Conservatism in belief revision is defined as the human tendency to cling to prior beliefs or initial impressions, often revising views less than normatively predicted, even when presented with new, contradictory evidence. This bias explains why changing deeply held views or values is so difficult for people. We cover the ongoing debate between foundations and coherence theories of belief revision. The foundations theory suggests beliefs require explicit justification, meaning losing a core belief can trigger a chain reaction of abandonment. In contrast, the coherence theory posits that beliefs are justified unless there's a specific reason to doubt them, advocating for minimal changes to increase overall coherence. While initial intuition might favour the foundations theory, people often exhibit belief perseverance, retaining beliefs even when their original evidence is discredited, which aligns more with the coherence theory's predictions. This is partly explained by a practical need to avoid mental clutter and not meticulously track every justification for one's beliefs. Another significant finding is that experimental demonstrations of conservatism might not indicate a normative fault, but rather reflect a rational Bayesian response to information perceived as coming from a less-than-fully-reliable source. The model of "bounded revision" is introduced as a two-dimensional operation that fills the space between conservative and moderate revision, satisfying key AGM and Darwiche-Pearl axioms, unlike previous models. Finally, from a veritistic perspective, while the merits of individual conservatism are ambivalent, a less conservative, "verificationist" approach can collectively improve beliefs within an epistemic community, suggesting that individual caution in belief formation pays off socially

Ver todos los episodios