Accuracy and unbiased beliefs don’t pay. Overconfidence does. Preposterous but probably true:
Overconfidence is a major puzzle in evolutionary biology, economics and political science, because despite causing costly errors and policy failures, it remains a widespread bias in human judgement and decision-making. Animals too, in examples seen during conflict behaviour, are liable to overconfidence.
In the “Evolution of Overconfidence”(1), Dominic Johnson and James Fowler use game theory to model the situations under which overconfidence is an advantage, and find that they occupy a large part of the parameter space. They also show that overconfidence and conflict tend to increase with greater resource availability. As long as the prize at stake sufficiently exceeds the cost of competing for it, it seems, fortune favours the brave.
(1) Johnson, Dominic D. P., and James H. Fowler. “The Evolution of Overconfidence.” Nature 477, no. 7364 (September 15, 2011): 317–20. doi:10.1038/nature10384.