MODELING SOCIALLY INTELLIGENT AGENTS

Abstract

An approach to modeling boundedly rational agents is described. This approach uses the parallel evolution of a population of mental models for each agent based on the genetic programming paradigm. Then some of the characteristics one would expect of a society of agents (as opposed to a collection of interacting agents) is argued for. These include the ability to identify, model, and communicate with specific other agents in a heterogeneous way. An example model is described, which extends Brian Arthurs' “El Farol Bar” model with learning and communication. The model allows the co-evolution of the agents' populations of models. The results are then analyzed to show the differentiation that results between the agents. A specific case study of the situation at the end of the simulation is then examined in depth. This indicates that some of the problems in analyzing human communication will also occur with such models and that explanations based on Wittgensteinian language games and use in practice may be most appropriate.