Religion in the Law: The Disaggregation Approach

Should religion be singled out in the law? This Article evaluates two influential theories of freedom of religion in political theory, before introducing an alternative one. The first approach, the Substitution approach, argues that freedom of religion can be adequately expressed by a substitute category: typically, freedom of conscience. The second, the Proxy approach, argues that the notion of religion should be upheld in the law, albeit as a proxy for a range of different goods. After showing that neither approach adequately meets crucial desiderata for an inclusive theory of religious freedom, the Article sets out the Disaggregation approach and defends against the alternatives.

expresses our essential nature as God-created creatures; and therefore deserves special deference and respect. 2 Liberal political philosophers, following John Rawls, start from the idea that a just political order should be responsive to a pluralism of ethical principles and ways of life. 3 It should not entrench any particular (here, theistic) conception of the good life as worthy of politico-legal protection. The second possible approach is that endorsed by most critical scholars of religion. It denounces, not simply the notions of religion and freedom of religion, but the core ideals of normative liberalism (secularism, religious freedom, the separation between public and private, state sovereignty, and so forth) as ethnocentric, mystifying and oppressive. The project of liberal political philosophy, by contrast, is a normative one. Its response to the defaults of particular laws and institutions is not to throw its hands in despair and lament that religious freedom, or liberal justice, are 'mission impossible '. 4 It tries, rather, to articulate appropriate standards that can serve as benchmarks to evaluate (and reform) existing state of affairs.
The political theorists' approach is normative in a further sense. It seeks to identify the core values that should be protected by the law. As a result, it eschews purely descriptive or semantic approaches to legal terms. When it considers freedom of religion, it is not concerned with defining what religion is -an elusive project at best, as critical scholars of religion have amply shown. 5 Rather, it rejects any essentialist or semantic approach; and is concerned with identifying the core values that the law can properly express. 6 This is important because not all values can, or indeed should, be expressed by the law. Just as we would not want the law to express the whole of the value of 'the family', for example, so we would not want the law to capture of the whole of the value of religion. At best, the law will put forward an interpretive notion of 'the family', or of religion. That a particular law or theory does not capture what religion really is, therefore, is not, in itself, a sufficient objection to it. What matters is that the law, or the theory, expresses and protects the correct underlying values. It is at this more fundamental level that interpretive approaches must be assessed and evaluated.
With these preliminaries in mind, let me set out the aim of this paper. In what follows, I evaluate two influential interpretive theories of freedom of religion in political theory, before articulating and defending an alternative one. The first, which I call the substitution approach, argues that freedom of religion can be adequately expressed by a substitute category: typically, freedom of conscience. The second, the proxy approach, argues that the notion of religion should be upheld in the law, albeit as a proxy for a range of different goods. After showing that neither approach adequately meets crucial desiderata for an inclusive theory of religious freedom, I set out my preferred approach -the disaggregation approach -and defend it against the alternatives. I. THREE DESIDERATA Before describing the various approaches, I begin, in the first section, by identifying the desiderata that an inclusive theory of religious freedom must meet. I do so by surveying common criticisms of the existing law of religious freedom, both from critical and normative standpoints. They are all formulated against the implicit or explicit background of a theory of fairness as inclusiveness. Three different lines of critique have been developed: religious freedom is construed too narrowly to protect a range of valuable religious practices adequately; religious freedom is rooted in a sectarian view that religion itself is a special good; and the privileges of religious freedom treat non-religious citizens unfairly. Let me explain these in more detail.
