The following post is entirely based on Carl Sachs' illuminating comments on my last post (see here).
"McDowell notes, in his response to Bernstein, that the "intermible oscillation" isn't between coherentism and bald naturalism, but between coherentism and the Myth of the Given: between idealism and empiricism, if you will. Davidson figures as the coherentist, and Gareth Evans is cast as the unwitting defender of the Myth of the Given. (Although MOG also appears in Quine, and is at the heart of the entire empiricist tradition from runs from Locke to Russell.)
Bald naturalism and the alternative, "rampant platonism," are attempts to halt the oscillation. The former does so by reducing the space of reasons to the space of laws (reducing sapience to sentience, perhaps?), whereas rampant platonism divorces the space of reasons from all human conceptual capacities.
I agree that McDowell's notion of experience is too "thin." Like the post-analytic philosophers with whom he's engaged (Quine, Davidson, Selllars), McDowell thinks of experience as percpetual recognition of objects. (Although this is also where Kant starts digging.)
The narrowness of McDowell's conception of experience is, I think, well revealed in the Bernstein essay. But he reveals it partly through considerations of our animal embodiment. This takes Bernstein towards Adorno, not Habermas (which, of course, is a big part of his current project). I don't know what Habermas has to say about embodied animal experience. Certainly the impression one has of him is that everything normative and rational occurs at the level of language; animal embodiment simply drops out of the picture. There is a difference between the exercise of reason and the spontaniety of animal life, but there is also, if not identity, at least continuity."