A part of Lewis's "locus classicus" I tend to avoid discussing
This is from the third chapter of Miracles.
CSL: But', it will be said, 'it is incontestable that we do in fact reach truths by inferences'. Certainly. The Naturalist and I both admit this. We could not discuss anything unless we did. The difference I am submitting is that he gives, and I do not, a history of the evolution of reason which is inconsistent with the claims that he and I both have to make for inference as we actually practise it. For his history is, and from the nature of the case can only be, an account, in Cause and Effect terms, of how people came to think the way they do. And this of course leaves in the air the quite different question of how they could possibly be justified in so thinking. This imposes on him the very embarrassing task of trying to show how the evolutionary product which he has described could also be a power of 'seeing' truths.
VR: Here Lewis presents the Darwinian objection to his argument from reason. Antony Flew's statement of this response is here:
Flew: A]ll other things being equal and in the long run and with many dramatic exceptions, true beliefs about our environment tend to have some survival value. So it looks as if evolutionary biology and human history could provide some reasons for saying that it need no be a mere coincidence if a significant proportion of men’s beliefs about their environment are in face true. Simply because if that were not so they could not have survived long in that environment. As an analysis of the meaning of ‘truth’ the pragmatist idea that a true belief is one which is somehow advantageous to have will not do at all. Yet there is at least some contingent and non-coincidental connection between true beliefs, on the one hand, and the advantage, if it be an advantage, of survival, on the other.
VR: Lewis replies in this way. I have de-emphasized this line of response because I am inclined to suppose that this response begs the question. Can he be defended here?
But the very attempt is absurd. This is best seen if we consider the humblest and almost the most despairing form in which it could be made. The Naturalist might say, 'Well, perhaps we cannot exactly see--not yet--how natural selection would turn sub-rational mental behaviour into inferences that reach truth. But we are certain that this in fact has happened. For natural selection is bound to preserve and increase useful behaviour. And we also find that our habits of inference are in fact useful. And if they are useful they must reach truth'. But notice what we are doing. Inference itself is on trial: that is, the Naturalist has given an account of what we thought to be our inferences which suggests that they are not real insights at all. We, and he, want to be reassured. And the reassurance turns out to be one more inference (if useful, then true)--as if this inference were not, once we accept his evolutionary picture,  under the same suspicion as all the rest. If the value of our reasoning is in doubt, you cannot try to establish it by reasoning. If, as I said above, a proof that there are no proofs is nonsensical, so is a proof that there are proofs. Reason is our starting point. There can be no question either of attacking or defending it. If by treating it as a mere phenomenon you put yourself outside it, there is then no way, except by begging the question, of getting inside again.
Steve Lovell responds: Lewis evidently saw the possibility of such a response, but thought that it begged the question. He claimed that it amounted to the naturalist arguing for the reliability of his cognitive faculties, and that if those faculties really are in doubt such an argument should not persuade us. But this seems like a poor response to Flew. Flew is not trying to remove doubts about our cognitive faculties, he is attempting to stop those doubts from arising in the first place.
Well, that's what I would have thought, and that is why I have not emphasized this aspect of Lewis's argument. Have I missed something here? I believe Jason Pratt thinks that I am.
I am linking to Lovell's excellent discussion of the AFR.