Simple and Clever Decision Rules in Single Population Evolutionary Models

This paper compares two decision rules available to myopic players who are repeatedly randomly matched to play a 2 x 2 symmetric game. Players using the simple decision rule evaluate the strategies by comparing their current playoffs to those of an opponent currently playing the other strategy, while those following the clever decision rule assess the strategies under the assumption that opponents' actions are fixed. It is shown that while populations of simple players can fail to learn to play Nash equilibria or even dominant strategy equilibria, populations of clever players always learn to play approximate Nash equilibria.