As anti-empiricists, Chomsky and McDowell differ, however, in that whilst both think that knowledge (more than which empiricists would allow) requires having a particular kind of mind or subjectivity, the subject matter for Chomsky can be allowed to depend on, or itself be shaped by, that subjectivity. McDowell cannot allow that and this presents a prima facie problem. How can it be that the truths a thinker thinks might depend on features of their subjectivity without collapsing into a crass subjectivism? The answer, you won’t be surprised to hear, is a form of occasionalism.
In passing the points that characterise empiricism here are these:
[F]irst of all, a position arrived at a priori. It’s guiding notion, put one way, is that we are universal thinkers: we enjoy no cognitive capacities, so see nothing of the world, that would not be shared by any thinker with our sensory sensitivity to the stimuli that impinge on us (such things as light, sound, pressure).
Second, for a given domain, the empiricist will claim to identify those procedures, or abilities, which are the knowledge-yielding ones with respect to that domain. What these are is, from the empiricist perspective, something to be arrived at a priori. They will be just those capacities enjoyed by any thinker at all with relevant sensory sensitivities.
Third, the empiricist will hold that we can know a fact in the relevant domain only where that fact is provable, or ascertainable (with sufficient certainty) from privileged facts by application of the specified knowledge-yielding procedures.
Fourth, an empiricist may claim that there are facts in the relevant domain only insofar as these are derivable from privileged facts according to the principles defining the correct operation of those knowledge-yielding procedures. Typically, such an empiricist will hold that the facts in the chosen domain are far fewer, and less interesting, than we would have supposed. [Travis 314-7]
This is initially a surprising way of describing empiricism insofar as Epistemology and Metaphysics 101 goes. But the centrality of experience of the traditional Ep&Met view has to be fleshed out with some substance and Travis’ account dovetails with Quine whose views of knowledge of language is one of the two key foci of the paper and is a helpful foil to McDowell’s position
So back with the question raised above: How can it be that the truths a thinker thinks might depend on features of their subjectivity without collapsing into a crass subjectivism? I’ll drop some numbers into the quote of how the problem is set up.
Trivially, a statement is true just when things are the way they are according to it. That suggests, innocently enough, that a statement’s truth depends on precisely two factors : first, how things are according to it; and, second, how things are. Innocence ends if one supposes that one can specify how things are according to a statement—which way it speaks of things as being—in such a way that the truth of any statement which speaks of things as that way can depend only on whether that is, in fact, a way things are.  Let P be a way a statement might thus represent things. Then, accepting that idea,  we may still innocently allow that the way given thinkers think decides whether some one of their statements stated that P, or, say, that Q, where that is another such way for a statement to represent things.
 But one cannot, accepting this idea, allow that, where a statement spoke of things as being P, whether it thus stated truth depends on how a particular (sort of) thinker thinks.  For what thinkers could thus decide, in thinking as they do could only be, within this framework, how things were: whether that which is so according to any statement which states P is so. That would be mind-dependence of the worst sort. Yet, with the end of innocence in place, anti-empiricism is under pressure to say just that.  For, given the role it assigns to sensibilities, it seems, where thinkers think in terms of, say, things being chairs or not, still, for all that, whether things are as they say (on some occasion) in saying such-and-such to be a chair depends in further ways on how they are designed, or equipped, to think. [ibid: 377-8]
So 1 is innocent: a statement’s truth depends on the two factors one would expect. 2 introduces an idea. If there is some slippage between a statement and how it might represent things, so if there is a way of thinking of the statement without appeal to the occasion of its use, we might label one way things might be P. And it remains innocent that what makes P what a statement states rather than Q a matter for features of the speaker.
So the problem comes with 4. ‘But one cannot... allow that, where a statement spoke of things as being P, whether it thus stated truth depends on how a particular (sort of) thinker thinks.’
But why would we? We’ve already allowed those features of subjectivity to fix whether it was P or Q stated in a given statement. Thereafter, why are we not back with the innocence of 1? The answer is 6: ‘it seems, where thinkers think in terms of, say, things being chairs or not, still, for all that, whether things are as they say (on some occasion) in saying such-and-such to be a chair depends in further ways on how they are designed, or equipped, to think.’
At this point I can’t help thinking that if we’ve bought into the significance of occasionalism, its need at this point might seem obvious. But I’m not there yet. A little earlier we have this helpful vignette:
Sid buys a DIY chair kit. On bringing it home he discovers that it is much more difficult to assemble than he had imagined. It remains a neatly stacked pile of chair parts in his spare room. One day, someone, pointing at the pile, asks, ‘What’s that?’ ‘It’s a chair’, Sid replies, ‘I just haven’t got around to assembling it yet.’ On a later occasion, Sid and Pia, with guests, find themselves a chair short for dinner. ‘There’s a chair in the spare room’, Sid says helpfully. But there is still only the pile. Recognisably, Sid spoke truly the first time, falsely the second. It just takes a different way of thinking of being a chair to see the truth of that first thing from the way it takes to see the falsehood of the second. Such contrasting ways of thinking are a common everyday part of our way of dealing with the world. [ibid: 336]
Given our understanding of the requirements of a chair in a hurry at a dinner party, the chair kit is not a chair. It is not true to say that there is a chair, on that understanding, in the spare room. If this is described sufficiently to explain – in yet another context – so that Sid’s comment, and its falsity, is clear to an audience, could that articulation of the content of the utterance not be called say P by contrast with the Q of the first understanding (where chair as kit is fine)? With that in place, why can’t we say 3 again? I’m not getting the problem. I would be much less puzzled if the very idea of labelling understandings as P and Q were supposed to be the end of innocence (though, in a context, occasionalism should not threaten mere labelling). But the key link seems dark. I will press on with the paper on psychologism and see whether it helps.Travis, C. (2002) ‘Frege’s target’ Royal Institute of Philosophy Supplement 51: 305-343