Sampling dynamics applied to the traveler's dilemma reveals non-Nash behavior.

The traveler's dilemma is characterized by a strict Nash equilibrium that is Pareto-inefficient. Even though a coordinated choice of more efficient options would be preferred by both players, any individual deviation from Nash behavior will induce a lower payoff. Yet experimental behavior reported in the literature demonstrates that deviations from Nash behavior toward Pareto-better outcomes are typical. An application of the sampling dynamics to this dilemma further justifies the non-Nash behavior. Due to the fact, that the sampling dynamics is not payoff-monotonic, even dominated strategies can survive the selection process. It is shown that for a generalized traveler's dilemma an interior equilibrium attracts almost all trajectories, given that there are sufficiently many options. The limit for infinitely many options is derived. [ABSTRACT FROM AUTHOR]