Title: | Bounded Rationality via Recursion |

Author: | Maciej Latek, Robert Axtell, and Bogumil Kaminski |

Book Tittle: | Proc. of the 8th Int. Conf. on Autonomous Agents and Multi-Agent Systems |

Editor: | Decker, Sichman nad Sierra, and Castelfranchi |

Pages: | 457--464 |

Year: | 2009 |

Crossref: | aamas09 |

Abstract: | Current trends in model construction in the field of agentbased computational economics base behavior of agents on either game theoretic procedures (e.g. belief learning, fictitious play, Bayesian learning) or are inspired by artificial intelligence (e.g. reinforcement learning). Evidence from experiments with human subjects puts the first approach in doubt, whereas the second one imposes significant computational and memory requirements on agents. In this paper, we introduce an efficient computational implementation of n-th order rationality using recursive simulation. An agent is n-th order rational if it determines its best response assuming that other agents are (n−1)-th order rational and zero-order agents behave according to a specified, non-strategic, rule. In recursive simulations, the simulated decision makers use simulation to inform their own decision making (search for best responses). Our goal is to provide agent modelers with an off-the-shelf implementation of n-th order rationality that leads to modelconsistent behaviors of agents, without requiring a learning phase. We extend two classic games (Shapley’s fictitious play and Colonel Blotto) to illustrate aspects of the n-th order rationality concept as implemented in our framework. |

Cited by 3

@InProceedings{latek09a, author = {Maciej Latek and Robert Axtell and Bogumil Kaminski}, title = {Bounded Rationality via Recursion}, booktitle = {Proc. of the 8th Int. Conf. on Autonomous Agents and Multi-Agent Systems}, crossref = {aamas09}, pages = {457--464}, year = 2009, editor = {Decker and Sichman nad Sierra and Castelfranchi}, abstract = {Current trends in model construction in the field of agentbased computational economics base behavior of agents on either game theoretic procedures (e.g. belief learning, fictitious play, Bayesian learning) or are inspired by artificial intelligence (e.g. reinforcement learning). Evidence from experiments with human subjects puts the first approach in doubt, whereas the second one imposes significant computational and memory requirements on agents. In this paper, we introduce an efficient computational implementation of n-th order rationality using recursive simulation. An agent is n-th order rational if it determines its best response assuming that other agents are (n−1)-th order rational and zero-order agents behave according to a specified, non-strategic, rule. In recursive simulations, the simulated decision makers use simulation to inform their own decision making (search for best responses). Our goal is to provide agent modelers with an off-the-shelf implementation of n-th order rationality that leads to modelconsistent behaviors of agents, without requiring a learning phase. We extend two classic games (Shapley’s fictitious play and Colonel Blotto) to illustrate aspects of the n-th order rationality concept as implemented in our framework.}, url = {http://jmvidal.cse.sc.edu/library/latek09a.pdf}, cluster = {185669515192180813} }Last modified: Wed Mar 9 10:16:57 EST 2011