Individual and cultural learning in stag hunt games with multiple actions

We compare cultural learning and individualistic, belief-based learning in a class of generalized stag hunt games. Agents can choose from among multiple potentially cooperative actions or can take a secure, self-interested action. We assume that a proportion of the cooperative actions prove effective, while others can be undermined by a predatory action. For this class of games, the set of stable equilibria is identical under the two learning rules. However, we show that the basins of attraction for the efficient equilibria are much larger for cultural learning. We further show that as the stakes grow arbitrarily large, cultural learning always locates an efficient equilibrium while belief-based learning never does.