73 Confirmation Bias

After I learned about confirmation bias, I started seeing it everywhere. —Jon Ronson

Confirmation bias is the human tendency to search for, interpret, and recall information in a way that confirms what we already believe or want to believe. Once an initial opinion has formed, we naturally seek out evidence that supports our preconceptions and find it difficult to see things differently. People rarely look for evidence that might prove them wrong, and when contradictory information does appear, they often ignore, deny, or dismiss it. Francis Bacon captured this centuries ago: “The human understanding when it has once adopted an opinion … draws all things else to support and agree with it.” The result is that we easily fool ourselves—and may even feel we are being completely rational while doing so.

Far from being a mere cognitive shortcut, confirmation bias is actively motivated by deeper psychological drives. One is the protection of self-integrity: we defend our beliefs to maintain a sense of intelligence and professionalism. Contradictory information triggers a defensive reflex, leading us to question the validity of the data rather than our own views. We also rely heavily on self-serving evidence, latching onto anecdotes or one-sided facts that support our stance while ignoring the bigger picture. These dynamics show that confirmation bias is fueled by our desire to feel competent and consistent, not simply by laziness.

Much of this bias operates automatically. Our minds are wired to accept what we encounter as true by default; understanding an idea requires constructing a mental model of what the world would be like if it were true. Doubting and un-believing are additional, effortful steps. When our thinking is otherwise occupied or we are tired, that second step often fails, making us substantially more gullible and prone to accept empty persuasive messages—a dynamic advertisers exploit when they run commercials during late-night television. This automatic tendency to believe what we hear, combined with a lazy or busy capacity for critical thinking, is the engine that drives confirmation bias.

Associative memory reinforces the bias. When we ask ourselves a question focused on one side of an issue, our minds automatically retrieve matching evidence. If you ask “Is Sam friendly?” different instances of Sam’s behavior will come to mind than if you had asked “Is Sam unfriendly?” The very framing of the question prompts a lopsided search of memory, predisposing us to answer in the affirmative. Similarly, if you consider the probability of a tsunami hitting California, your mind will call up vivid images of tsunamis rather than all the years one didn’t occur, leading you to overestimate the likelihood. This selective recall—remembering the hits and forgetting the misses—feeds directly into the availability error, but it also perpetuates confirmation bias by making consistent evidence effortlessly accessible. When our brains look for clues, we naturally grasp at the familiar, predisposing us to affirm what we already think we know.

The real-world manifestations of confirmation bias are everywhere. To defend the idea that Santa is real, one can point to millions of believers and unexplained presents under Christmas trees—both are some evidence, but considering only that evidence is a path to self-delusion. News consumption follows suit: conservatives watch Fox News, liberals watch MSNBC, each choosing channels that primarily confirm their worldview. Christians read Christian authors; atheists read atheist authors. People are far more likely to skip straight to criticisms of a book they disagree with than to read the book itself. More than 90% of college professors believe they are better than their colleagues; 70% of college students see themselves as above-average leaders; 25% of students place themselves in the top 1% for getting along with others. Gamblers swear by ‘hot’ and ‘cold’ streaks, convinced by patterns they have selectively remembered. Even sky gazers see faces of God in cloud formations, seeking meaningful shapes that confirm expectations while discarding the abundance of shapeless vapor.

In daily life, the bias seeps into mundane judgments. Imagine a father meeting his daughter’s boyfriend for the first time. Convinced the young man is bad news, Dad notices the untucked shirt and spiked hair and thinks, “He doesn’t even care enough to make himself presentable.” When the boyfriend mentions he dropped out of college to work at an ice cream parlor, Dad concludes, “No future.” Throughout the evening, Dad overlooks every good quality and sees only what he decided to see before the door opened. The same pattern plays out when we assume a homeless man is intoxicated because his speech is slurred from dehydration, or when we assume the colleague who works late is a neglectful parent when he misses a recital. In each case, we collect the data that fits our first impression and ignore the rest.

