexual harassment by a colleague, there should also by a legal remedy to fight against this kind of conduct, not just the option of leaving. Thus, I conclude that a primary capability for economic security should be on the list, comprising the ability to earn or receive a sufficient level of income and wealth but going beyond this to include other safeguards for independence (I leave open the details of these here). A third category is those items which do not figure on Rawls s list primary goods, but which should arguably be included on the list of primary capabilities. Here we take inspiration from those which criticized Rawls for not acknowledging certain goods as primary, and see if they are candidates for primary capabilities as well. I will restrict myself to two of them (there might be more). Rawls has been reproached not to have added health care and education on his list. Whether or not that is correct, these must surely figure on our list of primary capabilities (Daniels 2010; Brighouse and Unterhalter 2010). Opportunities for education and health care are two of the main ingredients which safeguard the personal abilities necessary for autonomy (Raz s first condition). Good education helps develop the cognitive, emotional and volitional qualities which together make it possible for adults to form a conception of the good and live one s life according to it. Good health care helps to ward off threats against these abilities; in as far as this is within human competence at all. As 16

17 primary capabilities, the political obligations towards education and health care can be finetuned to take account of individual variations in people s possibilities to take advantage of these resources. 17 Finally, after we have established a list of primary capabilities there is the question of the relevant distribution rule. Nussbaum has argued that the capability approach is a sufficientarian approach, which seeks to guarantee each of the basic capabilities up to a certain threshold level (Nussbaum 2006). This fits well with the idea that all of the conditions of autonomy admit of degrees. As is a commonplace in the autonomy literature, one can have the personal abilities, independence and adequate range of options to a smaller or larger extent. Accordingly, one can be more or less autonomous. This idea naturally leads to a sufficientarian distribution rule for the primary capabilities. We will need to set a minimal threshold level of autonomy, which the state is to provide for its citizens and then translate this into thresholds for each specific primary capability. Where exactly these thresholds should lie, is beyond the reach of this paper. We can only determine after we have established a more definitive basic list, the exact importance of each basic capability to the overall goal of autonomy, and above all, what level of autonomy itself we judge to be minimally acceptable in our society. Conclusion This paper has argued for taking the concept of autonomy as central to the development of a capability theory of justice. In such a theory a conception of autonomy serves to select a list of primary capabilities as well as certain indirect political obligations towards secondary capabilities. In addition it leaves room for people s autonomous choice to convert these capabilities into actual functionings, thereby deepening their politically guaranteed capacity for autonomy into a substantively autonomous life, to the extent they judge fitting. Thus the notion of well-being loses its separate theoretical standing in the capability theory. Arguably, an autonomous life brings well-being to the person living it because of it being autonomously chosen. This however is not the primary focus of the theory. Just societies are not directly responsible for the well-being of their citizens, but merely for their freedom to live a good life. 17 Apart from health-care and education, arguably the personal abilities for autonomy are also developed in the most intimate private sphere, the family (and in a circle of friends, neighbours, etc.). Direct interference in this sphere is a more delicate matter from a political point of view. So we would have to recognize some indirect political obligation to step in once this sphere fails heavily to provide what it should in this respect. This is probably not best conceptualized as a primary capability itself. 17