William James, in the chapter on “Attention” of his Principles of Psychology, is very interesting on the question of whether paying attention to something is a free act or an entirely (physiologically) determined one:

If feeling is an inert accompaniment [to neural activity], then of course the brain-cell can be played upon only by other brain-cells, and the attention which we give at any time to any subject… is the fatally predetermined effect of exclusively material laws. If, on the other hand, the feeling which co-exists with the brain-cells’ activity reacts dynamically upon that activity, furthering or checking it, then the attention is in part, at least, a cause. It does not necessarily follow, of course, that this reactive feeling should be ‘free’ in the sense of having its amount and direction undetermined in advance, for it might very well be predetermined in all these particulars. If it were so, our attention would not be materially determined, nor yet would it be ‘free’ in the sense of being spontaneous or unpredictable in advance. (page 424)

Translating this into contemporary terms, James’ point is that one can reject eliminativism without thereby buying into some sort of idealism or mind/matter dualism. Even if “feelings” are impossible in the absence of a physical correlate, and even if can indeed be correlated with specific physiological states, this need not entail that they have no causal impact and are merely epiphenomenal. My feelings may indeed have an impact on what I think and do, without this being translated into necessary causal relations among the physical states correlated with these feelings, thoughts, and actions.

James immediately goes on to note that, as to the choice between eliminativism/determinism and the causal efficacy of feelings, “the question is of course a purely speculative one, for we have no means of objectively ascertaining whether our feelings react on our nerve-processes or not.” He does not pretend to offer an answer. However, several pages later he warns us against “the strange arrogance with which the wildest materialistic speculations persist in calling themselves ‘science’” (page 429).  This is because the eliminativist view is no better than “an argument from analogy… The consciousness doesn’t count, these reasoners say; it doesn’t exist for science, it is nil; you mustn’t think about it at all” (pages 429-430). Eliminativism solves the problem of consciousness by ruling it out a priori. “Such conduct may conceivably be wise,” James says; “but scientific, as contrasted with metaphysical, it cannot seriously be called” (page 430). This is not to reject science, but to remind us that, even as we rely on it, we cannot thereby eliminate metaphysical considerations as well.