This paper uses an agent-based model with an adapted stag hunt style scenario to explore the role of the social transmission of correct information about stag hunting and potentially incorrect information about the costs of defection on cooperation in a small artificial society. The computational architecture of the model draws upon Daniel Sperber and Hugo Mercier’s concept of epistemic vigilance as well as Brian Skyrms’ work on cooperation in stag-hunt scenarios. In the model, communities of 100 hunters begin with no knowledge of stag hunting or punishment for defection and via imperfect social learning, guided by source or content vigilance, move toward a stag hunting or hare hunting equilibrium, where stag hunting may be motivated by the expectation of cooperation or by the fear of punishment. Most successful communities end up using content vigilance to determine their beliefs regarding stag hunting but use source vigilance to determine their beliefs regarding punishment, as predicted in the theoretical work of Konrad Talmont-Kaminski. These findings contribute to the ongoing debate in a variety of disciplines about the conditions under which—and the mechanisms by which—cooperation emerges and is maintained in human societies.