Image 01

Posts Tagged ‘Randall Beer’

Information and dynamics in minimally-cognitive agents

Monday, September 5th, 2011

Randall Beer
Cognitive Science Program
School of Informatics and Computing
Indiana University
USA

Whatever else they may be, the notions of information and dynamics are mathematical concepts grounded in information theory and dynamical systems theory, respectively. Too often, debates regarding these concepts misconstrue or completely ignore these mathematical underpinnings. As mathematical theories, they can be applied to any system that takes the proper form. Thus, they intrinsically make no scientific claim as to “what’s really going on” in a given system. The more interesting question is what kinds of insights and explanations do these different mathematical languages provide and, perhaps most importantly, how do these distinct explanations relate when both languages are applied to the same system? In this talk, I compare and contrast the explanations that arise from applying both information theory and dynamical systems theory to the analysis of an evolved model agent capable of a solving a simple relational categorization task.

Presentation Slides [pdf]

 

Randall Beer: Information and dynamics in minimally-cognitive agents from eSMCs on Vimeo.