I offer a new theory both of separation of state and religion (secularism) and of freedom of religion (exemptions). I argue that we should dispense with the semantic concept of religion in legal and political theory and, instead, “disaggregate” religion—a complex notion covering conceptions of the good, conscientious obligation, constitutive identity, mode of human association, vulnerability to discrimination, totalizing institution, and inaccessible doctrines.

In what sense, if any, should there be a separation between state and religion? Instead of drawing on vague notions of neutrality or secularism, I identify three central liberal values, and map them onto three specific dimensions of religion or the good. With regard to the three central liberal values, the justifiable state appeals to the idea that the state should only be justified by reasons that are accessible to citizens. The inclusive state honors the equal status and citizenship of all, while the limited state respects individual self-determination in private matters. Each of these three values, in turn, picks out a different feature of disaggregated religion: religion as nonaccessible, as divisive, and as comprehensive. Disaggregating religion in this way allows me to specify that religion is not uniquely special: nonreligious ideologies and practices can also be inaccessible, divisive, and comprehensive. This means that the state need not be separate from religion when religion is not divisive, inaccessible, or comprehensive.

When we think about the justification of exemptions from the law, we need to pick out a different feature of religion. The ethically salient category here is integrity-protecting commitments (IPCs). These further subdivide into obligation-IPCs and identity-IPCs. I articulate two principles of fair exemptions, which broadly map onto each IPC. Under disproportionate burden, the pursuit of some state regulatory interest makes it impossible for some citizens to fulfill an obligatory requirement of their faith or culture, yet makes it possible to relieve them of the burden without excessive cost. Under majority bias, minority citizens are unable to combine the pursuit of a socially valuable opportunity with an identity-IPC, whereas the equivalent opportunity set is available to the majority. I argue that exemptions are only legitimate if they do not violate basic rights and do not unreasonably shift burdens onto others.

I also argue that justifying religious collective rights is not as straightforward as many have thought. Even though the state does not share sovereignty with other institutions, it must respect associational autonomy. I then apply my theory of disaggregation to the general puzzle of collective religious exemptions from antidiscrimination laws. I argue whatever rights religious associations have should be derived from the liberal value of freedom of association. However, freedom of association itself is an internally complex idea. I disaggregate the values it protects, so as to justify some of the collective rights claimed by religious groups. I set out two salient associational interests: coherence and competence interests. While many associations can appeal to coherence-related interests to defeat the application of some general laws, only some can, in addition, appeal to competence-related interests. Disaggregating associational interests in this way allows me to explain why religious associations (but not only they) can have some latitude in choosing their personnel—including via exemptions from nondiscrimination legislation. By contrast, nonidentificatory commercial associations do not have the relevant interests, and therefore should not benefit from freedom of religion protections.

My preferred conception of liberal justice is one that allows only a restrictive scope for religious exemptions from general laws, in particular when these exemptions undermine the rights of others. But this conception of justice is only one among a family of reasonable conceptions of liberal justice. There is more permissible variation in the justice of state-religion relations, it turns out, than most Western liberals have so far recognized. In an attempt to come to terms with the scope and depth of reasonable pluralism about liberal justice itself, I present liberal theory as a dualist normative theory. I describe fundamental questions about whether a state is liberal at all as questions of liberal legitimacy. A state is legitimate if it meets a number of desiderata concerning the justification of its constitutional framework, the inclusiveness of its political arrangements, and its entrenchment of key liberal principles. But agreement on the basic foundations of this minimal secularism is compatible with large disagreements about justice, and minimal secularism can accommodate both a progressive liberal state (Secularia) and a conservative liberal state (Divinitia).

This, I hope, is a response to those who worry that liberalism, far from being a potentially universal framework for the democratic and fair resolution of conflicts about religion, is in fact the sectarian, comprehensive ideology of Western progressives—the religion of liberals.