Confirmation Bias — Why Educated Indians Still Believe What They Want to Believe

The argument had been going on for forty minutes. Arjun, 31, a data analyst at a Gurugram fintech company, had arrived at his parents' Sunday dinner with three printed articles, two YouTube links saved on his phone, and the quiet confidence of someone who has prepared thoroughly for a debate he believes he cannot lose. The topic was a political one — it does not matter which one, because the specific topic is not the point — and Arjun had spent the previous week assembling what felt to him like an airtight case. Every source he had found confirmed his position. Every counterargument he had encountered, he had been able to rebut. His father, a retired professor of economics, held the opposing view with equivalent confidence and had assembled equivalent evidence for it. By the end of the evening, neither had moved. Both had, if anything, emerged more certain than before. Arjun drove home convinced that his father simply refused to look at the evidence. His father sat in the kitchen thinking exactly the same thing about his son. Both of them were right about the other. Neither of them was right about himself.

This is confirmation bias at full operational capacity — not in someone who has not thought about the issue, not in someone who lacks information, but in two educated, intelligent, evidence-citing people who have both done the cognitive work of constructing a case and arrived, through that work, at the conclusion they started with. The specific irony of confirmation bias in educated populations is that the skills education develops — the ability to find evidence, to construct arguments, to anticipate objections — are the same skills that, when deployed in the service of a prior conclusion, become the machinery of motivated reasoning. Arjun was not failing to think carefully. He was thinking carefully in entirely the wrong direction, and the care made the error harder to detect rather than easier.

What Confirmation Bias Actually Is — and What It Is Not

Confirmation bias, as defined by Peter Wason in his foundational 1960 research at University College London, is the tendency to search for, interpret, favor, and recall information in a way that confirms or supports one's prior beliefs or values. Wason's original 2-4-6 task — in which participants were asked to discover a rule governing number sequences and consistently tested only sequences that confirmed their initial hypothesis rather than sequences that might disconfirm it — demonstrated that intelligent people systematically avoid the tests most likely to reveal they are wrong. This finding has been replicated across hundreds of studies in the six decades since, and it remains one of the most robust and depressing results in all of cognitive psychology: the mind, left to its own devices, is not an instrument of inquiry. It is an instrument of confirmation.

The bias operates through three distinct mechanisms that produce different kinds of errors. Selective exposure: people preferentially seek out information sources likely to confirm their existing views, constructing an information environment that increasingly reflects back what they already believe. Selective interpretation: when people encounter ambiguous information — and most real-world information is ambiguous — they interpret it in the direction of their prior beliefs, so that the same evidence is read as supporting opposite conclusions by people with opposite priors. Selective recall: people remember confirming information more readily and more accurately than disconfirming information, so their felt sense of the evidential record is systematically skewed. These three mechanisms compound each other: the information you selectively seek is the information you selectively interpret favorably, which is the information you selectively remember — and the accumulated memory then feels like evidence for seeking more of the same.

What confirmation bias is not is simply a matter of preferring agreeable information. Daniel Kahneman's framework, developed across decades of research at Princeton and Hebrew University and synthesized in Thinking, Fast and Slow, distinguishes between System 1 thinking — fast, intuitive, emotionally driven — and System 2 thinking — slow, deliberate, analytical. Confirmation bias operates primarily through System 1: the gut sense that something is true, the immediate comfort of recognition when familiar information appears, the instinctive resistance when contradictory information is encountered. System 2 is capable of overriding this, but doing so requires effort, motivation, and the specific willingness to experience the discomfort of discovering that a prior belief may be wrong. Most of the time, that willingness is absent — and the System 2 resources that could catch the bias are recruited instead to rationalize what System 1 has already decided. This is the cognitive architecture that produces Arjun's printed articles: the sophisticated, effortful, evidenced case built in service of a conclusion that was reached before any article was printed.

Why Education Does Not Fix It — and Often Makes It Worse

Nandini, 33, is a senior research associate at a pharmaceutical company in Hyderabad. She has a master's degree in biochemistry, spends her professional life evaluating evidence, and considers herself a careful, data-driven thinker — which she is, in her professional domain. In conversation, she describes a pattern she has noticed and named in herself over several years: that the rigorous skepticism she applies to clinical trial data at work is entirely absent from the way she evaluates information about certain topics she holds strong views on. A WhatsApp forward claiming a health benefit for a traditional remedy she grew up using does not receive the same methodological scrutiny as a pharmaceutical study she is paid to evaluate. A news report that confirms her view of a political party is read and shared; a news report that complicates it is dismissed as biased reporting. She has the tools to catch this. She does not, in those moments, use them.

