Humans Are Not Very Good at Detecting Lies, Scams, and Fraud

Why Humans Fail to Detect Lies

Author’s Note

Trust makes ordinary social life possible, and criminals exploit that necessity through deliberate deception. Scam victims deserve explanations grounded in human psychology rather than accusations about intelligence or character. The distinction between sensing dishonesty and verifying information matters: a feeling of credibility is an experience, while independently checked evidence provides a basis for action. Recovery and prevention both benefit from that distinction.

Overview

Most people believe they can tell when someone is lying to them.

They trust their instincts, watch for shifting eyes or a nervous laugh, and feel confident that a lie would show itself. Decades of research tell a different story. Human beings are poor lie detectors, and the gap between how well people think they detect deception and how well they actually do it explains a great deal about why fraud succeeds.

The clearest evidence comes from a landmark review by psychologists Charles Bond and Bella DePaulo, who combined the results of 206 studies involving 24,483 judges. Their finding was striking: people achieve an average of 54% correct lie-truth judgments, correctly classifying 47% of lies as deceptive and 61% of truths as nondeceptive. Pure guessing produces 50%.

Human judgment, in other words, beats a coin toss by only four percentage points, and it catches fewer than half of the lies people actually hear.

This result is not a fluke of one study or one method. Later reviews describe the 54% average as a reliable and reproducible finding, and accuracy stays remarkably consistent across settings. The pattern holds whether the liar has much at stake or little. Motivated liars (54 percent) yielded about the same degree of accuracy as liars lacking motivation (53 percent).

No matter how researchers arrange the test, human lie detection lands in the same narrow band just above chance.

Several Forces Combine to Produce this Result

The first is that humans lean toward belief

People correctly identified 61% of truths but only 47% of lies, which reveals a built-in tilt toward accepting what others say.

Communication researcher Timothy Levine calls this the “truth default.” In ordinary life, most of what people hear is true, so believing others works well most of the time. Constant suspicion would make cooperation, friendship, and commerce impossible. The truth default is not a flaw in human reasoning. It is the foundation of social life, and it leaves people exposed whenever someone deliberately exploits it.

The second force is that people watch for the wrong signals

Popular culture teaches that liars avoid eye contact, fidget, stammer, or touch their faces. Research does not support these beliefs. A large review of 120 studies examined 158 possible signs of deception and found that most behaviors were not linked or were only weakly linked to deception. Worse, the signals people rely on most are not the ones that matter. Research shows that the cues people think indicate deception are not actually strongly related to deception. Psychologists Maria Hartwig and Charles Bond concluded that people base their judgments on cues that are invalid. A nervous truth-teller looks guilty, and a calm, practiced liar looks honest.

The third force is that real signs of lying are faint and inconsistent

Liars do not share a universal tell. Some lying people appear anxious, but so do many honest people under pressure. A skilled liar rehearses the story, controls the voice, and maintains steady eye contact precisely because they know observers watch for those signals. The behavioral differences between lying and truth-telling, where they exist at all, are too small and too variable for the human eye to read reliably in the moment.

The fourth force is that experience and training do not solve the problem

One would expect police detectives, customs officers, judges, and intelligence professionals to outperform ordinary people. They do not. Empirical evidence suggests that experts such as police officers are not better at detecting deception compared to laypeople. Professionals often feel more confident in their judgments, but that confidence does not translate into accuracy. Years on the job teach people the same unreliable cues everyone else uses, and confident error is often more dangerous than humble uncertainty.

These limits grow sharper in the situations where fraud thrives. Laboratory studies typically test short, unrehearsed lies told by volunteers who have nothing to gain. Real-world deceivers differ in every important way. Professional scammers follow scripts refined across thousands of victims. They tell their lies over weeks or months, building familiarity that strengthens the truth default rather than weakening it. They often communicate through text and messaging apps, which strip away tone of voice and many of the few useful signals that exist; research shows that people are more accurate in judging audible than visible lies, and written messages offer even less to judge. A scammer also controls every piece of evidence the victim sees, from the photographs to the documents to the supposed bank statements.

