Rawls believes that a liberal view of human beings and liberal values can find, and need to find, support in a limited plurality of larger worldviews — what he calls "reasonable" comprehensive doctrines. Among these are his own secular Kantianism, but also certain versions of Christianity and, indeed, of Islam. So unlike many British liberals — for example, Polly Toynbee — Rawls sees in certain kinds of religion, not enemies, but important allies.

Nevertheless, Rawls believes that public discourse — the discourse of parliament and the law courts, and perhaps also of public rituals — should not involve religious references, but should be conducted in terms of "public reason". Public reason comprises the set of liberal moral values and such anthropological tenets as are necessary to make sense of it, upon which various "reasonable" comprehensive doctrines converge. That is to say, it comprises the "overlapping (moral and anthropological) consensus", which he believes can be made to float free of the various larger theological and metaphysical views that sustain it.

Here I part company with Rawls. While Kantian humanists and Christian humanists and Muslim humanists all affirm the liberal humanist value of human dignity, they do so in ways that are sometimes significantly different. Their common affirmation of human dignity does not prevent disagreement over how human foetuses should be treated or whether human beings should be permitted to control their dying by committing suicide. And these disagreements, these differences in interpretation, can be traced back to their deeper religious and philosophical worldviews. Public reason, therefore, is not entirely common. It is not neutral. It does not float free of the larger comprehensive humanist doctrines that support it. On the contrary, these larger doctrines give rise to significant disagreements within the common terms of public reason. Rawls himself implicitly recognised this, at least on the margins of his thought, where he acknowledged that public reason involves radical controversy as well as consensus. Why else would we need to have recourse, as Rawls acknowledges we do, to decision by majority vote?

So, liberal humanist space is not indefinite. Nor should it be taken for granted; it is under threat from a variety of anti-humanisms. Liberal public institutions that would survive, then, cannot afford to take a neutral position on ethics and anthropology. Nor can they afford to be neutral with regard to which larger views of the world dominate public culture, since some of these are positively subversive of liberal ethics and anthropology. Liberal public institutions therefore need to foster worldviews that commend the virtues necessary for liberal public discourse to flourish. They need to do this because, as Rawls rightly observed, there are illiberal barbarians inside the gates; and within living memory their number has been known to grow to democratically dominant proportions.