Interactivism: A functional model of representation for behavior-based systems

Abstract

In this paper we present the interactive model of knowledge representation and examine its relevance to behavior-based robot control. We show how the interactivist position that representation and motivation are different aspects of the same underlying ontology of interactive dynamic systems emerges within the framework of our Behavior-Network architecture. The behavior-network architecture consists of a collection of task-achieving behaviors that interact amongst themselves and with the environment generating global behavior that aid survival in its environment while carrying out various tasks. The representational and motivational aspects of the underlying control structure are implicit in the connectivity between the behavior modules and in the current activity levels over those connections. The interactivist stance enables us to address the issue of representation in behavior-based systems and suggest directions for future inquiry.