Human altruism seems to be problematical in a Darwinian world. How can organisms that reduce their fitness to promote the interests of their competitors survive natural selection? This dissertation formulates this problem as four propositions, each of which seem true, but the four are contradictory. The dissertation shows that one of these propositions, "Psychological altruism is a trait that results in an organism performing selectively disadvantaged acts" is false as stated. Instead, the dissertation argues that altruistic acts can be selectively advantaged if the altruistic agent's reputation results in its participating in additional profitable transactions. The dissertation first develops a precise definition of altruism suitable for mathematical modeling. The mathematical nature of this definition provides a way of making some of altruism's complexities and nuances more explicit. Then the dissertation develops a modification of the Prisoner's Dilemma and uses this game in mathematical models and a computer simulation to analyze the emergence of altruism in two societal situations. The models develop criteria on the frequency and magnitude of altruistic acts that will allow altruistic agents to invade and persist in a population of cooperators and egoists. The dissertation also presents a model that shows that even egoistic agents must consider altruistic acts if they wish to maximize their expected fitness, because fitness may depend upon the stability of the agent's group.

An implicit assumption of all models that demonstrate the evolution of pro-social behavior is that there are groups of individuals with sufficient cohesion to enable a large number of interactions. The dissertation presents a computer simulation based upon a model that explains foraging behavior in the great apes that specifies the conditions that enable formation of coalitions amongst egoistic independent organisms. A more detailed simulation explores the kind of proto-altruistic behavior that increases the likelihood of survival of these coalitions.