Identity Claims: Why Liberal Neutrality is the Solution,Not the Problem

The Rejection of Neutrality

According to Kymlicka:

Or in the even stronger words of Taylor:

Kymlicka's and Taylor's charges against neutrality are in fact three-pronged, and other similarly minded theorists have latched on to one, two or all three prongs. The first prong is that real liberal states are not neutral. That is, if we look at Canada, for example, particular languages are privileged (English and French) and some — such as its indigenous citizens — have rights that are not available to all. This privileging of both majority and minority practices is not neutral, but nor as the argument goes is it necessarily illiberal. Nor is Canada unique; most actual liberal states, albeit to varying degrees, do not simply ‘stand back’ in the cultural marketplace, but are very much involved in protecting and promoting various cultures, be they languages, religions or practices, and of both the majority and minorities.

The second prong of the attack is more theoretical; that is, liberal states cannot be neutral. For a state to function it will have to privilege some ways of life and not others. The most common example given here is language, with the argument that for a state to function adequately it needs a lingua franca. There are other areas that are commonly thought of as unavoidably non-neutral too. Veit Bader (1997), for example, argues that liberal democratic states require a particular political culture in order to function. In even stronger form, some argue that states will require not just a generic liberal democratic culture, but a particular national culture whose accompanying feeling of togetherness provides support for public goods, especially redistributive justice (e.g. Miller, 1995; Soutphommasane, 2012; Tamir, 1993).

Accordingly, then, if liberal states are not and cannot be neutral, we should stop pretending they are or ever could be neutral. Further, if the good of neutrality lies in the freedom of citizens to follow their own ways of life, then, as the argument also goes, neutrality is failing and is better replaced by some non-neutral, or at least much more active, form of state policy.

Yet we need not move so fast in the rejection of neutrality. The first problem with these two ways of rejecting neutrality is that they fail to see it as a political ideal. Ideals are ideals because they are never fully realised, both in practice and in ideal theory. If we look at the ideal of democracy, for example, no contemporary state is democratic according to any of the various notions of ideal democratic theory out there. This does not mean that we cannot call particular states democratic, nor does it mean that we should abandon the concept of democracy altogether. What it does mean is that we can critique existing states using our particular normative ideal of democracy and encourage the state to come closer to this ideal while still balancing other values. Neutrality here is no different: no actual state will be neutral, but this does not mean we cannot (or should not) critique its practices against this ideal.

This balancing of other values leads to the more theoretical critique that liberal states cannot be neutral. Almost no ideal theory promotes a world of just one value. Even Rawls' promotion of justice as the ultimate value is one that already has other considerations (namely efficiency and security) built into it. Further, the sort of neutrality discussed here (that is, between ways of life) is not an ultimate value, but a derivative value — other values are prior to neutrality and thus justify it, such as the freedom of individuals to live their lives as they see fit. As a derivative value it must be balanced by other considerations. ¹ If, for example, the state really cannot provide a just distribution of resources without privileging a majority culture, then it should privilege that culture. This does not mean it can do anything to privilege this culture, but that it is justified in deviating from neutrality as much as, and no more than, is necessary to achieve this outcome. Neutrality, like other ideals, particularly derivative ideals, needs to be balanced with other things that are held as important. As such, in both practice and theory it is not a reasonable critique of neutrality to say it cannot be realised (compare Carens, 2000, p. 11, pp. 52–87).

Yet ultimately this is not the main criticism of liberal neutrality. While issues of feasibility are used to soften neutrality up, it is its desirability (or lack thereof) where the main blow is landed. This is the third prong: the liberal state should not be neutral in matters of identity as state neutrality either fails to lead to the accommodation of minority ways of life or at least makes some minority ways of life comparatively and significantly more difficult to follow. It is this argument that liberal neutrality cannot adequately and fairly accommodate difference that I will address in the remainder of the article. In response, I will argue that liberal neutrality has been wrongly characterised, and that a ‘reflective’ rather than a ‘starting gate’ notion of neutrality best encapsulates the ideal. This reflective, or difference-sensitive notion of neutrality, while not able to bring about the accommodation of all differences, more adequately and fairly creates the space for people to live their lives as they see fit than the alternatives do. Yet liberal neutrality can imply several different things, so let me first briefly lay out three basic conceptions, before explicating my own position in the section that follows. ²

