2.1 Identity as an object of reasoned choice

Sen's Reference Sen1998 Romanes lecture Reason Before Identity covers a great deal of ground, which is further developed in his books The Argumentative Indian and Identity and Violence and related works (see Sen Reference Sen1998, Reference Sen2000, Reference Sen2005a, Reference Sen2005b, Reference Sen2006a, Reference Sen2009a, Reference Sen2009b, Reference Sen2014 inter alia). A central theme of this lecture, and indeed of Sen's subsequent writings, is that we can make choices relating to identity. On Sen's view, identity is not merely ‘discovered’. It is not – in his words – a ‘destiny’ that each of us can simply claim. For Sen, the notion of identity as a ‘destiny’ is an ‘illusion’, as is the idea that there is only one identity for each person. These twin illusions – the illusions of ‘destiny’ and of the ‘uniqueness’ of identity – are, on his view, problematic at a number of levels. The idea that we each only have one identity can be used as an instrument of violence (Sen Reference Sen2006a: 175) in a world where groups might have conflicting interests, or where radical ideologies might flourish through a simplified or all-encompassing notion of identity. The crude identification with a single nation or ethnic or religious group might crowd out or minimize the weight given to other affiliations based on profession (e.g. those involved in being a lawyer or economist), intellectual, political or moral views (e.g. being a logical positivist, a member of the Conservative party or a vegetarian), each of which can be important to how we see ourselves.

Sen explicitly opposes two positions that he associates with communitarianism in political philosophy, represented by Michael Sandel and Alisdair MacIntyre inter alia. The first is the idea that identity is ‘discovered’. Here he cites Michael Sandel's Liberalism and the Limits of Justice. Sandel writes of citizens that ‘community describes not just what they have as fellow citizens but also who they are, not a relationship they choose (as a voluntary association) but an attachment they discover, not merely an attribute but a constituent of their identity’ (Sandel Reference Sandel1998: 150 cited in Sen Reference Sen2006a: 36). If Sandel takes each person to belong to a unique community, then he would appear to have fallen prey to both illusions Sen cites – of ‘uniqueness’ and of ‘destiny’.Footnote ¹ However, Sen's own criticism here is restricted to the claim that Sandel invokes the notion of discovery and so lends credence to the ‘illusion of destiny’.

A second claim Sen (Reference Sen2006a: 33–4) associates with communitarians is that notions of rationality are community-relative, and indeed that norms cannot be criticised from a position which is not itself within the community. He objects to this line of thought on at least two grounds: that it leaves little space for the criticism of norms across different communities or cultures (an argument he has made elsewhere at greater length);Footnote ² and that there is a considerable variety of views within each culture, so that, as he puts it, ‘cultures need not involve any uniquely defined set of attitudes and beliefs that shape our reasoning’ (Sen Reference Sen2006a: 35). However, he does not dispute the claim (also associated with communitarians in political philosophy) that we do not ‘reason from nowhere’, or that our reasoning is influenced by our social affiliations.

In the light of this discussion of communitarianism, identity emerges as an object of choice in Sen's work, with choices to be made subject to the various constraints people face in the standard way economic theory sees them as made from a set of feasible options or commodity bundles (Sen Reference Sen2006a: 5–6). The mere fact that there are constraints on choices, Sen suggests, does not undermine the fact that choices are made, or have to be made. To clarify Sen's claims, there are three conceptions of ‘identity’ which might helpfully be distinguished. These are: the way a person sees herself, her self-perception; the way a person is perceived by others, her perceived identity; and finally the specific social affiliations or identifications a person has – as a member of a family, cult, political party, or other group – which are usually termed ‘social identities’ (Appiah Reference Appiah2005: 65). Sen's view that identity is a matter of choice is partly a consequence of the recognition of the multiplicity of a person's social affiliations. Salma might, at the same time, be a Punjabi, an economist, a woman, a left-leaning liberal and a Buddhist. Her affiliations are in part a matter of choice. Yet the role of choice in how she sees herself – her self-perception – goes further than this and involves a range of decisions about the relative importance to be given to these distinct social affiliations according to context. At an economics conference, her social affiliation as an economist might be given priority or may be salient in a way that being a Buddhist or Punjabi is not. In Sen's writings, the issue of how these social affiliations are defined is not highlighted as a matter of personal choice as such, though as we shall see, it is not neglected altogether.

To explain the sense in which Sen understands identity choice as reasoned, I must briefly introduce one aspect of his account of rationality which differs from standard accounts in economics. On Sen's view, people's objectives and values are themselves subject to scrutiny in a way that they are not in conventional rational choice theory (Sen Reference Sen2002a: 40). As he has recently expressed it: ‘the essential demands of rational choice relate to subjecting one's choices – of actions as well as objectives, values and priorities – to reasoned scrutiny’ (Sen Reference Sen2009b: 180).Footnote ³ Since this view is non-standard, I avoid use of the term ‘rational choice’ in referring to choice based on what Sen calls ‘reasoned scrutiny’, but instead refer to ‘reasoned choice’.

The underlying model of identity choice seems, nonetheless, in important respects to follow the traditional economic model of choice from a menu or feasible set. There are (at least) two stages in the choice: (1) a person can be seen as having a set of alternative social affiliations from which she can choose a subset; and (2) she can choose the priority she gives to each of the selected social affiliations in particular contexts. As a consequence of (2), her self-perception, and indeed her perceived identity, can shift according to the context she finds herself in because of the salience of, or the weight she gives to, different social affiliations in different contexts.

