Multi-Agent Classifier Systems and the Iterated Prisoner’s Dilemma

  • K. Chalk
  • G. D. Smith
Conference paper


This paper describes experiments using multiple classifier system (CS) agents to play the iterated prisoner’s dilemma (IPD) under various conditions. Our main interest is in how, and under what circumstances, co-operation is most likely to emerge through competition between these agents. Experiments are conducted with agents playing fixed strategies and other agents individually and in tournaments, with differing CS parameters. Performance improves when reward is stored and averaged over longer periods, and when a genetic algorithm (GA) is used more frequently. Increasing the memory of the system improves performance to a point, but long memories proved difficult to reinforce fully and performed less well.


Classifier System Finite State Automaton Multiple Classifier System Round Robin Tournament Finite State Automaton 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    R. Axelrod. Effective choice in the prisoner’s dilemma. Journal of Conflict Resolution, 24:3–25, 1980.CrossRefGoogle Scholar
  2. [2]
    R. Axelrod. More effective choice in the prisoner’s dilemma. Journal of Conflict Resolution, 24:379–403, 1980.CrossRefGoogle Scholar
  3. [3]
    R. Axelrod. The evolution of strategies in the iterated prisoner’s dilemma. In L. Davis, editor, Genetic Algorithms and Simulated Annealing, pages 32–41. Morgan Kaufmann, 1987.Google Scholar
  4. [4]
    J. Bendor, R.M. Kramer, and S. Stout. When in doubt: Cooperation in a noisy prisoner’s dilema. Journal of Conflict Resolution, 35(4), 1991.Google Scholar
  5. [5]
    K.G. Binmore and L. Samuelson. Evolutionary stability in repeated games played by finite automata. Journal of Economic Theory, 57, 1992.Google Scholar
  6. [6]
    A. Carbonaro, G. Casadei, and A. Palareti. Genetic algorithms and classifier systems in simulating co-operative behaviour. In R.F. Albrecht and C.R. Reeves, C.R. Steele, editors, Artificial Neural Networks and Genetic Algorithms, pages 479–483. Springer-Verlag, Wien New York, 1993.CrossRefGoogle Scholar
  7. [7]
    P.H. Crowley. Evolving cooperation: Strategies as hierachies of rules. BioSystems, 37:67–80, 1996.CrossRefGoogle Scholar
  8. [8]
    G. Ellison. Learning, local interaction, and coordination. Econometra, 61(5), 1993.Google Scholar
  9. [9]
    A. Fairley and D.F. Yates. Improving simple classifier systems to alleviate the problems of duplication, subsumsion and equivalence of rules. In R.F. Albrecht, C.R Reeves, and N.C. Steele, editors, Artificial Neural Nets and Genetic Algorithms. Springer/Verlag, 1993.Google Scholar
  10. [10]
    D.B. Fogel. On the relationship between the duration of an encounter and the evolution of cooperation in the iterated prisoner’s dilemma. Evolutionary Computation, 3(3):349–363, 1996.CrossRefGoogle Scholar
  11. [11]
    J.R. Hoffmann and N.C. Waring. The localisation of learning and interaction in the repeated prisoner’s dilemma. Draft Copy, University of East Anglia, 1996.Google Scholar
  12. [12]
    J.H. Holland. Processing and processors for scemata. In Jacks E.L., editor, Associative Information Processing, pages 127–146. American Elsevier, New York, 1971.Google Scholar
  13. [13]
    J.H. Holland. Properties of the bucket brigade algorithm. In Grefenstette J.J., editor, Proceedings of the First International Conference on Genetic Algorithms and Applications. Morgan Kaufmann, 1985.Google Scholar
  14. [14]
    J.H. Holland. The effect of labels (tags) on social interactions. Santa Fe Institute Discusion Paper 93-10-064, 1993.Google Scholar
  15. [15]
    O. Kirchamp. Spatial Evolution of Automata in the Prisoner’s Dilemma. PhD thesis, 1995.Google Scholar
  16. [16]
    Y. Mor, C.V. Goldman, and J.S. Rosenschein. Learn your opponents strategy (in polynomial time)! In: Adaptation and Learning in Multi-Agent Systems, (G. Weiss and S. Sen), pages 164–176, 1996.Google Scholar
  17. [17]
    J.F. Nash. Non-cooperative games. Annals of Mathematics, 54:286–295, 1951.MathSciNetMATHCrossRefGoogle Scholar
  18. [18]
    W. Poundstone. Prisoner’s Dilemma. Oxford University Press, 1993.Google Scholar
  19. [19]
    B.R. Routledge. Co-evolution and Spatial Interaction. PhD thesis, 1993.Google Scholar
  20. [20]
    A. Rubinstein. Finite automata in the repeated prisoner’s dilemma. Journal of Economic Theory, 39, 1986.Google Scholar
  21. [21]
    T.W. Sandholm and R.H. Crites. On multi-agent q-learning in a semi-competetive domain. In G. Weiss and S. Sen, editors, Adaptation and Learning in Multi-Agent Systems, pages 191–205. 1996.Google Scholar

Copyright information

© Springer-Verlag Wien 1998

Authors and Affiliations

  • K. Chalk
    • 1
  • G. D. Smith
    • 1
  1. 1.School of Information SystemsUniversity of East AngliaNorwichUK

Personalised recommendations