Undercurrent research13 Apr 202612 min read

Undercurrent · Deep Dive · Psychology · Part I

The
Baseline

The body language industry sells certainty that the science has never supported. Trained professionals detect lies at 54 percent — barely better than a coin flip. Here is what reading people actually means, and why almost everything you have been taught about it is wrong.

PsychologyBehaviourDeceptionIntelligence

54%

Lie detection accuracy

across 24,483 judges (Bond & DePaulo)

$900m

Spent on TSA's SPOT programme

performed no better than chance

20/41

Behavioural indicators

negatively correlated with real threats

1/25s

Microexpression duration

faster than conscious masking

You cannot spot a liar. Neither can your friend who watched a TED talk. Neither can the FBI agent with twenty years on the job. The entire industry that tells you otherwise — the body language experts on YouTube, the airport security training programmes, the corporate workshops — is selling certainty that the science has never supported.

What you can do is read change. And change only has meaning when you know what someone looks like at rest.

The Coin Flip

Trained professionals detect lies at 54 percent accuracy

In 2006, Charles Bond and Bella DePaulo published the most comprehensive study ever conducted on human lie detection. They synthesised 206 separate research documents covering 24,483 individual judges — laypeople, police officers, customs agents, parole officers, judges. The aim was simple: how good are humans, in real time, without technology, at telling truth from lies?

The answer was 54 percent. Across every category of observer, regardless of training or experience, accuracy hovered just above chance. A coin flip would get you 50.

But the headline number hides something more important. When the meta-analysis broke the data down by what was being judged, a pattern emerged that explains almost every failure of intuitive lie detection in modern life.

What was judged	Accuracy	What this means
Truthful statements	61%	Default assumption: people are honest
Deceptive statements	47%	Worse than random chance
Visible cues only	Lower	Body language actively misleads
Audible cues only	Higher	Voice carries more useful signal

Bond & DePaulo (2006), Personality and Social Psychology Review

Humans operate with a built-in truth bias. We assume the people we are talking to are telling the truth, because society could not function if we did not. The cost of this bias: when someone is actually lying, we miss it more than half the time.

The deeper problem is that the moment we are told to look for lies, we start looking for the wrong things.

The Motivation Paradox

Truth-tellers who care about being believed look exactly like liars

When you ask someone what a liar looks like, they describe the same person every time. Anxious. Avoiding eye contact. Fidgety. Hesitant. Sweating. This is the moralistic stereotype Bond and DePaulo identified, and it is the operational vulnerability that makes lie detection close to impossible.

The stereotype assumes that lying causes psychological torment. It does, sometimes — but only for people who feel guilty about lying. Practiced liars, sociopaths, and confident manipulators feel almost nothing, and so display almost nothing. Meanwhile, an innocent person being interrogated about a crime they did not commit experiences acute fear, looks anxious, breaks eye contact, and fidgets.

The autonomic nervous system that produces these responses does not differentiate between the fear of being caught in a lie and the fear of being disbelieved while telling the truth. The physiological output is identical.

"The Othello error occurs when an investigator correctly observes fear in a suspect, then incorrectly concludes that fear is evidence of guilt — when in fact it may be the natural response of an innocent person terrified of being disbelieved."

— Paul Ekman

Othello is the Shakespeare play in which a husband murders his wife after misreading her terror as the guilt of an unfaithful woman. She was innocent. He saw her fear and built a story around it that fit his suspicions. The error is named after him because investigators have been making the same mistake for four centuries.

The $900m Mistake

The TSA spent nearly a billion dollars proving it does not work at scale

In 2003, the United States Transportation Security Administration launched a programme called Screening of Passengers by Observation Techniques. SPOT trained 3,000 specialised Behaviour Detection Officers and deployed them to 176 airports across America. Their job was to observe passengers in security queues and identify likely terrorists from their body language alone — typically within thirty seconds.

The officers worked from a checklist of 94 behavioural indicators. Each indicator carried a point value. Above a threshold score, the passenger was pulled for questioning. The programme cost roughly £150 million per year and ran for a decade.

Beyond Google

In 2013, the US Government Accountability Office reviewed the entire SPOT programme. They examined 400 peer-reviewed studies covering 60 years of behavioural detection research. Their conclusion: human ability to identify deception from behavioural indicators is "the same as or slightly better than chance." Of the 41 indicators on the SPOT checklist they could empirically test, 20 were negatively correlated with real high-risk outcomes — meaning passengers exhibiting those behaviours were statistically less likely to be threats. The TSA was actively profiling innocent travellers experiencing normal airport anxiety.