2.2 The minimal constraints on identity choice

While Sen emphasizes the importance of choice, he also accepts that there are aspects of one's identity one does not choose. There are a number of ways in which choice can be constrained which are relatively uncontroversial: an elderly Indian Muslim woman cannot plausibly take on the social identity of a white boy scout. Or to take his own example, Sen cannot ‘readily choose the identity of a blue-eyed girl from Lapland’ (Sen Reference Sen2006a: 30). Choice is constrained because, even if there is a multiplicity of social affiliations one can choose from, some of these are mutually exclusive and the way we see ourselves – our self-perception – is generally constrained because of this, as is, on his account, our perceived identity – the way others see us. In other words, a person's self-perception and perceived identity are usually constrained by logical possibility (e.g. one cannot usually, at one and the same time, be both a girl and a man) and by certain biological and social facts (e.g. about eye colour or country of origin), many of which are quite contingent. Of course, this is not invariably so: ill-informed others may make mistakes, which might lead to cases of ‘mistaken identity’, so that perceived identity might not be constrained by facts.Footnote ⁴ Some people's identity – both their self-perception and their perceived identity – may reflect their ambivalence about their gender, and such people may also, in turn, take on both a male and a female identity (e.g. by cross dressing).Footnote ⁵ It is, nonetheless, hard to contest the idea that, in general, informed and reasoned judgements of self-perception and perceived identity are constrained by logical possibility as well as biological and social facts. Finally, it is worth noting that these facts combined with logical possibility constrain the range of social affiliations which one can choose from and thus relate to the first stage of choice distinguished above. Restrictions of that range then shape informed self-perception and perceived identity.

In what follows, I refer to constraints imposed by logical possibility and biological and social facts as the minimal constraints on identity choice. I argue that Sen's own account of identity as an object of constrained choice actually suggests no restrictions on choice of self-perception beyond these constraints. Before making that argument in detail, it is worth noting that I am here concerned with constraints on informed and reasoned identity choice in Sen's writings. Choice may in practice be constrained because there is lack of information or inadequate reasoning because (for example) there is limited information on the range of available social affiliations. This point plays a central role in Sen's discussion of multiculturalism and education, to which I turn in section 3.

2.3 Perceived identity and social context

Sen thinks there are further constraints on choice. Another way identity choice can be constrained, for example, has to do with the way others see us. The examples he gives typically involve people living under conditions of discrimination or oppression. The most obvious such case is the Jewish community in Nazi Germany. Arguably another involves Muslims in certain countries in the aftermath of the 9/11 attacks. An example might run as follows. In the post 9/11 world, Salma, a woman born into a Muslim family, might be seen as a ‘Muslim’, even if she has renounced her religious views, become a Buddhist and also renounced her social affiliation as a Muslim in various other ways. Sen's claim seems to be that people do not have much control over the social affiliations that are attributed to them: they may not be able to choose their perceived identity in this sense. But then it is hard to see how perceived identity can be an object of choice in the first place. Salma might, as Sen suggests, try to convince others that there are alternative ways of seeing her, or that they are incorrectly describing her social affiliations (Sen Reference Sen2006a: 6–8). But perceived identity is not obviously an object of constrained choice on the part of the person who is perceived, rather than something she can merely try to influence through her choices and acts, including acts of persuasion. Indeed, I suggest that it does not make sense to see a person's lack of control over her perceived identity as a constraint on identity choice.

Sen also uses this kind of example to suggest that social context can constrain or influence identity choice. He notes that a ‘person of Jewish origin in Nazi Germany may not be able to alter that identity as he or she wished’ (Sen Reference Sen2005a: 351). But here again he is primarily concerned with perceived identity, and not – as I understand him – with a person's self-perception or the range of affiliations from which the person can choose. He writes that: ‘[o]ur freedom in choosing our identity, in terms of the way others see us, can sometimes be extraordinarily limited’ (Sen Reference Sen2005a: 351, italics added). So this kind of example does not show that certain contexts impose constraints on choice of self-perception or one's own social affiliation(s) unless they are accompanied by some further explanation or account of how choice of self-perception or social affiliations can be restricted by perceived identity.Footnote ⁶ Sen does not provide any explicit explanation or account of the sort that is required. These considerations do not, in themselves, suggest any constraints on informed and reasoned choice of self-perception or social affiliation in addition to the minimal constraints mentioned earlier.

Social context might influence identity in another way, which primarily relates to how social affiliations come about in the first place. In this context, because Sen is concerned with how identities emerge, he partially addresses how they come to be defined. To take one of his examples (Sen Reference Sen2006a: 26), if someone wears size 8 shoes, even if she is classified as a ‘person with size 8 feet’, that does not necessarily amount to her having a ‘size 8’ identity of some sort. But context might, as Sen suggests, transform the situation. If for some reason there is a shortage of size 8 shoes, due to some systemic failure in, say, a socialist economy, people with size 8 feet might gang up together, form clubs and so on.Footnote ⁷ A group with a shared interest might grow up and this new group might influence both the range of social affiliations on offer, people's self-perceptions, as well as how certain people are perceived.