1. Too narrow A number of critical scholars have argued that the legal treatment of religion still bears the marks of the ethnocentric, western, textualist, RELIGION IN THE LAW Protestant and belief-based understandings of religion that have accompanied the rise of the modern, secular euro-Atlantic state. As a result, religions that are more practice-, tradition-or ritual-based have fared badly under the liberal law of religious freedom. There are many illustrations of this distortion. Saba Mahmood and other writers influenced by Talal Asad have argued that liberal law struggles to protect embodied practices of piety -paradigmatically, Islamic veiling practices. 7 Winnifred Sullivan has shown how the elitist, textualist Protestant bias of US judges renders them incapable of capturing the popular, unruly, ritualised religiosity which she saw at work in the baroque funerary displays in a Florida cemetery. 8 Constitutional commentators have castigated a US Supreme Court decision which failed to see government logging plans through a Native American sacred territory as an infringement of religious freedom (Lyng v Northwest Indian Cemetery Protective Association, 1988). 9 This criticism is valid and important, provided it is clarified in two crucial ways. First, for reasons adduced in the introduction, it has to be presented as an interpretive not a semantic critique. In other words, the claim should not be that the existing law does not protect all that is religious, according to some ordinary-meaning, semantic understanding of the term. Rather, the claim is that the law fails to protect practices which exhibit those normative values -still to be specified -which are valuable in religion. Second, the critique must be formulated carefully. Some versions of it suggest that what is wrong with the liberal law of freedom of religion is that it protects only beliefs, and not practices. 10 If that is the claim, it is mistaken. While it is true that canonical liberal accounts of freedom of religion  University Press, 1996), pp. 196-199. 10 This criticism can, however be validly applied to the European Court of Human Rights (ECtHR) jurisprudence on freedom of religion. The Court has over-emphasised the distinction between forum internum and forum externum, notably in order to deny religious freedom protections to 'manifestations' of religious belief involving Islamic dress and symbols. See, eg., Carolyn Evans, 'The ''Islamic Headscarf'' in the European Court of Human Rights', 7 Melbourne Journal of International Law, 2006; Isabelle Rorive, 'Religious Symbols in the Public Space: In Search of a European Answer ', 30 Cardozo Law Review (2008-2009), pp. 2673-2674 -from John Locke onwards -distinguish (with good reason) between freedom of thought and belief on the one hand, and freedom of expression and practice on the other, this does not mean that the latter is left unprotected. That religious life has an essentially expressive dimension has been at the core of struggles over religious freedom for centuries. And the modern cases of religious exemptions from liberal laws have all concerned religious practices, not merely religious belief. This is true, for example, of Sabbatarian exemptions, conscientious objection to military service, and accommodation of religious dress in workplaces, to name just a few.
If that is correct, what does the Too Narrow critique amount to? In my view, it should be formulated as a more precise claim, as follows: the standard law of religious freedom tends to protect practices that are believed to be a matter of compulsory obligation. For example, Quaker pacifists see themselves as under a stringent duty to refuse to bear arms; Sabbatarians see themselves as obligated to honor the God-designed day of rest, and so forth. The Too Narrow critique here, then, is that a range of non-Christian, or non-Protestant practices, when they do not directly express a belief in a divine injunction, are less likely to be protected by the liberal state. One example of such a narrow view of religious practice is the ruling of some European judges, in Islamic veiling cases, that because the hijab is not a compulsory requirement of the Muslim faith, it does not fall under the protection of religious freedom. Another example is the Florida burial displays studied by Sullivan. Insofar as much of religious activity is rooted in orthopraxy rather than orthodoxy, the Too Narrow critique is effective. An important desideratum of an inclusive theory of religious freedom is that it is not narrowly biased in favour of obligation-based belief and practices.

Sectarian
The critique here points in a different direction. It is that freedom of religion protects a sectarian good -a good whose value is not universally recognised. It is a relic from an earlier age, one that has little justification in contemporary pluralistic societies where religion is only one of the things that people value. Some political theorists, for example, point out that the liberal state should protect generic capacities or moral powers, such as people's ability to form, revise and live by their conception of the good, whatever it might be. 11 The liberal state, then, should content itself with protecting generic rights and freedoms, such as thought, belief, expression, and association. It should not favour one specific way in which the good is pursued, namely, the religious life. The liberal state, to be properly inclusive of all its citizens, should only appeal to values and ideals that all can, in principle, recognise and adhere to. For this to be the case, the relevant values have to be abstract rather than concrete; general rather than specific; ecumenical rather than sectarian.
3. Unfair to non-religious people This criticism follows from the previous one. One consequence of the sectarian privileging of religion in the law, it is argued, is that non-religious citizens are unfairly treated by the law of religious freedom. This is particularly the case in the area of legal exemptions and accommodations. In modern, highly regulated states, citizens are subjected to a wide array of laws and regulations -ranging from health, safety and educational requirements to non-discrimination on grounds of gender or sexuality, or from parking and zoning regulations through to regulation on dress and uniform. Yet, typically, the law provides exemptions from these burdensome laws for religious citizens, but not for non-religious citizens.
