Political Polarization and Fractionalisation from Rational Values-Based Inference in an Agent-Based Graph Network

The rise in political polarization disrupts political consensus and causes individual harm. We build on a theoretical framework of political polarization that emerges from uncertain political identity inference and signaling mediated by moral values. The current computational model extends this framework with rational inference tools and graph theory to better capture the complex dynamics of value-based inference and group formation. We find that minimally constrained signaling and promiscuous inference and updating of moral values leads to general network homogeneity. This contrasts with previous models using the same overarching theoretical framework and highlights the influence of model implementation, which should be further explored to triangulate the necessary causes of polarization. We discuss future extensions to the model to explore what facilitates political polarization as found in previous studies and the real world.