This paper reconsiders evidence from experimental common pool resource games from the perspective of a dynamic model of sampling. Despite being parameter-free, the model is able to replicate some striking features of the data: monotonic frequency distributions, the persistent use of strictly dominated actions, and stable heterogeneity in choices. We argue that these patterns cannot be fully accounted for by existing theories based on other-regarding preferences and norms, and that the dynamics of sampling provide a useful complementary explanation for behavior in social dilemmas.