Multi-Agent Based Modeling and Simulation of Consensus Formations in Arguments

We propose a two layered multi-agent model of a consensus formation in arguments consisting of micro level and a macro level. The micro level simulates the opinion dynamics induced by mutual interactions among individual agents. We design this micro level agent system using a cellular automaton by defining its local function based on social impact theory. On the other hand, the macro level, consisting of mediator agents, makes a decision by integrating the opinions given by the micro level simulation. We discuss what information of micro level is essential for decision-making in arguments among three opinions.