In a laboratory experiment, we use a public goods game to examine the hypothesis that human subjects use an involuntary eye-detector mechanism for evaluating the level of privacy. Half of our subjects are “watched” by images of a robot presented on their computer screen. The robot—named Kismet and invented at MIT—is constructed from objects that are obviously not human with the exception of its eyes. In our experiment, Kismet produces a significant difference in behavior that is not consistent with existing (...) economic models of preferences, either self- or other-regarding. Subjects who are “watched” by Kismet contribute 29% more to the public good than do subjects in the same setting without Kismet. (shrink)

Previous research in North America has supported the view that male involvement in committed, romantic relationships is associated with lower testosterone (T) levels. Here, we test the prediction that undergraduate men involved in committed, romantic relationships (paired) will have lower T levels than men not involved in such relationships (unpaired). Further, we also test whether these differences are more apparent in samples collected later, rather than earlier, in the day. For this study, 107 undergraduate men filled out a questionnaire and (...) collected one saliva sample (from which a subject’s T level was measured) at various times across the day. As in previous studies, men involved in committed, romantic relationships had lower salivary T levels, though only during later times of the day. Furthermore, additional analysis of the variation among unpaired subjects indicated that men without prior relationship experience had lower T levels than experienced men. Finally, while paired men as a group had lower T levels than unpaired men, those men at the earliest stage (less than six months) of a current relationship had higher T levels than unpaired men as well as men in longer-term relationships. These results suggest that variation in male testosterone levels may reflect differential behavioral allocation to mating effort. (shrink)

Although provocative, the data reported in Henrich et al.'s target article suffer from limitations, including the fact that the “selfishness axiom” is not an interesting null hypothesis, and the intrinsic limitations of quasi-experimental designs, in which random assignment is impossible. True experiments, in the laboratory or in the field, will continue to be crucial for settling core issues associated with human economic behavior.