Nandini's experience is not an anomaly. It is precisely what the research predicts. A landmark 2012 paper by Dan Kahan at Yale Law School, examining the relationship between cognitive ability and polarization on contested scientific topics, found that higher cognitive ability predicted stronger polarization rather than weaker — the more cognitively capable the person, the more effectively they deployed their skills in support of the position their identity group held. Kahan called this identity-protective cognition: the use of intelligence not to reach accurate conclusions but to protect the social identity associated with a particular set of beliefs. A subsequent 2017 study by Kahan and colleagues, published in Advances in Political Psychology, extended this finding across multiple contested topics and found that the pattern was robust: numeracy, verbal reasoning, and general cognitive ability all predicted stronger, not weaker, politically motivated reasoning in domains where beliefs were identity-linked.

The mechanism is not difficult to understand once it is named. Education develops reasoning skills that are value-neutral: they can be used to reach accurate conclusions or to defend inaccurate ones with equal technical effectiveness. In a context where a person's social identity, group membership, and emotional commitments are invested in a particular set of beliefs — which, in India, describes a great many beliefs about politics, religion, caste, region, and family — those cognitive skills are deployed in the service of those beliefs rather than in their critical examination. Nandini does not fail to evaluate the WhatsApp forward carefully because she lacks the skill. She fails to evaluate it carefully because in that moment, the belief the forward confirms is not a hypothesis under investigation. It is a commitment under protection.

The Indian Context: Why Confirmation Bias Has Specific Local Amplifiers

Vikram, 38, is a software architect in Chennai who describes himself, without apparent irony, as someone who always follows the data. He reads extensively, maintains strong opinions across a wide range of topics — political, economic, historical, social — and engages with those topics with genuine intellectual energy. In a conversation about how he forms his views, he is asked a simple question: when did he last significantly change his mind about something he had held strongly? He thinks for a long moment. He mentions a technical architecture decision from work. He cannot produce an example from politics, history, or social questions — domains in which he holds his strongest views and in which he reads most extensively. The reading, it turns out, has not been changing his mind. It has been furnishing his existing positions with better arguments.

While confirmation bias is universal, its specific expression in the Indian educated middle class is shaped by cultural and structural factors that amplify it in identifiable ways. The first amplifier is the architecture of Indian social identity itself. In a society organized around intersecting axes of caste, religion, region, and language — each identity-constitutive in ways with deep historical roots — the beliefs associated with these identities are not held as provisional hypotheses. They are held as constitutive commitments: to question them is not intellectual inquiry but social threat. Henri Tajfel and John Turner's Social Identity Theory, developed across research at Bristol University in the 1970s and 1980s, established that people's sense of self is substantially constituted by their group memberships, and that this creates a systematic motivation to view in-group beliefs and positions favorably relative to out-group alternatives. In the Indian context, where multiple powerful identity groups are simultaneously active in a single person's self-concept, this motivation operates across more domains simultaneously than in most other cultural contexts — which means more beliefs are identity-linked, and therefore more resistant to evidence-based updating.

The second amplifier is the WhatsApp information ecosystem. A 2023 study by researchers at the Reuters Institute for the Study of Journalism documented that messaging app ecosystems — particularly in countries like India where WhatsApp functions as a primary news distribution channel for a large portion of the educated population create closed information loops in which content is shared selectively within ideologically homogeneous groups, receives social validation through group endorsement, and is rarely subjected to external verification before being accepted as credible. The social endorsement of a shared forward the fact that it comes from a trusted person within the group functions as a credibility signal that bypasses critical evaluation. Nandini, the biochemist who scrutinizes clinical trials for a living, accepts a health claim from a family WhatsApp group without applying a fraction of that scrutiny, not because she is inconsistent but because the social context of the information trusted source, group affirmation, emotionally resonant content has already done the work of making it feel credible before her analytical faculties are engaged.

The third amplifier is the cultural valorization of argumentative performance. Across many educated Indian family and social contexts, the ability to argue a position effectively to cite evidence, to rebut counterarguments, to demonstrate the superiority of one's view through rhetorical and evidential skill — is a valued social performance. This creates an incentive structure that rewards the ability to find evidence for a predetermined position and penalizes visible belief-updating, which can appear as inconsistency or weakness. Arjun's Sunday dinner was not an inquiry. It was a performance. Both participants knew it. The goal was not to discover what was true. The goal was to demonstrate that one's existing position was defensible. In that context, confirmation bias is not merely a cognitive tendency. It is a social requirement.

