Here's a hypothetical dialogue between an agnostic and a theist.
A: Evolutionary theory (whether genetic or mimetic or a combination) can explain why you believe in theism. This explanation has nothing to do with the truth of theism. Hence, you should not be a theist.
B: The explanation has much to do with theism. For a good creator would want us to believe in him, and hence, if our beliefs arose through evolutionary means (which you grant!), he would have likely set up the evolutionary pressures in such wise that they should favor belief in theism.
A: You're begging the question by depending on the theism that my argument puts into question.
B: Are you a sceptic about our empirical, logical and mathematical knowledge?
A: I would be really stupid to be such, since scepticism about any of these areas would undercut my belief in evolutionary theory.
B: Good. But now consider this claim: One can give an evolutionary explanation (genetic or mimetic or a combination) of your empirical, logical and mathematical belief.
A: Yes, I can, and I see where you're heading. There is, however, a crucial asymmetry between the theistic case and the empirical, logical and mathematical ones. The explanation of why we hold these empirical, logical and mathematical beliefs depends on the truth of these beliefs. It is useful to believe tigers are dangerous because they are dangerous and the environment is such that true beliefs about dangerous things are useful.
B: But it is useful to believe that God exists because God has set up an environment in which having theistic beliefs is useful for the dissemination of genes or memes.
A: But in your "God has set up" conjunct, you're depending precisely on one of the beliefs put in question by my evolutionary argument.
B: And in the same way in your "the environment is such that" conjunct, you're depending precisely on one of the beliefs put in question by the evolutionary argument for scepticism about empirical, logical and mathematical beliefs.
A: But I can argue a priori that in most environments, having true empirical beliefs is useful.
B: I doubt you can. Maybe in the case of most environments governed by laws like ours you can argue for this, but then you will need a posteriori knowledge of the laws. But the claim that this is so in most logically possible environments? Arguing for that seems nigh impossible. And even if you succeed, what useful thing have you done? The a priori beliefs on which such an argument would be based have been put in question by the evolutionary argument just as the a posteriori ones have. You have failed to show an asymmetry between the theistic case and the empirical, logical and mathematical ones.
A: How about this asymmetry? I should not worry about sceptical arguments that put into question beliefs that all reasonable people should hold, and certain basic empirical, logical and mathematical beliefs are among these.
B: I agree with the principle, but it is not clear that it is of any use. For I think that all reasonable people should either accept theism outright or should accept all the premises of the cosmological argument. It is unreasonable not to do so.
A: You beg the question here.
B: This is exactly what the opponent who doubts basic empirical, logical or mathematical truths will say to you about your assumption that all reasonable people accept them.
A: Let's go back a bit. I see another disanalogy between the two cases. You are claiming that theistic beliefs are selected for because they are true, since the ultimate explanation of the selective pressures involves God's setting them up to produce theistic belief. But here we have an asymmetry between the theistic case and the other cases. For one cannot explain why belief in the dangerousness of tigers was selected for apart from saying that tigers are dangerous. But I can explain why belief in theism was selected for without making use of theism.
B: I can explain why belief in the dangerousness of tigers is selected for without saying that tigers are dangerous. You see, whenever there is a tiger in a given area, it is dangerous to stay in that area. So leaving that area is selected for, and a belief in the dangerousness of tigers leads one to leave the area.
A: Why is it dangerous?
B: Because there is a dangerous animal there.
A: You mean a tiger?
B: Yes, of course.
A: So, you see, you had to make reference to the tigers' dangerousness after all.
B: Yes, but only when pressed for further explanation. I gave you a perfectly good explanation of why it is good to leave the area, because it is dangerous to stay in that area. But this explanation raised a new question for you, namely why the area was dangerous. To this I responded that there was a dangerous animal. This raised a new question for you, namely why there was a dangerous animal. And to this I responded by saying that the animal was a dangerous tiger. Yes, we've arrived at the dangerous tigers, but only by asking further questions. The initial explanation was a perfectly correct explanation--if there is a tiger in an area, it's a dangerous area. But we can ask more and more questions here.
A: I'm losing sight of where this is leading.
B: Sorry. The point is that this is exactly parallel to the theism case. Why is belief in theism selected for? Because it correlates with certain kinds of behaviors which are beneficial, let's say. But now this raises a new question, namely why these behaviors are beneficial? Let's say one answers by citing facts about the environment. But then one can ask why these facts obtain. And eventually, I claim, one will get to a question where the explanation is going to be: "Because God set up the environment so as to select for theistic belief." So there is a parallel. In the case of both the theistic and non-theistic beliefs, a non-ultimate explanation of selectivity can be given without making reference to the truth of the claims. But in both cases, an ultimate explanation makes reference to the truth of the claims.
A: Hmm. How about this response? I can give an alternate ultimate explanation of the universe, an explanation that is non-theistic, and that implies selectivity for theistic beliefs. This story, maybe, involves the Big Bang happening out of a quantum vacuum.
B: I doubt you can plausibly flesh out your alternate ultimate explanation in a way that will stand up to criticism. In fact, the best atheistic responses to the cosmological argument involve denials that there is an ultimate explanation. But suppose you can. I will simply deny this explanation, on the grounds of my theism.
A: There is still an asymmetry. For while I can give a putative non-theistic ultimate explanation of why belief in theism is selective, you cannot give even a putative ultimate explanation of why belief in, say, the dangerousness of lions is selective without talking of the truth of that belief.
B: True, because I am not in fact a sceptic about empirical, logical and mathematical beliefs. But if I were one, I could easily come up with a dozen putative ultimate stories here, involving various demons or the like. These stories wouldn't be true by your lights, but neither is your story true on my lights.
A: Should I thus become a sceptic about empirical, logical and mathematical claims?
B: By no means. But you should abandon your claim that there is an asymmetry between the evolutionary argument for scepticism about theism and the evolutionary argument for scepticism about empirical, logical or mathematical truths. If one succeeds, you have no argument against the success of the other. If it is acceptable to make use of the beliefs put into question to argue that we would be likely to non-coincidentally come to true belief through the evolutionary process in order to refute the evolutionary sceptical argument, then this response will work for both. If one begs the question, so does the other.
A: But maybe this is the asymmetry: Accepting scepticism about empirical, logical and mathematical knowledge would undercut my belief in evolution, and hence would undercut the scepticism. This scepticism is, thus, self-defeating, unlike scepticism about theism.
B: I actually suspect that theism underwrites the scientific method, because the scientific method depends on simplicity being a good guide to truth, and the only plausible reason to think that simplicity is a good guide to truth is to think that a good God created the world, and either created us with doxastic faculties adequate to the world or maybe created a simple world because simplicity is innately good. So evolutionary scepticism about theism would also be self-defeating. But I see you're not impressed. Let me then take a different tack. One could, no doubt, choose a carefully delineated area of empirical knowledge which is irrelevant to your reasons for your beliefs in evolutionary theory--e.g., empirical knowledge about distant galaxies. If the only reason you have for rejecting the sceptical argument is its self-defeat, you will have to accept the sceptical hypothesis about this area. And surely you don't want to do that.
A: No, I see I shouldn't do that.