Significance of Agent-Based Simulation in Social System Theory

Since the last decade of the 20th century, agent-based simulation method has been developed and becoming a powerful tool for social science. This approach has attracted attention as a new trend, which clarifies the social dynamics and complex human relations. Especially, the following is an important advantage of this approach. This simulation is performed in bottom-up and process-traceable ways, therefore we can clarify how the macroscopic form and social order are generated from the interactions of individual agents. The agents only receive a small number of constraints, each interacting autonomously in the computer space. Then we can reproduce by simulation the manner of forming the ordered whole (social system) from individual behaviors.

    The advantage of agent-based simulation lies in that we can explore the micro-macro link between individual and society by a bottom-up procedure. There has been a deep division between the methodological individualism and collectivism. To bridge this division has been the most difficult work so far. Because of the emergent property of a macro level, it has been regarded as almost impossible to derive the characteristics of macro (society) from the micro (individual). In fact, while efforts to the problem of micro-macro link have been made in sociology, meaningful results have not been achieved.

    In the presentation, I argue the micro-macro problem from three aspects based on the viewpoint of agent-based simulation. First is to examine the mechanism of emergence with reference to Schelling’s “A Self-Forming Neighborhood Model.” Second is unintended consequence of action by referring to Yamamoto’s model regarding a trap of egalitarianism in the logic of social contract. Third is mathematically unsolvable solution with reference to Axelrod’s “Tit for Tat” strategy in the iterated Prisoner's Dilemma game.