Of mice and madmen: typical causal roles and radical indeterminacy

Abstract Following Lewis, many functionalists hold that functional roles should be understood as kind-relative typical roles—e.g., pain causes flinching for the typical human, though for some of us it might not. However, an indeterminacy problem arises if the relevant notion of typicality is cashed out in statistical terms. In this paper, I show how the problem arises in the context of Lewis’ analytic functionalism. Under reasonable assumptions, Lewis’ view implies that for each individual and any coherent attribution of mental states to that individual, there is likely to be a maximally fitting scheme of interpretation according to which that individual has those mental states. The problem, in other words, is not just one of radical indeterminacy, but maximal indeterminacy: anyone can be interpreted as having any internally consistent system of beliefs, as having any desires whatever, as being in pain, in pleasure, and so on. Any functionalist solution to this problem requires some change to how the constraints on fittingness are understood. I discuss some possible fixes, including a new potential role for naturalness within the Lewisian theory.