Computational implementation

Some things compute, others do not. Digital computers do, and perhaps minds, but not rocks. Or so it seems. Determining what conditions a physical system must satisfy in order to compute is the focus of theories of computational implementation, or physical computation. In this chapter we explore some implications of these theories for the computational theory of mind (CTM). After highlighting some general features of implementation, we review the theories on offer and how they address the challenge of pancomputationalism, namely, the thesis that every physical system computes. We argue that satisfying minimal desiderata for theories of implementation – to avoid the most damaging form of pancomputationalism – sharply limits the degree to which psychology is autonomous from neuroscience. In particular, if the mind is computational and physically implemented, then psychology constrains the kinds of structure and organization to be found in the nervous system, and neuroscience constrains the kinds of computation to be posited by psychologists.