The Summer '05 issue of Philosophia Christi contains three papers on Alvin Plantiga's evolutionary argument against Naturalism (EAAN, since philosophers are obsessed with acronyms). The argument runs something like this: According to evolutionary Naturalism (N&E), human cognitive powers developed by a process of natural selection acting on random genetic variation. Since what natural selection "selects" for is survival/reproduction, that means that the purpose of our reasoning ability (R) is to allow us to leave more offspring. But if this is the case, says Plantiga, then R, our reasoning ability, need not be truth-based at all. Natural selection will select for some R that produces reproductively advantageous beliefs, regardless of their truth-value. Therefore (skipping a few steps), P(R|N&E) (read: the probability that our reasoning faculties are reliable given naturalism and evolution) is either low or inscrutable. And if this is the case then N&E is self-defeating: if you accept it, you affirm that the reasoning process used to arrive at that conclusion that N&E is true is unreliable.
A few years ago, Plantiga spoke at Cal State San Bernardino, and Libby and I got to hear him speak. His talk wasn't about the EAAN, but he brought it up during the Q&A almost as an aside: "Oh, by the way, Naturalism is incoherent." The reaction among the mostly secular audience was, I think, one of suspicious incredulity - similar to my reaction after hearing the ontological argument for God's existence for the first time: "No way it was that easy... something must have been slipped past me under the table."
Plantiga's argument has been attacked from just about every conceivable angle by philosophers who think they know what Plantiga is smuggling past them. I think the two main lines of attack are to deny that P(R|N&E) is really low/inscrutable, and to deny that Naturalism is irrational even though P(R|N&E) is low.
The latter argument is surprising to me because it seems, well, indefensible. Nevertheless, the two arguments against Plantiga mentioned in PhilChristi take this approach. The more interesting of the two, put forth by a guy named Beilby, argues that while P(R|N&E) is low under current evolutionary models, there is no reason to assume that no suitable evolutionary model will be discovered which provides a high probability for the reliability of our reasoning faculties. So P(R|N&E) is not really low, even though it is low according to current evolutionary theory. We ought not to abandon evolutionary theory, furthermore, because it is a fruitful research program and it is unrealistic to require a theory not to have holes which require further study. In order to be successful, according to Beilby, "Plantiga must argue not only that P(R|N&E) is low, but that we have good reason to think it cannot be high. This is a truly formidable quest."
Call me crazy, but I don't think the quest is quite so formidable. What undiscovered resources could evolutionary theory possibly posses that would produce faculties aimed not at survival of the fittest but at something completely unrelated, truth? Any such resource would, by definition, not be evolutionary.
Perhaps truth-based faculties are themselves reproductively advantageous? Again, this seems absurd. Does being really smart improve your chances of out-reproducing the other guy? I don't think so.
There is a third option: perhaps there is a high probability that evolutionary processes directed at survival of the fittest produced truth-based faculties as a byproduct. But here again we are at a loss to describe how exactly reliable abstract reasoning could possibly arise from a process directed only at survival. The problem is not that evolutionary theory does not yet know how reliable cognitive faculties evolved, the problem is that evolutionary theory provides no resources with which to evolve them.