A Naturalistic Theory of Perceptual Representation

Abstract

I propose a theory of representation concerning the perceptual events that are posited and studied by the cognitive and neuro-sciences. The theory is intended to help explain relationships between the perceptual and executive systems, and to place metasemantic constraints on future accounts of the semantics of natural languages. I begin by setting out desiderata for the theory. In particular, I intend the theory to be naturalistic at least in accordance with a specified kind of epistemological naturalism, to give priority to explaining the properties of the representing events themselves rather than their contents, to avoid the widespread lack of clarity among similar theories when it comes to identifying contents, to apply to human-like systems with executive functions and language, to be compatible with constraints imposed by natural selection, and to posit narrow contents that are capable of figuring in a certain kind of autonomous causal explanation. The suggested theory for meeting these desiderata is based on a definition of perceptual states by ceteris paribus effects on the motor control system, which contrasts with the orthodox description of tokened perceptual states as carrying information about their external causes. I then propose that the representational content of a perceptual event is specified by the motor control system effects that define the state it tokens, but only when this event affects the executive systems. Intuitively, these representations are constructions out of the behavioural dispositions that are mediated by perceptual events, such that these constructions are used by the executive systems in the trialling of potential behavioural outputs. While this behavioural model theory of perceptual representation satisfies the desiderata, I argue that it warrants scepticism about manifest objects and their properties. I conclude with a brief discussion of the implications of the theory.