Superiority Bias and Communication Noise Can Enhance Collective Problem Solving

Error affects most human judgments and communications. Here we consider two types of error: unbiased noise and directional biases, and consider their effects in the context of collective problem solving. We studied an agent-based model of networked agents collectively searching for solutions to simple and complex problems on an NK landscape. We implemented superiority bias as a reluctance to adopt solutions used by others unless they were substantially better than one’s own solution. We implemented communication error by injecting noise into solutions learned from others. These factors both reduce the short-term efficiency of social learning, as individuals are less likely to faithfully copy superior solutions. We find that when a team faces complex problems, both communication noise and superiority bias have a positive effect on the overall quality of the team’s collective solution, at the cost of increased time and resource usage. We find that when a team faces simple problems, a moderate level of communication noise leads to a decrease in the required time and resources for a team. We discuss these results in terms of tradeoffs between the quality of a collective solution and the time and resources needed to reach that solution.