An Enactivist-Inspired Mathematical Model of Cognition
arXiv Preprint Archive June 10, 2022 Vadim Weinstein, Basak Sakcak, Steven M. Lavalle
This work identifies five core principles of enactivist cognitive science from the literature and builds a mathematical framework for modeling cognitive systems—both artificial and natural—that adheres to those principles. The framework avoids attributing contentful symbolic representations to agents and treats brain, body, and environment as an inseparable whole. The central concept is a sensorimotor system, a type of transition system. The authors introduce the notion of sufficiency as a foundational concept, prove a uniqueness theorem about minimal sufficient refinements (which correspond to optimal attunement of an organism to its environment), and relate sufficiency to existing concepts like sufficient history information spaces. The work aims to make enactivist ideas accessible to computer scientists, AI researchers, and psychologists, while providing philosophers a mathematical tool.