This dissertation investigates belief polarisation and opinion diversity through computational models of argumentation. Key findings show that even epistemically rational agents can polarise through deliberation, with egocentric argumentation strategies driving polarisation while allocentric strategies reduce it. The research also reveals that diverse groups face challenges in collective decision-making, as majority voting often yields inconsistent outcomes. The study introduces “taupy,” a new Python implementation for dialectical structures analysis, and demonstrates how computational modeling can reveal dynamic aspects of philosophical concepts that traditional methods cannot access.
