Please note: This PhD seminar will take place online.
Aarti Malhotra, PhD candidate
David R. Cheriton School of Computer Science
Supervisor: Professor Jesse Hoey
Emotions are known to influence humans’ feeling, thinking, behaving and surviving in general. Artificial Intelligence (AI) research has largely focused on rational thinking, decision-making, goal achievement, and reward maximization. AI agents have been given deliberative capabilities to fulfill tasks and to interact with the environment, but there has been relatively limited focus on making them emotionally intelligent and able to handle a social world. Emotions are studied more from a detection and generation perspective, but only some studies focus on utilizing emotions in decision-making and behavior.
We systematically survey existing computational models of emotions in decision-making that are implemented in the form of computer simulations or systems. This brings forward a unified perspective to this interdisciplinary topic, while highlighting practical usability of such systems. Based on 129 articles analyzed, we found 102 unique models and extracted features related to the emotions, decision-making and the link between them. We synthesize the extracted data into four conceptual model types, viz. Matching, Appraisal and Coping, Decision-theoretic and Parametric and provide a process view of each type. These range from simple to complex computational mechanisms. Additionally, we extract supplementary implementation information for all the articles. This survey may help researchers and practitioners to advance affective computing, in making socially intelligent agents more emotionally aligned.