Confirmation Bias As Misfire Of Normal Bayesian Reasoning

A short, sharp epistemics reframe: confirmation bias isn't a special faulty mode but a misfire of normal Bayesian reasoning. The coyote-vs-polar-bear-in-Berkeley example shows that heavily discounting evidence against a strong prior IS correct updating; the anti-vaxxer is doing normal reasoning but started from a wrongly-strong prior and underweights new evidence. Notes refinements (emotion-locked priors, evolved prior-overweighting) might collapse back to the standard model, but argues it's still more useful to treat confirmation bias as normal reasoning and ask when it fails.

Quality 70 · Strong. Strong, upper (70). Punches above its 524 words: a genuinely clarifying, fairly original reframe of a standard concept, crisply illustrated (coyote / polar bear). Capped by brevity and Scott's own hedge that the refinements may make the objection 'kind of pointless.'

Claude’s paradigm shift 54 · Moderate. Moderate, upper (54). Reframing confirmation bias as a Bayesian-updating misfire is a fresh, non-obvious angle that updates how informed readers think about it -- though 'biases as misfires of normal cognition' is itself a known LW theme.

Real-world impact 2 · Minor. 2 -- minor/within-blog. A clarifying conceptual point circulated within the rationalist / epistemics sphere; no material footprint.