The programme's own validation study, completed in 2011, was found by the GAO to be methodologically compromised. Officers knew which passengers had been selected by SPOT and which by random sampling, destroying the blind nature of the test. Data collection across airports was uneven. The non-probability sampling did not support the conclusions drawn.

Most damaging of all: referral rates varied enormously between airports. A behaviour treated as suspicious by an officer in one city was routinely ignored by an officer in another. There was no baseline. There was no consistency. There was no science. Just three thousand people watching travellers and making decisions that affected millions of journeys, based on a checklist that did not work.

What Reading People Actually Means

Behaviour only has meaning relative to a baseline

The professional behavioural analysts who do this work for a living — FBI agents, intelligence officers, experienced clinical interviewers — do not look for lies. They look for change. Specifically, they look for sudden shifts in someone's normal pattern of behaviour, triggered by specific topics or questions.

This is the foundational technique called baselining, and it is what separates real behavioural analysis from the body language industry. Before you can spot meaningful behaviour, you need to know what someone looks like at rest. Their normal vocal pitch. Their normal blink rate. Their normal posture. Their normal use of filler words. Their normal level of eye contact.

How professionals baseline a subject

Stage 1 — Rapport phaseBenign questions about weather, biographical data, neutral topics

Stage 2 — ObservationRecord vocal pitch, blink rate, posture, gesture frequency, breathing pattern

Stage 3 — Probe phaseIntroduce relevant questions while continuing to observe the same channels

Stage 4 — Deviation analysisNote clusters of change tied to specific topics — never single isolated cues

A subject who fidgets constantly during the rapport phase is not exhibiting a tell when they fidget during the probe phase. That is their baseline. A subject who maintains steady eye contact for ten minutes and then breaks it the moment a specific name is mentioned has produced a deviation. The deviation does not prove deception. It identifies a topic that produces stress, which the interviewer can then explore with calibrated follow-up questions.

The mechanism is the opposite of how the body language industry sells it. The pop-psychology version says: here is a list of universal tells, learn them, spot them in others. The professional version says: there are no universal tells, only individuals, and you have to learn the individual before you can read them.

The Cluster Rule

One signal means nothing. Three signals together mean something

Even after you have a baseline, single signals are statistically meaningless. A nose scratch could be a histamine response, a dry room, a half-conscious habit. A break in eye contact could be a passing thought, a noise in the hallway, a moment of recall. The body produces noise constantly, and isolated cues sit indistinguishably within it.

Professional analysts apply the cluster rule: behavioural change only becomes diagnostic when multiple distinct signals fire simultaneously or in rapid sequence, all triggered by the same stimulus. A subject who simultaneously crosses their arms, drops their vocal pitch, increases their blink rate, and produces a microexpression of fear at the moment a specific name is mentioned has produced a cluster. That is intelligence. That is a topic worth probing.

Single cue

Subject scratches nose

Could be allergies, dryness, habit, idle gesture. Diagnostic value: zero.

Cluster

Nose scratch, vocal drop, arm cross, blink spike

All triggered by one specific question. Diagnostic value: high. Worth investigating.

Paul Ekman's work on facial expressions is real science. The Facial Action Coding System he developed in 1978 with Wallace Friesen is the definitive anatomical taxonomy of human facial movement, used in psychological research and animation studios alike. Microexpressions — those involuntary flashes of emotion lasting between 1/25th and 1/5th of a second — exist and are measurable.

What microexpressions reveal is concealed emotion. They do not reveal lies. A microexpression of fear means the subject is feeling fear they are trying to hide. It does not tell you why. The innocent suspect terrified of false accusation produces the same microexpression as the guilty suspect terrified of being caught. Ekman has been warning about this misapplication of his own research for decades.

The New Frontier

The future of lie detection is not the body. It is the brain

Aldert Vrij at the University of Portsmouth has spent two decades building a different model of deception detection. His framework discards the search for emotional leakage entirely. Instead, it exploits a structural fact about lying: fabrication is cognitively harder than recall.

Telling the truth is easy. You access episodic memory — the mental tape — and read it back. Telling a lie is hard. You suppress the truth, invent a plausible alternative, monitor it for internal consistency, watch the listener's reaction, and simultaneously control your body to avoid looking nervous. Every one of these tasks competes for the same finite working memory.

Vrij's insight: investigators do not need to wait for a liar to leak. They can deliberately overload the liar's cognitive capacity until the lie collapses on its own.

