Abstract

I examine quantitative theories that make clear predictions about > long-term play in the laboratory based solely on the experimental instructions and information about the population. Such theories, even if they are correct only some of the time, enables us to cumulate knowledge about human behavior rather than having a series of disconnected results and theories. I refer to such theories as benchmark theories. I show that psychological theories with hard-wired social preferences fare poorly, while welfare maximization that allows for both noisy play and willingness to sacrifice for the common good does relatively well.

Practical information