Bayesian Inference
A conclusion depends not only on the evidence in front of you, but also on the belief you brought to it. Rational updating combines both, changing confidence only as much as the new evidence warrants.
Two people can see the same evidence, reason correctly, and still reach different conclusions. The difference may not be logic or honesty; it may be what each believed before the evidence arrived.
E1Evidence does not start from zero
Bayesian inference begins with a prior: your existing estimate of what is likely. New evidence then pushes that estimate up or down according to how strongly the evidence favors one possibility over another. The result becomes your new prior for the next decision. This is why evidence alone does not determine belief: identical information can produce different updates when it enters different starting models. The discipline in Bayesian Updating is therefore not to discard prior knowledge, but to expose it and revise it proportionately.
E1A prior can inform—or imprison
Different conclusions are not automatically equally sound. A useful prior reflects relevant knowledge, and useful updating requires evidence capable of challenging it. If the prior is badly biased, or if Believing Is Seeing distorts how you perceive the evidence, the machinery can preserve error instead of correcting it.
E1State the starting odds
Before evaluating an important new claim tomorrow, write down what you currently believe and how confident you are. Then specify what result would materially raise or lower that confidence; after seeing the evidence, record the revised estimate rather than merely declaring the claim true or false.
Episodes that teach this
-
From Instinct to Insight - How Do We Really Decide? Bayesian Thinking - Future IQ
· explained at 2:00
7,456 views
"A decision that you make... doesn't just depend on the evidence... It depends on what you believed before you saw the evidence."