The Timing of Social Comparison in Crowds

Abstract

Abstract : Models of crowd behavior facilitate analysis and prediction of the behavior of large groups of people, who are affected by each other's presence and actions. For instance, in defense and security applications, generative models of crowd behaviors are used for decision-support, simulation, and training. Most existing approaches for modeling crowd behavior have focused on algorithmic and mathematical approaches, which generate simulations which are qualitatively or visually appealing, but have not been tied to social psychology, nor to cognitive architectures. In previous work, we proposed a novel model of crowd behavior, based on Social Comparison Theory SCT, a popular social psychology theory. The SCT model relies on simulated entities (agents) to compare themselves to others, but the timing of these comparisons is not well understood: People clearly do not imitate others all the time, yet there is evidence that shows that people (and therefore, the agents), do some comparison at all times (but do not at on these comparisons). While some progress has been made to address this question, it remains open. In this report, we present an extension of the SCT model to address this open question. We argue that comparisons take place all the time (i.e., differences are perceived and processed), but the cognitive architecture limits actions taken to minimize differences to cases where the comparisons yield significant differences. We use both toy domain experiments as well as movies of human pedestrians to argue for our position. Two corollaries of our work are (i) implications for the role of agent modeling and plan recognition in cognitive architectural mechanisms, and (ii) initial steps in accounting for group size in social comparison.