 |
Center for Nonlinear Studies |
This is the first of several lectures presenting the fundamentals of how game theoreticians view the world, together with some extensions reflecting recent experimental results, and reflecting more of a physics / machine learning view of the world.
In this lecture, I will first review single-stage noncooperative games, giving several classical examples of "full rationality" player behavior. I will then introduce two recent models of bounded rational behavior, the Quantal Response Equilibrium, and Level K satisficing models.
I will then show how to combine Level K satisficing models with Bayes nets, to predict behavior in multi-stage games. I will illustrate this hybrid model by using it to predict the behavior of pilots in near mid-air collisions.
Host: Robert Ecke and Eddy Timmermans, CNLS, 667-1444