Two Approaches to Transparency about Self-Knowledge
According to transparency views of self-knowledge, we learn about our own mental states not by turning our attention inward to detect the presence or absence of those states (as "inner sense" and "self-monitoring" views suggest) but rather by turning our attention to the outside world. In Gareth Evans's (1982) example, if someone asks me if I think there will be a third world war, in answering that question, I don't think about myself; rather, I think about the state of the world.
Suppose I answer "yes" to Evans's question. I have reached some sort of judgment. But what exactly have I reached a judgment about? There are two very different options insufficiently distinguished in the literature. Option 1: I am reaching a judgment about the world. In the context of the question (which was, literally speaking, about what I think), that judgment about the world serves a self-ascriptive function. Option 2: I am reaching a judgment about my mind. I'm not attending to my mind as a means of reaching that judgment, but the judgment is still a self-directed one. Call Option 1 the Topic Shifting approach to transparency and Option 2 the Self-Judgment approach.
Topic Shifting and Self-Judgment have complementary virtues and vices. Topic Shifting fits nicely with the intuitive sense in Evans's and others' examples that I'm not really thinking about my own mind; but then it's not clear why the result is supposed to be self-knowledge. It doesn't even seem to be self-belief. Conversely, on the Self-Judgment approach it's clear why the conclusion might count as self-knowledge, but we seem to have abandoned the core idea that I am thinking about the world, not my own mind, in answering the question.
Why not have it both ways? Combining Options 1 and 2: The transparency procedure produces a judgment that is both about my mind and about the world. This Dual Content approach shares with the Self-Judgment approach that it's clear how the product of the transparency procedure could be self-knowledge. And yet we can retain much of the original transparency intuition: The conclusion of my reflections does involve, perhaps even is mostly, a judgment about the world.
Think about avowals. An avowal, as I intend the term, is an assertion with a dual fucntion: If I avow some proposition P (say "the world is flat") I am doing two things. I am asserting that the world is flat, and I am asserting that I believe (or judge) that the world is flat. This self-attributive aspect of avowals distinguishes them from simple assertions. On the Dual Content approach, the transparency procedure generates avowals.
Consider a spectrum from simple assertion to self-alienated confession: The simple assertion that P is not at all an assertion that I believe that P. The self-alienated confession that P is not at all an assertion that P is true, but only that I (seem to) believe it. Through the middle is a range of avowals with different degrees of emphasis on asserting P vs. asserting belief that P. Assertions containing self-ascriptive phrases like "I think" might tend, on average, to be somewhat more toward the confession side than assertions without self-ascriptive phrases.
Final thought: The public visibility of blog posts, Facebook status updates, and the like creates an atmosphere of self-observation that tends to convert simple assertions into avowals. We are thus becoming an avowal society.