The following argument is valid:
- (Premise) Every reasonable desire can be fulfilled.
- (Premise) The desire for moral perfection is reasonable.
- (Premise) Moral perfection requires being such that one is morally responsible and yet cannot do wrong.
- (Premise) If naturalism is true, a state that entails moral responsibility and an inability to do wrong is not attainable.
- Therefore, naturalism is false.
The argument being valid, the question is whether it is sound. I think (1) is plausible if we take "reasonable" in a strong enough sense. It is easy to argue for (4), since our best theories involve such a degree of indeterminism that, if they are complete descriptions of human beings, the possibility of doing wrong will always be there. That leaves (2) and (3). There is an argument from authority for (2): Kant thought so (and made an argument somewhat similar to this one). It does seem that a part of the moral life is the pursuit of moral perfection, and the moral life is reasonable in a strong sense.
That leaves (3). Let's consider two alternate views of moral perfection.
"Moral perfection only requires that one be morally responsible and never any longer actually do wrong." This is too weak, surely. It would mean that anybody whose existence ends with a morally responsible choice to do something right achieves moral perfection just prior to that choice.
"Moral perfection requires having all the virtues to a complete degree. Having the virtues to a complete degree is incompatible with self-initiated wrongdoing, but is not incompatible with losing the virtues or being forced through neurological manipulation into wrongdoing." This view is plausible, but I think the argument can still be run on this view, albeit with some complications. The challenge is whether an analogue of (4) is still true. I think it is. The morally perfect person is not blind to temptation—i.e., she is aware of the goods that temptation offers. (Courage is not achieved by not noticing danger.) But if she is aware of these goods and naturalism holds, then it is surely possible for her to choose these goods, where the choice is constituted by an indeterministic event in the brain, even if she has brain structures that are virtuously pointed the right way. And such a choice would be, surely, a morally responsible one, being a choice of a (lesser) good that comes from one's appreciation of that good. (If it be said that only deterministic choices are morally responsible, then moral responsibility is not available given naturalism in our indeterministic universe, and, again, moral perfection is unattainable.)