Against a deceiver with this much preparation and control, the modest 54% accuracy seen in laboratory tests becomes an optimistic estimate.

Human memory and emotion add one more layer. Once a person trusts someone, they interpret new information in light of that trust. An odd detail gets explained away as a misunderstanding, a delayed meeting as bad luck, and a request for money as an emergency between people who care about each other. Each step feels reasonable from the inside. This is not foolishness. It is the same pattern of interpretation every person uses to maintain relationships with family, friends, and colleagues, and scammers deliberately engage it.

The research leads to one firm conclusion about scam victims: being deceived is not a sign of stupidity, carelessness, gullibility, foolishness, naivete, or weak character. It is the expected outcome when an ordinary human mind meets a skilled and determined liar. Everyone carries the same truth default, relies on the same misleading cues, and performs at the same level just above chance, whether that person is a retiree, a physician, a police officer, or a psychologist. The responsibility for a scam belongs entirely to the criminal who designed the deception, studied the victim, and exploited a universal human trait for profit. The victim did not fail a test that others pass. Humans are simply bad at detecting deception, and the people who prey on that fact are the ones who deserve blame.

What the Truth Default Is

The “truth default” comes from communication researcher Timothy Levine’s Truth-Default Theory, developed over decades of deception experiments and laid out in his 2020 book Duped. Its central claim: when people communicate, they don’t actively evaluate whether each statement is true. They passively accept it. The question “is this person lying?” never comes up unless something specific raises it.

The default is not the same as believing someone after weighing the evidence. It’s the absence of any weighing at all. People don’t think “I believe you”; they just process what was said as information and move on.

How it works: triggers and thresholds

Levine’s model says people stay in the truth-default state until a trigger knocks them out of it. The triggers include:

• A visible motive to lie, such as a salesperson praising their own product
• Behavior that seems off, like nervousness, evasiveness, or odd timing
• Inconsistency: a story that contradicts itself or known facts
• Third-party information, such as a friend’s warning or a news report
• Lack of coherence, where the account doesn’t hang together

Even when a trigger appears, it has to pass a threshold. A single small oddity usually gets explained away (“he’s probably just tired”). Only when suspicion accumulates past that threshold does a person begin actively assessing honesty. And even then, people often slide back into the default once the trigger fades.

Levine argues this is rational, not a flaw. In research by Serota, Levine, and Boster, most people reported telling no lies on a given day, and a small minority of prolific liars accounted for most of the lies told. In a world where most statements from most people are true, assuming truth produces correct judgments most of the time. That also explains the “veracity effect“ in lie-detection studies: people are good at identifying truths and poor at identifying lies, because their default matches the usual reality.

The weakness is just as clear. The truth default works only when liars are rare. A professional scammer is precisely the rare, prolific liar the system isn’t designed to catch.

The Psychology: why belief comes first

Believing is automatic; doubting takes work. Psychologist Daniel Gilbert argued that the mind accepts a statement in the act of understanding it, and only afterward, with effort, tags it as false. In his experiments, people under distraction or time pressure remembered false statements as true, because they lacked the mental resources to complete the “unbelieving” step. Fatigue, stress, grief, and loneliness all drain those resources, which is why scammers press for quick decisions.

Several well-documented biases reinforce the default:

• The illusory truth effect: repeated statements feel more true, whether or not they are. Daily messages repeating the same story build belief through familiarity alone.
• Processing fluency: information that’s easy to absorb, such as a polished, coherent story, feels more credible.
• Confirmation bias: once a person trusts someone, they notice evidence supporting that trust and discount evidence against it.
• Commitment and consistency: after investing time, emotion, or money, people strongly prefer to stay consistent with their earlier choices. This connects to the sunk-cost fallacy, where past investment becomes a reason to continue.
• Cognitive dissonance: admitting deception means admitting a painful error, so the mind finds explanations that preserve belief.
• The halo effect and liking: an attractive, warm, or admirable person is judged honest across the board.
• Authority and social proof: apparent credentials, official logos, and testimonials from “others” lend credibility without being verified, which is the fallacy of appeal to authority and appeal to popularity in action.

