A systematic validation of evolutionary techniques in the field of bargaining is presented. For this purpose, the dynamic and equilibrium-selecting behavior of a multi-agent system consisting of adaptive bargaining agents is investigated. The agents' bargaining strategies are updated by an evolutionary algorithm (EA), an innovative computational method to simulate collective learnin g in societies of boundedly-rational agents. Negotiations between the agents are governed by the well-known``alternating-offers' protocol. Using this protocol, the influence of various important factors (like the finite length of the game, time preferences, exogenous breakdown, and risk aversiveness) is investigated. We show that game theory can be used successfully to interpret the equilibrium-selecting behavior observed in computational experiments with adaptive bargaining agents. Agreement between theory and experiment is especially good when the agents experience an intermediate time pressure. Deviations from classical game theory are, however, observed in several experiments. Violent nonlinear oscillations may for instance occur in the single-stage ultimatum game. We demonstrate that the specific evolutionary model governing agent selection is an important factor under these conditions. In multiple-stage games, the evolving agents do not always fully perceive and exploit the finite horizon of the game (even when time pressure is weak). This effect can be attributed to the boundedly-rational behavior of the adapting agents. Furthermore, when the agents discount their payoffs at a different rate, the agent with the largest discount factor is not able to exploit his bargaining power completely, being under pressure by his impatient opponent to reach an early agreement. Negotiations over multiple issues, a particularly important aspect of electronic trading, are studied in a companion paper cite{Gerding:00. We are currently investigating the behavior of more complex and powerful bargaining agents.

Optimization (acm G.1.6), Learning (acm I.2.6), Problem Solving, Control Methods, and Search (acm I.2.8)
Relaxation oscillations (msc 34C26), 2-person games (msc 91A05), Noncooperative games (msc 91A10), Games in extensive form (msc 91A18), Multistage and repeated games (msc 91A20), Evolutionary games (msc 91A22), Rationality, learning (msc 91A26), Applications of game theory (msc 91A80)
Software (theme 1), Logistics (theme 3), Energy (theme 4)
Software Engineering [SEN]
Intelligent and autonomous systems

van Bragt, D.D.B, Gerding, E.H, & La Poutré, J.A. (2000). Equilibrium selection in alternating-offers bargaining models: the evolutionary computing approach. Software Engineering [SEN]. CWI.