Reading Solum's excellent paper on public reasons, I am a bit puzzled about this (at *753):
Finally, the behavior of public officials in their official capacity should be governed by the principle of excluding nonpublic reasons. Public officials are different from private citizens because they personify the state; the statements of public officials in their official capacity are, in a real sense, the statements of the state and hence of the public at large. For this reason, it would be unfair to allow public officials to express their own deep convictions about the good as the official reasons for state action. Allowing public officials to advance nonpublic reasons would violate the requirement of treating all citizens fairly.
What I find puzzling is that this really doesn't work given the way nonpublic reasons are characterized. Nonpublic reasons aren't necessarily "deep convictions about the good"; they are any reasons that don't count as public reasons. And this causes a problem for the above claim, because one of the things that cannot count as public reasons is expert assessment in controversial matters (see the Rawls quote toward the beginning). And it can hardly be right that public officials should never make policy in controversial matters based on expert opinion.
Further, it seems clear to me that there are many cases in which public reasons are not precise enough for policy-making, and public officials are nonetheless constrained to form policy on the basis of more specific nonpublic reasons. Consider abortion. Given that public reasons are the result of overlapping consensus, lack of consensus in this area makes it very difficult, perhaps impossible, to exclude nonpublic reasons for taking a particular course of action (e.g., not trying to get around Roe v. Wade or trying to get around Roe v. Wade). I find it difficult to see why Rawls thought, for instance, that the Supreme Court was the exemplar of public reason; Supreme Court decisions are motivated by public reason, in that there are public reasons for the Supreme Court considering the cases it does, and there are public reasons that serve as general background for the decisions. But the actual decisions are very, very often not decided on public reasons at all. At least, I can't be the only person who has found Supreme Court reasoning to be dubious on occasion. But despite the dubiousness, it is, as I said, clearly defensible by public reasons for the Supreme Court to handle most of the cases it handles, even if it occasionally has to use nonpublic reasons to do so. And if you aren't convinced about this with regard to the Supreme Court, Congress and state legislatures are even better examples. Sometimes policy is needed despite lack of consensus on principles that would actually decide either way; unless we're just going to have people appealing to public reasons for their action in ways that are only hazily relevant, we need to have some direct connection of public reasons with the actual policy. And this will almost always involve reasons that do not fall within the field of public reasons.
And this is not a problem because: (1) Human beings can come to the same policy conclusions even if they share no premises. That is, the policy itself can become part of the constructed field of public reasons, simply by being reached from radically different, and even incommensurable perspectives. It is possible for an atheist to affirm a given policy (say, related to separation of Church and State) for entirely atheistic reasons, and for a theist to affirm the same policy for entirely theistic reasons (say, Baptist theology). In such a case the conclusion is a public reason, but the only ways it is supported is by nonpublic reasons. One really can't exclude the latter, unless one is really willing to say that we can go around not giving any reasons for policies. We can see this sort of thing working in cases like Maritain's brilliant solution for how to make the UNESCO Charter, which was carried over into the drafting of the Universal Declaration of Human Rights: even if we can't agree on reasons, we can sometimes come to create reasons on the basis of those reasons. The Soviet Union can agree to conclusions put forward by a Catholic country for reasons no Catholic country could accept; and vice versa. And this is often all one needs. This is part of what happens in some cases of enlightened self-interest; and this, along with the public discussion (a more fundamental notion than that of public reason, since public reason is a result of public discussion) is much more central to the notion of liberal democracy than any particular view on whether nonpublic reasons should be excluded or not. (2) It is possible to have nonpublic reasons that nonetheless are quasi-public, because they involve patterns of thought that are widespread. For instance, if you have a community of theists who make policy on reasons derived from their view of providential moral order, and an atheist comes into the community, there is suddenly no consensus to make public reason in the strict sense possible. But they can easily still make policy if the atheist is willing to concede that, while he will never agree that there's a providential moral order, there is something that he can allow to be vaguely analogous to it, because the atheist can allow that the theists are supporting policy with reasons that are at least somewhat analogous to what he could accept, and vice versa. In other words, analogy and family resemblance is often enough, even if there is no commeasure, no definable common ground. So in addition to public reasons we should allow that there are quasi-public reasons; and it can be deucedly difficult to tell them apart at times.
But then, Rawls is not someone I read much, so I could easily be missing something. I do want to reiterate how much I like the paper, though; and Solum's inclusionary principle is interesting and worth thinking about.