A real-world example of how social context can influence identity choice might be the case of those who were perceived as being, or who saw themselves as, Jewish in Nazi Germany. Whatever their self-perception was before the rise of Nazism – i.e. whether or not they saw themselves as Jewish – the fact that they became the object of discrimination on the basis of perceived identity may have influenced how they saw themselves and indeed their social affiliations, because there was now a reason for solidarity with others who were similarly perceived.

One might make a similar argument about the relation between self-perception and perceived identity for some Muslims after 9/11. Potential discrimination on the basis of perceived identity after 9/11 may have provided some Muslims with a reason for solidarity, which might in turn have influenced their self-perception. But, still, choices have to be made. Even if factors which influence the formation or disintegration of groups are not a matter of individual choice, they can influence that choice. It is not obvious, however, that the relevant dynamics or factors constitute constraints on choice itself. So again it is not clear that these examples establish any further constraints on identity choice.

What of the possibility of perceived identity constraining choice of self-perception in conditions of oppression or discrimination? Issues raised by choice of identity in the face of an atmosphere of, say, anti-Semitism or Islamophobia appear to have less to do with constrained choice and more to do with the way in which social affiliations are defined. In a well-known discussion, Jean-Paul Sartre argues that an anti-Semite can define what it is to be Jewish (Sartre Reference Sartre1954: 83).Footnote ⁸ Sen (Reference Sen2006a: 7) cites one context in which Sartre makes this sort of point. Here Sen focuses on common misdescriptions involved in the anti-Semite's view of the relevant characteristics of Jewish people and the relevance of their other characteristics (including the importance of other, non-Jewish, social affiliations). One way of seeing this phenomenon relates to the way choice is presented or framed. For example, it might be that anti-Semites understand, or frame, what it is to be Jewish in such a way that it is incompatible with being French.Footnote ⁹ The choice might then be presented as follows: ‘you can be either French or Jewish but not both’. That is, the choice is presented in such a way that the two options appear mutually incompatible as a matter of logic. This sort of example need not only relate to how people are identified by others (i.e. perceived identity) but is also relevant to aggressive assertions of social affiliation and self-perception by members of certain groups.

In their discussion of Islamophobia, Haleh Afshar, Rob Aitken and Myfanwy Franks (Reference Afshar, Aitken and Franks2005) give examples of both phenomena in the post 9/11 climate in Britain. On the one hand, they quote the Foreign Office minister Denis MacShane as claiming that ‘it is “time for the elected and community leaders of British Muslims to make a choice” between being British and Muslim’ (Afshar et al. Reference Afshar, Aitken and Franks2005: 263). Some Muslim groups accept the terms of the debate by insisting that their members give up their British identity and become ‘just Muslims’ (Afshar et al. Reference Afshar, Aitken and Franks2005: 274). This would suggest that the terms of the debate themselves can restrict choice. And the terms of the debate are rarely a matter of choice. Yet as Afshar and her co-authors also point out, other Muslims attempt in this context to find ways of being ‘both Muslim and British’ (Afshar et al. Reference Afshar, Aitken and Franks2005: 274). So we can reject the ways in which the terms of the debate are presented by, and the underlying perceptions of, some of the participants. That, too, can be a matter of reasoned choice.

This all suggests that even if one has limited influence on the social context in which one makes choices, choice about identity extends to the very description of one's social affiliation. This is particularly the case where issues relating to social affiliation are contested. Indeed, in his reading of Sen, Kwame Anthony Appiah (Reference Appiah2008: 349) focuses on precisely this point. He suggests that when the conditions that must be met for identifying someone as a member of a particular group (say as Jewish or Muslim) are contested, then there is some choice over whether one sees oneself as, or is perceived as, having that affiliation. In these cases, there is ambiguity about how the objects of choice (social affiliations in this case) are individuated or defined, which, in turn, leaves room for choice and which can have implications for whether various social affiliations are seen as mutually exclusive. The menu of options is not, as economists sometimes see it, purely ‘given’ or fixed, but itself potentially contested and the subject of reasoned discussion and choice.

When the definition of a social affiliation is contested, there will inevitably be the possibility that one's self-perception (e.g. the way in which one sees oneself as a Jew or Muslim) may not be the same as one's perceived identity (e.g. the way others perceive a person as a Jew or Muslim) because of different characterizations of the relevant social affiliation. So while it is easy to see why Sen interprets Sartre's point about the way a group (e.g. anti-Semites) can in effect sometimes define an identity (e.g. the Jewish identity) in terms of a constraint on choice, there is a level at which the definition of affiliations is itself an object of choice. Given the important roles of choice and reasoning in the definition of affiliations and self-perception, and the fact that one cannot meaningfully be said to choose one's perceived identity, the main constraints on the possibility of informed and reasoned identity choice that Sen identifies in this discussion remain – I suggest – the minimal constraints identified above.

