Within epsilon of optimal play in the cultaptation social learning game

Social learning, in which members of a society learn by observing the behavior of others, is an important foundation for human culture, and is observed in many other species as well. It seems natural to assume that social learning evolved due to the inherent superiority of copying others' success rather than learning on one's own via trial-and-error innovation. However, there has also been substantial work questioning this intuition [3, 5, 1, 6, 4]. For example, blindly copying information from others is not useful if the information is wrong---or if it once was right but has since become outdated. Under what conditions does social learning outperform trial-and-error learning, and what kinds of social-learning strategies are likely to perform well?