The Evolution of Good and Evil

Abstract

How can we explain why cognitive biases are not wiped away by natural selection? Can other-regarding agents evolve out of self-regarding agents? This article develops a model of the co-evolution of morality (i.e. tendency to internalize externalities of one's own actions) with overconfidence (i.e. tendency to overestimate one's own ability). Nature presents agents with the choice to accept or reject a lottery with randomly drawn parameters. Parameters of the lottery determine its objective expected individual and group payoffs. Each agent's bias parameter determines how the objective expected payoffs of the lottery are transformed into subjective expected payoffs on which the agent's decision rule operates. The agent's morality parameter determines the weight of payoffs to others in the agent's decision rule. The model shows how the individual objective payoff-maximizing level of morality varies with cognitive bias, resulting in a positive level of morality as long as individuals are sufficiently biased and covariance of group and individual payoffs is positive.