Modeling the Effects of Behavior Moderators for Simulation-Based Human Factors Design

Abstract

Designing systems, interfaces, procedures and artifacts in simulated environments before they are developed and deployed has the potential to greatly decrease the costs of design and development and, in some cases, can provide significant safety advantages. Creating realistic models of humans is an important aspect of the modeling problem, but existing models tend to model typical humans and fail to account for the significant differences seen from person to person or even by the same person in different circumstances. In the modeling literature, models of the factors that lead to such differences (including personality, affect, training, etc.) are typically called behavior moderators or performance moderators. This paper describes the MINDS (Modeling INdividual Differences and Stressors) project, which builds on previous work in behavior moderator modeling by supporting richer representations of moderators, moderator dynamics, and moderator interactions and by providing moderator-integration approaches for common behavior-modeling technologies, including production rules, fuzzy logic, and Bayesian networks. We provide a demonstration scenario from a military-operation domain.