Limited evidence that reputation-based partner choice facilitates information sharing in humans.

A necessary prerequisite for the accumulation of beneficial knowledge, or 'cumulative cultural evolution', is the sharing of information via social learning. Yet little work in the field of cultural evolution has examined the mechanisms that support information sharing in the face of exploitative information free-riding and information hoarding. We ran a series of online interactive experiments (N = 716) combined with computational reinforcement and social learning models to test whether the mechanism of reputation-based partner choice can effectively support information sharing. Participants in groups chose whether to (i) engage in costly innovation and (ii) whether to share the resultant knowledge. Sharers received increased reputations for sharing and participants could use reputations to select recipients of knowledge. We found strong priors for information sharing that persisted throughout the experiments in participants from both the UK and China (study 1). However, partner choice was generally too weak to explain the presence of widespread information sharing, which persisted even when we reduced the benefit of innovation (study 2), introduced exploitative bots (study 3) and removed reputations altogether (study 4). Our results suggest that indiscriminate, group-based sharing is more important for facilitating cumulative cultural evolution than discriminate reputation-based sharing.