Polarization and Conformity in the Coevolution of Beliefs and Networks
This paper sets out to understand the behavioural foundations of polarization – extreme and opposing views of different groups in a society – and conformity - the tendency of people to match their beliefs to those of others. These social phenomena are studied by generalizing the coevolutionary model developed in Arifovic, Eaton, and Walker (2015) to allow for a wider range of learning behaviours. Unlike when learning follows a process of weighted averages, learning involving weighted medians can generate bifurcated networks with polarized beliefs. Learning via weighted medians can also lead to connected, conformist networks.