An Agent-Based Model of Social Impact with Bayesian Updating Rules

The process of social influence and individual opinion formation can aptly be constructed as a complex adaptive system. Individual people formulate their opinions through some unknown process that considers social pressure and their own individual agency. What emerges from this process is a global opinion distribution, or "public opinion," that is a nonlinear epiphenomena of local agent action. Opinion dynamics models postulate simple agent rules that approximate this unknown local process so as to generate global behavior mimicking observed phenomena. This proposal is concerned with the development of an agent-based network model of social impact that (1) considers variation in the way different agents respond to social influence (2) allows agents a Bayesian updating procedure and (3) employs a network topology that is a better approximation of social networks than random graphs.