The dynamics by which groups polarize have been well studied across the social sciences. There has been relatively little focus, however, on why groups polarize over matters of fact, including over scientific beliefs. In this talk, Weatherall presents a set of models aimed at addressing why this might happen. In particular, he shows that when individuals trust the evidence of scientific peers less as their shared beliefs diverge, polarization can emerge endogenously even among agents who are semi-rational truth-seekers, and who gather evidence about the world. In addition, he explores the endogenous emergence of epistemic factions who hold multiple, polarized beliefs. As he argues, this can happen when agents use multiple beliefs to ground trust in the evidence produced by their peers.