BAYESIAN UPDATE

Bayes' theorem, informally: your degree of belief in something should be proportional to your prior belief times the likelihood of the new evidence given that belief. New evidence doesn't replace your prior; it updates it. The size of the update depends on how surprising the evidence is given each hypothesis.

The practical version doesn't require math. It requires honestly stating: what did you believe before? How likely is this new piece of evidence on each hypothesis you're entertaining? Then: how much should you move?

1. 1.Pick a question with two or three candidate answers (hypotheses).
2. 2.Before the new evidence: assign each hypothesis a rough probability that adds to 100%. E.g.: 'I was 60% sure my partner is upset about X, 30% about Y, 10% something else.'
3. 3.Name the new evidence: what just happened?
4. 4.For each hypothesis, ask: 'How likely is this evidence GIVEN this hypothesis?'
5. 5.The hypothesis under which the evidence is more LIKELY gets boosted; the others shrink. Reassign probabilities.
6. 6.Notice: did your beliefs move? By how much? Was the movement proportional to the evidence?