If beliefs don’t affect behavior (epiphenomenalism), or they affect behavior but not by virtue of their content (semantic epiphenomenalism), then it is hard to see how evolution could select for mechanisms that produce True* beliefs, for evolution would seem to be blind to belief content just as Plantinga has suggested. However, if beliefs do affect behavior by virtue of their contents, then Plantinga’s crucial first premise is very likely false. As Plantinga himself has said:
Plantinga does think that our sensory organs could evolve naturally to reliably indicate certain environmental states of affairs, so some of Harry’s beliefs could also be true in some ways. For instance, Harry may believe that a green tree is in front of him and that he had better run around it or risk serious injury. Is his belief that the tree is green True? The tree merely absorbs all wavelengths of light except green. It reflects green wavelengths which are then detected by the indicators in his retina, leading to a belief that the tree is green, but the only thing that is green is his mental representation of the tree. Nevertheless, his idea that there is something in front of him and that colliding with it will cause injury surely is true. Does it matter that the tree isn’t really green but that his mental representation of it is? Evolution doesn’t seem to care about it so long as Harry sees the tree and avoids injury. Should we really care? That’s just how we experience certain truths about our environment. As I argued in part 2 of this series, that’s just truth to us.
* In this 4 part series, I use lower case t 'truth' to denote what seems true to us and capital T 'Truth" for what's actually or ultimately true. More on this here.