Abstract

This article addresses the issue of compositionality of mental representations from the perspective of a foundational framework for cognitive science. The dynamical cognition framework is inspired partially by connectionism and partially by the persistence of the problem of relevance within classical computational cognitive science. It treats cognition in terms of the mathematics of dynamical systems: total occurrent cognitive states are mathematically/structurally realized as points in a high-dimensional dynamical system, and these mathematical points are physically realized by total-activation states of a neural network with specific connection weights. The framework repudiates the classicist assumption that cognitive-state transitions conform to a tractably computable transition function over cognitive states. Computational Theory of Mind states that the causal role of a mental representation is syntactically determined, but this idea of syntactic determination of causal role is ambiguous.