A Multi Scale Cognitive Architecture to Account for the Adaptive and Reflective Nature of Behaviour

Abstract

We present a multi-agent cognitive architecture in which reactive and sequential processes can dynamically influence each other to insure that behaviour is responsive to current context (behavioural and situational) while sensitive to the longer term behavioural sequences needed to solve complex problems. We present the implementation of such a system using an agent based approach and illustrate its processing trough the simulation of two psychological tasks.