Selection Dynamics and Adaptive Behavior without Much Information

This paper investigates whether behavior in a coordination game changes when subjects are limited to the information used by reinforcement learning algorithms. In the experiment subjects converge to an absorbing state at rates that are orders of magnitude faster than reinforcement learning algorithms, but slower than under complete information. Usually, this state is very close to a mutual best response outcome. All of the subjects are within a dime of giving a best response and 82.5 percent of the subjects gave a best response to the behavior of the other subjects in their cohorts without any information about their own best response function. The stability conditions derived from the best response dynamic are to conservative both under complete information and reinforcement information.