When Rituals Fail: Rationalization, Bayesianism, and Predictive Processing

Why do rituals persist in human societies despite their frequent and observable failures to produce intended outcomes? This paper advances a two-part argument to explain this resilience. First, at the individual level, I argue that belief in ritual efficacy is maintained through Bayesian-rational processes, where the invocation of auxiliary hypotheses absorbs disconfirming evidence and shields central beliefs from significant revision. Importantly, such protection is not complete. Each failure produces a small but non-zero erosion of individual confidence. Second, I address the resulting population-level puzzle: why does such incremental doubt not accumulate into widespread skepticism and the eventual collapse of ritual systems? I argue that social features and informational dynamics (e.g., memory biases, the underreporting of failure, pluralistic ignorance) as well as the protective "design" of rituals themselves systematically inhibit the aggregation of doubt across individuals and generations. By linking individual cognition with population dynamics, this account explains the remarkable resilience of ritual systems in the face of persistent empirical failure.