3.1 Capability and Identity

In his discussion of multiculturalism, Sen also worries about constraints on identity choice. To elucidate these passages, it is helpful to explain relevant aspects of Sen's work on ‘capability’. Earlier work has attempted to map Sen's views of identity onto his work on capability (see Davis Reference Davis2006; Kirman and Teschl Reference Kirman and Teschl2006; Teschl and Derobert Reference Teschl, Derobert, Comim, Qizilbash and Alkire2008), and he himself has made the link (Sen Reference Sen2009b: 244–7). Nonetheless, there are some difficulties of interpretation because the idea of ‘capability’ can be interpreted in different ways. In its most basic form, Sen's capability approach suggests that the quality of life, egalitarian claims and development might be seen in terms of what people are able to be or do (see Sen Reference Sen and McMurrin1980, Reference Sen1985, Reference Sen1992, Reference Sen, Nussbaum and Sen1993, Reference Sen1999, Reference Sen2006b and Reference Sen2009b inter alia). Sen calls the relevant ‘doings’ and ‘beings’ ‘functionings’. At an intuitive level, the ability to achieve various functionings is a person's capability, and the person's well-being is then constituted by the functionings that are actually achieved. Development is conceived of as capability expansion, and egalitarian claims can be judged in the ‘space’ of capability inter alia.

In spite of the intuitive simplicity of the capability concept, there are two different ways in which it can be presented, and in its more technical formulations it is perhaps less intuitive. On the one hand, capability is primarily seen as an opportunity concept: various lives are constituted by combinations (or n-tuples) of functionings, and the range of lives a person can choose from is her capability (e.g. Sen Reference Sen, Nussbaum and Sen1993: 31). On the other hand, capability also has a meaning closer to its standard use in English; it relates to a person's power or ability (Sen Reference Sen, Agarwal, Humphries and Robeyns2003: 323; Reference Sen2009b: 233) – what a person is actually able to do or be. In Qizilbash (Reference Qizilbash2006: 21), I suggest that in Sen's work these two senses of capability are run together, yielding a somewhat technical sense of ‘capability’: a person is able – in the relevant sense – to do and be precisely those combinations of things that constitute lives or lifestyles she has opportunity to choose from. Her capability constitutes a kind of freedom: ‘the substantive freedom to achieve various functioning bundles (or, less formally put, to achieve different lifestyles)’ (Sen Reference Sen1999: 75). However, there is a tendency for Sen sometimes to focus on specific abilities in the second (ability or power) formulation, so that ‘capabilities’ refer not so much to combinations (or n-tuples) of functionings, but rather to specific abilities, such as the ability to appear in public without shame or to be socially integrated or to participate in the life of the community (Sen Reference Sen, Nussbaum and Sen1993: 36–7). Sen has recently recognized the fact that the term capability is used in these different ways in his work (Sen Reference Sen2009b: 233). A considerable literature now focuses on what list (or lists) of functionings or ‘capabilities’ understood in this second sense is (or are) relevant to the evaluation of equality, development etc. in particular contexts (Nussbaum Reference Nussbaum1988, Reference Nussbaum, Douglass, Mara and Richardson1990, Reference Nussbaum1992, Reference Nussbaum, Nussbaum and Glover1995, Reference Nussbaum, Crocker and Linden1998, Reference Nussbaum2000, Reference Nussbaum2003; Reference Nussbaum2006; Reference Nussbaum2011; Qizilbash Reference Qizilbash1998; Alkire Reference Alkire2002; Robeyns Reference Robeyns2003 inter alia). Sen has famously avoided advancing any definitive, context-free or all-purpose list of functionings or capabilities, while stressing the importance of public reasoning and discussion in making evaluative decisions for social decision making (Sen Reference Sen1999: 76–81).

It is relatively easy to see how identity features in Sen's writings on the second – power- or ability-based – way of defining capability, which typically involves mentioning certain specific abilities. The ability to be socially integrated and the ability to participate in community life are frequently cited by Sen and imply membership of social groups with which a person would identify. If these capabilities are important to the evaluation of egalitarian justice, the quality of life or development, then identity is also relevant to evaluation of each of these. Sen does indeed mention such capabilities in discussing the relationship between people's capabilities and the communities they belong to or live in.

However, as already noted, Sen's capability approach does not provide any definitive list of functionings or capabilities and has been criticized on these grounds (Qizilbash Reference Qizilbash1998; Nussbaum Reference Nussbaum2003; Sen Reference Sen, Agarwal, Humphries and Robeyns2003, Reference Sen2005b inter alia). Furthermore, it does not provide any formula for weighing different functionings or capabilities. So there is no implied weight given to functionings or capabilities that involve specific identifications or social affiliations. How important the social dimension and any implied social affiliations are in the evaluation of a person's quality of life or justice is thus not made explicit. But the mere fact that Sen repeatedly cites these abilities suggests that his position is not especially ‘individualistic’. He does not see men as ‘islands’, even if the capability approach has been criticized for being somewhat ‘individualistic’ (Gore Reference Gore1997; Stewart and Deneulin Reference Stewart and Deneulin2002; Sen Reference Sen2002b; Alkire Reference Alkire, Comim, Qizilbash and Alkire2008; Deneulin Reference Deneulin, Comim, Qizilbash and Alkire2008; Robeyns Reference Robeyns, Comim, Qizilbash and Alkire2008; Sen Reference Sen2009b).Footnote ¹⁰

In his writings on identity, development and culture, Sen takes this approach further, suggesting that there is an argument for seeing cultural identities or affiliations as objects one has freedom to choose, not characteristics one is simply born with. He argues that ‘there is a strong case for including cultural freedom among the human capabilities people have reason to value’ (Sen Reference Sen2006a: 112). His focus on ‘cultural liberty’ is related to cultural diversity in the context of multiculturalism as well as to issues such as tradition and development. If the preservation of tradition and ‘economic development’ – say reduction of some form of poverty – are in conflict, Sen suggests that ‘it is the people directly involved who must have the opportunity to participate in deciding what should be chosen’ (Sen Reference Sen1999: 32). In particular, Sen's discussion of development suggests that it is not for ‘experts’ or ‘guardians’ to decide on which traditions must be observed.

