Internal Evolution for Agent Cognition

Abstract

Agent-Based Modeling (ABM) is a powerful simulation technique with applications in several fields, in particular social sciences. Artificial Stock Market (ASM), introduced by a group of researchers at Santa Fe Institute (SFI) in the mid 90’s, is one of the pioneering works in which the application of agent-based modeling is examined being used to model a stock market and to study economic behavior. A number of heterogeneous agents form a market in which they buy and sell shares of a single introduced asset and they make their decision upon their expectations of the market which is determined from their aggregate expectations, while they improve their predictions based on the response they get from the market. The computer simulation of the model gives a market dynamics which has similarities with real market data. The internal evolution of agents has a direct effect on the market dynamics, and the learning speed of agents controls the overall qualitative characteristics of the market. With emphasis on the role of evolution of agents in this artificial stock market, an implementation of the model has been done and a few issues have been studied.