Seminario Datalab

Behavioral Models of Opponents in Competitive Interactions: Theory and Applications

Ponente:  William Caballero (United States Air Force Academy)
Fecha:  lunes 22 de noviembre de 2021 - 15:00
Lugar:  Online - (ID: 821 7068 1298)


Over the last decade, adversarial risk analysis (ARA) has proven itself to be an effective decision support tool in non-cooperative settings. Whereas conventional game theory treats the competition as a system and seeks to find its equilibrium, ARA provides tailored decision support to a given agent under Bayesian assumptions. However, when solving an ARA problem involving human opponents, analysts should remain cognizant of empirical testing that has illustrated their bounded rationality. The perfect rationality assumptions frequently utilized within traditional game theory has often been the basis of its criticism. Depending on the setting, ARA may be susceptible to the same critique. Therefore, we discuss some of the most prevalent descriptive models of choice from the behavioral economic and behavioral game theory literature as well as their potential incorporation into ARA modeling. We also present recent literature that has leveraged such behavioral models for decision support and discuss its relation to the growing ARA body of literature. In so doing, we provision ARA researchers with additional modeling tools for use when encountering boundedly rational agents.