Multi-agent DSIT Simulation: Reexamination of Robustness of Prediction

Abstract

The Dynamic Social Impact Theory (DSIT) predicts emergence of spatial clustering and other group-level phenomena as a result of independent interactions between individuals. In order to investigate the role of uniformity of distances among people in the cellular automata framework in the original studies, we took a multi-agent model approach by placing agents in a two-dimensional space. By examining roles of both the function that determines attitude changes in agents and the methods of placing the agents, we found that attitude scores of the agents became cohesive and spatially clustered. At the same time, complete agreement was rarely observed, and diversity was preserved. We concluded that concurrence of clustering and continuing diversity were indeed a robust phenomenon under practical assumptions such as cohesiveness of people’s locations and nonlinearity of attitude changes. To apply the knowledge we learned here, it will be crucial to conduct empirical studies to find out when and how these assumptions of nonlinear attitude changes occur and to learn more about how communications are determined by distances between individuals and other factors.