Conceptions of Neutrality

The three most common and basic conceptions of state neutrality in relation to the ways of life within a polity are neutrality of justification, of intent and of outcome. ³ Under neutrality of justification, no law or policy should be justified by the rightness of any particular way of life. This is a form of procedural neutrality, where the laws and policies that citizens should live by should be equally justifiable to all. The most famous example of this type of neutrality is John Rawls' political liberalism (2005), in which the principles of justice that regulate a polity must be consistent with public reason, that is, the form of reason that is not particular to any one conception of the good, and can reasonably be shared by all citizens. Thus Rawls' principles of justice are intended to have neutral justification; the rightness of no particular way of life is used to justify them, and they are intended to support the widest possible ways of life. ⁴

The second type of neutrality is of a different order. Here what matters is not so much how a law or principle has been justified as much as its intention. Under neutrality of intent, a neutral institution or policy should not intend to favour (or hinder) any particular way of life. For example, a state that provided funding and support to opera but not pop music, or to Olympic rowing but not rugby league, or to Catholicism but not Islam, would not be neutral in this sense. These types of policy decision intentionally help one way of life and do not assist, and perhaps even hinder, its competitors. Neutrality of intent can be enacted through either act or omission. That is, a state can be neutral in this way by either standing back and doing nothing — what is sometimes called ‘benign neglect’ — or by offering all parties similar support, for example by funding and supporting all religions in the same way. The former is a form of ‘hands off’ neutrality and the latter a form of ‘hands on’ neutrality.

The third type of neutrality is effectively the flipside of neutrality of intent. Neutrality of outcome is concerned not only that institutions or policies do not intend to favour any particular way of life, but they do not actually favour — even unintentionally — any way of life. Thus, to use the opera/pop music example, a neutral (in intent) policy may support neither form of music. Yet such a policy is unlikely to have neutral effects. Pop music by its very nature is likely to be well supported, while opera is usually less popular and is comparatively much more expensive to produce. The effect on these two art forms would not be neutral. Thus, unlike the other two conceptions of neutrality, in neutrality of outcome the state will most likely have to offer support to some ways of life; it is entirely unlikely that neutrality of outcome will be achieved by the state simply standing back.

It should be noted that none of these three conceptions of neutrality necessarily calls for complete neutrality. Rawlsian justificatory neutrality is not, even in his own words, neutral with respect to everything (Rawls, 2005, p. 192). ⁵ While it is neutral between conceptions of the good, it is not neutral about ultimate values (conceptions of the right) — and it is certainly not meant to be neutral to the ‘unreasonable’. It is hard to see how any political authority could be totally neutral in this respect. This range aspect of neutrality is also present in neutrality of intent and neutrality of outcome. State institutions and policies will not be neutral with respect to any way of life, but only neutral, at most, in regard to justice-respecting ways of life. Neutrality in all three conceptions is a range concept where everything within certain boundaries is to be treated neutrally. ⁶ Thus a claim that ‘there is no such thing as complete neutrality’, while true, is only trivially so; no actual conceptions of neutrality strive for it, and it would be hard to envisage any remotely liberal political theory advocating such a position (even if it was somehow possible).

