Mind & Psychology · Thinking & decisions
Confirmation bias
People look for, notice and remember evidence that fits what they already think. See the number puzzle that exposes the habit, why it may not be a simple flaw, and what helps against it.

What you will learn
- Describe Wason's 2-4-6 task and the mistake most people made in it
- Explain why a test that could prove you wrong tells you more than one that fits
- Describe how confirmation bias shapes what people search for and what they remember
- Explain how both sides of the death penalty debate read the same studies as support
Lesson 1 of 4 · 4 min
The 2-4-6 puzzle that most people get wrong
In 1960 the psychologist Peter Wason gave 29 students three numbers, 2, 4 and 6. The numbers fit a simple rule he had in mind. Each student's job was to find it.
A student wrote down any three numbers with a reason, and Wason said whether they fit the rule. Students could test as many sets as they liked. They were told to announce the rule only when they felt highly confident.
One student tried 8, 10, 12, then 14, 16, 18, then 20, 22, 24, then 1, 3, 5. Every set fit. She announced that the rule was to add two each time, and Wason told her she was wrong.
Most students did what she did. They tested sets that fit their own idea, heard yes each time and grew sure of a rule that was too narrow. Only 6 named the right rule on their first announcement.
Try to recallWhy could no number of yes answers prove the rule "add two each time" in Wason's 2-4-6 puzzle?
Every set that adds two also fits the real rule, any three rising numbers. So both rules predicted each yes, and the answers couldn't tell them apart. Only a set that breaks your rule, like 1, 2, 3, can show that it's wrong.
In the 2-4-6 puzzle a pile of yes answers felt like proof and wasn't. To test an idea, try the case your idea says should fail.
Lessons in this topic
The 2-4-6 puzzle that most people get wrong
Try Peter Wason's number puzzle and see why testing only the guesses that fit your idea can leave you sure of a wrong answer.
How the bias shapes what you look for and remember
See confirmation bias at work in the questions people ask and the memories they keep, and in a famous study where both sides of the death penalty debate read the same evidence.
Is confirmation bias a flaw or a feature?
See why checking where you expect a yes is often a sound shortcut, and why some researchers think reasoning evolved to win arguments.
What actually helps against confirmation bias
Learn why knowing about the bias doesn't switch it off, and which methods have reduced it in experiments.
Three quick stories from this topic
Bacon's early warning
Francis Bacon described confirmation bias in 1620, centuries before psychologists named it
In his Novum Organum he wrote that the mind, once it adopts an opinion, draws everything else in to support it. He blamed the habit for beliefs in astrology, dreams and omens.
A name in the 20th century
The psychologist Peter Wason coined the term confirmation bias after his 1960 number puzzle.
Still the starting point
In 1984 psychologists traced several biases to the failure Bacon described and tested a fix, asking people to consider the opposite.
Same study, opposite verdicts
Death penalty supporters and opponents read the same studies, and each side kept its view
In a 1979 Stanford experiment, every student saw one invented study for each side. A supporter faulted one study for covering too short a period, while an opponent defended the same study. People demand more of evidence they dislike.
Invented on purpose
The researchers swapped which kind of study showed deterrence, so the method alone couldn't decide which study a student liked.
A higher bar
Psychologists call it disconfirmation bias. Welcome evidence gets "Can I believe this?" and unwelcome evidence gets "Must I believe this?"
The 2-4-6 puzzle
Only 6 of 29 students solved Peter Wason's 2-4-6 number puzzle on their first try
The hidden rule was just any three rising numbers. Most students tested only sets that fit their own narrower guess, such as adding two each time, so every answer came back yes and they grew sure of the wrong rule.
A rule too wide to guess
Every set that fits "add two" also fits "rising numbers," so a yes couldn't tell the two rules apart.
The test that works
Only a set that breaks your own rule, like 1, 2, 3, can show that the rule is wrong.
Sources
- On the failure to eliminate hypotheses in a conceptual task, Quarterly Journal of Experimental Psychology / University of Florida
- Confirmation bias, Wikipedia
- Confirmation bias, a ubiquitous phenomenon in many guises (Review of General Psychology, 1998), Raymond Nickerson, copy hosted at UC San Diego




