Imagine a society the great majority of whose members are strongly committed to religious toleration. They have read Locke, Bayle, Voltaire and other canonical proponents of toleration. They have also noticed the bloodshed and suffering caused by religious intolerance both in the past and in their own age. In light of their reading and experience, they are fully committed to religious toleration: they believe that all citizens should enjoy toleration in religious matters.Footnote 1 Their society is democratic so that, unlike their forbears who had to plead the case for toleration with monarchs or oligarchs, they can fashion their community as they see fit. How then should they act on their shared commitment to toleration?

The obvious answer is by framing laws and designing institutions that secure religious toleration for the members of their society. We can therefore suppose that they will enact laws that secure for citizens the right to pursue their own religion in their own way and that simultaneously impose upon each citizen the duty to allow others to pursue their own religion in their own way. They may go further and entrench the freedoms demanded by toleration in a constitutional bill of rights, perhaps modelled on the ECHR, particularly article 9. They will also ensure that their public institutions are even-handed with respect to people as religious believers, including as ‘unbelievers’, so that no citizen’s opportunity to participate in public life is prejudiced by his religious convictions (unless his religious convictions themselves forbid that participation). As well as having read the classics of toleration, they may have read John Rawls’s Political Liberalism (1993), in which case they may resolve that no use should be made of political power, of any kind, purposely to promote or to disadvantage a particular faith or variant of faith or lack of faith.

What adjective should we use to describe the political community that these individuals create? The obvious answer is ‘tolerant’, and their community is tolerant not primarily because of the attitudes of its population but because toleration is instantiated in, and secured by, the society’s laws and institutions. We might contrast this society with a second that models intolerance. The intolerant society permits the preaching and practice of only a single variant of a single faith and forbids any manifestation of any other religious belief. It might also burden ‘heretics’ and ‘infidels’ with penal taxes, exclude them from public office, and brand them publicly as degenerate. Or it might simply expel or execute them.

But consider now a third society. Its members are hesitant about the merits of toleration. They fear the suffering that might accompany state-sponsored intolerance but they also fear for the survival of their own faith if other faiths were to be tolerated. Because of their ambivalence, they refrain from putting in place any of the laws and institutional arrangements that characterise the first society. Citizens remain free to tolerate or not to tolerate the religious lives of their fellows as they see fit. The society’s population is divided between two faiths. Some in each faith group behave tolerantly towards members of the other. But many in each group do not; they violently attack members of the other group, desecrate their places of worship, use terror tactics to deter them from assembling for collective acts of worship, and discriminate against them at every opportunity.

Given my descriptions of these three societies, it seems just obvious that the first society is more tolerant than the second, while the third falls somewhere in between the two. Yet, as Peter Balint (2012) points out, if we adhere to the academic orthodoxy on toleration, that ranking would seem to be mistaken. The first society will have little claim to be more tolerant than the second, let alone the third. According to orthodoxy, toleration has two necessary or ‘possibility’ conditions. First, toleration entails disapproval or dislike; in the absence of some sort of negative appraisal, the tolerator has no occasion to tolerate the tolerated. That is how toleration differs from endorsement, acceptance or indifference. Secondly, a person tolerates only if he has the power to do otherwise. If A disapproves of B’s conduct but is powerless to affect it, A can neither tolerate nor not tolerate B’s conduct. Toleration is an option for A only if intolerance is too.

Now consider how the first society performs in relation to these possibility conditions. That society, on my account, is tolerant in virtue of the laws and institutional arrangements it possesses. Yet laws and institutional arrangements cannot themselves disapprove or dislike. Their authors can disapprove or dislike but that suggests it is those authors who will be tolerant (or intolerant) rather than the rules and arrangements they establish. A similar objection can be mobilised in relation to power as a possibility condition of toleration. Laws and institutions (institutions considered as inanimate structures without reference to those who staff them) cannot wield power. They can be the instruments only of somebody’s power and it is that somebody, rather than the laws or institutional structures themselves, who will be the source of toleration or intolerance.

