Computational models of emotions for autonomous agents: major challenges

Abstract

A great number of computational models of emotions (CMEs) have been developed to be included in, or as part of, cognitive agent architectures. These computational models have been designed to provide autonomous agents with appropriate mechanisms for the processing of emotional stimuli, the elicitation of synthetic emotions, and the generation of emotional responses. The research on CMEs has allowed for improvements in several application domains and contributed to progress in areas such as human-computer interaction and artificial intelligence. Nevertheless, despite the wide variety of CMEs proposed in the literature and their success in multiple areas, the complexity and quality of current and future human-centered applications require the development of more flexible and robust CMEs. In this sense, CMEs have yet to face a series of challenges in order to meet such types of requirements. In this paper, we explore key aspects of the development and applications of CMEs for autonomous agents, discuss four major challenges facing their development process, and present a novel approach to deal with these challenges.