A Motivation-Based Self-organization Approach

Abstract

In this paper we propose a self-organizing mechanism to model adaptive multi-agent systems. The mechanism is inspired in a psychological approach of motivation. Under this approach motivation is an internal state or condition that serves to activate or energize behavior and give it direction. Thus, we propose an intrinsic motivation or reason for engaging in a particular behavior called goal-achievement, in this way agents will have a drive to reach a clearly defined state. To increase the strength of this particular behavior we propose a reinforcement mechanism based on agents past experience as a feedback. The intention of this mechanism is to mimic the adaptation exhibited by entities of complex systems composed of multiple autonomous entities that make local decisions leading to a global organized behavior for the system. Such organization is reached autonomously in different ways, thus adapting to task and environment.