Population Learning in a Model with Rugged Payoff Landscapes and Endogenous Networks