How Confirmation Bias Operates Across Specific Belief Domains

The specific domains in which confirmation bias is most consequential in the Indian educated context are worth examining individually because the mechanisms differ slightly even when the underlying cognitive error is the same. In the domain of political beliefs, it operates primarily through selective exposure: people construct information environments — through their choice of news channels, social media feeds, and social circles — that reinforce what they already believe, and they experience contradictory news not as information to be evaluated but as evidence of the source's bias. Vikram, the Chennai software architect, watches one news channel and dismisses another as propaganda. He has not evaluated the second channel's content. He has evaluated its political alignment, found it to conflict with his prior, and pre-emptively categorized its output as unreliable. This is a fully self-sealing system: any evidence against his view confirms, through the dismissal process, that the sources of that evidence are not to be trusted.

In the domain of medical and health beliefs, confirmation bias takes a form that is particularly visible in the Indian educated middle class and particularly worth examining carefully. Preethi, 41, is a general physician in Coimbatore who describes with professional frustration what she observes in educated patients who simultaneously hold degrees, practice evidence-based medicine in some contexts, and maintain firm commitments to practices with no validated evidence base — specific dietary beliefs, astrological guidance on medical timing, untested supplements promoted within family networks. The specific cognitive mechanism here is what researchers call the Texas Sharpshooter Fallacy: the tendency to notice and remember hits — times when the practice appeared to work — and to not notice or quickly forget misses — times when it did not. The patient who took a traditional remedy and recovered remembers the recovery as evidence of the remedy's effectiveness. The patient who took it and did not recover explains the failure as a matter of incorrect dosage, insufficient compliance, or the illness having been too advanced. The belief is maintained not because the evidence supports it but because every outcome, positive or negative, is interpreted through the belief.

In the domain of caste and social hierarchy beliefs, confirmation bias operates through the selective attention and recall mechanisms that Tajfel and Turner's Social Identity Theory predicts. The person whose social identity is invested in a hierarchical belief about caste-correlated capability or character will notice and remember individual instances that appear to confirm it, and will explain away or fail to encode instances that contradict it. This is not experienced as bias by those who hold it. It is experienced as an accurate reading of the evidence — because the evidence itself has been filtered through a prior that determines what counts as evidence and how each data point should be read. The specific invisibility of this process to the person inside it is what makes it most resistant to challenge.

The Specific Injury of Smart People Arguing Badly

Kavitha, 29, is a corporate lawyer in Mumbai who describes, with the specificity of someone who has spent time examining it, a professional moment in which she recognized confirmation bias operating in herself in a context where it had measurable professional consequences. She had been building a legal argument in an internal review and had assembled what she describes as a comprehensive set of supporting precedents. When a colleague raised three contrary precedents in the review meeting, her first instinct was not to evaluate them. It was to explain why they were distinguishable — to find, immediately and fluently, the reasons why the contrary evidence did not apply. She was able to do this convincingly, and the review proceeded in the direction she had argued for. It was only later, reviewing the decision alone, that she registered that she had never actually evaluated the contrary precedents on their merits. She had evaluated them on their threat to her existing argument, and the evaluation had been conducted entirely in the service of neutralizing that threat rather than determining whether they were relevant.

This is the specific injury that confirmation bias in educated people produces that differs from the injury produced by simple ignorance: it creates highly capable defenders of inaccurate positions. The argument that confirmation bias builds containing real evidence, citing real sources, anticipating obvious objections, articulated with the fluency that education produces is genuinely difficult to counter. The person on the other side, who may hold a more accurate position but has not assembled an equivalently sophisticated case for it, often finds themselves losing the argument in the social sense even when they are right. This produces a discourse environment in which being more confident, better prepared, and more rhetorically skilled is rewarded over being more accurate. Arjun's father, the retired economics professor, was not wrong to observe that his son refused to engage seriously with contrary evidence. But neither was he examining his own equivalent refusal. Both were performing the argument rather than conducting the inquiry, and the performance rewarded the better performer, not the more accurate reasoner.

What the Research Says About Debiasing and Why It Is Hard

The research literature on cognitive debiasing is considerably less encouraging than the research on the bias itself. Several interventions have been studied in controlled settings: consider-the-opposite instructions, which ask people to actively generate reasons why their belief might be wrong; accuracy prompts, which direct attention to the importance of being correct rather than consistent; and perspective-taking exercises, which ask people to articulate the strongest version of the position they oppose. These interventions show modest positive effects in laboratory settings where the beliefs are low-stakes and the motivation to defend them is minimal. A 2022 meta-analysis published in Psychological Bulletin, examining debiasing interventions across 52 studies, found that the interventions most consistently effective in real-world settings shared a common feature: they worked not by providing contradictory information which typically activated defensive reasoning rather than genuine updating but by changing the evaluative framework within which information was being processed.

