Where it sits
It starts in which question you ask - and carries on through how you read the answer.
Tell it apart
| Aspect | Confirmation bias | Motivated reasoning |
|---|---|---|
| Driven by | Test strategy | Desired conclusion |
| Needs a stake | No | Yes |
| Conscious | Rarely | Sometimes |
| Fix | Test the opposite | Name the want |
Confirmation bias is about which evidence you look at; motivated reasoning is about wanting a particular answer. The distinction matters because you can show confirmation bias with nothing at stake at all — Wason's number triples had no emotional content whatsoever.
What it doesn't mean
It's mostly unconscious. Wason's participants weren't cheating — they generated tests their hypothesis would pass, a habit researchers call a positive test strategy. You can be sincerely trying to be fair and still do it.
Wason's subjects were intelligent young adults and most still failed. Later research finds it in analysts, clinicians and scientists — expertise gives you more evidence to use selectively, not less bias.
What it really means
Peter Wason's 1960 experiment is one of the cleanest demonstrations of a bias ever built, and its cleverness is that it strips out everything people usually blame. No politics. No ego. No money. Just three numbers on a card — 2, 4, 6 — and an instruction: work out the rule by proposing triples of your own, and I'll tell you yes or no.
Almost everyone forms the same hypothesis immediately: even numbers ascending by two. And then almost everyone tests it the same way. 8-10-12. Yes. 20-22-24. Yes. 100-102-104. Yes. Three confirmations, growing confidence, wrong answer. The actual rule was any three ascending numbers, and the single test that would have exposed it — trying something the hypothesis forbids, like 1-2-3 — is the one hardly anyone runs.
That's what makes confirmation bias worse than the usual telling of it. It isn't wilfully ignoring inconvenient facts; that would be easier to catch. It's that the mind, asked to test an idea, reliably generates the experiments the idea will pass. Every result comes back positive, which feels exactly like being right. And expertise doesn't rescue you — it mostly enlarges the pool of evidence you can select from.
Where it comes from
Latin — to make firm
The Latin is literal about it: confirmare is 'to make firm, strengthen'. English narrowed the noun long before the psychology arrived — by the late 14th century confirmation meant verification and supporting evidence, which is the sense Wason's term uses.
- 1960
Peter Wason gives 29 people the triple 2-4-6 and asks them to find the rule by proposing triples of their own. The rule was simply 'any three ascending numbers'. Only 6 reached it without first announcing a wrong one.
- 1998
Raymond Nickerson's review consolidates the field, arguing the bias is largely unintentional and appears in science, medicine, law and everyday reasoning alike.
In real life
- Convinced a new colleague is lazy, you notice every long lunch and miss the evenings she stays late.
- Thinking about buying a particular phone, you search why this phone is the best rather than problems with this phone, and the results agree with you.
- A fan who believes the referee is biased against their team remembers every bad call against them and none of the ones that went their way.
- Someone who believes a full moon makes hospital wards busier recalls the chaotic full-moon shifts and forgets the quiet ones.
- After reading one article that supports your view, you share it at once; an article that challenges it gets checked for flaws first, if you read it at all.
Questions
Q01What is confirmation bias in simple terms?
It's the habit of collecting evidence that agrees with you. Decide a new colleague is disorganised and you'll register every missed deadline while quietly discounting the three reports they filed early — and finish the month more certain than you started.
Q02What was the 2-4-6 experiment?
Wason showed people 2, 4, 6 and asked them to work out the rule by testing triples. Almost everyone tested 8-10-12, 20-22-24 — sequences fitting their guess of 'even numbers going up in twos'. The real rule was 'any three ascending numbers'. Trying 1-2-3 would have cracked it instantly. Hardly anyone tried.
Q03How do you counter it?
Test what would prove you wrong rather than what would prove you right. Before buying the car you've fallen for, don't search its name plus 'review' — search its name plus 'problems', and set the question as 'what would make me walk away?'
Sources
- Confirmation Bias in Psychology — Simply Psychology.
- Effects of task structure and confirmation bias in alternative hypotheses evaluation (2024) — PMC.
- confirmare (etymology) — Wiktionary.