Partial Imitation Rule in Iterated Prisoner Dilemma Game on a Square Lattice

t] A realistic replacement of the general imitation rule in the Iterated Prisoner Dilemma (IPD) is investigated with simulation on square lattice, whereby the player, with finite memory, can only imitate those behaviors of the opponents observed in past games. In contrast to standard practice where all the possible behaviors of the opponents are accessible, the new partial imitation rule assumes that the player can at most access those behaviors of his opponent observed in past few moves. This partial imitation of the behavior in IPD shows very different outcomes in the long time behavior of the games, such as the ranking of various strategies. In particular, the well known tit-for-tat (TFT) strategy loses its importance.

• Type: Journal
• Year: 2010
• Where: NICSO
• Authors: Degang Wu, Mathis Antony, Kwok Yip Szeto
• Keywords: Artificial Intelligence, Behavior, Imitation Rule, NICSO 2010, Partial Imitation