As egalitarian theorists of religious freedom have pointed out, this is unfair. It is unfair, in particular, if citizens with comparable beliefs are denied the protections enjoyed by religiously-motivated citizens. Thus secular pacifists deserve the same level of protection as religious conscientious objectors -as the US Supreme Court recognised in celebrated Vietnam war cases (United States v Seeger (1965) and Welsh v United States (1970). And doctors refusing to abstain from lifeendangering abortions in Catholic hospitals should be able to appeal to their conscience in the same way as Catholic doctors refusing to perform abortions. 12 We have, then, three desiderata that an inclusive law of religious freedom must meet. It must not be (i) narrow, (ii) sectarian, or (iii) unfair to non-religious people. An immediate objection arises. Can these desiderata be met simultaneously? On the face of it, it seems 12 Taylor and Maclure, Secularism and Freedom of Conscience; Eisgruber and Sager, Religious Freedom and the Constitution; Micah Schwartzman, 'What If Religion Is Not Special?', Micah Schwartzman, 'Religion as a Legal Proxy', paper presented to the APSA Annual Meeting, Washington D.C. 28 August-1 September 2014 (available at SSRN http://ssrn.com/abstract=2416254). that they point in opposite directions: one cannot satisfy them all at the same time. Thus, if the law does not single out the seemingly sectarian good of religion (ii) it is unlikely that it will be able to protect all valuable religious practices (i). In what follows, I argue that two existing strategies -the substitution strategy, and the proxy strategy -are indeed vulnerable to this 'trilemma', in that they fail to reconcile the three desiderata. Either they are non-sectarian but too narrow; or they accommodate all valuable religious practices but are sectarian in their valuing of religion itself. I then suggest that my proposed alternative -the disaggregation strategy -succeeds in meeting the three desiderata at once.

II. THE SUBSTITUTION STRATEGY
The substitution strategy is favoured by prominent egalitarian theorists of religious freedom. 13 The general structure of the theory is clear enough. It is the human capacity for moral or spiritual agency, not for leading good lives with a determinate, perhaps religious, content, that grounds the respect that the state owes to persons qua persons. Rawls, for example, argued that what the liberal state protects is a generic ability or moral power: people's capacity to form, revise, and live by their conception of the good. More recently, Ronald Dworkin has suggested that religious freedom is not sui generis and is only one implication of a right to ethical independence. 14 But now egalitarian theorists of religious freedom face a problem. If religious freedom is broadened and dissolved into a general right of ethical independence, it will only ground a general right of moral freedom. It will not ground any special right of exemption. As Dworkin has argued, if the liberal state is to be neutral between different ways of pursuing the good (be it a life of religious piety, scholarly pursuit, or consumerist materialism), it becomes impossible to carve out a specific area of protection from the law. Yet while Dworkin is ready (at least on principle 15 ) to bite this particular bullet, 13  no other egalitarian theorist does so. 16 All seek to justify the principle of exemptions from the law. This means that they need to distinguish, among the vast range of commitments and conceptions of the good that people hold in pluralistic societies, those that deserve special protection. It is not sufficient to appeal to a 'thin theory' of the good à la Rawls or Dworkin. Egalitarian theorists need a more substantive theory of the specific good that is protected by freedom of religion in accommodation cases. They need what Charles Taylor, in Sources of the Self, called 'strong evaluations': evaluations about better or worse, important or trivial conceptions of the good life, views which are not reducible to mere preferences, desires and inclinations, but are rather the standards by which desires, preferences and inclinations can be judged. 17 Typically, egalitarian theorists find this good in the value of living by the demands of one's conscience. The good protected by freedom of conscience is the ability to act in accordance with one's perceived moral duties, which are seen as categorical. 18 Rawls, for example, justified the 'lexical priority' of freedom of conscience by hypothesising that the parties in the original position would under no condition agree to sacrifice the pursuit of their non-negotiable, binding commitments. Maclure and Taylor argue that religious convictions ought to be legally protected as convictions of conscience understood as 'meaning-giving beliefs and commitments'. 19 Convictions of conscience give a moral orientation to people's life, they are the fundamental beliefs and commitments that make it possible for persons to have a moral identity and to make moral judgments. Egalitarian theorists, then, substitute conscience for religion -or, to put it differently, they identify acting conscientiously as the core moral value that is traditionally protected by freedom of religion. Religion is the semantic term, but conscience is the interpretive value. If that is the case, then, the law should protect freedom of conscience, instead of freedom of religion. 16  How does the Substitution Strategy fare in relation to our three desiderata? I argue that it meets (ii) and (iii) but fails to meet (i). The Substitution Strategy meets (ii) the Non-Sectarian requirement. This is because, as Paul Bou-Habib has argued, freedom of conscience is rooted in a good that many people, both religious and non-religious, recognise as a valuable good. This is the good of living a life of integrity: a life where one's actions cohere with one's ideas about what is right for one to do. When people are forced to act against these convictions, they experience a loss of personal integrity, a feeling of self-alienation, a sense that their own actions violate the moral principles that define who they are. 20 The Substitution Strategy also meets (iii): it is not unfair towards non-religious citizens. Having identified the feature in virtue of which religious practices are protected, egalitarian theorists are then able to extend protection to non-religious practices which exhibit the same feature. So they are able to equalize relative comparable burdens. Eisgruber and Sager justify the draft exemptions granted to secular pacifists during the Vietnam War in this way. 21 Maclure and Taylor would treat secular and religiously-motivated vegetarians equally 22 ; while Schwartzmann would grant secular doctors in Catholic hospitals the right to act as their conscience dictates in tricky abortion cases. 23 All these practices are relevantly similar and comparable because they are rooted in stringent obligations of conscience.