The implication is significant. Showing Arjun better evidence for his father's position will not change Arjun's mind. It will give him better evidence to argue against. The error is not informational. It is structural located in the framework within which Arjun is processing information, not in the information itself. What changes the framework is changing the social norm around belief-updating: from one that values consistency and argumentative performance to one that values accurate calibration. Research by Philip Tetlock at the University of Pennsylvania on superforecasting the study of people who are reliably more accurate than average at predicting future events finds that the single most consistent feature of good forecasters is not intelligence, not information access, and not domain expertise. It is actively open-minded thinking: the genuine willingness to update beliefs on new evidence, which Tetlock operationalizes as the measurable frequency with which a person actually revises their probability estimates in response to new information. The superforecasters are not smarter than Arjun. They are more committed to being right than to being consistent.

What Actually Helps: Reasoning Habits That Reduce the Bias

The first and most foundational habit is what psychologist Jonathan Baron at the University of Pennsylvania calls actively open-minded thinking a deliberate orientation toward the possibility that one's current belief is wrong, operationalized as a genuine search for evidence that would update it. The specific practice Baron identifies as most diagnostic is asking, for any belief held with strong confidence, "What would I need to see to change my mind?" Not as a rhetorical question designed to demonstrate that no such evidence exists, but as a genuine specification of the conditions of revision. Arjun, preparing for the Sunday dinner, did not ask this question. If he had asked it seriously, what evidence encountered in the next week would genuinely shift my confidence in this position? and found that he could not specify any, he would have located the boundary of his confirmation bias: a belief he has decided to hold regardless of what the evidence turns out to be. The inability to specify conditions of revision is the most reliable available signal that a belief is being held for reasons other than evidence.

The second habit is steel-manning rather than straw-manning opposing views, engaging with the strongest possible version of a position one disagrees with rather than its weakest or most extreme expression. Hugo Mercier and Dan Sperber's argumentative theory of reasoning, developed in research published in Behavioral and Brain Sciences, proposes that human reasoning evolved primarily for social purposes to construct arguments and evaluate those of others in a group context rather than for individual truth-seeking. This architecture means that the mind naturally generates weak versions of opposing positions, which are easier to rebut, and strong versions of supporting positions, which are easier to affirm. The deliberate practice of finding the most sophisticated, evidence-supported articulation of an opposing view, seeking it out in the actual literature rather than in the caricature familiar from one's own side's criticisms, is cognitively effortful precisely because it works against this default. Kavitha, the Mumbai lawyer, has built this into a professional habit: before finalizing any significant legal argument, she requires herself to write the strongest version of the opposing brief, not to be convinced by it, but to ensure that what she is arguing against is the actual position rather than the version of it that is easiest to dismiss.

The third habit is the deliberate monitoring of one's information environment for homogeneity, the honest periodic assessment of whether the sources one reads, watches, and shares represent genuine diversity of serious perspective on contested questions, or whether they primarily represent confirmation of a prior view. Nandini, after recognizing her own pattern, instituted a specific rule for herself: for any topic on which she holds a confident view, she reads one serious, carefully reasoned source that reaches a different conclusion before considering her position settled. Not a fringe source, not a source she already knows she disagrees with and can dismiss on methodological grounds, but a source that is doing the epistemic work seriously and arriving somewhere different. The point is not to become convinced. The point is to ensure that the conclusion she holds is one that has survived engagement with the strongest available opposition, which is the minimum standard for holding it in a way that is responsive to evidence rather than merely defended against it.

Frequently Asked Questions

Q1. If confirmation bias is universal, why does education seem to make it worse in some cases?

Because education develops general reasoning skills, the ability to find evidence, construct arguments, and articulate positions that are value-neutral and can be deployed either to reach accurate conclusions or to defend inaccurate ones. When a belief is identity-constitutively tied to social belonging, group membership, or emotional investment, the cognitive skills education provides are recruited by the motivated reasoning system to defend the belief rather than to examine it. Dan Kahan's research at Yale on identity-protective cognition demonstrates that higher cognitive ability predicts stronger polarization on identity-linked beliefs, not weaker, precisely because more cognitively capable people are more effective at deploying their skills in service of conclusions that were reached before the reasoning began. Arjun's printed articles are not evidence of careful reasoning. They are evidence of careful reasoning's resources being directed by motivated reasoning's agenda.

Q2. What is the difference between confirmation bias and simply having a well-supported belief?

