Multi-agent systems and role games : collective learning processes for ecosystem management

TLDR: This chapter proposes a method, develops abstract models, and presents first results of a modeling tool used to understand the dynamics of natural and renewable resources and their management in the field of simulating societies in interaction with their environment.

Abstract: For several years now, a field of research on simulating societies in interaction with their environment has been taking shape Methods such as multi-agent systems (MAS) can be used to create virtual societies The effects of interactions among different behaviors on resource dynamics and associated feedback are simulated Modelers use such methods to create computer representations of dynamics observed in the field [] After using operational research methods to identify the solution to a given problem, researchers turned to ways of representing a stakeholder's strategic and tactical decisions through expert systems or simulation methods Then, multiagent systems provided a new tool for the modeling of complexity and particularly the modeling of collective decisions Our team' has been using MAS in several ways 1 We developed abstract models, also called artificial societies that help to understand the generic properties of interacting processes We mainly developed models on non-merchant exchanges and reputation, models on economic tools for the regulation of economic exchanges and models on spatial dynamics 2 We also developed applied models to understand the dynamics of natural and renewable resources and their management 3 We developed a modeling tool (CORMAS, common-pool resources and multi-agent systems) 4 We also worked on defining the possible uses for these MAS tools within the very wide framework of collective decision support The objective of this chapter is to propose a method and present our first results (Resume d'auteur)