I. Introduction
Whereas much of affective computing focuses on inferring a person’s emotions from expressive displays, often without regard to the situation that evoked these feelings [3], computational models of emotion seek to reason about situations, including how features of a situation may evoke certain emotions and how these emotions might change subsequent decisions, behavior and beliefs [1, 4]. Computational models have many uses. They predict how individuals respond in emergencies [5]. They model the way emotion influences decision-making [6, 7]. They drive the expressive behavior of virtual humans [8] and social robots [9]. They help inform psychological theory [10]. And they might help improve the accuracy of emotion recognition methods by allowing them to incorporate both expressions and information about situations [11].