The Influence of Evolutionary Selection Schemes on the Iterated Prisoner's Dilemma
We study Genetic Algorithms (GA) to simulate the emergence of cooperation in nonzero-sum and noncooperative competitions between different agents. The evolution of cooperation is not obvious in here since, in "nonzero-sum" competition, the benefits of one agent are not necessarily equal to the penalties of the other. Further, in "noncooperative" situations both participants have no information about their opponent's strategy. A famous and elegant example of a nonzero-sum/noncooperative competition is the so-called prisoner's dilemma (PD). A very interesting variate of the single-round PD game is the iterated PD (IPD) game, in which two players repeatedly play a PD. Under these conditions, the optimal strategy for a player depends on the policy of the opponent, as is often the case in real-life negotiations (e.g., oligopolistic price setting in the economic field or the nuclear arms race in the political arena). The analysis of IPDs using GAs has proved to be a very appealing technique to many. In previous studies, however, only strictly generational selection schemes are considered: after each generation, all parent strategies are replaced by offspring strategies. As is well known, the particular choice of GA selection scheme strongly affects the evolution of the population and the quality of the evolved strategies. Hence, in this study, we systematically assess the strong influence of various alternative selection schemes on the evolution of IPD strategies. The performance of an elitist selection scheme (only accepting offspring superior to parents) is evaluated. We also apply schemes intermediate to strictly generational and elitist schemes in which parents have a finite maximum life span. Simulations show the strong impact of the various evolutionary selection schemes on the dynamics of the population. This indicates the possible sensitivity of the level of cooperation in economic markets to the actual selection of the individual agents. Further work, especially using more detailed economic models, is hence strongly recommended.
1. Check below under "Related research" whether another version of this item is available online.
2. Check on the provider's web page whether it is in fact available.
3. Perform a search for a similarly titled item that would be available.
|Date of creation:||01 Mar 1999|
|Contact details of provider:|| Postal: CEF99, Boston College, Department of Economics, Chestnut Hill MA 02467 USA|
Web page: http://fmwww.bc.edu/CEF99/
More information through EDIRC
Please report citation or reference errors to , or , if you are the registered author of the cited work, log in to your RePEc Author Service profile, click on "citations" and make appropriate adjustments.:
- Ken Binmore & Nir Vulkan, 1999. "Applying game theory to automated negotiation," Netnomics, Springer, vol. 1(1), pages 1-9, October.
- Miller, John H., 1996. "The coevolution of automata in the repeated Prisoner's Dilemma," Journal of Economic Behavior & Organization, Elsevier, vol. 29(1), pages 87-112, January.
- Thomas Riechmann, 1999.
"Learning and behavioral stability An economic interpretation of genetic algorithms,"
Journal of Evolutionary Economics,
Springer, vol. 9(2), pages 225-242.
- Riechmann, Thomas, 1997. "Learning and Behavoiral Stability - An Economic Interpretation of Genetic Algorithms," Hannover Economic Papers (HEP) dp-209, Leibniz Universität Hannover, Wirtschaftswissenschaftliche Fakultät.