In a highly controversial paper by Robert Aumann(see here), it is stated as a theorem:

In PI games, common knowledge of rationality implies backward induction.

If we stick to the strong and controversial definition of rationality in the paper

Rationality of a player means that he is a habitual payoff maximizer: that no matter where he finds himself, at which vertex,he will not knowingly continue with a strategy that yields him less than he could have gotten with a different strategy.

Can we have some other implication, like, If a game admits a unique Nash equilibirum, does common knowledge of rationality implies Nash equilibirum?

  • $\begingroup$ The first quote should be "In perfect information games, common ...", which makes it much less controversial. $\endgroup$ Commented Apr 4, 2015 at 21:46
  • $\begingroup$ @TheAlmightyBob You're right! The controversies are different in two settings. $\endgroup$ Commented Apr 4, 2015 at 21:55

1 Answer 1


No. In static games, common knowledge of rationality is equivalent to rationalizability. Bernheim, in "Rationalizable Strategic Behavior" (Econometrica, July 1984) gives an example on page 1012 in Figure 1 of a normalform game in which there is a unique Nash equilibrium, yet multiple strategies are rationalizable.

  • $\begingroup$ You are looking at David Pearce's paper, but you should look at Doug Bernheim's paper in the same issue of Econometrica. $\endgroup$ Commented Apr 6, 2015 at 18:41
  • $\begingroup$ My bad. Sorry : ) $\endgroup$ Commented Apr 6, 2015 at 18:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.