Newtonian Emotion System

Abstract

The main goal of our work is to provide virtual characters with the ability of emotion expression. Our approach is based on the idea that emotions serve a fundamental function in human behavior, and that in order to provide artificial agents with rich believable affective behavior, we need to develop an artificial system that serves the same purpose. We believe that emotions act as a subsystem that enhances human behavior, by stepping up brain activity in arousing circumstances, directing attention and behavior, establishing importance of events and act as motivation. This paper summarizes the psychological theories that our computational emotion model is based on, defines the key concepts of the Newtonian emotion model that we developed and describes how they work within an agent architecture. The Newtonian emotion model is a light-weight and scalable emotion representation and evaluation model to be used by virtual agents. We also designed a plug-and-play emotion subsystem to be integrated into any agent architecture.