When is Similarity-biased Social Learning Adaptive?

Cultural evolution theorists have suggested that humans employ similarity-biased social learning - we tend to imitate people who are like us. Informal evolutionary explanations have been offered for the bias. Similarity-biased learning might enable the smooth acquisition of social roles which sustain conventions and norms. This paper describes a formal model designed to uncover when similarity-biased learning is, and is not, adaptively advantageous. Whether we should expect similarity-biased social learning to evolve strongly depends on assumptions about the adaptive function of social roles, the initial conditions, a variety of parameter settings, and the population structure. Making small changes to these assumptions can collapse the explanation. The results suggest we should be very cautious about claims suggesting there is a universal, evolved tendency towards similarity-biased learning