Of more interest is the likelihood of a policy or institution being neutral in one way, but not in another. Thus a neutral justification may lead to a policy that has non-neutral intent. And a policy of neutral intent may lead to non-neutral outcomes. For example, we can imagine an official state language being justified neutrally. The existence of a lingua franca might be justified on grounds of efficiency, social cohesion and perhaps as a prerequisite for social mobility (Barry, 2001). That is, polities function better, people feel more connected, justice can be both served and seen to be done, and people may be more likely to avoid being ‘victims’ of their birth when there is a core language. ⁷ These neutral justifications will certainly not lead to neutrality of intent — indeed quite the opposite as one language is to be promoted ahead of others. Whichever language is chosen, it is unlikely to be as easy for every citizen to converse in it. Indeed, as already mentioned, this is one of the charges against the non-neutrality of actual liberal states. As far as neutrality of outcome is concerned, we do see in practice that when particular languages are promoted, other minority languages tend to die off. ⁸

It should also be clear that neutrality of intent will commonly be in tension with neutrality of outcome; giving equal resources to all religions, for example, does not mean they will be equally successful. More interesting, though, is the way that over time neutrality of intent may fail on its own terms. If we take the example of religious holidays, we can imagine a society where several different denominations of Christians reside, perhaps even with quite a high degree of animosity, and all of which recognise Sunday as the day of rest. The state enacts a policy that formally recognises this as the day of rest and does so with neutral intent — it does not intend to favour any of these competing ways of life (and indeed the outcome is neutral too). Yet over time, other religions with different days of rest begin to be practised in this society and some Christians abandon religion altogether. This policy, while being neutral in intent when first implemented, is now very biased towards particular ways of life and there is a good case that it unfairly impinges on some individuals' lives more than others.

For the liberal concerned with dealing fairly with diversity, there are now two possible problems with neutrality. First, neutrality of justification can come into tension with neutrality of intent, and second, neutrality of intent may, over time, take us a long way from what we may commonly consider neutral treatment. This would mean that, as its critics charge, though not necessarily for the same reasons, these two traditional understandings of neutrality may not be adequately and fairly accommodating of diversity.

Neutrality Defended

Liberals, whether they argue for differentiated rights or for a state that stands back in the cultural marketplace, place significant value on the importance of individuals living their lives as they see fit. For some liberals these life plans must be autonomously chosen and/or autonomy promoting (e.g. Raz, 1986), while for others they simply need to be expressed preferences (e.g. Kukathas, 2003). At this stage this distinction matters much less than the shared premise that, whatever ways of life are deemed legitimate, they should not be subject to unnecessary restrictions as the result of state actions. The question of which state actions unnecessarily restrict legitimate ways of life requires an answer of how to balance competing ways of life. It is here, the neutralist argues, that by not favouring any particular (legitimate) way of life all possibilities and their adherents have been treated fairly, and in an important sense equally. One part of this fairness consideration is resource focused: why, against a background of justice, should resources that you could meaningfully use to support your own life be used to support ways of life you neither favour nor benefit from (Patten, 2012, p. 263)? But the other part is simply about the difficulty, or otherwise, of living your life as you see fit in an environment where yours is not among the favoured ways of life. Some lives are hard to live without sufficient adherents — this is particularly true where languages are concerned — and when other ways of life are promoted this can provide incentives to move towards these favoured ways of life and reduce the number of adherents in others. At other times the issue is more symbolic, with the potential psychological harm that can occur when a particular way of life is held as less important than another. And at other times it is the social pressures (noted by J. S. Mill, 1998 [1859]) that an emboldened majority may use to make a way of life more difficult.

None of this is to say that these considerations always matter — we regularly tax people to pay for things they neither favour nor benefit from, and government decisions often provide disincentives for particular ways of life. Instead, what matters is that when the state decides on laws or policies it should be careful not to favour unjustifiably any legitimate way of life; that is, it should give these freedom-protecting considerations due weight, and attempt to treat these ways of life fairly and equally.

Nothing of what I have said so far necessarily takes a side in the neutrality/perfectionist debate. While autonomy-promoting perfectionists may claim to have rejected neutrality, it need not be entirely abandoned. So while perfectionist liberals are explicitly not neutral between ways of life that realise the value of autonomy and those that do not, they are still left with a range of ways of life that are consistent with autonomy; and it would seem that they should still insist that the state remain neutral among these ways of life. On this reading, what is at stake in the neutrality/perfectionist debate is the range of ways of life the state should treat neutrally, with ‘neutralists’ including all permissible ways of life, and ‘perfectionists’ only including valuable ways of life. This would mean that one needs to be careful of overstating the distinction between neutralists and perfectionists, and that understanding neutrality, in whatever form, is still necessary even among (most of) those who explicitly reject a neutral state (compare Wall, 2010). ⁹

