As a grad student in Behavior Analysis (the psychological science that grew out of radical behaviorism), I'd say that the consistent naturalist's only answer to this argument would be to redefine "rational" as a class of "private events" -- unobservable internal behavior. And this is essentially what B.F. Skinner and Arthur Staats do with all cognitive constructs. The painful thing about reading G. H. Clark's criticism of behaviorism from memory is that he really doesn't get the name of the game here; to the behaviorist, "memory" is just what we commonly call that which in reality is a class of internal conditioned responses. In other words, the consistent naturalist would have to say that there is no such thing as rationality qua rationality. Man is essentially irrational matter that behaves in a way that appears rational -- the word "rational" being merely a label we apply to certain classes of behavior.
So from my perspective, cih1355, your particular criticism begs the question. It kind of works with the naturalist who hasn't thought his philosophy of mind the whole way through. However, I don't think it works against behaviorists, who, in my assessment, are more consistent as atheists.
The obvious problem that the behaviorist's escape valve creates, though, is that even concepts of right and wrong and of truth and falsehood become completely arbitrary. If those things are just classes of behavior, with no external existence that imposes itself on us, then science is no more intrinsically valuable than mysticism. Using science to help people is no more intrinsically moral than wallowing in ignorance. It may be that, in the external world, naturalistic evolution is what we might call "true." But who cares that theists are wrong? Why write books about how they're ruining the U.S.? The universe is what it is, and there is no "should" about it.