Stubborn Agents: examining the underutilization of social learning

Abstract

**Author(s):** Restrepo Ochoa, N; Acerbi, A; Introduction: Individual exploration and social learning play a key role in cultural evolution. If culture is the key to human success, social learning is the mechanism that allows for the transmission and refinement of cultural tools. We have evidence that people are keen foragers of social information. However, recent work also highlights a contradictory finding: people, in many circumstances, underuse social information – i.e. they copy others less than what would be predicted by theoretical models. In this talk, I will present a model that examines this tension. Methods: This is a simple model, where agents live in an environment that consists of a fixed number of "traits". Each trait changes with a specific probability, and agents try to track these environmental changes by either exploring individually or learning from one another. We vary the extent to which the environment changes, as well as the accuracy of social learning. Results: We show that, when we parametrize the model using rate of changes comparable to the experimental conditions under which social learning has been examined, exploration and social learning are worthwhile. However, we also explore more stable cultural environments, simulating distributions where few traits change frequently and the majority change rarely, plausible given current evidence about the rate of cultural change. In these situations, learning can often be counterproductive, especially if it is subject to error and/or if there is the possibility of deception, and the best strategy is not changing much at all. Discussion: Given realistic conditions of environmental change, our agents use social learning at rates comparable to the rates observed experimentally. A key implication is that we need more empirical work that seeks to measure rates of environmental change (and not only cultural change).