Abstract
A rationale based on adaptive learning is used to explain inefficient bargaining under private information. A dynamic bargaining model using genetic algorithms is developed in which independent but interacting negotiators coevolve their offers. A computer simulation analysis is conducted that compares bargaining outcomes under complete and incomplete information. Results show that inefficient bargaining (i.e., delays and failures to agree) may be due to problems in the agents' adaptive processes that arise because of incomplete information and that adaptive failures may occur even under complete information. Adaptive learning is identified as both a mechanism for resolving conflict and a potential source of conflict.