This is another variant of Leon Porter's 1983 argument against semantic naturalism. Suppose:
- Unless the reliability of our doxastic faculties can be explained scientifically, naturalism is false.
- As a matter at least of nomic necessity, every grammatically correct sentence in the language of science (i.e., the language that a completed and correct science would use) expresses a proposition which is either true or false.
- Only claims that can be made in the language of science can be given a scientific explanation.
- If a language can express the claim that our doxastic faculties are reliable, then the language has a predicate T equivalent to (minimally: "of nomic necessity coextensive with") a truth predicate for utterances, and a predicate U equivalent to the predicate "is an utterance".
- Some habitable locations in spacetime can be specified in the language of science solely in terms of relations to inorganic phenomena.
- If x is any habitable location in spacetime that can be specified in the language of science solely in terms of relations to inorganic phenomena, and A and B are predicates in the language of science, then there is a nomically possible world at which there is a description D of x in the language of science, and an utterance at x of a sentence equivalent to "No item satisfying both A and B is located at any place satisfying D", with there being no other utterances at x.
I think the controversial claims are (1)-(3). Claim (1) is not a conceptual truth. But if the reliability of our doxastic faculties cannot in fact be explained scientifically, then naturalism fails to explains something that theism can explain easily and simply. Hence we have good reason to prefer theism to naturalism then. What makes naturalism in regard to us at all plausible is the possibility of evolutionarily explaining central features of us. But our doxastic reliability is one of those features.
Claim (2) is, perhaps, best seen this way. The language of all our modern scientific theories is such as to make (2) be true if these theories are true. It is extremely plausible that the same will be true of the language of completed science. If we think that the language of completed science will be mathematical, (2) is very plausible.
While (3) is plausible, one might worry a little about the conjunction of (1) and (2). One's worry might take this. Perhaps the explanation of our doxastic reliability is done in two steps. The first step is to philosophically establish a bridge between some scientific claim and the claim that our doxastic faculties are reliable, in such a way that one can argue philosophically that if the scientific claim is true, then our doxastic faculties are in fact reliable. I think this is only plausible, however, if the scientific claim includes predicates equivalent to "is true" and "is an utterance". (One might think that only the truth of beliefs, not of utterances, needs to be talked about scientifically. But I think this is mistaken in light of the way that language is in fact involved in our doxastic functioning.)