However, the Substitution Strategy fails to meet desideratum (i): it remains too narrowly tailored to a certain kind of obligation-based practice. As a result, it is vulnerable to the objection we set out above, namely, that many valuable religious practices are not matters of obligation. Crucially, it does not mean that they thereby fail to express the core value that religious freedom expresses. Assuming -with Bou-Habib -that this core value is ethical integrity, it is obvious that many (though not all) 24 24 Here I have in mind a range of activities of religious organisations-such as the hiring and firing of staff, the provision of public services, etc. There is a tendency to define all these activities as 'religious' (Cf. Hosanna-Tabor, 2012 and Burwell. V. Hobby Lobby, 2014). In my view, the rights of religious associations should fall under the right of associations more generally. I make the full case elsewhere. ethical integrity, even though they are not rooted in conscience. 25 Consider, for example, the following practices: the ingestion of peyote in Native American ceremonies, the wearing of hijab by pious Muslim women, or funerary displays in Sullivan's cemetery. They are valuable religious practices: valuable because they allow individuals to live with integrity. But they may not be demands of conscience strictly speaking. 26 Of course, there has been a tendency, among legal practitioners, to re-describe these practices in the language of conscientious obligation, so as to accommodate them under the label of freedom of religion. But this has been rightly castigated as a Protestantisation of non-Protestant religion, and this re-description will not at any rate be relevant to all contested instances. Paradigmatic of these difficulties is, again, the Lyng case. However hard one tries, it is difficult to reconcile the intuitions that the protection of sacred lands is a core dimension of Native American 'religious' life (insofar as it allows members to live with integrity) with the egalitarian postulate that what freedom of religion protects is the value of conscience. As we shall now see, the proxy strategy offers a solution to this particular problem.

III. THE PROXY STRATEGY
Let us first set out the general structure of the proxy strategy, as recently articulated by Andrew Koppelman. 27 The proxy strategy accepts the general criticisms leveled against the concept of religion in the law, and endorses the shift from a semantic to an interpretive approach. In particular, Koppelman argues that there is no essence to 25 Bou-Habib himself seems to concede that the value of integrity cannot be exhausted by conscience. He writes: 'as well as depending on compliance with perceived duty, integrity may depend on one's attempting continuously (which is not to say incessantly) to discover what one's duties in fact are. It seems to me that religious conduct that aims at achieving communion with a divine will or ultimate reality may have a claim to accommodation under this heading. The ingestion of peyote in the sacramental worship of NAC is an example'. Bou-Habib, 'Theory of Religious Accommodation ', p. 123 religion, and that there is no directly identifiable good associated with religion that the law should aim to identify and protect. Therefore, the proxy strategy rejects the reductionist move of the substitution strategy, and points out that whatever is protected under the label 'religion' cannot be reduced to the value of conscience. Yet the fact that the term 'religion' does not refer to one identifiable good but, rather, to a loose cluster of different goods, does not mean that it is a useless legal term. On the contrary, the use of the term 'religion' can be seen, in Wittgensteinian fashion, as a fairly stable linguistic practice exhibiting sufficient 'family resemblance' to protect a broad range of activities. Inevitably, like all legal proxies, the term 'religion' will be sometimes under-inclusive (when it fails to protect secular conscience) or over-inclusive (when it protects religious activities which do not express any particular good). But, Koppelman argues, it is still better than any other alternative. He draws an analogy with compulsory driving licenses. The aim of the law is to ensure safe driving, and it uses driving licenses as a proxy. Even though there is no guarantee that those holding a valid driving license, and only they, will be safe drivers, the proxy is still the best way to approximate safe driving. 28 How well does the Proxy Strategy fare in relation to our three desiderata? I argue that it meets (i) but it fails to meet (ii) and (iii).
