Accounting for Terrorist Behavior in Allocating Defensive Counterterrorism Resources

Today's de facto game‐theoretic models assume payoffs given by the von Neumann–Morgenstern expected utility theory. This is not necessarily descriptive of terrorist behavior in real‐world contexts. Terrorists often exhibit a type of strategy‐induced substitution or behavior referred to as transference. Modified prospect theory decision weights are proposed to model transference. To account for the defender's incomplete information, terrorists’ target choice probabilities are modeled with a logit distribution using endogenously determined decision weights. The strategic logit risk analysis (SLRA) method is developed as a framework for integrating the two behavioral models. It accounts for the feedback loop between the allocation of defensive resources and attack probabilities characteristic of adversarial dynamics. For the resource‐allocation optimization problem, SLRA provides a link between traditional safety probabilistic risk analysis (PRA) and game‐theoretic models. A simple but realistic example consisting of three potential targets is used to explore the impact of the proposed behavioral models and differences between minmax, Nash equilibrium, traditional safety PRA, and SLRA strategies. There are profound differences between these models with significant impact on the allocation of counterterrorism defensive resources.