The operational difference is testability, specifically, whether the person holding the belief can specify conditions under which they would revise it. A well-supported belief is one for which the person can articulate what evidence, if it appeared, would shift their confidence. A confirmation-biased belief is one that has been made unfalsifiable: every piece of contrary evidence is explained away by the same prior that produced the belief in the first place. If contradictory reporting is always dismissed as biased media, if contrary research is always attributed to methodological failure or political motivation, and if every negative outcome is explained as a special case that does not apply to the general principle then the belief cannot be disconfirmed by any evidence, which is not a mark of its strength but of its insulation from the process by which beliefs are supposed to be evaluated. Vikram's inability to identify a single significant belief he has updated through reading is the diagnostic signal: not that his views are wrong, but that his reading has not been serving an inquiry.

Q3. Why is WhatsApp specifically such an effective confirmation bias amplifier in India?

Because it combines three features that each individually would amplify confirmation bias and that together are particularly powerful. First, content is shared within closed groups that are ideologically and socially homogeneous, which means the information environment is pre-filtered to contain content the group already agrees with. Second, content receives social endorsement from trusted sources within the group before it is encountered, which provides a credibility signal that bypasses critical evaluation. Nandini accepts the health claim because her trusted aunt sent it, not because she has evaluated it. Third, the format favors shareable, emotionally resonant, conclusion-forward content over complex, caveated analysis, which means the content most likely to circulate is also the content most likely to confirm prior beliefs at the expense of accuracy. The Reuters Institute research on messaging app ecosystems documents that this combination produces substantially higher levels of misinformation acceptance than almost any other studied information environment, specifically in populations with high educational attainment precisely because the educated person's critical faculties are bypassed by social trust rather than overcome by argument.

Q4. Is confirmation bias the same in all cultural contexts, or does the Indian context produce it differently?

The underlying cognitive mechanism is universal, but the specific expression in the Indian educated middle class is amplified by local factors that make it particularly intense in certain domains. The intersecting, identity-constitutive nature of caste, religion, region, and language in Indian social identity means that more domains of belief are identity-linked and therefore subject to identity-protective motivated reasoning than in most other cultural contexts. The collectivist social structure, in which group belonging and social approval carry substantial weight in individual self-evaluation, raises the stakes of belief-updating in ways that increase resistance to it: to update a belief publicly is to risk the social cost of appearing inconsistent, disloyal, or corrupted by outside influence. And the argumentative performance culture of many educated Indian family and social contexts, in which winning the debate is a valued social outcome, creates incentives that reward the accumulation of confirming evidence over the honest engagement with disconfirming evidence.

Q5. Can confirmation bias be completely eliminated through effort and awareness?

No, and the research literature is clear on this point. Confirmation bias is a fundamental feature of human cognition the tendency to use prior beliefs as interpretive frameworks for new information is what makes perception and decision-making possible at the speed that life requires. The goal of debiasing is not elimination but reduction: specifically, the reduction of its most consequential expressions in high-stakes belief domains where accuracy matters. Awareness of the bias helps modestly at the margin. Knowing the term and having encountered the research did not prevent Kavitha from dismissing the contrary precedents in real time, though it did allow her to recognize what she had done afterward. The more reliable interventions are structural: building specific habits that interrupt the default toward confirmation, deliberately diversifying information environments, and most importantly changing the evaluative norm from one that rewards consistency and argumentative performance to one that rewards accurate calibration. The latter is a social rather than individual change, which is why it is also the harder one.

Q6. What is the single most effective thing an educated person can do to reduce confirmation bias?

Based on the research on actively open-minded thinking by Jonathan Baron and on superforecasting by Philip Tetlock, the single most practically effective habit is the regular, honest practice of specifying for beliefs held with strong confidence what evidence would change one's mind and then genuinely looking for it. Not as a rhetorical exercise designed to demonstrate that no such evidence exists, but as a real search conducted with the willingness to find it. This requires Vikram to ask not only what he believes about the political or historical or social questions he holds his strongest views on, but what he would accept as sufficient evidence to revise those views and whether he has actually encountered that evidence and dismissed it or simply avoided encountering it in the first place. The person who cannot complete this exercise for any of their strongly held beliefs on contested topics has found the boundary of their confirmation bias: the commitments they have decided to maintain regardless of what the evidence turns out to be. That boundary is worth knowing, even when especially when crossing it feels like too much to ask.

The relationship between confirmation bias and the curated self we present in evaluative and social contexts and what genuine intellectual honesty costs in environments that reward performance over accuracy is explored in The Person I Am Alone vs The Person I Show the World. And for the specific way that family and cultural contexts in India shape not just what we believe but how fiercely we defend it, Talking to Parents About Your Relationship — The Indian Challenge examines that dynamic in depth.