First, because it sees religion as a set of loosely connected practices rather than as an essence, it is able to accommodate the critique that standard understandings of religion are too narrow. Koppelman argues, following Greenawalt, that, in practice, courts have rarely found it difficult to identify what religious practices are. Some judges may hold a narrow, Protestantized view of the demands of religion, but this is a feature of their idiosyncratic training and bias, not of the polysemic and flexible notion of 'religion' that has become prevalent in the law. On the family resemblance view, there is no reason why they should be biased in this way. On the contrary, a vague notion allows an ongoing conversation about the meaning of religion, and suggests the inclusion of unusual or minority religious beliefs and practices by analogical reasoning. 29 Thus practices that have no connection to a transcendental god or to individual conscience 28 Koppelman, 'Neutrality and the Religion Analogy', p. 12. 29 See also Greenawalt, Religion and the Constitution, Part 1, pp. 139-142.
-from Buddhist rituals to Native American traditional practices -can be brought under the protection of freedom of religion. The Lyng case -the 'blind spot' of the US First Amendment definition of religion -was wrongly decided only because judges took too narrow a view of what counts as a burden on religion. 30 Second, however, it is not clear that the Proxy Strategy can escape the charge of being sectarian. This is not only (or necessarily) because it protects a good that is a good only for some but not for others. It is, more fundamentally, because it does not aspire to protect a clearly identifiable, valuable good (or bundle of goods): it is not clear what religion is a proxy for. Consider, in this context, the limits of Koppelman's analogy with the driving license case. In this case, the law identifies a valuable aim -safe driving -and sets out a proxy to achieve it, however imperfectly. The structure of freedom of religion, however, is different. The problem is not that the proxy achieves the good imperfectly or indirectly (this, after all, is the point of proxies). The problem, rather, is that it is not clear that the law protects a good at all. If religion, in the ordinary-meaning semantic sense, is a complex bundle of things, not all of them good, then it is not clear that the legal protection of religion actually expresses any interpretive value at all. 31 Third, the proxy strategy does not adequately protect non-religious practices which, on the face of it, are as valuable as religious practices. Consider again the Lyng case. Koppelman's reading of the case would suggest that as long as a practice can be called 'religious', in the ordinary sense of the term, it will enjoy the protection of religious freedom. But, as critics have pointed out, the First Amendment's singling out of religion has led to a radical distortion 30 Koppelman, 'Neutrality and the Religion Analogy', p. 11. 31 To clarify one possible misunderstanding. When I say that religion does not protect only 'good' things, I do not refer to religious practices which, say, grievously infringe on the rights of others. When liberals talk about good and bad religion, they usually have in mind such cases. Religion is not religion, they claim, when it demands that abortion clinics or shopping centres be bombed, or that children be denied access to life-saving medicine. I disagree. Insofar far as these actions are expressive of individual's ethical integrity, they correctly express religious values. This does not mean to say, of course, that they should ipso facto be tolerated: in liberal societies, the pursuit of the good is constrained by the demands of the right. When I say that some religious activities are not 'good', what I have in mind, instead, are practices that are claimed as exercises of religion by religious organisations, yet only have a tenuous connection to the normative value of ethical integrity. When religious organisations spend money, run businesses, hire staff, and provide secular services, it is not clear that they ipso facto express the value of ethical integrity. On the disaggregation theory I favour, as I suggest below, such activities are best protected under generic rights of freedom of association, with no special concern for the religious nature of the organisations. The good pursued, here, is associative freedom, not freedom of religion.
of Native American practices. Aboriginal peoples typically see all their daily practices as suffused with their community way of lifepraying, but also singing and dancing; respecting sacred burial sites, but also food gathering, hunting and fishing. When the Pueblo Indians fought for recognition under the First Amendment, they were forced arbitrarily to separate their way of life into 'religious' and 'non-religious' domains. One effect of the singling out of religion is arbitrarily to favour certain types of activities over more broadly defined 'cultural' or 'traditional' practices. 32 Such distortions are less likely to occur under constitutional settlements, such as the Canadian constitution, where religion is not pro tanto more valued than culture -insofar as both allow their members to live with integrity. 33 The worry here is that the privileging of religion over culture in the US constitution is unfair to those with meaning-giving, integrityprotecting cultural commitments.