The power condition is at odds with my portrait of the tolerant society in another, potentially more damaging, way. That society has rules and arrangements that deprive its citizens of the option of intolerance; citizens are not allowed to interfere with or impede the religious freedom of their fellows. That prohibition, in removing from citizens the freedom to interfere in one another’s religious lives, may seem to remove from them the power that is a possibility condition of toleration. If we deprive people of the possibility of behaving intolerantly, we also deprive them of the possibility of behaving tolerantly. So, rather than incarnating toleration, the first society would seem to have eliminated it.

The first society might still claim to be less intolerant than the second (‘intolerant’) society but, as Balint points out, not being intolerant is not always the same as being tolerant. However, judged according to the orthodoxy, the first society will be less tolerant than the third. In the third society, citizens remain legally free to impede one another’s religious activity and, for that reason, toleration remains an option for them: they can choose to refrain from using their power to impede or persecute those whose faith or denomination they reject. If some citizens opt to be tolerant, that will render their society more tolerant than the first society, even if the citizens who tolerate are only a small minority of the society. The third society is therefore more tolerant than the first, and the first is no more tolerant than the second.

That conclusion is plainly fatuous and it is no less fatuous for the simple logic with which the orthodox conception of toleration delivers it. No one outside the academy is likely to recognise the third society as more tolerant than the first, or the first as no more tolerant than the second. Even if we were to suppose that every last person in the third society opted to behave tolerantly, the first society would still outperform the third on the score of toleration since it guarantees toleration for its citizens, while the third leaves each individual’s toleration entirely at the mercy of others. The toleration secured by the first society is therefore superior to that available in the third precisely in the way that freedom conceived as non-domination is superior to freedom conceived as mere de facto non-interference (Pettit 1997).

How, then, can we avoid the absurdity of accounting the third society more tolerant than the first? The answer is by recognising that toleration can be approached from more than one perspective. As well as conceiving it from the perspective of tolerator or tolerated, we can conceive it from a third-party perspective. We do so when we think about the sort of toleration that ought to obtain amongst the members of a society. Consider the position of the citizens of the first society when they are on the point of constructing rules and arrangements for their society. They take a view on the toleration citizens should extend to one another and they enact laws accordingly. If they believe citizens should enjoy a generous measure of toleration, they will establish an arrangement that requires citizens to extend that toleration to one another. However, in establishing that arrangement, they will not themselves engage in toleration. They are not tolerating anyone or anything; they are simply deciding upon and establishing rules that instantiate the demands of toleration. Once those rules are in place, their society will be tolerant not because it, as a society, extends toleration to someone or something but in virtue of the mutual toleration it secures for its members. Moreover, the toleration it secures will be ‘horizontal’ toleration amongst its citizens rather than ‘vertical’ toleration between ruler and subjects.

Although I associate this third party perspective particularly with liberal democracy, it is a perspective that could have been assumed by a post-Reformation monarch and its third-party nature is perhaps more readily apparent in that case. If a monarch were ‘an enlightened prince’ of the sort contemplated by Kant (1991, p. 58), he would regard the religious lives of his subjects as none of his business and allow them complete freedom in matters of religion. Having decided upon his own position, he might then direct his attention to his religiously divided subjects, determine that they should be tolerant of one another, and legislate accordingly. In subjecting his population to a regime of mutual toleration, he would occupy the role of neither tolerator nor tolerated; rather he would stand in a third-party relation to his subjects and the toleration he secured would be, as before, horizontal toleration amongst his subjects rather than vertical toleration extended by himself to them.

