Cell Assemblies as an Intermediate Level Model of Cognition

Abstract

This chapter discusses reverberating circuits of neurons or Cell Assemblies (CAs) derived from Hebbs 9 proposal. It shows how CAs can quickly categorise an input and make a quick decision when pre- sented with ambiguous data. A categorisation experiment with a com- putational model of CAs shows that CAs categorise a broad range of patterns. This chapter then describes how CAs might be used to implement the primitives of an symbolic cognitive architecture. It also shows how a system based on CAs is theoretically capable of fast learning, variable binding, rule application, integration with emotion and integration with the external environment. CAs are thus an ideal mechanism for further research into both compu- tational and cognitive neural models. Our medium to long-term plan for exploration of thought via CAs is described. If humans use CAs as a basis of thought, then studying how biological systems use CAs will provide information for computational models. The reverse is also true; computational modelling can direct our research activity in biological neural systems.