IV. THE DISAGGREGATION STRATEGY
One possible way in which we could address the limitations of the two strategies surveyed above is by combining them. Many scholars seem to veer towards this mixed position, which consists either in a dual-tracked 'freedom of religion and conscience', 34 or a more radical protection of 'people's sense of obligation and … core beliefs and identity.' 35 These are promising strategies. However, in my view, they face a dilemma. Either they do not go far enough and, by singling out freedom of religion as a specific kind of freedom, they reproduce the same difficulties as the proxy strategy. Or they suc-32 Wenger, We Have a Religion. 33 For a comparison, see Lori G. Beaman, 'Aboriginal Spirituality'. Beaman complains that 'In Canada, where group rights and the correcting of systemic disadvantage are constitutionally possible, aboriginal claims are framed as treaty rights… resulting in the minimization or marginalization of issues concerning religious freedom'. It may be the case that Aboriginal rights are not given sufficient weight, but the argument that this is because their ''religious freedom'' has been denied assumes what has to be demonstrated, namely, that the category of religious freedom is the most suitable to protect Aboriginal rights. 34 Micah Schwartzmann, ''Religion as Legal Proxy''. See also Koppelman's rejoinder: ''Religion as a Bundle of Legal Proxies. Response to Micah Schwartzman'', San Diego Law Review 51 (2014). As Schwartzmann notes, the double protection of religion and conscience is in fact the actual law in many jurisdictions, including in the text of international conventions, such as the ECHR. For an early defense in the US context, see Rodney K. Smith, ''Converting the Religious Equality Amendment into a Statute with a Little ''Conscience'''', BYU L. Review (1996), pp. 645-688. 35 Alan Patten, ''Three Theories of Religious Liberty'', paper presented to the APSA Annual Meeting, Washington D.C. 28 August-1 September 2014, p. 11. ceed in providing a non-sectarian description of the plurality of values that underpin freedom of religion but, in that case, they become indistinguishable from the disaggregation strategy. In what follows, therefore, I go straight to sketching my own version of disaggregation, without ruling out the possibility that a radical mixed strategy, if suitably formulated, would be equally compelling.
The disaggregation strategy builds on an innovative proposal made by James Nickel in a 2005 article entitled 'Who Needs Freedom of Religion?'. 36 The gist of Nickel's argument is that freedom of religion does not need to be singled out by the law as a special freedom. Religious activities and practices can be adequately protected through generic liberal rights of belief, thought, expression, privacy, association, conscience, and so forth. This means that religious freedom is derivative, like scientific freedom or artistic freedom: it is implied and entailed by basic liberal freedoms, and justified on the same grounds as them. The fact that a liberty is derivative does not mean that it is less important. While there is much to learn from Nickel's approach, it is, in my view, deficient in two ways. First, Nickel suggests that freedom of religion can be removed from the list of derivative rights. The disaggregation strategy, by contrast, makes no such claim: it merely aims to interpret the notion of religion in law (regardless of whether the category of freedom of religion is upheld or not). Second, Nickel does not specify in sufficient detail the different dimensions of religion which freedom of religion protects; nor does he explain how his reductionist approach to religion relates to other normative values such as equality, nondiscrimination and non-establishment. 37 The starting point of the disaggregation strategy is to suggest that different parts of the law should capture different dimensions of religion for the protection of different normative values. Consider the following, non-exhaustive list of legally-relevant dimensions of religion.
