If one bias deserves first place, it is confirmation bias: the tendency to search for, interpret, favor and remember information in ways that support what we already believe. It affects experts and beginners alike, usually without any sense that we are being unfair.
The idea is old. In 1620 the English philosopher Francis Bacon wrote in his Novum Organum that “the human understanding when it has once adopted an opinion ... draws all things else to support and agree with it.”
The 2-4-6 task
In 1960 the British psychologist Peter Wason told participants that the sequence 2, 4, 6 followed a rule he had in mind. They could propose other sequences, and he would say whether each fit, until they felt sure enough to announce the rule. Most people guessed something like “numbers going up by two,” then tested sequences such as 8, 10, 12 or 20, 22, 24. Every one fit, so they grew confident, and wrong. The actual rule was simply any three increasing numbers.
The lesson is that confirming examples can fit many different ideas. A better strategy is to test cases that could prove your idea wrong, such as 1, 2, 3 or 6, 4, 2. A “no” teaches you far more than another “yes.”
Three forms of the bias
- Biased search. We seek out sources, friends and questions likely to agree with us.
- Biased interpretation. We read ambiguous evidence as support, and we hold agreeable evidence to a lower standard than disagreeable evidence.
- Biased memory. We remember the hits and forget the misses, like the times a “lucky” shirt seemed to work.
They saw a game
After a rough 1951 football game between Dartmouth and Princeton, the psychologists Albert Hastorf and Hadley Cantril later showed a film of the game to students from each school and asked them to note rule violations. Princeton students saw the Dartmouth team commit more than twice as many violations as Dartmouth students did. The study showed that the same evidence can look different depending on group loyalties.
Motivated reasoning
Motivated reasoning is reasoning steered by what we want to be true rather than only by accuracy. The psychologist Ziva Kunda argued in 1990 that people do not simply believe whatever they like. They still feel they need reasons. But desire shapes which evidence they look for, how hard they look, and when they stop.
The psychologist Thomas Gilovich described the result as a double standard. For a conclusion we welcome, we tend to ask whether we may believe it, and almost any support will do. For a conclusion we dislike, we ask whether we must believe it, and we hunt for any flaw. The result is an asymmetric standard: easy entry for welcome claims, a high bar for unwelcome ones.
Belief perseverance
Beliefs can outlive their evidence. In a 1975 study led by Lee Ross, students received made-up feedback on how well they judged which of a set of notes were genuine. Later they were told the feedback had been assigned at random. Many still believed they were unusually good or bad at the task. This is belief perseverance: beliefs surviving even after their supporting evidence has been discredited, often because we have already built explanations around them.
Why it persists
Confirmation bias is not pure foolishness. Nobody can recheck every belief, and beliefs tie us to friends and communities, so changing them can feel like disloyalty. Some researchers argue that reasoning evolved partly for arguing with others, which may explain why we find flaws in other people's arguments far more easily than in our own. The practical remedy: let others test your reasoning.
Worked example: testing a lucky mug
Priya believes she does better on exams when she drinks from her lucky mug. She can recall several great results with the mug in hand. Is she right?
First, notice the three forms of confirmation bias at work. Biased memory: the mug-and-success occasions stand out, while the mug-and-flop days fade. Biased interpretation: when she did badly with the mug, she explained it away (“that exam was unfair”). Biased search: she never looked at how she did without the mug.
A fair test needs all four combinations, not just one. Suppose Priya checks her records for her last 20 exams:
- With the mug: 12 exams, 8 good results and 4 poor ones. That is 8 out of 12, about 67 percent good.
- Without the mug: 8 exams, 5 good results and 3 poor ones. That is 5 out of 8, or 62.5 percent good.
The difference is small, and with only 20 exams a gap this size could easily arise by chance. Priya was attending only to one cell of the table, mug and success, which is exactly what confirmation bias encourages. Her memory supplied eight vivid “proofs” but hid the comparison that mattered.
A stronger test follows the 2-4-6 lesson: write down in advance a prediction that could fail, then record every result.
The mug might still help indirectly, by making Priya feel calm. The point is not to mock it but to notice how easily evidence seems to pile up for an idea we already like.