Franz Knappik

Bayes and the first person: consciousness of thoughts, inner speech and probabilistic inference

On a widely held view, episodes of inner speech provide at least one way in which we become conscious of our thoughts. However, it can be argued, on the one hand, that consciousness of thoughts in virtue of inner speech presupposes (unconscious) interpretation of the simulated speech. On the other hand, the need for such self-interpretation (even if unconscious) seems to clash with distinctive first-personal characteristics that we would normally ascribe to consciousness of one’s own thoughts: a special reliability; a lack of conscious ambiguity and incomprehensibility; and a sense of causal agency. I try to resolve this puzzle by proposing an account for the requisite self-interpretation of inner speech in terms of Bayesian probabilistic inference. Drawing on “perceptual loop” accounts of speech control, I argue that such interpretive probabilistic inferences are used for the control of inner speech, and that as a consequence of this function, they are biased toward the correct interpretations. I conclude by showing how this model can explain the first-personal characteristics of consciousness of one’s own thoughts. In the case of the sense of causal agency, the resulting explanation yields novel accounts for “audible thoughts” and thought insertion.

Self-knowledge about attitudes: rationalism meets interpretation

Recently influential “rationalist” views of self-knowledge about our rational attitudes hold that such self-knowledge is essentially connected to rational agency, and therefore has to be particularly reliable, immediate, and distinct from third-personal access. This approach has been challenged by “theory theory” or (as I prefer to call them) “interpretationist” views of self-knowledge: on such views, self-knowledge is based on the interpretation of information about ourselves, and this interpretation involves the same mindreading mechanisms that we use to access other persons’ mental states. Interpretationist views are usually dismissed as implausible and unwarranted by advocates of rationalism. In this article, I argue that rationalists should revise their attitude towards interpretationism: they can, and ought to, accept themselves a form of interpretationism. First, I argue that interpretationism is correct at least for a substantive range of cases. These are cases in which we respond to a question about our attitudes by a conscious overt or inner expression of our attitude, and form a self-ascriptive belief on the basis of that expression. Second, I argue that rationalists can adopt interpretationism without abandoning their basic tenets: the assumption that both approaches are incompatible is unfounded.