Nevertheless, my interest here is less in whether justifications should be neutral (for ease of argument I assume from now on that justifications should be neutral, and thus side with neutralists on this matter), nor in which justifications might be considered neutral, but in what happens after this initial justificatory stage. It is here at this policy implementation or institutional design stage that I want to suggest a way in which the freedom for individuals to live their lives as they see fit can still be fairly upheld. Of course one way to do this would be somehow to look into everybody's life plans and desires and then try to fulfil them. This would be a complete neutrality of outcome. The most basic problem with this approach is that it would be apolitical. If it really was possible for everybody to do what they wished (even if we exclude things that harm others), then politics would have no role. So while neutrality of outcome may be relevant in a limited domain (perhaps between a limited number of discrete things, such as organised religions), it is not relevant on a general scale. There are also good reasons for not using neutrality of outcome as an unrealisable, but still action-guiding ideal: it fails a basic choice-sensitivity test and may undermine individual responsibility for life choices (Kymlicka, 1989, pp. 884–5), and would also seem to require continuous interference in people's lives. This is not to say that outcomes are not relevant — as I will argue in a moment they may be important indicators — just that a general concern with neutrality of outcome seems misplaced in any political context.

Instead I want to show how neutrality of intent can be made sufficiently and fairly accommodating, and thus rebuff the argument that liberal neutrality (at least at this level) cannot adequately deal with diversity. My point here is to argue on the terms of (at least most of) the critics of neutrality, and to show how it is neutrality that brings about fair and equal accommodation of difference; and further, that a commitment to fair and equal accommodation is a commitment to a principle of neutrality. In the previous section, I suggested that a neutral justification may lead to non-neutral intent, and that over time what started out as neutrality of intent may no longer appear particularly neutral. Yet this does not mean that in these cases we should abandon these other forms of neutrality altogether. Regardless of the justification that is used for a particular policy or institution, it should still aim to be as neutral as possible between ways of life (remembering that neutrality is an ideal in this domain). This means two things. First, if, for example, a common language is required, then its choice and use need to be implemented to just the level required to achieve its goal, and no more. It is hard to say in the abstract how far this should be, but the French history of nation building in the nineteenth century, involving the deliberate wiping out of regional dialects and the punishment of those who continued to speak them, would certainly be going too far (see Weber, 1977). The ideal of neutrality between ways of life needs to be balanced with the particular policy goal (even if justified neutrally) thereby to keep it in check. If, for example, there are sufficient reasons of social coordination to favour one way of doing things over others, then on a neutral justification this is permitted. Neutrality of intent is not then nullified, but relegated to a balancing value.

Second, and equally important, in order to deal with the potentially changing nature of the range of ways of life over time, this balancing act needs to be ongoing rather than simply one-off. A requirement for police officers to wear a uniform could have been both justified neutrally and initially implemented with neutral intent in relation to the ways of life of a particular state's citizens. However, if the freedom of individuals to live their lives as they see fit is taken seriously, this cannot be the end of the story. People change, fashions and tastes evolve, and migrants with different ways of life enter a political community. The existing uniform regulations may become much less neutral than was initially intended. A form of neutrality that treats fairly the freedom of individuals to live their lives as they see fit needs to be able to respond to changing circumstances and change its policies to maintain a more common-sense neutrality. This is not neutrality of outcome, but nor is it neutrality of intent as usually understood. It is instead a realisation that a policy that is both knowingly and avoidably not neutral in intention at that point in time should be changed.

This notion of state neutrality is range sensitive. As stated above, one is never neutral in the abstract, but instead neutral among certain things: in this case, ways of life. On this understanding of neutrality, the state needs to be sensitive to the changing nature of the things among which it is neutral. In the police uniform example this neutrality can be realised by expanding and changing the range of possible uniform variations: skirts as well as trousers, long sleeves as well as short ones, alternative headwear in the same colour as the existing hats, etc. This conception of neutrality takes seriously the idea of ‘not intentionally favouring’ any particular ways of life by applying this condition beyond simply the design and setting up of policies and institutions.

