From PhilPapers forum Philosophy of Mind:

2009-05-17
The 'Explanatory Gap'
Reply to Jamie Wallace
Stevan,

I think you are wrongly assuming that the "problem" generated by uncomplemented categories indicates a problem which exists outside of the grammar in which those categories are defined.  In other words, the only problem here is the desire to take the notion of uncomplemented categories seriously.

You write:

"But the category "feeling" is one of a family of special cases (each of them causing conceptual and philosophical problems) because they are "uncomplemented categories" -- a kind of "poverty of the stimulus" problem arising from the fact that they are based (and can only be based) exclusively on positive instances: In contrast, the category "redness" is perfectly well-complemented: I can sample what it feels like to see red things and non-red things, no problem. But not so with the category "feeling": I can sample what it feels like to feel: I do that every time I feel anything."

To feel is to feel some X, so that any knowledge of feeling is knowledge of feeling some X.  Knowledge of feeling cannot be separated from knowledge of X.  There is thus no uncomplemented (and no "Cartesian") knowledge of feeling, just as their is no uncomplemented (and no Cartesian) knowledge of thinking.

Feeling is not an object of knowledge, but rather a way of knowing.  Thus there is no uncomplemented category to worry about here.  The problem you have been discussing is not a "hard problem" with which philosophy or science must reckon, but a simple problem with your grammar, with your categorizing "feelings" as objects of knowledge, and not ways of knowing. 

This error underlies your entire discussion, explaining your incoherent distinctions between Cartesian and non-Cartesian knowing and between functing and feeling.  It also explains the contradiction between your allegiance to physicalism and your insistance that feelings are somehow non-causal.  (Consider, if this contradiction is not already clear, that the term "physical" implies functional/causal congruity with respect to predictive models, and that this is a property which you deny feelings.  Consider also the contradiction implied by the fact that your argument here is motivated by the existence of feelings; for if feelings cannot causally influence behavior, how could they motivate it?)  Once the original category error is corrected, all of these problems disappear. 

Of course, you could try to argue that feelings really are uncomplemented categories.  Perhaps you wish to claim that one can feel without feeling some X, or that one could know that one was feeling without knowing that one was feeling some X.  But I don't see how you could support such a position.  So far, the only support you have provided is an appeal to common knowledge (as though it was just obvious that feeling could be separated from feeling some X) and your claim that anybody who denies this fact is disingenuously begging the question.  These tactics are no more persuasive than the theistic arguments they resemble.  The bottom line is, your claim invites contradiction without compensation, and I think the error is easy enough to correct.