Emergence in multi-agents systems : cognitive hierarchy, detection, and complexity reduction part 1: methological issues.
Abstract
In a pioneering book on ”artificial society” and multi-agent simulations in social sci-ences, (Gilbert and Conte 1995) put the emphasis on ”emergence” as a key conceptof such approach: ”Emergence is one of the most interesting issues to have beenaddressed by computer scientists over the past few years and has also been a mat-ter of concern in a number of other disciplines, from biology to political science”(op.cit. p.8). More recently, Agent based Computational Economics (ACE) put theemphasis on the question of emergence, following for instance (Tesfatsion 2002a) or (Axtell and Epstein and Young 2001) The present paper provides a formal definitionof emergence, operative in multi-agent framework designed by Agent Oriented Pro-gramming, and which makes sense from both a cognitive and an economics pointof view. Starting with a discussion of the polysemous concept of emergence, thefirst part of this paper is dedicated to clarifying the question by focussing on theproblem of modelling cognitive agents in artificial societies. The key questions areintroduced by way of a paradigmatic example. The second part of this paper is dedi-cated to introducing and discussing operative definitions and related implications. Inorder to illustrate our formal definition of emergence, a companion paper (Phan andGalam and Dessalles, 2005) discusses the ACE population game model of (Axtelland Epstein and Young 2001) and builds a multi-level-model based on the formalframework introduced in this paper.
Origin : Files produced by the author(s)