Confirmation bias: how our beliefs filter reality

Confirmation bias is the tendency to favor information that validates our pre-existing beliefs, while downplaying or ignoring information that contradicts them. It is not a rare malfunction: it is a universal mechanism, rooted in our cognitive architecture.

The psychologist Daniel Kahneman describes two speeds of information processing. System 1, fast and automatic, relies on intuitions, reflexes and deeply held beliefs. System 2, analytical and rational, is too costly in time to be used routinely. Confirmation bias takes hold in System 1, generating distortions in information processing without our being aware of it.

Described as an omnipresent phenomenon, in many forms[1], this bias belongs to a vast family: the Cognitive Bias Codex by Buster Benson and John Manoogian III organizes a very large number of them. Among them, confirmation bias is one of the most frequent and most impactful on the quality of decisions.

The bias operates in three main contexts: social relationships, information consumption and the workplace. Each has its own reinforcement mechanisms.

Social selection and interpersonal relationships

We naturally choose to surround ourselves with people who share our opinions. This social selection reinforces homogeneous circles where convictions are consolidated without being questioned. At the collective level, this phenomenon fuels the polarization of opinions and groupthink: disagreements become less frequent, certainties more rigid, judgments about outsiders more clear-cut.

Information consumption and social media

Information filtering operates on two levels. First, individually: we avoid sources that contradict our convictions and consume those that confirm them. Second, algorithmically: digital platforms amplify this phenomenon by offering content similar to what we have already viewed, creating information bubbles that reinforce confirmation bias with every login.

Professional and strategic decisions

In a professional environment, this bias takes a particularly costly form. A leader who has validated a market hypothesis will naturally interpret data in line with that hypothesis, without testing contradictory scenarios. Investment decisions and hiring are especially exposed: favorable signals are retained, warning signs are downplayed.

Anxiety and phobias: when fear looks for its proof

Confirmation bias also helps explain how a fear persists. If you dread elevators, you will notice every story of a breakdown and forget the hundreds of uneventful rides. By avoiding the situation, you also deprive yourself of any chance to find out the opposite. This is the cycle that cognitive behavioral therapy works on: examining the evidence for and against a thought, then confronting it with the facts through exposure.

The practitioner's perspective

The medical field is not immune. In an experiment conducted with 75 psychiatrists and 75 medical students, 13% of the psychiatrists and 25% of the students mainly sought information confirming their initial diagnosis, and these participants less often reached the correct diagnosis[2].

Company stories are often used to illustrate this bias. They are after-the-fact readings: other factors played a role, and the share of the bias cannot be isolated. They do, however, show what a conviction that is never put to the test can cost.

Kodak: the ignored invention

Kodak developed the first digital camera in-house as early as 1975. Its executives nevertheless set the technology aside, convinced that film photography would remain dominant. Contrary signals existed. They were downplayed. The result: a century-old company filing for bankruptcy in 2012, unable to revise its certainties in the face of a market disruption it had itself initiated.

Blockbuster: the underestimated threat

Faced with the emergence of Netflix, Blockbuster maintained its conviction that the physical in-store rental model would remain relevant. Attendance data and customer behavior were sending clear signals. The confirmatory bias of the leadership teams led them to ignore these warnings. In 2010, the company declared bankruptcy.

Nokia: fatal technological certainty

Nokia dominated the global mobile phone market in the early 2000s. The internal conviction that its technological lead was unassailable slowed its adaptation to the rise of smartphones. Economic strategies were built on hypotheses validated internally, without serious contradictory testing. Its market share subsequently collapsed.

These three cases have one thing in common: the selective search for evidence in situations where the stakes were high.

Cognitive biases do not operate in isolation. They chain together and reinforce one another, degrading the quality of decisions cumulatively. Here are the four most frequent interactions with confirmation bias.

Anchoring effect and first impressions

The anchoring effect refers to our tendency to stick with the first idea we received on a subject. Once this anchor is set, confirmation bias takes over: we gather information that validates it and set aside information that invites us to revise it. Revising our judgment becomes difficult, even in the face of strong evidence.

Survivorship bias and overgeneralization

Survivorship bias consists of concentrating attention on the cases that succeeded while ignoring the failures. The result: we build distorted models from exceptions. Confirmation bias amplifies this phenomenon by retaining only the success stories that confirm the initial theory, systematically making counterexamples invisible.

Dunning-Kruger effect and overconfidence

First described in 1999, the Dunning-Kruger effect[3] describes the tendency of people with low skill to overestimate their abilities, for lack of being able to measure their gaps. This mechanism ties in directly with confirmation bias: the more competent we believe ourselves to be, the more we validate our own ideas without submitting them to the test of contradiction. Ultracrepidarianism is its extreme form.

Confusing correlation with causation

A misinterpretation of statistical links: two correlated phenomena are perceived as connected by a cause-and-effect relationship. Confirmation bias reinforces this confusion by steering the search for evidence toward data that seem to confirm a nonexistent causality. A classic example: believing that a repeated coincidence proves a mechanism, when no cause-and-effect link exists.

Reducing the influence of confirmation bias takes concrete strategies, applied at two levels: individual and collective.

Individual strategies:

• Actively seeking contrary evidence: before validating a hypothesis, deliberately look for the data that contradicts it.
• Pre-mortem: imagine that the decision has failed before making it, then look for the reasons why.
• Diversifying sources of information: expose your convictions to opposing viewpoints, drawn from varied and independent sources.
• Keeping a decision journal: note the initial hypotheses, the data taken into account and the data set aside. Going back over these notes afterwards reveals filtering patterns.
• Asking the inverted question: for each conviction, explicitly ask "What could prove that I am wrong?"
• Slowing down System 1: take the time to engage Kahneman's System 2 on high-stakes decisions, by imposing a period of reflection before any validation.

The collective approach: adversarial validation

At the level of a team or an organization, critical thinking is institutionalized through structured processes. Appointing a "devil's advocate" in strategic meetings, whose explicit role is to defend the opposing thesis, helps bring objections to the surface.

Peer reviews, decision audits and multidisciplinary committees serve the same purpose: exposing hypotheses to outside eyes that have no stake in validating them.

In clinical practice, actively looking for what contradicts an initial diagnosis helps avoid this pitfall[2]. Virtual reality environments such as those of C2Care also allow the practitioner to review the same situation under comparable conditions, adjusting the parameters, and to compare their impressions with what they observe. This does not eliminate the bias: it is one more reference point in their clinical judgment.

Training in cognitive biases is another avenue: recognizing the mechanism does not eliminate it, but can create a critical distance.

Confirmation bias is one of the most widely studied cognitive biases. It operates silently, in social relationships, in the consumption of information and in strategic decisions, filtering reality to make it fit our pre-existing beliefs.

It can be costly: the stories of companies that failed to revise their certainties illustrate this, even if they do not prove it on their own. It also helps explain why a fear persists.

It interacts with other cognitive biases to amplify distortions of judgment. Active strategies of adversarial validation, both individual and collective, can limit its effects. Recognizing this mechanism is already a first form of critical thinking.