This preprint develops a theoretical model of political polarization based on the concept of stabilized inference. It argues that polarization may arise not only from disagreement over evidence, but from asymmetries in evidential uptake, where identity-linked expectations shape what counts as evidence in the first place. The paper draws on predictive processing to formalize this distinction and explores its implications for political judgment and institutional dynamics. This version is made available to facilitate early discussion and feedback.
Miguel dos Santos Pereira (2026) studied this question.