This filtering shapes not only personal judgments but also how we perceive our closest relationships. In a troubled marriage, if one partner labels the other as “selfish,” that label becomes self-reinforcing: selfish acts become easy to spot, while generous acts go unnoticed. The therapist Aaron T. Beck found that simply asking couples to keep a “marriage diary” chronicling things their mates did that pleased them reversed the dynamic. One week, Karen noted that Ted sympathized about a difficult client, helped clean the house, kept her company during laundry, and suggested a walk—acts that had previously been erased from memory by her negative frame. A study by Mark Kane Goldstein found that 70% of such couples reported an improvement in their relationship; all that had changed was their awareness. The same dynamic corrupts workplaces: a colleague speaks against our idea in a meeting, and if it happens twice, we brand him a “brown-noser,” thereafter noticing every ambiguous action that fits the label.

Rather than collecting balanced evidence, we often pretend we want the truth when we’re really seeking reassurance. “Do these jeans make me look fat?” does not invite an honest answer. Smokers in the 1960s were more eager to read an article headlined “Smoking Does Not Lead to Lung Cancer” than one with the headline “Smoking Leads to Lung Cancer.” Singers who audition for reality TV shows with no discernible talent are crushed when judges reject them—because they’ve only ever solicited feedback from supportive friends and family, locking their spotlights on affirmation and creating a wildly distorted pool of data. In business, someone considering a new car or computer will search for reasons to justify the purchase but won’t be as diligent about finding reasons to postpone. Professor Dan Lovallo observes, “Confirmation bias is probably the single biggest problem in business, because even the most sophisticated people get it wrong. People go out and they’re collecting the data, and they don’t realize they’re cooking the books.”

The Dreyfus affair stands as a monumental case study. In 1894, French army officers confronted a torn-up memo revealing a spy in their ranks. They quickly fastened on Alfred Dreyfus, the only Jewish officer on the general staff. Dreyfus was cold and arrogant, a gambler, rumored to keep mistresses—hardly a trustworthy character. Handwriting experts were consulted; the first confirmed a match, the second was equivocal. The officers dismissed the second expert because he worked for the Bank of France, a world they associated with powerful Jewish financiers, a conflict of interest they used to bolster their case. Dreyfus was convicted of treason, publicly degraded amid anti-Semitic jeers, and shipped to Devil’s Island in solitary confinement. Only years later was he exonerated, after another officer confessed and the real spy was identified. The tragedy illustrates how a quest to confirm a suspicion can marshall anecdotes, dismiss disconfirming evidence, and ruin lives.

Yet confirmation bias can appear highly scientific. We gather data, but we are selectively spotlighting the scenes that affirm our gut. A meta-analysis of 91 studies with over 8,000 participants found that people are more than twice as likely to seek confirming information over disconfirming information. This ratio climbs dramatically in emotional domains, which is why CEOs who receive flattering press coverage subsequently pay higher acquisition premiums—each major-media puff piece ratchets the premium up by 4.8%, as they come to believe their own myth.

Our deliberate reasoning often defaults to a “positive test strategy”—we consciously look for data that would confirm our current belief rather than seeking evidence that might disconfirm it. This is the opposite of what sound hypothesis-testing requires. Consider the classic experiment: the experimenter thinks of a rule governing three-number sequences and tells you that “2, 4, 6” fits the rule. Most people’s instinct is to test hypotheses like “ascending even numbers” or “ascending odd numbers” by offering sequences that confirm those guesses—such as “8, 10, 12.” But the secret to discovering the rule is to offer sequences designed to disprove your hypothesis, such as a descending sequence or a mixed-order set. The actual rule (“any three ascending numbers”) is discovered only when subjects actively try to falsify their assumptions; very few spontaneously do so. This ingrained tendency to seek confirmation is the default, but there are instructive exceptions. Chess grand masters habitually look for where a speculative move might be weak, while novices look only for confirmatory moves. The financier George Soros, when making a bet, searches for instances that would prove his initial theory wrong. These counterexamples suggest that true confidence lies in the ability to scrutinize one’s own ideas without needing constant ego-stroking.