The Neuroscience: trust as the brain’s resting state

The brain research is younger and less settled than the behavioral research, but several findings fit the truth-default picture:

• Doubt depends on the prefrontal cortex. In a 2012 study by Asp and colleagues, patients with damage to the ventromedial prefrontal cortex were more credulous toward misleading advertising than other patients. The researchers proposed that this region supplies the “false tag” that marks information as doubtful. When it’s impaired, or simply overloaded by stress and fatigue, belief stands unchallenged.
• Social connection is rewarding. Attention, affection, and belonging engage the brain’s reward circuits, the same dopamine pathways involved in other rewards. A scammer’s constant attention becomes something the brain seeks out.
• Love dampens critical judgment. Brain imaging of people in love, by Bartels and Zeki, showed heightened activity in reward regions alongside reduced activity in areas associated with critical social assessment.
• Rejection hurts like pain. Social exclusion activates brain regions overlapping with physical pain processing, according to research by Eisenberger and colleagues. That makes doubting a loved one, and risking the relationship, genuinely painful.
• Oxytocin and trust: a 2005 study found that oxytocin increased trust in an economic game. Later replication attempts have been mixed, so this one belongs in the “promising but unproven” category.

The Anthropology: why humans evolved to trust

Language itself depends on the truth default. If listeners doubted every statement, communication would carry no value, and the cooperation that let humans survive in small foraging bands would collapse. Hunting, sharing food, raising children, and warning each other of danger all required taking each other’s word.

Cognitive scientist Dan Sperber and colleagues describe the other half of the system as epistemic vigilance: humans trust by default but stay alert to signs that a source is unreliable, checking a statement against what they already know and the speaker’s reputation. In a band of a few dozen people, that vigilance worked well. Everyone had a known history, gossip spread warnings quickly, and a liar faced exclusion. The default could be generous because the community supplied the checks.

The modern world keeps the default but strips away the checks. An online stranger has no history, no community watching, and no reputation to lose. And notice what scammers do with the trigger list: they offer a consistent, rehearsed story; they hide their motive behind love or friendship or greed; they isolate victims from the friends and family who would provide third-party warnings; and they manufacture urgency so the effortful work of doubting never happens. They don’t defeat the truth default by accident. They engineer around every one of its exits.

Why this matters for victims

The truth default isn’t gullibility. It’s a core feature of the human mind that makes language, cooperation, and relationships possible, and it serves people well in nearly every interaction they’ll ever have. Scam victims weren’t missing a faculty that others possess. They were using the same one everyone uses, against someone who had studied how to keep it switched on.

Detecting Deception Online

If people barely beat chance when they watch a liar face to face, online deception sounds hopeless. And it mostly is.

The person on the other end is a profile, a voice, or a stream of text messages. There is no handshake, no room to share, and often no way to confirm that the face in the photographs belongs to the person typing. Yet the research on lie detection points to a useful conclusion: the most reliable way to catch a liar never depended on reading faces in the first place. Researchers who study deception have largely moved away from watching behavior and toward testing information. That shift works just as well online, and in some ways better.

The first principle is to stop judging the person and start checking the facts

Psychologists Aldert Vrij and Galit Nahari developed what they call the Verifiability Approach, built on a simple observation: truth-tellers offer details that others can check, such as names, places, dates, and people who were present, while liars avoid checkable details because each one creates a risk of exposure. Online, this translates directly into practice. A real person living a real life produces a trail of verifiable facts. A deceiver produces a story that stays vague exactly where verification would be possible: an employer that cannot be called, a base that cannot be named, a family that cannot be contacted, and a location that always shifts out of reach.

