Entropic Selection of Nash Equilibrium*
Mehmet Barlo (Sabancı University)
Date: Friday 2 November 2012,
Hour: 13:30
Place: Natuk Birkan Building
Murat Sertel Faculty Lounge
Abstract: Observing that more variable best responses necessitate more precise anticipation and imply more complex behavior, we follow Shannon (1948) of information theory to measure the information content, i.e. variations, in best responses via a weak form of entropy. This delivers a refinement based on complexity aversion: The entropic selection of Nash equilibrium. We prove that given any normal form game this selection is non-empty, admits desirable properties, and does not have containment relations with perfection, properness and persistence. In fact, in well-known games that display important insights about virtues/problems of various refinement notions, our notion displays none of the criticisms associated with these examples.
*-joint with Zeynel Harun Alioğulları
Boğaziçi Üniversitesi
Ekonomi Bölümü
Natuk Birkan Binası K.2
Tel: 0212 359 6505
Faks: 0212 287 2453
