More Realism in the Prisoner's Dilemma
One important assumption made in the standard Prisoner's Dilemma game, and one for which Robert Axelrod's (1984) work has been criticized, is the assumption that an individual has only two extreme choices when interacting with another individual. Robert Trivers (1985) also stated an interest in seeing if the same results could be achieved in a more complex model. In response to these ideas, a computer simulation was run with an expanded payoff (5×5) matrix that was created by interpolating values from the original payoff matrix used by Axelrod.
A PDF version of the paper is downloadable.