(Joint work with Prof. Venkat Anantharam) Cumulative prospect theory (CPT) is a framework pioneered by Kahneman and Tversky, which is believed, based on extensive experimentation with human subjects, to form a better theory with which to model human behavior when faced with choices than is expected utility theory (EUT). Given the pervasive role of humans as agents in networks (e.g. social networks) and markets (e.g. labor markets) building mechanisms based on presumably more accurate models of human behavior is of great interest both for increasing human welfare and for building more efficient commercial systems that interact with humans. It is important to emphasize that CPT includes EUT as a special case and therefore provides a strict generalization of existing modeling techniques. Game theoretic models are pervasive in the study of interactions between autonomous agents. There is a rich theory for EUT agents. Game theory for CPT agents, however, is barely getting off the ground. In this talk, we present some of our findings concerning the following three topics that have occupied a central role in game theory research with a twist where the interacting agents have CPT preferences: (i) Game equilibrium when agents are strategic, namely, Nash equilibrium and correlated equilibrium. (ii) Repeated game dynamics when agents employ learning strategies. (iii) Optimal resource allocation over networks. The talk will give a gentle introduction to the underlying ideas of CPT, which may be unfamiliar to many.
Soham is a fifth year graduate student in the EECS department at UC Berkeley, advised by Prof. Venkat Anantharam. He graduated from IIT Bombay with a B.Tech. degree in electrical engineering (advised by Prof. Vivek Borkar) with honors and a minor in computer science in 2015. He is a recipient of the Best Paper Award at the 9th EAI International Conference on Game Theory for Networks (GameNets 2019). His research interests include cooperative and non-cooperative game theory, network economics, behavioral economics, decision analysis, and optimization.