Is cognition a matter of representations? Emulation, teleology, and time-keeping in biological systems

摘要

Contemporary literature distinguishes two ways to defend the claim that cognition is a matter of representations: first, cognition involves representation-hungry tasks; second, cognition involves a complex form of informational covariation between subcomponents of a system with an adaptive function. Each of these conceptions involves a different notion of representation, and promotes a particular view of the architecture of cognition. But despite the differences, each of them aims to support the claim that cognition is a matter of representations on architectural constraints. The objectives of this article arc twofold: first, it is argued that architectural constraints do not entail either of those two ways to defend the claim that cognition is a matter of representations; second, it is claimed that both notions of representation share an objectionable common element-namely, the idea of a model that grounds the representational reading-that must be abandoned, in favor of a more economical explanation in terms of causal relations, in order to get a clear view of cognition.