Update beliefs by degrees, not wholesale
Treat new information as evidence that shifts probabilities, not as proof that changes everything.
Why it works
When people update beliefs in response to new information, they tend to either ignore it (conservatism bias) or overcorrect to near-certainty (representativeness). Bayesian updating — the mathematically correct approach — requires multiplying the prior probability by the likelihood ratio of the new evidence. Even an informal version of this (how much more likely would this evidence be if my hypothesis is true, versus false?) produces more calibrated beliefs.
How to do it
- State your current probability estimate before looking at new evidence.
- Ask: "How much more (or less) likely is this evidence if my hypothesis is correct vs incorrect?"
- Shift your probability in proportion to that ratio — not to 0% or 100% unless the evidence is truly definitive.
Evidence
Bayesian reasoning is the normative standard in probability theory. People systematically deviate from it in both directions (over- and under-updating) depending on whether the evidence is vivid or abstract. Training in Bayesian reasoning improves calibration. (observational)
Formal Bayesian computation is not practical for everyday decisions; the goal is the habit of proportional updating, not precise calculation.
Sources
- Kahneman (2011), Thinking, Fast and Slow — review of representativeness and Bayesian deviation research
Common mistake
Treating a single confirming data point as near-proof — which is the representativeness heuristic exactly: a story that matches your hypothesis feels like it proves the hypothesis.
Practice this with IX Coach
7 days free, then $40/month (~$1.30/day).
More practices for Base-Rate Neglect: Why We Ignore the Odds
- Ask the base rate before evaluating the specific case
Before judging any individual instance, first establish how often this kind of thing happens in general.
- Use reference class forecasting for project estimates
Estimate how long similar projects have taken historically before estimating your specific project.
- Always identify the denominator when evaluating risk or success
When a number or story is striking, ask: "Out of how many total cases?"
- Deliberately invoke the outside view for important decisions
For any high-stakes prediction, force yourself to start with how things typically go, not how your situation feels.
- Practice probabilistic calibration by tracking your predictions
Assign explicit probability estimates to your predictions and track whether they come true at the right rate.
- Expect regression to the mean in extreme outcomes
Unusually good or bad performance tends to be followed by more average performance — not because of what you did.
Related concepts
- Thinking, Fast and Slow, Made Usable
Two systems, the biases they create, and when to slow down
- Bayesian Thinking: How to Update Beliefs Rationally
Holding beliefs as probabilities and updating them when evidence arrives
- Anchoring Bias in Negotiation and Judgment
Why the first number wins — the mechanism, and how to set and resist anchors