A similar line of thought runs through Sen's discussion of cultural diversity. One way of promoting cultural diversity, he thinks, involves the ‘freedom to pursue ethnically diverse lifestyles, for example, in food habits or music, [which] can make a society more culturally diverse precisely as a result of cultural liberty’ (Sen Reference Sen2006a: 116). Here, implicitly, choice of a variety of social affiliations, which can include ‘cultural identities’, and self-perceptions is promoted as a consequence of cultural liberty. The importance of cultural diversity derives from the value of cultural liberty. Sen wants to distinguish this approach from one which primarily promotes cultural diversity while at the same time implicitly endorsing a cultural conservatism that encourages people to stick to their ‘own cultural background and not try to consider moving to other lifestyles’ (Sen Reference Sen2006a: 116).

In a specific application of this line of thought to education policy in the UK, Sen worries about the expansion of faith-based schools when such schools ‘give children rather little opportunity to cultivate reasoned choice in determining the priorities of their lives. . .[and] fail to alert students to the need to decide for themselves how the various components of their identities . . . should receive attention’ (Sen Reference Sen2006a: 118). Finally, he argues that the vision of a multi-cultural Britain as a ‘federation of communities’ might box children into particular singular identities, with ‘little opportunity for reasoning and choice’ (Sen Reference Sen2006a: 118). He adds that (he has elsewhereFootnote ¹¹ argued that) this vision ‘tends to reduce the development of human capabilities of British children from immigrant families’. This line of thought explicitly links capability to culture and identity. It involves seeing the freedom to choose one's social affiliations as intrinsically important but implicitly goes further in arguing that suppression of this freedom might hinder the development of other capabilities one has reason to value. In the UK context, the worry concerns the quality of life of immigrants who might be disadvantaged in various ways.

While Sen here relates identity choice to capability by focusing on the specific notion of cultural liberty, the arguments just developed are more directly available in his capability approach when capability is framed in terms of opportunity. If a person's capability refers to collections (or n-tuples) of functionings that constitute lives or lifestyles she can choose from, then an expansion of capability will necessarily involve a greater range of lives or lifestyles people can choose from. The new lives or lifestyles on offer can involve identities that were previously not available. In the case of gender, in the process of capability expansion, women may have new opportunities that involve new identity choices, e.g. about whether to be housewives or career women or about the range of professions open to them. Similarly, as the capability of immigrants grows, one would expect that they can choose from a wider range of identifications involved in the broader range of lives on offer.

A wider range of identity choices seems to go hand in hand with a growing set of opportunities. Furthermore, when a person chooses the collection (or n-tuple) of functionings which constitutes her life, she must make choices about her various affiliations, their relative importance, and the importance of these affiliations as compared with other aspects of her life. On this way of relating the capability approach to identity, evaluations of egalitarian justice, the quality of life and development must directly engage with identity choice, since egalitarian justice, the quality of life and development are seen inter alia in terms of the lives from which people can choose. For this reason alone, a model of multiculturalism that restricts reasoned choice of identity can be criticized through the lens of the capability approach because it can restrict opportunity. So when capability is articulated as an opportunity concept, this approach contains the conceptual armour to criticize such a model of multiculturalism without invoking any specific freedom, such as cultural liberty.

Seen from this opportunity perspective, Sen's arguments hark back to John Stuart Mill's remarks in The Subjection of Women distinguishing ‘old’ and ‘modern’ institutions and ideas. Mill wrote that, in the world of modern institutions and ideas, ‘human beings are no longer born to their place in society, and chained down by an inexorable bond to the place they are born to, but free to employ their faculties’ (Mill Reference Mill and Okin1988: 17). So when Sen argues that ‘[b]eing born in a particular culture is obviously not an exercise of cultural liberty, and the preservation of something which is stamped, simply because of birth, can hardly be, in itself, an exercise of freedom’ (Sen Reference Sen2006a: 117), he echoes Mill's remarks made in the context of gender equality. Indeed, Sen's argument for reasoned choice of identity and cultural liberty falls within the tradition of equality of opportunity to which Mill's arguments for gender equality also belong. His argument is a straightforward extension of his arguments in favour of equality of capability in the context of theories of justice (see Sen Reference Sen and McMurrin1980, Reference Sen1992, Reference Sen2009b).

3.2 Constraints on information and opportunities for reasoning

In his discussion of multiculturalism, Sen is clearly concerned with ways in which restrictions or constraints on identity choice can lower capability or opportunity. But since we have already seen that Sen's own discussion of identity as an object of reasoned and informed choice subject to constraints implies only minimal constraints on choice, the question arises of how such choice is further constrained on his view. There seems to be a tension between these elements in his writings on identity. In the context of multiculturalism and education, I suggest that the important issues Sen raises relate to the nature of choice: he stresses that choice must be reasoned and informed. He objects to the model of multiculturalism, and the faith-based education it might promote, in part because they can restrict the information or opportunities for reasoning people have. His writings on capability and justice can help to clarify, develop and qualify his thoughts here.

