Models to the issue of altruism which rely on externalities of well-being are rarely used explicitly. In this paper we compare such utility-based approaches with the standard approach on altruism which is based on externalities of income. Testable differences of both types of models are derived in the case of incomplete information. More specifically, applied to the Dictator Game and the Impunity Game both played under incomplete information, the utility-based based approach predicts dictators to change their behavior in comparison to Dictator Games under complete information. Under the income-based approach, behavior should not differ in the three versions of the Dictator Game.
Altruism Incomplete Information Consistent Expectations