Norm Adaptation and Revision in a Multi-Agent System

In this paper we address the question of assigning social norms to agents: should we attempt to ascribe social norms to agents that will act in complex dynamic environments, or is it possible to allow the agents to adapt to new situations as they arise, and choose their norms accordingly? We argue that adaptation is preferable to prescription, in that agents should be allowed to revise their norms on the fly. A system is constructed in which the performance of multiple agents operating in the same environment can be assessed. Experimental results concerning alternative norm selection strategies are presented and discussed.

• Type: Conference
• Year: 2003
• Where: FLAIRS
• Authors: Nicholas Lacey, Henry Hexmoor
• Keywords: Artificial Intelligence, Complex Dynamic Environments, FLAIRS 2003, Multiple Agents, Social Norms