There is a strong case, on Sen's view, for seeing cultural liberty as a capability we have reason to value (Sen Reference Sen2006a: 113). In the context of faith-based schools, he is concerned that we enhance ‘the capability of children to live “examined lives” as they grow up in an integrated country’ (Sen Reference Sen2006a: 160). Elsewhere he stresses the importance of ‘our ability to consider alternative options, to understand what choices are involved, and then to decide what we have reason to want’ (Sen Reference Sen2006a: 114). Sen's concern to stress what we have ‘reason to value’ or ‘reason to want’ is part of his broad view of rationality as reasoned scrutiny and is motivated, to some extent, by a worry about the way in which the desires and aspirations of certain people – notably various underdogs and the disadvantaged more generally – can adapt to the circumstances they find themselves in. If they adapt in this way, they may be satisfied or happy with small mercies but may not have a good quality of life. As a consequence, ‘utility’ understood as desire satisfaction or happiness can be a misleading currency of well-being or advantage. This is the ‘adaptation problem’ (Qizilbash Reference Qizilbash2006, Reference Qizilbash2007, Reference Qizilbash, Basu and Kanbur2009a). But equally, people who have adapted in this way may have a limited or parochial perspective on life, a perspective that might change or be broadened if they are exposed to other views and voices.

In his recent writings on justice, one of Sen's key claims is that public reasoning, by introducing a range of perspectives, might limit such parochialism by broadening people's horizons. Such reasoning, in his view, can also be strengthened by introducing ‘distant voices’, notably those from outside any particular community or nation. As he puts it (Sen Reference Sen2006c: 237), introducing distant voices can ‘be critically important for the reach of public reasoning’ and ‘can help to avoid undue dominance of local interest as well as possible parochialism of local reasoning shaped by the influence of established conventions and limited informational frameworks’.Footnote ¹² Clearly, education can either broaden or narrow people's perspectives and ‘informational frameworks’. Sen's worries about faith-based schools can be understood in this context. However, public discussion about what we have reason to value can, on this account, presumably also act as a corrective to education which either narrows, or fails to broaden, horizons. Sen's emphasis on public reasoning in his writings on capability and justice thus not merely illuminates but also qualifies and complements his discussion of faith-based schools and multiculturalism.

Sen's position can be criticized on the grounds that his work might be applied in a paternalistic or illiberal manner with theorists or society judging what is best for people in a way which might potentially lead to restrictions on liberty. Richard Layard (Reference Layard2005) and Robert Sugden (Reference Sugden2006, Reference Sugden, Bruni, Comim and Pugno2008, Reference Sugden2010) have developed this sort of criticism in different ways, and Sen has responded to some of these criticisms (see Sen Reference Sen2006b, Reference Sen2009b; Qizilbash Reference Qizilbash, Basu and Kanbur2009a, Reference Qizilbash2011a, Reference Qizilbash2011b, Reference Qizilbash and Clark2012). His responses are also relevant to the cases of people who live in minority or immigrant communities whose preferences may be shaped by the experience of living as underdogs. If their desires or preferences adjust to their disadvantageous living conditions and are restricted by the circumstances of their formation, they may not necessarily fit with what, one might claim, they have reason to value.

One of the most charged examples that Sen cites in relation to choice of identity can be understood in this light. It involves what he calls ‘reactive self-perception’ (Sen Reference Sen2006a: 88–93), where self-perception typically involves defining oneself in opposition to others, particularly in the context of colonialism, or of a colonial inheritance. In his discussion of the mistreatment of people during colonialism (and ‘the colonised mind’), he writes that: ‘[i]t cannot make sense to see oneself as primarily someone who (or whose ancestors) have been misrepresented, or treated badly, by colonialists, no matter how true that identification can be’ (Sen Reference Sen2006a: 89). He suggests that ‘to lead a life in which resentment against an imposed inferiority from past history comes to dominate one's own priorities cannot but be unfair to oneself. It can also vastly deflect attention from other objectives that those emerging from past colonies have reason to value and pursue in the contemporary world’ (Sen Reference Sen2006a: 89). This point can no doubt be contested and might fuel a concern about paternalism. Sen fears that the experience of the oppressed can restrict their horizons or perspectives so that either they do not see that there are certain things they have reason to value or they do not give adequate weight to certain valued ends.Footnote ¹³ Yet, on Sen's own account, people are free to choose how they see themselves and the priorities they assign to different social affiliations: why should we object to their self-perceptions or priorities when these seem unreasonable to us?

In Sen's works on capability and justice, the relevant evaluative issues are a matter of public discussion, and perhaps his views about reactive self-perception are best seen as contributing to that discussion. The central claim must be that people who are brought up in a particular way may have a restricted view of the identities open to them or be predisposed to give some specific identity – such as a traditional religious, nationalistic or ethnic identity – a high or overwhelming weight. The key objection cannot be to the fact that people are making a free choice, but to the range of identities or weights that are seriously considered in making the choice. It is this range and the evaluations which are made in identity choice that can be influenced by education and public reasoning. These are the key factors – in Sen's discussion – influencing the information a person has, and the opportunities for reasoning, about these matters. Certain sorts of education that narrow, or fail to broaden, people's perspectives and restrictions on public reasoning and discussion, which might limit its capacity to undermine certain sorts of parochialism, then become central constraints on reasoned and informed identity choice beyond the minimal constraints identified in section 2.

