The short version
You have a belief. You go and check it. You find support — and the checking feels exactly like checking, which is the part that makes this so hard to catch.
The failure isn’t that you ignored the evidence against. It’s that the search never went anywhere the evidence against could have been.
Where the name comes from
Peter Wason, 1960, and it is still the cleanest demonstration in psychology.
He gives you three numbers — 2, 4, 6 — and tells you they follow a rule he has in mind. You may propose other triples, and he’ll tell you whether each one fits. When you’re confident you know the rule, say it.
Almost everybody forms a hypothesis immediately — ascending even numbers, or n, n+2, n+4 — and tests it. 8, 10, 12? Yes. 20, 22, 24? Yes. 100, 102, 104? Yes. Three for three, confidence rising, and the rule announced.
The rule was any three numbers in ascending order. Every test came back yes because very nearly anything would have. The triple that would have settled it — 1, 2, 47, or anything the hypothesis forbids — is the one almost nobody proposes, because it feels like wasting a turn on something you already know won’t work. Only about a fifth of Wason’s subjects named the rule on their first attempt.
Why it doesn’t work
Three different things wear this one name, and it helps to know which you’re doing:
- The search — you ask a question shaped like your answer
- The reading — the same study is decisive when it agrees and shaky when it doesn’t
- The memory — what comes back later is what fitted
- “Is there evidence for this?” is a normal question
- The methodological criticisms are usually correct
- Nothing was suppressed; it just didn’t resurface
Every one of those is a thing a careful person does. The bias isn’t in how well the question got answered. It’s in which question got asked.
And it survives because most of the time it works. Testing whether your idea holds up is a perfectly sound way to proceed when the idea is roughly right — which yours often is. It fails in the specific case where the world could be arranged so that your test passes anyway, and from inside the test those two situations look identical.
This is not confirmation bias
- Testing a hypothesis by looking for what would fit. A positive test is not automatically an error. If it is genuinely capable of coming back negative, you are doing the thing properly. Wason’s subjects fail because their tests couldn’t fail, not because they ran them.
- Making a case. Assembling supporting evidence for an audience is advocacy, and it’s what an argument is for. This is about what happens to your own search, before anybody is watching.
- Relying on a settled prior. Nobody re-derives germ theory before washing their hands, and refusing to reopen everything constantly is how a working mind stays working.
- Being right. People who are right also find confirming evidence. That’s the uncomfortable part: the presence of support tells you almost nothing by itself.
The test is about the shape of the search, not its result: what would have changed your mind — and did you go anywhere that thing might have been found?
How to call it out
Almost entirely inward, because as an accusation it’s cheap and unfalsifiable, and everybody now knows the phrase.
- Name the disconfirmer in advance, in writing. “If X turns out to be true, I’m wrong about this.” Then go and check X specifically, on purpose, before you check anything else.
- “What’s the strongest evidence against this — and did I go looking for it, or did I wait for it to arrive?”
- Ask somebody who disagrees for the best version of their case, and sit with it for a day before answering. The answering is where the bias re-enters.
Wason’s lesson fits in one sentence, and it’s a strange instruction that turns out to be the whole discipline: propose the triple your rule says shouldn’t work. If you never do, you will end up confident, and you will not be finished.