Here, then, is a conception of neutrality that takes seriously the fact that neutrality is a range concept, and is in fact difference sensitive. It acknowledges that the range of differences over which the state is neutral may expand and adapt ¹⁰ — the details of the range of tolerable differences cannot be decided once and for all at a single point in time, and the details of what occurs within that range cannot be known ahead of time. While in justificatory neutrality, at least of the Rawlsian variety, this range manifests itself through the exclusion of ways of life altogether such that only general principles are permitted (omission), in the case of neutrality of intent, the intention is not to privilege any actual ways of life. While this may be moderately easy to do when a policy or institution is first implemented — there will be an existing range of ways of life — over time this range will change and so too must the policy if it is to remain neutral in this way.

In the police uniform example, the policy of uniform wearing is feasibly justified on the grounds that police need to be recognisable, and that a common uniform helps provide some discipline for them as well as authority in the eyes of the general public. These are seemingly neutral justifications. Yet these goals can be achieved with some variation (as now happens in most Western liberal multicultural jurisdictions), and thus there is insufficient justification for the uniform to be strictly uniform, so to speak.

The argument that liberal state neutrality should be thought of as dynamic and difference sensitive is consistent with two forms of liberalism: that of recognition and an active state, and that of withdrawal and a smaller state. What I will call difference-sensitive neutrality can thereby be realised by the state either withdrawing support for privileged ways of life (being ‘hands off ’), or maintaining this support and extending the privilege to the previously excluded (being ‘hands on’). ¹¹ ‘Hands on’ neutrality is perhaps the closest rival to difference-sensitive neutrality, and in the next section I want to add a third element to this conception of neutrality (in addition to it being an ideal and responsive to range), which brings it much closer to a more traditional ‘hands off’ understanding of neutrality. This involves a strong preference for moving closer towards neutrality by removing rather than adding support in situations where non-neutrality needs addressing.

Why ‘Hands Off’ is Better than ‘Hands On’

Difference-sensitive neutrality could be interpreted as either a ‘hands on’ or a ‘hands off’ version of neutrality. Anna Elisabetta Galeotti (2002), for example, puts forward a ‘hands on’ version of neutrality with similar features to mine. She argues that when it is realised that the state is acting non-neutrally, the state should not simply expand its range of neutrality but that, in so doing, it should also publicly declare the previously excluded way of life to be of equal worth. That is, because some ways of life were previously privileged, the excluded ways of life should now receive special treatment as a form of balancing out. This approach does not advocate removing recognition or privileged status, but adding new ways of life to the list. In this sense neutrality is still the intent. Thus this is also a conception of neutrality that is sensitive to difference.

This ‘hands on’ form of difference-sensitive neutrality is, however, generally inferior to a more ‘hands off’ form of neutrality with a strong preference for withdrawing support rather than offering support when it is realised that neutrality is being unjustifiably violated. To give an example, many Western countries have recently had ‘gay marriage’ debates. One of the main arguments given by advocates of gay marriage is that because the state recognises heterosexual marriage, it is unfair to homosexuals that their way of life is not equally recognised (both symbolically and/or legally). In other words, it is a non-neutral policy. Whereas those who argue for recognition or for ‘hands on’ neutrality are committed to offering homosexual couples the option of marriage, difference-sensitive neutrality does not require this move. Instead of offering marriage as an option to homosexual couples, it would countenance withdrawing the recognition of marriage by the state altogether; marriage would become a private religious or cultural matter, and from a legal point of view the state would simply register partnerships. The state should recognise that it is avoidably, and seemingly unjustifiably, privileging one way of life over another and thus needs to become more neutral. Yet both responses — recognising gay marriage and withdrawing marriage as a state issue altogether — can be seen as moving closer to the ideal of neutrality. So why favour the first approach?