It is informative here to consider Sen's response to one critic who argues that the capability perspective is potentially paternalistic. In responding to Richard Layard's criticism which cites the adaptation problem (Layard Reference Layard2005: 113–21; Qizilbash Reference Qizilbash, Basu and Kanbur2009a, Reference Qizilbash and Clark2012), Sen writes that: ‘[w]hat the critics of unreasoning acceptance of persistent deprivation want is more reasoning about what ails perennial underdogs, with the expectation that, with more scrutiny, the “well-adapted” deprived would see – and “feel” more reason to grumble’. He illustrates this point by considering the position of disadvantaged women in India. He writes that: ‘the obedient and unagonised acceptance by women of their subjugation in traditionalist India has been giving way over the decades to some “creative discontent”, demanding social change . . .’ He adds that ‘[t]he role of interactive public discussion on the toleration of chronic deprivation plays a big part . . .’ in this process (Sen Reference Sen2009b: 275). Clearly, this line of argument is relevant to identity choice, especially in relation to factors that constrain informed or reasoned choice and restrict the opportunities open to minority groups.

4.1 Akerlof and Kranton's Framework

Having critically examined how, and to what extent, Sen's discussion of identity represents choice as constrained, it is worth comparing that discussion with Akerlof and Kranton's attempt to formalize identity choice in an economic model (Akerlof and Kranton Reference Akerlof and Kranton2000, Reference Akerlof and Kranton2002, Reference Akerlof and Kranton2005, Reference Akerlof and Kranton2010). There are several similarities between Sen's discussion and Akerlof and Kranton’s. For one, in Akerlof and Kranton's framework there are constraints on, or limits to, choice of ‘identity’. Unfortunately, however, they define ‘identity’ in a number of distinct ways in their work, which makes interpretation more difficult. It is variously defined as ‘a sense of self’, ‘self-image’ and ‘social categories’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 715, 716, 718). The notions of self-perception and social affiliation are, as we shall see, sometimes explicitly run together.

To explain Akerlof and Kranton's approach, I must first outline some elements of their framework. Firstly, in Akerlof and Kranton's framework, individual j's utility function, U_j, has three arguments: j's identity or self-image, written I_j, the actions of that person, written a_j, and the actions of others, a_-j:

(1)\begin{equation} {\rm U}_j = {\rm U}_j ({\bf a}_j ,\,{\bf a}_{ - j} ,I_j ).\end{equation}

They write that ‘identity is based on a set of social categories, C’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 718). Each person j has an assignment of people to these categories, c_j. Akerlof and Kranton add that one person's mapping of another individual into various social categories need not be consistent with the other’s, and that categories need not be mutually exclusive, so that people may be mapped into several social categories. They also introduce the notion of ‘prescriptions’, written P (Akerlof and Kranton Reference Akerlof and Kranton2000: 719), which indicate the behaviour appropriate for people in different social categories in different situations. These ‘may also describe the ideal for each category’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 718). Finally, ‘[i]dentity further depends on the extent to which j's own given characteristics є_j match the ideal of j’s assigned category, indicated by the prescriptions P’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 719). They accept that they ‘use the word identity to describe both a person's self-image as well as her assigned categories’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 718). Since it is both oneself and others who assign a particular person to particular categories, these assigned categories seem to relate to how a person is perceived by herself and others in terms of membership of specific groups. So the distinction between perceived identity and self-perception comes into play here, and perceived identity is again relevant in this account.

To explain their terminology, Akerlof and Kranton illustrate it in the case of gender. Here they write ‘there is a set of categories, C, “man” and “woman”, where men have a higher social status than women’, whereas ‘c_j describes j's own gender category as well as j's assignment for everyone else in the population’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 718). Finally, P ‘associates to each category basic physical and other characteristics that constitute the ideal man and woman as well as specifies behaviour in different situations according to gender. E.g. the ideal woman is female, thin and should always wear a dress. . .’. Given this terminology, Akerlof and Kranton (Reference Akerlof and Kranton2000: 719) propose the following ‘representation of I_j’:

(2)\begin{equation} I_j = I_j ({\bf a}_j ,{\bf a}_{ - j} ,{\bf c}_j, {\bm \ukrie}_j,{\bf P})\end{equation}

The ‘identity’ of a particular person depends on her own actions, those of others, her assigned categories c_j, her characteristics є_j and prescriptions, P. Akerlof and Kranton claim that ‘the individual is likely to have some choice over identity, as indeed people may even have some choice over their gender’, where the relevant choice may be more or less conscious (Akerlof and Kranton Reference Akerlof and Kranton2000: 720). Essentially the choice here is seen as the agent's partial ability to determine the social categories, c_j she is assigned to. So, for example, a woman can choose whether she is a career woman or a housewife (Akerlof and Kranton Reference Akerlof and Kranton2000: 724). That is simultaneously also a choice of self-image.

