The effect of matching mechanism on learning in games played under limited information

Nick Feltovich and Sobei H. Oda

We report results from a laboratory experiment that examines the effect on learning of the mechanism used to match individuals to opponents in two-player games. Several games are played repeatedly under either a fixed-pairs or a random matching treatment. Unlike most economics experiments, the games are played under limited information: subjects are never shown the games' payoff matrices nor given information about opponent payoffs. We find that behavior in the treatments, while similar in early rounds, diverges over time. Depending on the game, the matching mechanism can affect which of multiple equilibria is likely to be reached, the speed of convergence, or other aspects of outcomes. Usually but not always, fixed-pairs matching is associated with more likely coordination on a pure-strategy Nash equilibrium, more likely play of a higher-payoff Nash equilibrium, and faster convergence toward pure-strategy play.