Laboratory experiments have generally supported the theorem that, in classical property rights environments, noncooperative behavior in markets yields efficient social outcomes. Experiments, however, regularly fail to support the game theoretic prediction of noncooperative behavior in small-group strategic interaction and in large-group public good environments. In these two types of experiments subjects frequently achieve more efficient social outcomes–they collect more money from the experimenter–than noncooperative game theory predicts. Many subjects in these experiments exhibit reciprocity even in single-play games. Evolutionary psychologists hypothesize that humans have evolved mental algorithms for identifying and punishing cheaters in social exchange. (JEL A11, A12, B41, C70, C72, C92)