Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
The data from experiments with the Voluntary Contributions Mechanism suggest five stylized facts, including the restart effect. To date, no theory has explained all of these facts simultaneously. We merge our Individual Evolutionary Learning model with a variation of heterogeneous other-regarding preferences and a distribution of types to provide a new theory that does. In addition, our theory answers some open questions concerning the data on partners–strangers experiments. One interesting feature of the theory is that being a conditional cooperator is not a type but arises endogenously as a behavior. The data generated by our model are quantitatively similar to data from a variety of experiments, and experimenters, and are insensitive to moderate variations in the parameters of the model. That is, we have a robust explanation for most behavior in VCM experiments.