Here's the key thing re the latter paper - I think I can show that, even if semantic properties of neural structures are epiphenomenal, as Plantinga supposes they very likely are given naturalism, it turns out surprisingly enough that natural selection will still favour some belief contents over and others, and will in fact favour beliefs that are true. That sounds impossible, doesn't it? Surely natural selection can only favour those properties of neural structures that causally effect behaviour, right? And semantic epiphenomenalism entails that such properties as being a true belief cannot causally affect behaviour. But actually, given one very obvious possibility that a naturalist might well sign up to, it's not impossible for natural selection to favour true belief. In fact it's probable.
That said, the EAAN is certainly ingenious. I have a lot of time for Plantinga.