Joachim Andersen

Accelerated Learning

Confirmation Bias: Why We See What We Expect to See

A vintage brass magnifying glass with a horn handle on dark velvet at night, enlarging a single gold marble among dozens of black onyx marbles, with a warm spotlight falling on the few gold marbles
We find the gold we are looking for and overlook the rest.


Confirmation Bias is the tendency to search for, interpret and remember information in a way that confirms what I already believe. It rarely feels like bias from the inside. It feels like I’m simply noticing the evidence, while the evidence against me somehow seems weaker, rarer or easier to explain away.

I notice it most when I have already made up my mind about a person, a product or an investment. Every new detail seems to fit the picture, and I have to make a deliberate effort to see the details that don’t.

The essence of Confirmation Bias is to ask: “What would I expect to see if I were wrong, and have I actually looked for it?”


Origin: An Old Observation With a Modern Name

People noticed this long before psychology had a word for it. Around 400 BC, the Greek historian Thucydides wrote that it’s “a habit of mankind to entrust to careless hope what they long for, and to use sovereign reason to thrust aside what they do not fancy”. In 1620, Francis Bacon wrote in Novum Organum that the human understanding, once it has adopted an opinion, “draws all things else to support and agree with it”.

The modern term came from the English psychologist Peter Wason, who in the 1960s ran experiments showing that people test their ideas by looking for confirming cases rather than cases that could prove them wrong. I describe his best-known experiment in my article on Falsifiability. Since then, hundreds of studies have shown the same pattern in how we search for information, how we interpret it and what we remember.


Key Concepts

  • Biased Search
    We tend to look for information that supports my view, and to ask questions that can only be answered in a way that confirms it.

  • Biased Interpretation
    When evidence is mixed or unclear, we read it in the direction we already lean, and we hold opposing evidence to a much higher standard.

  • Biased Memory
    We remember the hits that support our beliefs and forget the misses, so my experience seems to confirm my views more than it really does.

  • Belief Perseverance
    Beliefs can survive even after the evidence that created them has been shown to be false.

  • Not Only for the Uninformed
    Intelligence and expertise don’t protect against it. Skilled people are often better at finding reasons to keep believing what they want to believe.


Some Examples

  • Same Studies, Opposite Conclusions
    In 1979, Charles Lord, Lee Ross and Mark Lepper at Stanford gave supporters and opponents of the death penalty the same two fictional studies, one appearing to support deterrence and one appearing to undermine it. Each side rated the study that agreed with them as more convincing, found detailed flaws in the other, and in the end reported that their views had become more polarized. Identical evidence pushed the two groups further apart, although later research suggests such polarization happens less often than this famous study implies.

  • Questions That Answer Themselves
    In a 1978 study by Mark Snyder and William Swann, participants were asked to find out whether a person was an introvert or an extrovert by choosing interview questions. Those testing for introversion picked questions like “What do you find unpleasant about noisy parties?”, while those testing for extroversion picked questions like “What would you do to liven up a dull party?” The questions themselves made confirmation almost inevitable.

  • The Fingerprint that Matched Too Well
    After the Madrid train bombings in 2004, the FBI declared a fingerprint found on a bag of detonators a “100% match” to the American lawyer Brandon Mayfield, who was arrested and held for two weeks before Spanish authorities identified another man. The Justice Department’s Inspector General later found that the examiners had reasoned backward from Mayfield’s prints and begun to “find” features in the latent print that were not really there.

  • Good at Math, Except When It Matters
    In a study published in 2017, Dan Kahan and colleagues showed people the same table of data and asked what it showed. When it was presented as a test of a skin cream, the most numerate people did best. When the same numbers were about a gun control ban, answers split along political lines, and the polarization was greatest among the most numerate, who used their skills selectively to reach the answer that fit their views.


Applying Confirmation Bias

Charles Darwin described a habit he called a golden rule: whenever he came across a fact or thought that went against his general results, he wrote it down at once, because he had found that such facts were “far more apt to escape from the memory than favourable ones”. Inspired by that, this is the simple process I follow:

  1. Define What Would Change My Mind
    Before I look at the evidence, I write down what result would show that I’m wrong.

  2. Ask Questions Both Ways
    I try to phrase questions so that either answer is possible, and I endeavor to ask what would be true if the opposite of my belief were correct.

  3. Write Down the Exceptions
    Like Darwin, I make a note of every fact that does not fit, because those are the ones I will forget first.

  4. Seek Out the Best Opposing Argument
    I look for the strongest version of the other side, and ideally someone who disagrees with me and is willing to say so. If I cannot find such a person, I try to steelman the argument (constructing and addressing the strongest, most logical version of an opposing argument).

  5. Hold Confirming Evidence to the Same Standard
    When something supports my view, I try to ensure asking the same tough questions about it that I would ask if it did not.


Pitfalls and Limitations

  • Knowing about the bias is not a cure. As the numeracy study shows, being smart can even make it stronger, because it gives me better tools to defend what I already believe.

  • Not every prior belief is a bias. Giving more weight to well-established knowledge than to a single surprising claim is often sensible. The problem is when no evidence could ever change my mind.

  • It works in groups too. Teams that share the same view can confirm each other’s beliefs and filter out dissent, which is why I value people who are allowed to disagree.

  • It can hide in tools. Search engines and social feeds that show me what I already like can do the biased searching for me, so I deliberately look outside them.


Related Mental Models

Confirmation Bias reminds me that the most dangerous evidence is the evidence I never go looking for. By deciding in advance what would prove me wrong and keeping a record of the facts that don’t fit, I give the truth a fair chance to change my mind.



To learn more about mental models and philosophy,
check out: Wise: How to Attain a Deep Understanding of the World, History, Our Place in It, and How to Live Well



📚 Sources and References