Validating a Hybrid Cognitive-System Dynamics Model of Team Interaction
Authors: Asmeret B. Bier, Michael L. Bernard
Abstract: Computational models of human behavior can lead to important insights regarding how people interact with each other and with their environments; but validation of these models is difficult and data is generally hard to collect. Better validation strategies could help to make these models more strongly justified and thus more useful. This paper describes an effort to study cooperation between teams in cyber security training exercises by building a model that captures the interactions between them. Real-world exercises provide a useful source of validation data and can serve to help calibrate the model. In this study we simulated two cyber scenarios where the primary difference was the intensity of cyber attacks experienced by two organizations. The model simulated the potential outcomes and decision-making processes involved in cooperative cyber security agreements designed to reduce redundant work. Insights learned from the model are intended to improve future versions of the cyber exercise. We also used validation data and insights from the exercise model to create and justify decision-making strategies in a model of a real-world counterpart to the situation exercises: information sharing programs for cyber defense.
Keywords: Cyber Security Training Exercise, Cooperation, Behavioral Influence Assessment, System Dynamics, Cognitive Model
Cite this paper: