Giving it all away: altruism and answers to the Wason selection task

The Wason selection task, a standard test of conditional reasoning, has featured prominently in experimental studies of cognitive adaptations for cooperation. The most prominent of these is Cosmides' investigations of cheater detection on social contract versions of the Wason selection task [Cognition 31 (1989) 187–276]. Subsequent to Cosmides' initial investigations, several researchers [Evol Hum Behav 21 (200) 25–37; Manage Decis Econ 19 (1998) 467–480; J Genet Psychol 163 (2002) 425–444; Evol Hum Behav 27 (2006) 366–380] have argued that people also are competent at detecting altruism on the Wason selection task, suggesting that there is nothing privileged about the detection of cheaters. However, an analysis of the selection tasks on which these claims are based suggests that participants may have solved these altruism-detection tasks correctly because the scenarios explicitly or implicitly provide the answer to the task in the scenario [Evol Hum Behav 21 (200) 25–37; Manage Decis Econ 19 (1998) 467–480; J Genet Psychol 163 (2002) 425–444], or due to confounds in the cheater-detection tasks leading to the (misleading) appearance of enhanced altruist-detection performance [Evol Hum Behav 27 (2006) 366–380]. We tested our conjecture by giving participants selection tasks with and without the answer embedded in the scenario. Performance dropped significantly on the altruism-detection tasks when the embedded answers were removed, whereas performance on cheater-detection versions was unaffected by the manipulation. A reanalysis of the findings of Oda et al. suggested that participants performed significantly worse on their altruism-detection problems than their cheater-detection problems — a finding that we replicate after removing confounds from the cheater-detection tasks of Oda et al. The results reaffirm the specificity of cheater-detection.