Unlike the ‘enlightened prince’ who stands outside the tolerant regime he creates for his subjects, the citizens of the first society will be subject to the rules of toleration they themselves create. As agents they will be obliged to comply with those rules and as patients they will benefit from the compliance of others. But we should not confuse their two roles: deciding upon the rules of the game is not the same as playing the game. In the latter role, citizens will be both tolerators and tolerated. In the former, they will not; rather they will stand in a third-party relation to their particular selves as prospective participants in the game they create.

That is not to say that anyone who legislates in relation to toleration must, or will, do so from a third party perspective. Post-Reformation monarchs typically legislated on religious matters from a first person perspective: they decided whom, if anyone, they would or would not tolerate and legislated accordingly. A democratic majority might similarly determine whom it will and will not tolerate and legislate accordingly. In both of these cases, ‘political toleration’ assumes a vertical form and conforms to a traditional tolerator-tolerated model.

I do not claim therefore that, in a democratic context, toleration must be provided for in a third-party way. I do claim, however, that a third party approach is more consistent with the ideas of democratic equality and liberal democracy. Balint is right to point out that even liberal democracy can engage in ruler-subject toleration; it can do so when it confronts illiberal groups, such as the white supremacists instanced by Balint. But, if we insist that political toleration can take only a ruler-subject form, we shall join Glen Newey (1999) and David Heyd (2008) in driving toleration to the margins of liberal democracy rather than locating toleration at its centre.

If we think about toleration in a third-party way, do negative appraisal and power lapse as possibility conditions of toleration? Both remain preconditions of toleration but not quite in the way they do for tolerator and tolerated. A regime of toleration will be necessary and possible only for a population characterised by some form of disagreement, disapproval or dislike. If that negative appraisal were to disappear, so would the need for toleration. Similarly, a regime of toleration would be unnecessary if, in its absence, a population were somehow powerless to be intolerant of one another. But while negative appraisal and power (not) to tolerate remain possibility conditions of (necessary ‘circumstances’ for) toleration, they are not features of a regime of toleration itself or of citizens as sponsors of that regime. Balint’s supposition that we must be able to find ‘sites’ of disapproval and forbearance somewhere within the regime itself, or amongst its sponsors, is therefore misplaced.

My original claim (2007, pp. 388–9) was that a liberal democratic regime of toleration provides for toleration not by enabling people to engage in acts of toleration but by holding intolerance at bay. My second claim was that, in so doing, the regime secures what matters about toleration, which is that people should not suffer intolerance rather than that they should engage in wilful acts of toleration. We could hold, contrary to that claim, that the value of toleration lies in its enabling tolerators to display the virtue of toleration, just as we might hold that the value of poverty relief lies in its enabling the rich to display the virtue of benevolence. But neither option has much plausibility or appeal.

I now want to add a further claim: there is no compelling reason why, in describing a liberal democratic regime of toleration, we should abjure the language of ‘toleration’ and speak only of the ‘absence of intolerance’. We can reasonably speak of people, under a regime of toleration, being legally required to behave tolerantly towards their fellows, where ‘legally required’ means that they have a legal obligation so to behave and ‘behave tolerantly’ describes their refraining from preventing acts of others that they find objectionable but that others are legally entitled to perform. If toleration entails forbearance and if forbearance remains ‘forbearance’ when it is legally required, why should toleration not also remain ‘toleration’ when it is legally required? Law does not render people incapable of doing what it prohibits; rather it requires them not to do what they remain capable of doing.Footnote 2

Why, finally, should we persist with the language of toleration in describing liberal democratic arrangements? Why not abandon it and make do with the language of freedom? Two considerations argue against that proposal. First, toleration relates to a wider range of issues than is normally captured by the language of freedom; it relates, for example, to issues concerning the use of public resources and to the prohibition of direct and indirect discrimination. Secondly, the political ideal of toleration arose in response to circumstances of plurality and disagreement, circumstances that also shaped liberal democratic thinking. It would be perverse to veto any mention of toleration in describing political arrangements that have been inspired by that idea and that have sought to realise it more fully than any political alternative.