Proceedings of the Annual Meeting of the Cognitive Science Society

Game Theory -based decision aids have been successfully em-ployed in real-world policing, anti-terrorism, and wildlife con-servation efforts (Tambe, Jiang, An, & Jain, 2013). Cognitivemodeling, in concert with model tracing and dynamic parame-ter fitting techniques, may be used to improve the performanceof such decision aids by predicting individual attacker behav-ior in repeated security games. We present three simulations,showing that (1) cognitive modeling can aid in greatly improv-ing decision-aid performance in the security domain; and (2)despite the fact that individual attackers will differ in initialpreferences and in how they learn, model parameters can beadjusted dynamically to make useful predictions for each at-tacker.