The evolution of optimism: A multi-agent based model of adaptive bias in human judgment

摘要

Human judgment under uncertainty has been shown to involve consistent departures from normative rationality, but it is hard to tell whether these biases are adaptive or not. We have attempted to throw light on this question by constructing a multi-agent based computer model. In the model, a variety of agents with different decision rules are allowed to compete in a variety of environments. We find that under certain environmental conditions biased agents who behave in ways similar to those observed in humans outperform the classically rational agent, who acts purely to maximize expected utility. These conditions are similar to those in which humans find themselves.