A friend of mine, Bubba, once said this about the subject:

Evolution, it is claimed, can account for changes in caterpillars and the lizards that our scientists have studied. Never mind the logical leap that it takes to go from changes in caterpillars to the caterpillars themselves: can evolution account for the scientists?

If the behavior of a scientist can be wholly explained as a byproduct of the spasms and collisions of an unguided universe, there is no reason to trust that behavior: the behavior may be useful in the sense of aiding gene propagation — and then again, it may not, as the number of evolutionary dead ends must be extraordinary — but there is no necessary connection leading from utility to logical validity.

The materialist looks to the heavens and argues that the stars are the result of unguided chaos, and he looks to the fields and posits that the cattle too are by-products of chance. And then he looks in a mirror and concludes, what? That he too is nothing more than the result of random chemical reactions?

If he does that, then there is no reason to trust his arguments about the stars and the cows. The thoughts and words that bubble up from his brain are as logically valid as trustworthy, and as rational as the bile produced by his liver: his thoughts may be produced by a more complex chemical reaction, but the slight difference in chemical complexity gives us no good reason to trust the result.

Turning his naturalism against himself, the naturalist destroys the reasons to trust the conclusions he reaches. He has, essentially, argued that all arguments are invalid and cut the very branch on which he sits.