The second principle is to verify through channels the other person does not control

Deceivers manufacture their own evidence, including photographs, documents, bank screenshots, and even the “friends” who vouch for them. Evidence supplied by the person being questioned proves nothing about that person. Independent checking means running a reverse-image search on profile photographs, looking up a company or agency through its official website rather than a link provided in conversation, calling a phone number found independently, and confirming a claimed job with the actual employer. A legitimate person survives this process without difficulty. A false identity rarely does, because the stolen photographs belong to someone else and the claimed life exists only in the conversation.

The third principle is to watch the story over time rather than the person in the moment

Investigators who use the Strategic Use of Evidence technique, developed by Maria Hartwig and Pär Anders Granhag, hold back what they know and let the subject commit to an account before revealing contradictions. Online relationships create a natural record for the same purpose. Messages, emails, and chats preserve every claim. A person reviewing that record later often finds details that shifted: an age that changed, a child’s name that varied, a deployment that moved from one country to another, a story about a dead spouse told two different ways. Liars carry a heavy memory burden, and the written record exposes it. Rereading the conversation from the beginning, with fresh eyes or with a trusted friend, reveals inconsistencies that felt invisible in the flow of daily contact.

The fourth principle is to ask for what a real person provides easily and a deceiver cannot

An unplanned live video call, at a time the deceived person chooses, remains one of the simplest tests, although artificial intelligence now produces convincing fake video, so a call alone no longer settles the question. A request to meet in person in a public place, a request to speak with a family member reached independently, or a question that requires local knowledge the claimed identity would have all test the story in ways a script cannot easily absorb. The response matters as much as the answer. Genuine people treat reasonable verification as normal. Deceivers respond with excuses, hurt feelings, anger, or urgency designed to make the question itself seem unkind.

The fifth principle is to recognize the structure of a scam rather than hunting for a single lie

Individual statements are hard to judge, but patterns are not. Fraud follows a recognizable shape: rapid intimacy or unusual friendship, a reason meeting stays impossible, a growing emphasis on secrecy, an introduction to money through a crisis or an opportunity, payment methods that are hard to trace, and pressure that rises whenever doubt appears. Any one of these elements has an innocent explanation. Together, they form the signature of organized deception. This is why structured tools such as the SCARS Institute’s IS IT A SCAM? diagnostics focus on combinations of behavior rather than on reading a person’s sincerity.

The sixth principle is to bring other people into the judgment

The deceived person is the worst-placed observer in any scam, because the scammer has spent weeks or months shaping that person’s trust and emotions. Friends, family members, bank staff, and support organizations stand outside the relationship and see the pattern clearly. This mirrors how small human communities have always handled deception: through shared knowledge and collective judgment rather than individual intuition. Scammers know this, which is why they work so hard to isolate their targets and encourage secrecy. Sharing the situation with someone trustworthy restores the protection that isolation removed.

None of these methods asks a person to become a human lie detector. They replace an unreliable skill, reading another person’s honesty, with reliable habits: checking facts independently, reviewing the record, testing claims, recognizing patterns, and seeking outside perspective. Online deception succeeds when people rely on feelings of trust that criminals manufacture. It fails when people rely on evidence the criminals do not control.

80% is Not Realistic

No technique that relies on human judgment reliably reaches 80% accuracy in the research. The methods that do exceed 80% either take the judgment out of human hands (machine detection) or replace detection with thorough verification.

Here’s what the research shows for each of your examples and for lie detection generally.

Human judgment, even with the best techniques, tops out around 70%

The strongest interviewing method, the cognitive approach (unexpected questions, asking for more detail, adding mental load), reached 71% overall accuracy versus 56% for a standard approach. That’s in live interviews, which don’t apply to most online contact. Even that figure is disputed: another meta-analysis looking only at human observers found 60% average accuracy, or about 55% after correcting for bias.

