CF Dictionary · Knowledge & Epistemology
Bayesianism
The view that rational belief is updated credences via Bayes' theorem. CF rejects it as analogous to credence and qualitative-error blindness.
Also: Bayes
Bayesianism is the family of views that hold rational belief consists of credences updated by Bayes' theorem. Bayesians assign prior probabilities to hypotheses, then update on evidence using the rule P(H|E) ∝ P(E|H)·P(H).
CF rejects Bayesianism for the same reason it rejects credences:
- Priors are arbitrary. Where do priors come from? "Arbitrary assumptions made up in advance before intelligent thinking happens."
- Updating requires multiplication across dimensions. Bayesians treat all evidence as additive in log-probability. CF shows this fails for different-dimensional factors.
- It hides decisive refutation. A decisive refutation should flip the status to refuted. Bayesianism just nudges the credence toward zero.
- It ignores breakpoints. A 1% drop in credence is treated as equivalent to a 99% drop. CF says qualitative differences matter.
What CF keeps
CF accepts that some situations have an irreducible statistical component (rolling dice, sampling a population). For those, you can use probability — but as a factor with its own dimension, not as a universal framework for belief.
"Bayesians / Bayesianism: People who advocate credences and update them using Bayesian probability math." — LessWrong summary of CF