To answer this, I want to return to what I take to be a shared premise underlying this dispute: the importance of living one's life as one sees fit and treating such ways of life fairly. A difference-sensitive neutrality that removes state action is more likely to increase the space for individuals to realise this freedom than one that increases state action. While recognising that gay marriage is better than only recognising heterosexual marriage, it is still not as neutral as it could be. If the aim is to be neutral towards people's significant relationships, then expanding the options from one to two still excludes and thus treats potentially unfairly other forms of legitimate and important relationships. The form these relationships could take will be difficult to list and then recognise in the abstract — surely there are many possible variations. It is much better for the state simply to stand back and treat whatever form of relationship people have in the same way, rather than privileging only some recognised forms. Legal recognition, and making marriage a private rather than a state affair, does not prejudge what forms of relationship individuals may hold as significant. It also allows particular religious groups to put quite restrictive parameters on marriage for their members, but also for those members easily to leave their group if they so wish.

Another example, and one where this standing back has at least partially occurred, comes from Australia where elections are held on Saturdays. Saturday polling may seem to impact unfairly and avoidably on some citizens, including religious Jews. One option would be to exempt religious Jews from Saturday polling, either entirely (in Australia voting is compulsory), or to give them another option, Sunday perhaps. This would be actively to recognise these Jews and offer a neutralising alternative. Yet neutrality can be achieved in another way that does not recognise any particular group or grant exemptions, and also allows other individuals greater freedom. The existing options of pre-poll and postal voting mean that any citizen can choose to vote prior to polling day. Another option, of course, would be to try to choose a different day altogether, say a Tuesday, on which no one had their Sabbath. But this last option, while perhaps neutral in intent, would seemingly fare worse on a neutral justification: maximising the number of people who vote.

In both cases, the existence of gay marriage advocates, or religious Jews wishing to exercise their civic duty, act as triggers for re-evaluating the neutrality of a particular policy or institution. Yet the state should not just respond singularly to claims of non-neutrality, but in its review of existing policies should try to find the most neutral solutions, and ones that are potentially neutral between more than just the current privileged way of life and those who are making a claim of non-neutrality. While those who make claims of non-neutrality upon the state often have genuine and important demands, it is likely that there are those less eloquent who also have important claims, and perhaps even those who are yet to realise or worry about whether their way of life is treated non-neutrally in some way. Indeed, the gay marriage issue seems to be partially one of ‘not yet realising’. In early gay rights advocacy, the issue of marriage was not even on the agenda; not only because there were much more important issues — such as overturning outright prohibition and the multiple sites of discrimination — but also because it was viewed as something that heterosexual rather than homosexual people found important, and something that institutionalised male supremacy and restrictive notions of monogamy (Altman, 2011).

The occasions when the particular policies and institutions of modern states change are quite rare. By removing privilege on these occasions, the new policies and institutions are likely to be neutral among a greater range of ways of life than if new kinds of recognition are simply added. In other words, the acknowledgement of avoidable non-neutrality in a particular area provides a unique opportunity for a modern state, and one that should be taken full advantage of. Difference-sensitive neutrality thus takes seriously that not only is the modern state a lumbering beast which changes very slowly, but also that it is unlikely to have good knowledge of people's actual ways of life, or predict how they will evolve in the future. This is in contrast to those who call for more activity and recognition, who seem to assume a state that has excellent knowledge, is dynamic, can change very quickly and, more importantly, do all this correctly. Thus the primary reason for ‘hands off’ rather than ‘hands on’ neutrality is based on a view of what the state can and should do given its limitations.

