Common reasoning in games (with Robert Sugden)
|Speaker:||Robin Cubitt, University of Nottingham|
|Date:||Friday 9 June 2006|
|Location:||Lecture Room D, Streatham Court|
The game-theoretic assumption of ‘common knowledge of rationality’ leads to paradoxes when rationality is represented in a Bayesian framework as cautious expected utility maximization with independent beliefs (ICEU). We diagnose and resolve these paradoxes by presenting a new class of formal models of players’ reasoning in which the analogue of common knowledge is provability in common reason. We show that a range of standards of decision-theoretic practical rationality can be assumed without inconsistency to be provable in common reason in models of this class. We investigate the implications arising when the standard of decision-theoretic rationality so assumed is ICEU.