Abstract

The proximity of introspection makes it difficult to explain. In what does our knowledge of our own beliefs and desires consist? Do we observe them with an inner eye? Do we infer their existence from premises concerning our actions and feelings? I reject both of these suggestions. Instead, I defend the view that facts about what we believe, and certain facts about what we want, are known by us in a direct or unmediated fashion. When one has genuinely introspective knowledge of a belief one's reason for believing that one believes that p is the very fact that one believes that p . ;According to this account, our knowledge of certain mental states is infallible relative to its grounds: there are no false introspective beliefs that have the same kind of justification as our typical, true introspective beliefs. A significant part of the dissertation is spent defending this claim against those who argue for the possibility of certain forms of self-directed error. I argue that it cannot be the case that a subject falsely believes that she believes that p because she instead either believes some other proposition or bears some other attitude toward p, and that these facts impugn observational and inferentialist accounts of self-knowledge. I also develop accounts of what beliefs and desires are to help explain the relative infallibility of introspection. ;The view of self-knowledge I offer is one according to which we have reasons for believing that we believe certain things and reasons for believing that we want certain things. I argue that any adequate account of self-knowledge must have this feature because justification is necessary for knowledge and we must have reasons to be justified. But I reject the view that only phenomenal or experiential states can endow a subject with reasons. Beliefs and desires are individuated by their causal roles---they are not purely phenomenal in nature---but the fact that one believes that p can nevertheless directly ground one's introspective beliefs. Introspective knowledge is both direct and grounded in reason