Agent-environment state machines

Tom Ziemke
http://www.ida.his.se/~tom/
Professor of Cognitive Science
School of Humanities and Informatics
University of Skoevde
Skoevde, Sweden

Abstract:

Traditional cognitive science and AI viewed cognition as, roughly speaking, the computational manipulation of internal representational knowledge of the external world through state machines (e.g., of the Turing machine type). Most of this is strongly questioned in recent work on embodied, situated and distributed cognition. However, using a number of concrete examples from robotic experiments, this talk will illustrate that the notion of state machines is also useful in describing distributed cognitive processes emerging from the interaction of embodied agents and their environments, where knowledge and states are not internal, but distributed over agent and environment.