Measuring Motivations for the Reciprocal Responses Observed in a Simple Dilemma Game.

We report on an experiment that uses a simple dilemma game to compare two characterizations of reciprocal behavior, differing on whether it is necesary to account for intentions in order to accurately predict and measure reciprocal actions. The design of the experiment allows us to study both positive and negative reciprocity in a simple framework. We find little evidence that intentions matter. Knowing the preferences for relative payoffs is sufficient to predict nearly all the reciprocity we observe.