Polarization Traps: A Formal Model of the Epistemic Structure of Moral Polarization

As individuals, we form our beliefs and opinions through exchange with other individuals and through outside influences. This often leads to groups of individuals holding relatively homogeneous positions that are diametrically opposed to other groups' opinions, ending in polarized states. Internet platforms add an additional dimension to this question, as the ways in which the platform curates and disperses information can intensify or mitigate polarization. In this work, we flesh out a proposal to develop theoretical models and algorithmic solutions for this phenomenon by linking philosophical and political literature to literature in statistics, opinion dynamics, and social learning. In this report, we narrow the scope of the problem we wish to study and define our goals for future work.