This series of talks was part of the Economics and Computation Boot Camp. Videos for each talk area available through the links above.
Speaker: Drew Fudenberg, Harvard University
When and why will observed play in a game approximate an equilibrium? What sort of equilibrium? To understand how and when equilibrium arises, we will look at the long-run behavior of plausible non-equilibrium dynamic processes. Lecture 1 will consider learning in static or 1-shot games, where Nash equilibrium has been found to be a good description of the outcomes of some (but not all) game theory experiment. Here subjects eventually play as if they have learned the aggregate distribution of play in the population. Lecture 2 will consider learning in extensive form games, where mistaken beliefs about opponents' play are more likely to persist (both theoretically and empirically) due to the tradeoff between exploration and exploitation.