Salient among mental states are (propositional) attitudes: thoughts, desires, and other states picked out by verbs that take sentence complements, e.g., ''that . . .'' clauses as their direct objects. These states admit of two orthogonal classifications: in terms of their content-what is expressed by the sentence complement, e.g. [(that) snow is white] vs [(that) grass is green]-and in terms of the specific attitude role involved-e.g., the thought vs the belief or the desire, say, that snow is white. These classifications are orthogonal insofar as most any content may be the object of most any attitude relation: most anything you can think, you can believe, desire, fear, hope. It is often an interesting question which of these two classifications is relevant to the understanding of some specific psychological state: thus, some have argued that what distinguishes the thought that Sam Clemens is funny from the thought that Mark Twain is, is not any difference in the content of the thought, but a difference in the role it plays in the architecture of someone's mind.
In his paper here and many others elsewhere, David Rosenthal proceeds in the opposite direction and makes the interesting proposal that what distinguishes conscious from unconscious mental states is largely a matter of content and not role. He agrees with the growing consensus in philosophy and psychology that there are unconscious thoughts-for the sake of the discussion, I will assume that Rosenthal thinks of the familiar postulations of, e.g., Freud, Chomsky, Weiskrantz, and Nisbett and Wilson as among them. 1 Rosenthal's proposal-or at least one of his proposalsis that what further makes a thought conscious is that it is the object of what he calls a ''higher order thought'' (''HOT''): to a first approximation, a conscious thought is a thought that is thought about (it is a ''target'' of the HOT). Much of his discussion