At this stage an objection from those who wish for a more active state may focus less on individual liberty and more on the problems of identity recognition. Indeed, many of those who argue for more ‘hands on’ approaches to neutrality do stress the importance of identity recognition. This is based on a highly plausible argument — espoused most prominently by Taylor (1994) — that, following Hegel, people's identities need to be recognised: first, for what they really are (i.e. ‘authentically’) and, second, as worthwhile. If someone's identity is misrecognised (for example, they are Welsh and are viewed as English), and/or their identity is viewed as being of lesser social status, they can suffer a serious harm. Thus it is argued that the state (and often the citizen too) must properly recognise all citizens' identities. If the state already successfully recognises the majority identity — indeed this is seen as almost unavoidable — and the identity of minorities is ignored and/or misrecognised, this is a serious problem that needs addressing by also recognising the minority identities. In other words, the state should be neutral in relation to identity by recognising minority identities, and failing to do so can cause serious harm. ¹²

But if misrecognition can cause such harm, then there is a practical danger of the state getting it wrong and doing more harm than good in its acts of recognition. While perhaps in a hypothetical polity with only a handful of individuals the state could correctly recognise all identities, it seems hard to imagine for anything approaching the size and diversity of a real polity. ¹³ As is often pointed out, one person's identity is not some static thing that remains the same for their whole lives, nor is an individual's identity the same as that of somebody else who may have the same identity markers — for example, there are many different identities contained within the marker ‘Muslim’. Moreover, most people have identities that are multifaceted; that is, their own individual identity is made up of many different singular identities whose weightings and meaning are likely also to be dynamic and fluid. If the state is not capable of being both cognisant of, and acting on, the subtleties of this identity fluidity, then it can misrecognise and/or ‘artificially’ fix or ‘reify’ particular identities (Fraser, 2000). That is, this political interference ends up creating incentives for particular identities, and thus instead of responding to problems of identity can end up creating them.

This suggests that, whatever the potential good of state recognition of identity, there are real practical problems for actually achieving it in any equitable fashion, and if the recognition thesis is correct (as I assume it is), then there is a danger that the state can cause real harm by incorrect and inequitable acts of recognition. This points to a preference for removing acts of non-neutral recognition rather than adding new acts of identity recognition. This does not mean that people's identities should not be recognised at all — freedom of association should be able to realise a reasonably dynamic culture of recognition — just that it seems better if the state in all its lumberingness is not the agent of such recognition.

Both false recognition and the reification of identity come together in what has been described as the ‘minorities within minorities’ problem. Here acts of recognition, either of identity or simply differential treatment (exemptions, rights, etc.), which recognise a whole group can fail to see diversity within (Eisenberg and Spinner-Halev, 2005). This can lead to oppression of those within that group who hold divergent views, or at least have divergent interests. The most common example here is the place of women in more traditional groups (Okin, 1997). If the state supports a particular group either by recognising their identity or in some more material way, it can give support to cultural spokespeople and to those with a more conservative view of the group's practice and identity. This can then silence or at least make more difficult the lives of those who have different views about the group's direction as well as those whose interests may partially pull against the direction of the group. This is no longer simply a problem of identity, but not all failures of neutrality will involve identity.

Thus, given the size of modern states, and the almost infinite amount of often changing diversity, it seems that trying to be ‘hands on’ will be more difficult than being ‘hands off’, especially if one is concerned with protecting individual freedom and not damaging identity. It seems that most of the time it will be easier and more viable to remain neutral by withdrawing support rather than constantly trying to juggle support for various minorities. Nevertheless, it may appear that this fluid and dynamic picture of the various ways of life within a polity also poses problems for my conception of difference-sensitive neutrality; it too needs to respond to the diversity that actually exists. Yet by expanding the space for diversity, difference-sensitive neutrality does not just respond to the initial potential trigger of non-neutrality. While these triggers, such as the gay marriage advocates mentioned above, inform the state of its non-neutrality, the state's response of withdrawing support for particular ways of life makes the policy or institution more neutral between actual ways of life. Yes, it might not always get it right, but by not recognising any particular way of life it is much less likely to fix or distort an identity, trap people within their groups or inhibit people's legitimate ways of life. Of course, not all ways of life will succeed, but if this is the yardstick, then as mentioned above, politics would have to be transcended.

To what extent, then, is difference-sensitive neutrality multicultural? It certainly does not prescribe identity recognition, exemptions from general laws, or differentiated rights. But nor does it prescribe assimilation or a privileging of majority ways of life. Difference-sensitive neutrality is not anti-multicultural, but at the same time does not support claims for diversity, and instead simply tries to make space for them.

