Granny versus game theorist
ambiguity in experimental games
pp. 183-204
in: Mohammed Abdellaoui, John D. Hey (eds), Advances in decision making under risk and uncertainty, Berlin, Springer, 2008Abstract
In standard game theory, strategic uncertainty in games is resolved in Nash equilibrium, at least for games with a unique Nash equilibrium. Given a player's equilibrium conjecture about opponents' play, she chooses a best response that conforms to the opponents' equilibrium conjecture about her play. What if players lack confidence in their equilibrium conjectures about opponents' play? This is plausible especially if the game is one-shot and players lack previous experience with the same opponents. Lack of confidence in probability judgements is modelled formally by the literature on ambiguity or Knightian uncertainty (Schmeidler, 1989; Gilboa & Schmeidler, 1989; Bewley, 1986). Recently, such approaches have been applied to strategic games (Dow & da Costa Werlang, 1994; Eichberger & Kelsey, 2000; Marinacci, 2000).1 Results on the comparative statics of equilibrium under ambiguity have been derived that should at least in principle be testable (Eichberger & Kelsey, 2002, 2005; Eichberger, Kelsey, & Schipper, 2006).