Integrative Modeling of Emotions in Virtual Agents

Abstract

To develop a robot that is able to recognize and show affective behavior, it should be able to regulate simultaneously occurring tendencies of positive and negative emotions. To achieve this, the current paper introduces a computational model for involvement-distance trade-offs, based on an existing theoretical model. The main mechanisms of this model have been represented as regression equations, using the LEADSTO modeling environment. A number of simulation experiments have been performed, which indicated that the model is adequate for simulating the dynamics of involvement-distance trade-offs and their influence on satisfaction. More specifically, the experiments confirmed the empirical finding that positive features do not exclusively increase involvement.