A human-like agent model for attribution of actions using ownership states and inverse mirroring

Abstract

This paper presents a neurologically inspired human-like agent model addressing attribution of actions to agents. It is not only capable of attribution of own actions to itself, but also to other agents, as for patients suffering from Schizophrenia. The mechanisms underlying the model involve ownership states and inverse mirroring to generate a mental image of the agent to which an action is attributed. The model is adaptive in that the inverse mirroring can develop based on Hebbian learning. The model provides a basis for applications to human-like virtual agents in the context of for example, training of therapists or agent-based generation of virtual stories. © 2012 Springer-Verlag.