Where this conception of neutrality also differs from the usual conceptions of multiculturalism is that it does not try to distinguish between types of difference. Its intention is to be neutral between mere preferences and deeply held beliefs, and not try to dig too deep or use some criteria as to why some people's differences may matter more than others. Expanding the range of neutrality, rather than the granting of exemptions or recognition, allows this lack of distinction — and this is a significant and important advantage. First, by avoiding the need to distinguish between types of difference it sidesteps the vexed questions of ‘authenticity’, ‘chosen-ness’ and ‘alterability’ which come with positive recognition. And second, for those, such as autonomy liberals, who think it matters that we have more rather than fewer viable options, it more easily allows ‘experiments in living’. Recognised exemptions and group-differentiated rights require officials to decide who is eligible and who is not, and thereby limit the flexibility of individuals to choose for themselves which identities or practices to subscribe to. On the conception of difference-sensitive neutrality I have put forward it is entirely up to the individual which way of life they identify with, and state officials do not need to examine their reasons. So, for example, if a uniform changes to allow a variety of headwear, then it is up to individuals which type of headwear they quite literally ‘try on’, and not a matter of being eligible (or not) for a recognised exemption. ¹⁴

Where numbers and intensity will likely matter, however, is when it comes to triggers for expanding the range of neutrality. Because neutrality is but one value that needs to be balanced with others, the way the case for change is made is likely to matter politically. Thus if there are a significant number of people who claim to be treated non-neutrally and in a significant fashion, this is likely to affect whether or not the range of neutrality is changed. This is not to say that these things should matter in a pure conception of difference-sensitive neutrality with some omniscient state, but to realise that changing policies and institutions is difficult and should not be done lightly. Yet whatever the trigger to expand the range of neutrality, the process of change should look beyond it to other potential ways of life that may also be being treated non-neutrally.

Finally, while the theoretical ideal of difference-sensitive neutrality involves a pulling back of unjustified non-neutral support, this may not always be possible in practice. This can be for several reasons. Sometimes it may be impractical — it might, for example, really be better to have only a small number of national languages. At other times there may be negative consequences of changing the status quo — Brian Barry (2001) suggests this in relation to removing what he sees as unjustified exemptions from humane slaughtering laws for Jews and Muslims. And at other times democracy may be a barrier, and it might not be politically feasible to reverse some areas of favouritism — subsidies for farmers or for Olympic sports are possible examples here.

In these cases, the lack of real possibility of removing support for particular ways of life does not mean that nothing should be done. In the first place, as I argued above, support should be reduced as much as is feasible in order to come closer to the ideal of neutrality. Failing this, there may be occasions when taking a more ‘hands on’ approach is the right thing to do. For example, in the marriage case, if it is not democratically possible to remove state support for heterosexual marriage, then homosexual marriage should be recognised by the state. This position would be less neutral than removing support altogether, but more neutral than only supporting heterosexual marriage, and is thus to be preferred.

Conclusion

I began by noting that many of those who argue for group-differentiated rights and the like used the failure of neutrality as a crucial first premise. Their claim was that because liberal states were not and/or could not be neutral, and more powerfully, should not be neutral, a variety of differentiated treatment was justifiable. In response, I have argued that neutrality is an unrealisable yet action-guiding ideal that necessarily needs to be balanced with other values; that neutrality is a range concept, and that neutrality of intent, in particular, needs to be sensitive to the changing nature of this range; and finally, that the values that justify neutrality, namely liberty, equality and fairness between ways of life, are best realised by a ‘hands off’ rather than ‘hands on’ form of neutrality. Accepting this argument pulls most, if not all, of the rug out from under multicultural and differentiated rights theory: the liberal state can have a policy of neutrality and still fairly and reasonably accommodate a variety of ways of life. In achieving this accommodation, the vexed questions of for whom, when and how various identities/practices/beliefs should be recognised are entirely avoided.