1. Religion as a conception of the good life 2. Religion as conscientious moral obligation 3. Religion as key feature of identity 4. Religion as mode of human association 5. Religion as vulnerability class 6. Religion as totalizing institution 7. Religion as inaccessible doctrine This plurality of dimensions of religion has been obscured by the two strategies we have considered so far. The proxy strategy tends to bundle them all together and make them the normative basis of a special kind of freedom, 'religious freedom'. Yet, it is unclear that freedom of religion is the right normative framework to capture all these dimensions. The substitution strategy, for its part, tends to collapse religion into conceptions of the good and conscience and, as a result, has little to say about the other dimensions. In what follows, I shall focus on dimensions 1-4, as they are directly relevant to the notion of freedom of religion which has concerned us here. 38 Recall that, on the interpretive theory that I favour, it is not enough simply to say 'religion is X and Y'. What is required is to identify the specific normative values which makes X or Y legally relevant. Just saying that a practice or institution is multi-faceted and internally complex, and irreducible to anything else (as is surely the case with religion) does not mean that it must be recognized as such in the law. 39 But, by parity of reasoning, nor does disaggregating the various empirical dimensions of religion in itself provide a reason for legal cognizance of any of them. So we need to know what kind of good is being protected in every case, and the good cannot be assumed to follow from the mere description of the empirical dimension of religion. With this in mind, let me briefly discuss the four first cases in turn. 38 Elsewhere, I show that seeing religion as a vulnerable class is crucial to theories of equality and non-discrimination; whereas seeing religion as a totalizing institution and inaccessible doctrine helps account for the value of non-establishment and (minimal) secularism. 39 This is the problem with Michael McConnell's argument, which derives from the fact that religion is uniquely complex the conclusion that it should be specially protected. But this is a non sequitur. He writes: ''Religion is a special phenomenon, in part, because it plays such a wide variety of roles in human life: it is an institution, but it is more than that; it is an ideology or worldview, but it is more than that; it is a set of personal loyalties and locus of community, akin to family ties, but it is more than that; it is an aspect of identity, but it is more than that; it provides answers to questions of ultimate reality, and offers a connection to the transcendent; but it is more than that. Religion cannot be reduced to a subset of any larger category. In any particular context, religion may appear to be analogous to some other aspect of human activity -to another institution, worldview, personal loyalty, basis of personal identity, or answer to ultimate and transcendent questions. However, there is no other human phenomenon that combines all of these aspects; if there were such a concept, it would probably be 1. When religion is conceived as a conception of the good, it should be tracked by generic liberal freedoms such as freedom of belief, thought, speech, and so forth. Freedom of religion, here, is simply derivative of a broader value, which itself justifies those generic freedoms -liberals usually refer to individual autonomy, or self-determination, or the capacity to pursue and develop the conception of the good that one in fact holds. The religious life, on this view, is one of the many ways for individuals to exercise their first 'moral power', to use John Rawls's phrase. For many purposes of legal protection of religion, the content of religion does not need to be specified, because freedom of religion is merely derivative of this more general right of ethical independencethe right to form, develop and pursue one' 2. Special legal protection for religion as conscientious obligation, in turn, is justified by more specific appeal to the value of integrity. As we have seen, we have good reasons to think that integrity is a non-sectarian, ecumenical value. There is a specific harm caused by being prevented from doing what you feel is right, as opposed to what you would prefer to do, or what would make you happy. 42 But while ethical integritydoing what you feel is right -is the correct interpretive value, it is not the case that it is only exhibited in conscientious action. Arguably, people act with integrity when they follow strongly valued practices, when they express 'core beliefs and identity', 43 to use Alan Patten's phrase, regardless of whether this is demanded by their conscience or not. 3. Special legal protection for religion as key feature of identity, therefore, can also be justified by appeal to integrity. The argument that, for purposes of exemptions, religion should be interpreted, not only as conscientious belief but also as one form of meaning-giving cultural commitment, has often been resisted by political theorists. 44 Yet this suspicion is rooted in two mistaken assumptions. The first is that religion essentially is different from culture. This may well be the case -and it is easy to provide theological, sociological, anthropological, political, and phenomenological accounts of this difference. The question that preoccupies us, though, is not a semantic or descriptive but, rather, an interpretive question. When we think of legal exemptions, the question we ask is as follows: What kinds of commitment are so important to people that their integrity would be threatened, were they prevented from acting on them? The onus is on critics to explain why certain kinds of cultural commitments would be less important to people than religious commitments. The second mistaken assumption is that if the law aims to protect all integrity-promoting practices, whether religious, cultural, ethical, etc., it will protect everything, and therefore nothing. There is something to this worry, but it is exaggerated. First, ethical integrity is defined so as to exclude the protection of trivial practices that only hold a marginal 42 See also Chandran Kukathas who writes that individuals have a basic human interest 'in living in accordance with the demands of conscience. For among the worst fates that a person might have to ensure is that he is unable to avoid acting against conscience -that he be unable to do what he thinks is right' in The Liberal Archipelago, p. 55. 43 Patten, 'Three Theories of Religious Liberty', p. 11. 