Pii: S1090-5138(99)00018-5

Reciprocal altruism in humans may be made possible in part by the existence of information processing mechanisms for the detection of overt cheating. However, cheating may not always be readily detectable due to the division of labor. Subtle cheating poses a serious problem for the evolution of altruism. This article argues that subtle cheating may have exerted selective pressures on early hominids to be sensitive to information regarding the genuineness of an altruistic act. In two experiments, subjects were required to complete Wason selection tasks designed to allow for the detection of altruism. Performance on the altruist-detection tasks was compared to performance on control Wason selection tasks (Experiment 1) and to performance on control and cheater detection tasks (Experiment 2). Participants were significantly better at solving cheater-detection and altruist-detection versions compared to control versions of the problems, and there was no significant difference between altruist-detection and cheater-detection. Results are discussed in relation to recent conceptual models for the evolution of altruism. Specifically, it is argued that non-kin altruism may be an evolutionarily stable strategy if altruists can detect one another and form mutually beneficial social support networks. © 2000 Elsevier Science Inc.