Music evolution in a complex system of interacting agents

The evolution of music, from random note strings to certain "pleasant" note sequences, is traced in a multi-agent computational model. A community of agents, with some musical guidelines and expertise from different aspects, compose their own and criticize other's music to improve individual music perfromance. Based on common musical interest, some social structure with Small World charactieristics is formed up. Through adopting general rule-based system and some evolutionary mechanisms, this model introduces a realistic communication scenario and shows an optimization process following local information. It provides an instructive computational framework to study music as well as other cultural phenomena, such as language. © 2005 IEEE.link_to_subscribed_fulltex