44 See Avigail Eisenberg, 'Religion as Identity', paper presented to the APSA Annual Meeting, Washington D.C. 28 August-1 September 2014. place in an individual's conception of the good or identity. One advantage of the approach is that it has the resources to accommodate the wearing of hijab as well as aboriginal practices; and to refuse accommodation to wearers of clown's hats. There will inevitably be hard cases, but -I would argue -they are not intrinsically more troublesome than those generated by alternative approaches. Second, it is not clear that the proposed approach would be less easily administrable than existing legal practice. Judges, when dealing with freedom of religion and freedom of conscience cases, already apply complex tests of sincerity, centrality, meaningfulness, and so forth. They do so, usually, in order to define whether a practice is properly religious or not. The disaggregation approach, by contrast, bypasses the need to settle on the semantics of what is religious and what is not. 45 Third, the more specific fear of exemption proliferation can also be alleviated. We can assume that only individuals with integrity-upholding commitments will go to the trouble of requesting accommodations in the first place. Furthermore, because (on my theory at least) ethical integrity does not carry the pro tanto weight conventionally accorded to claims of freedom of religion and conscience, it will only justify exemptions from a narrow range of laws and regulations (amongst which most cases of exemptions from dress codes). 46 4. Religion as a mode of human association. Consider the liberal freedoms that are broadly connected to this: freedom of association itself, free-45 In private correspondence, Andrew Koppelman has put the following objection to me. 'During Prohibition, the Volstead Act exempted sacramental wine. No attempt was made to examine individual Catholic priests and parishioners to determine the depth of their conviction. If 'religion' is not cognizable, it is hard to imagine how that could have been done… That is why proxies are indispensable'. This is an important objection. In response, I would say two things. First, this is better understood as a right of collective exemption -a right held by the church as a whole, so on my theory it would fall under freedom of association (and the use of sacramental wine could be linked to a core purpose of the association). Second, even if construed on the model of individual exemption, the argument would be that the use of sacramental wine during church services is central to the integrity of the communicant. It is quite possible that the most effective way to account for this is directly to appeal to what we know about the importance of religion in general, and sacraments in particular. But this is different from saying that it is freedom of religion as such that is doing the work here. Consider the following analogy. Take the case of drug control. If doctors can show that they need to use other illegal drugs, in a controlled environment, for important medical purposes, then doctors should get an exemption. By analogy, if a group can show that they use drugs, in a controlled environment (ie, without subverting the aims of the law which targets addiction etc), for central religious purposes, then this group should also get an exemption. We don't need freedom of religion to make this claim -even if 'religion' is useful as an explanation of the activities of the group (just as a 'medical' is useful as an explanation of the activities of doctors -but they don't need to invoke medical freedom as a special right). 46 This argument does not follow from the conceptual strategy of disaggregation but needs to be backed by a full theory of egalitarian justice. I develop an argument to this effect in Liberalism's Religion, manuscript in progress. dom of expression and speech, freedom of peaceful assembly, economic freedom, political participation. Arguably, these freedoms do not flow -at least, not directly or exclusively -from the value of ethical integrity (so they do not justify the extension of individual rights of accommodation to corporate rights of exemption). Yet such freedoms are crucial to the vitality of churches and religious associations, and to the diverse activities they take as central to the pursuit of their purposes. Some religions emphasize preaching and proselytising, others, charity work, yet others, successful business activities. Some focus on the preservation of community ways of life; others on the visible display of signs of religious membership; yet others, on personal ethics and conscience. If we take a broad view of the scope of religious activities, we can 'avoid the misconception that we have to find all protections for religious activity within a phrase like 'the free exercise of religion' or 'freedom of thought, conscience and religion'. 47 So the suggestion here is that even if we grant religious freedom as such lesser weight than existing approaches, we can still offer great scope for protection of religious practices and activities. The disaggregation strategy does not assume that all religious activities and practices must be protected in the name of freedom of religion. In particular, it rejects the tendency, within existing approaches, to capture the interpretive value of religion exclusively in terms of freedoms of belief, thought, or conscience.

V. CONCLUSION
The disaggregation strategy, I hope to have demonstrated, improves both on the substitution and proxy strategies. Unlike the proxy strategy, it does not single out freedom of religion as a special freedom protecting a loose cluster of activities and practices. Unlike the substitution strategy, it does not assume that all that freedom of religion protects should be protected under the label of freedom of conscience. As a result, the disaggregation strategy meets the three desiderata which I outlined at the start of this paper. It is not narrow, because it protects a broad range of associational, expressive activities, including those religious activities that are not obligations of conscience. It is not sectarian, because it is rooted in the ecumenical 47 Nickel, 'Who Needs Freedom of Religion? ', p. 951. value of ethical integrity, and in the normative justifications for generic liberal rights such as speech and association. And it is not unfair to non-religious citizens, because it respects their meaninggiving commitments, whether they are conventionally religious, cultural, or philosophical. The disaggregation approach, in a word, is religion-blind without being religion-insensitive, because it sees religion, not as a specialised and self-contained area of human belief and activity, but as a richly diverse expression of life itself.