Fake profiles: humans are at chance, and worse on fakes

A UK security research center tested 924 people across five studies. Participants judged real profiles correctly 79 to 86% of the time but fake profiles only 13 to 54% of the time, and no participant judged every profile correctly, with an average of 50% when shown a mix. That held whether people had limited or unlimited time. AI-generated profiles are harder still: in one study, people correctly identified generated profiles only 18.2% of the time on average, with overall accuracy of 48.9%, close to a random guess.

Machines do far better on the same task. A system built specifically to detect romance-scam dating profiles performed at 97% accuracy on held-out data. Most published fake-account detectors report accuracy above 90%, though those lab figures usually drop somewhat in real-world use.

If that is true, which it is, then why haven’t the social media and dating industry used these automated detection methods? Profit. Removing a significant number of profiles reduces their shareholder value and their ability to charge more to advertisers.

Impersonation by email (banking, government, employers): better, but still not safe

Phishing is the best-studied form of online impersonation. In one recent study, people’s overall accuracy was 82.1%, but they caught only 74.7% of phishing emails, missing about one in four. Other studies find people miss up to 47% of phishing emails even in a controlled lab. Training helps but doesn’t close the gap: after training that cut victimization by 40%, participants still fell for 28% of phishing emails. These are also lab conditions where half the emails are fake, and people know they’re being tested, so real-world performance is vastly lower.

Fake ads and phone impersonation

We found no reliable studies measuring how accurately people spot fraudulent ads or impersonated police, bank, or government callers. The research there focuses on scam prevalence and losses rather than detection rates, so we can’t give a defensible percentage. However, in both cases, the spontaneity and urgency factor would indicate a very low detection rate.

What actually works above 80%: verification, not detection.

The methods with the highest real success rates don’t ask anyone to judge honesty at all:

• Hanging up and calling back on a number found independently (the back of a bank card, an official website) defeats impersonation of banks, police, and agencies almost entirely, because the scammer controls only the number they gave you.
• Reverse image searches expose stolen profile photos, which is a fact check rather than a judgment.
• Platform and email filters, which some studies show catching up to 99% of phishing emails, stop most fraud before a person ever sees it.

These aren’t usually reported as “accuracy” percentages, because they work by bypassing judgment rather than improving it.

Review

No human lie-detection technique reliably exceeds about 70% accuracy, and people perform at roughly chance or worse when judging fake online profiles. The protections that consistently work, independent verification and automated screening, succeed precisely because they don’t depend on a person’s ability to sense deception. That supports your earlier point: victims aren’t failing at something others do well, because no one does it well.

Conclusion

Human beings depend on trust to communicate, cooperate, and maintain relationships. That dependence also creates an opening for criminals who deliberately manufacture credibility. A convincing story, steady eye contact, apparent affection, or professional presentation does not establish honesty. Confidence in judging another person’s sincerity offers little protection when the judgment rests on signals that do not reliably distinguish truth from deception.

For scam victims, understanding these limits changes the meaning of what happened. Believing a carefully constructed lie does not establish stupidity or defective character. Criminals exploit ordinary processes of attention, interpretation, familiarity, and attachment. They control the information presented, reinforce their claims through repetition, and discourage access to independent perspectives. Responsibility belongs to the people who organize and carry out that exploitation.

Effective protection requires a shift from evaluating appearances to examining evidence. Independently obtained contact information, confirmed organizational affiliations, documented inconsistencies, and outside review provide a stronger foundation for decisions than reassurance supplied by the person making the claim. Several warning signs deserve examination together rather than dismissal as unrelated misunderstandings. Automated screening adds another layer of protection, but its reported performance requires attention to the conditions under which it was tested.

Recovery does not require abandoning trust or treating every relationship as a threat. It requires separating the experience of feeling trust from the evidence supporting a particular claim. Scam survivors deserve practical safeguards, informed assistance, and freedom from blame. Verification strengthens those safeguards by placing decisions on information that a criminal does not control.