Computational Models of Emotion: A Brief Survey

Abstract

Computational models of emotion (CMEs) have emerged as powerful tools for enabling the development of emotion-inspired intelligent artificial systems and advancing affective science by formalizing theories and enabling systematic experimentation. This chapter provides an overview of CMEs, examining their development, theoretical foundations, implementations, and applications across multiple disciplines, including artificial intelligence, human–computer interaction, and psychology. The authors explore three primary motivations driving CME research: enhancing adaptive capabilities in artificial agents, creating emotion-aware systems for human interaction, and formalizing psychological theories of emotion. They discuss the predominant emotion theories underlying CMEs, particularly dimensional and appraisal theories, and review the main computational approaches, including logic-based systems, decision-theoretical frameworks and reinforcement learning, and data-driven machine learning techniques. They further examine specialized models addressing complex emotional phenomena such as reverse appraisal, emotional contagion, and emotion regulation. Despite considerable progress in CMEs, challenges persist in adapting models to new domains and evaluation methodologies. The chapter highlights how CMEs serve as valuable bridges between affective science and affective computing that could help advance our understanding of emotion’s role in cognition and behavior, facilitating not only practical applications in social robotics and human–computer interaction but also theoretical refinement through computational formalization and simulation experiments.