Akerlof and Kranton's discussion is broadly consistent with Sen's inasmuch as people are seen as having plural identities (i.e. social affiliations) and some choice over these in relation to how they see themselves.Footnote ¹⁴ To this degree, Akerlof and Kranton avoid Sen's illusions of destiny and of a unique identity. While they suggest that ‘to some extent people choose their identity; that is c_j’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 725), they nonetheless think that ‘identity choice’ is ‘very often limited’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 726). They write that ‘in a society with racial and ethnic categories, for example, those with nondistinguishable physical features may be able to “pass” as members of another group. But others will be constrained by their appearance, voice and accent’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 726). It should be clear here that while Akerlof and Kranton are in part concerned with self-perception, the constraints on, or limitations to, identity choice signalled by these examples relate primarily to perceived identity. In the example just given, it is one's ‘physical features. . .voice and accent’ which can block one's movement across social categories – and that clearly relates to whether or not one is perceived as being a member of a particular social category. They also discuss the potential costs of a change in identity. For an immigrant who becomes a citizen of the country she has moved to, the ‘decision is’, they tell us, ‘often fraught with ambivalence, anxiety and even guilt’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 726). In this example, the costs of a change of identity may be high, and the ‘price’ of seeing oneself in a particular way – as a citizen of a country one has moved to – may be high. Since the relevant ‘costs’ enter the calculus of ‘utility’, it is a matter of rational choice which reflects one's preferences, values or well-being (depending on how ‘utility’ is interpreted; see Broome Reference Broome1991 and Sen Reference Sen1991 inter alia), rather than a constraint. The choice might be tragic, or ‘limited’ in the sense that it is very costly to choose one of the options, but the options are feasible nonetheless and a choice is made. The key ‘limits’ on choice signalled by Akerlof and Kranton are thus either about perceived identity or relate to the costs of choosing specific options. Neither of these constitute genuine constraints on reasoned choice of self-perception.

4.2 Identity and Social Exclusion

To see how Sen's analysis may differ from Akerlof and Kranton's in the context of specific policies, it is worth considering Akerlof and Kranton's application of their identity model to the problem of social exclusion and minorities. They use a highly simplified model of social exclusion in which there are two groups, Greens and Reds, each with a single identity, with Greens being dominant so that the Reds might be seen as an excluded or disadvantaged minority. Prescribed behaviour for the Green group involves working, while for the Reds it involves not working. If members of each group depart from the prescribed behaviour associated with their group, they suffer a loss of identity and indeed of ‘utility’ as a consequence (Akerlof and Kranton Reference Akerlof and Kranton2000: 740–1). In this framework, the decision of members of the Red group not to work may be ‘rational’ (because it is ‘utility maximizing’), given their preferences. As Akerlof and Kranton put it, this ‘“self-destructive” behaviour is not the result of individual “irrationality,” but instead derives from low economic endowments and a high degree of social exclusion’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 742).

In this context, Akerlof and Kranton offer an interpretation of evidence to the effect that some ‘residential Job Corps programs might succeed while other training programmes fail’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 743). The reason for the success of the relevant programs, they think, is that ‘taking [Red] trainees out of their neighbourhoods would eliminate, at least for a time, the negative effects of interaction with those with Red identities’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 743). These negative effects would include the increased costs of changing identity or of acting contrary to the manner prescribed by a Red identity. Maximizing ‘utility’ while ‘rational’ in the conventional sense that economists refer to may not, in this case, promote well-being. Reds may be better off acting according to behaviour that is prescribed for the dominant Green group. As a result, Job Corps programmes which help them do this are more likely to be successful. In this model, the minority group (Reds) benefit from interaction with people from another group (Greens) because it lowers the costs of acting contrary to the behaviour prescribed within one's group, or the costs of changing identity. We earlier saw that these various costs are among the factors Akerlof and Kranton see as limiting identity choice. On this reading of their model, it is unsurprising that they conclude that ‘in a world of social difference one of the most important economic decisions that an individual can make may be the type of person to be. Limits to this choice would also be critical determinants of economic behavior, opportunity and well-being’ (Akerlof and Kranton Reference Akerlof and Kranton2000: 748).

While the last statement echoes Sen's central claims about identity choice, cultural liberty and capability in his discussion of multiculturalism, there are clearly significant differences in the analyses. As we saw, Sen thinks that if people are restricted in the perspectives and lifestyles they come across and know of in part because they are surrounded by people from the same religious or ethnic group and are taught a restricted traditionalist syllabus – as they might be in faith-based schools – they may be less able or free to make informed and reasoned choices about identity and consequently have fewer opportunities. For Sen, public reasoning can therefore play an important role in facilitating reasoned and informed identity choice for disadvantaged minority groups. Akerlof and Kranton's analysis, by contrast, leads one to focus on reducing the costs of acting contrary to prescribed behaviour and of certain choices of identity. The two analyses – and policy implications that might be drawn from them – differ in large part because they take different views of the nature of the constraints or limitations on rational or reasoned identity choice.

Nonetheless, the two approaches may sometimes converge at the policy level. For example, both would, in certain circumstances, encourage people from different social, cultural or religious backgrounds to mix either because this would promote informed and reasoned identity choice (on Sen's account) or because it would reduce the costs of acting in certain ways (on Akerlof and Kranton's model).