Autonomous acquisition of cooperative behavior based on a theory of mind using parallel genetic network programming

摘要

Understanding others as having intentional states, such as beliefs and desires, is called the theory of mind (ToM).To clarify the mechanism of the autonomous acqui sition of cooperative behavior based on the ToM, we con structed a functional model of the brain based on the functional parts combination (FPC) model. This model con sists of a set of functional parts and activation signals speci fying selective activated patterns, and activated modules can be executed in parallel based on the flow of control tokens. The module network and activation signals can be acquired by the evolutionary computation techniques used in genetic network programming and the genetic algorithm, respec tively. We used a hunter task as the task to be solved by agents, and encoded inherent activation signals into the genome as a first step. The result of a computer simulation shows the emergence of the pattern of the functional parts for processing the ToM through evolution characterized by punctuated equilibrium.