Trust, Fear, Reciprocity, and Altruism*

This paper uses a triadic experimental design to discriminate between actions motivated by preferences over the distribution of material outcomes and actions motivated by attributions of the intentions of others. Such discrimination is essential to empirical guidance for theory development because modeling intentions is quite different than modeling preferences. The triadic design includes the moonlighting game in which first-mover actions can motivate positivelyor negatively-reciprocal actions by second movers. Thus, first movers can be motivated by trust (in positive reciprocity) or fear (of negative reciprocity) in addition to selfish, altruistic, or inequality-averse preferences. Second movers can be motivated by altruistic, inequality-averse, or selfish preferences as well as positive or negative reciprocity. The triadic design includes specially-designed dictator control treatments to discriminate among actions with alternative motivations. Data from the experiment support the conclusion that first movers’ behavior in the moonlighting game is characterized by trust in positive reciprocity and an absence of fear of negative reciprocity. Furthermore, the first movers’ behavior is based on rational expectations because the second movers’ behavior is characterized by positive reciprocity but not by significant negative reciprocity.