When confronted with strong evidence against a cherished belief, the rational response is to change the belief. But what humans often do is dig in their heels. In a study by Brendan Nyhan and Jason Reifler, conservatives became more convinced that Saddam Hussein had weapons of mass destruction after being shown direct evidence that he did not. The doomsday prophet Harold Camping, when the rapture failed to materialize on May 21, 2011, didn’t admit error; he reinterpreted the non-event as an “invisible judgment day.” Conspiracy theorists are masters of this maneuver: any evidence against the conspiracy is twisted into evidence for it. Tell a 9/11 “Truther” that al Qaeda planned the attacks, and they’ll reply, “That’s what they want you to think”—an unfalsifiable defense. The “Ultimate Conspiracy Theory” illustrates the irrationality of such thinking: a clandestine group planted every conspiracy theory to throw you off the track; you can never prove it wrong, which is precisely why it is irrational. The human brain can respond to opposing views like a fortress under siege—raising drawbridges, releasing crocodiles into the moat—allowing little or nothing to penetrate. This can happen in fields like statistics as well: analysts will present datasets with no extreme events as “proof” that a Gaussian model is safe, confusing absence of proof with proof of absence. A single observation can reject the model, but millions of observations can never fully confirm it—an asymmetry that confirmation bias makes invisible.

This is why the opening quotation drips with irony. Learn a new word, and you’ll likely hear it multiple times that day. Start thinking the number 23 has significance, and you’ll see it everywhere. Once you learn about confirmation bias, you will start seeing it everywhere. That’s because you will be looking for confirmation and ignoring the counter-examples—just as the bias predicts. We remember the hits and forget the misses, which keeps the cycle spinning.

If your goal is truth, the antidote is clear but demanding. Actively try to prove your belief false; if you have only ever looked for supporting evidence, it is no surprise you found some. Genuinely seek out disconfirming information. Invite others to challenge your views. The more scrutiny a belief withstands, the more likely it is true. If solid evidence emerges against your position, change it. A good skeptic keeps an open mind and second-guesses even the most beloved conclusions, knowing that new information may demand revision. If future science were to demonstrate that star signs genuinely influence personality, the rational response would be to update that view, not cling to old certainties. Never wavering in the face of contrary evidence is a mistake; what matters is staying as closely aligned with truth as you can.

A suite of practical techniques can interrupt the automatic confirmation spiral. One is to deliberately “consider the opposite.” In organizations, this can take the form of assuming positive intent: when a colleague’s email seems disrespectful, you ask, “What’s another, more generous interpretation?” Indra Nooyi, the former CEO of PepsiCo, credits this rule—learned from her father—as the best advice she ever received: “When you assume negative intent, you’re angry. If you take away that anger and assume positive intent, you will be amazed.” At Industrial Scientific, employees were trained in this approach to bridge cultural gaps between French and Chinese subsidiaries where communication styles clashed.

The historical devil’s advocate role, used by the Catholic Church to argue against canonization, is another structured form of considering the opposite. After its elimination, canonization rates increased twentyfold. In business, Alfred Sloan, former CEO of GM, would purposefully delay decisions to encourage disagreement, knowing that having a contrarian voice reduces hubris. Roger Martin’s “What would have to be true?” technique goes further: when teams reach an impasse, he asks each side to identify the conditions under which the opposing option would be the best answer. This shifts the argument from adversaries defending positions to collaborators analyzing hypotheticals, allowing people to change their minds without admitting defeat.

A more radical commitment is to design deliberate mistakes. Paul Schoemaker, of Decision Strategies International, once challenged his firm to test one of its core assumptions by intentionally violating it. They chose to respond to a Request for Proposal (RFP), despite a long-standing policy that RFPs were a waste of time. The deliberate mistake resulted in more than $1 million in consulting business from a utility company. While most deliberate mistakes will fail, the practice signals that decisions will be made based on evidence, not folklore or political convenience. Even in personal life, such strategies can pay off: researcher John T. Molloy found that 20% of recently married women did not like their future spouse when they first met. Inspired by this, members of his team accepted dates with men they had repeatedly rejected; one such “mistake” ended in marriage, a small triumph over the confirmation bias.

Overcoming confirmation bias pays dividends beyond intellectual honesty. In conversation, rather than assuming you know what someone is about to say, pausing to listen and asking clarifying questions earns respect and often yields surprising insights. In the workplace, setting aside assumptions about how a project will go can open the door to groundbreaking information. These habits—regularly challenging your own notions, exposing yourself to unbiased sources, engaging with people who disagree with you, and structuring your environment to force consideration of the opposite—are not just tools for thinking better. They are, as Warren Buffett and Charlie Munger demonstrated, a pathway to better decisions. As David Hume put it, “A wise man … proportions his belief to the evidence.” And wise men and women know they must tirelessly work against their natural inclination to do otherwise.