An experiment was described to examine the choices made by rats in situations that can be modeled as 2×2 games, i. e, the three classes of constant-sum games and prisoner's dilemma. This paper deals with decision making in living systems at the level of the group. The rational choice in games of classes I and II is a pure strategy, which in repeated plays should manifest itself as the same choice throughout. The rational choice in games of class III is a mixed strategy, which in a single play must be determined by a chance device. In repeated plays, a mixed strategy can manifest itself in relative frequencies. Some learning models lead to fixation on either of the two strategies in prisoner's dilemma, depending on the initial condition. In this experiment, each subject played one of the four games 1000 times.