The reverse-order technique

Liars rehearse their stories chronologically. Truth-tellers have actual memories.

An investigator asks the subject to recount the event from end to beginning, in reverse. A truthful person can mentally walk backwards through a real memory — the room, the smells, the sequence. A fabricated story has no spatial or sensory depth to walk through.

Under reverse-order recall, deceptive subjects show measurable degradation: longer pauses, fewer sensory details, contradictory gestures, and what researchers call "body neglect" — the brain becomes so consumed with managing the verbal task that physical control deteriorates.

Standard interview accuracy

54%

Reverse-order interview accuracy

67%

Source: Vrij et al., University of Portsmouth, multiple replications

The accuracy improvement looks small. From 54 percent to 67 percent is thirteen points. In a field that has been stuck at chance for decades, it is the most meaningful empirical breakthrough in deception detection. It works because it stops asking the impossible question — "is this person lying?" — and starts asking a tractable one — "is this person's story holding up under cognitive pressure?"

The Practitioner Synthesis

Field operatives don't have time for science. They have heuristics

Academic research is slow, controlled, and unconcerned with deployment. Field intelligence work is fast, chaotic, and concerned with nothing else. The gap between the two is filled by practitioner frameworks — synthesised systems that take validated academic concepts and reassemble them into something an operative can use in real time.

The Six-Minute X-Ray, developed by former military intelligence operative Chase Hughes, is the most prominent modern example. It combines Ekman's facial coding, baselining principles drawn from FBI methodology, cognitive load indicators from Vrij's work, and elicitation techniques from the intelligence tradition. The aim: build a working profile of a subject's drives, decision style, and stress patterns within six minutes of conversation.

The framework's components — tracking blink rate spikes, identifying pacifying behaviours, mapping social drives like significance and acceptance, using elicitation rather than direct questioning — all draw from real research. As a structured heuristic for organising observation under time pressure, it is useful. As a validated scientific protocol, it is not. None of these proprietary frameworks have been subjected to the kind of independent peer review that established the original components they synthesise.

The distinction matters. The body language industry sells the synthesis as if it were the science. The science is the baselining, the cluster rule, the cognitive load research, the FACS taxonomy. The synthesis is a practitioner's shortcut — useful in the field, dangerous when mistaken for proof.

What This Actually Means

You can't read minds. You can read change

The honest summary of fifty years of behavioural science: humans cannot reliably detect lies from observation alone. The most decorated FBI agents perform at 54 percent. The TSA's billion-dollar programme performed at chance. The body language experts on YouTube are selling a product their evidence does not support.

What you can do is establish a baseline for the people you interact with regularly — partners, colleagues, clients — and notice when something deviates from it. Not a single twitch. A cluster. Multiple channels of behaviour shifting at the same time, triggered by something specific. That deviation does not tell you what is happening. It tells you something is happening, and that the topic deserves attention.

The professionals who do this for a living spend their careers learning two things. First, that confident certainty about reading people is the surest sign you are about to be wrong. Second, that the right question is never "are they lying?" — it is "what just changed, and what triggered it?"

You cannot detect a lie. You can only detect a change — and you need to know what someone looks like at rest before any of it means anything.

Sources & Further Reading

Bond, C. F., & DePaulo, B. M. (2006). Accuracy of Deception Judgments. Personality and Social Psychology Review, 10(3), 214–234.
US Government Accountability Office (2013). Aviation Security: TSA Should Limit Future Funding for Behavior Detection Activities. GAO-14-159.
Ekman, P. (2009). Telling Lies: Clues to Deceit in the Marketplace, Politics, and Marriage. W. W. Norton.
Ekman, P., & Friesen, W. V. (1978). Facial Action Coding System (FACS). Consulting Psychologists Press.
Vrij, A., Granhag, P. A., & Mann, S. (2011). Outsmarting the Liars: Toward a Cognitive Lie Detection Approach. Current Directions in Psychological Science, 20(1), 28–32.
Vrij, A., Mann, S., Fisher, R. P., et al. (2008). Increasing cognitive load to facilitate lie detection: the benefit of recalling an event in reverse order. Law and Human Behavior, 32(3), 253–265.
Navarro, J. (2008). What Every BODY Is Saying. HarperCollins.
Hughes, C. (2023). Six-Minute X-Ray: Rapid Behavior Profiling.
Barrett, L. F. (2017). How Emotions Are Made. Houghton Mifflin Harcourt.

You’ve looked beneath the surface.

Now follow the connection.

Psychology·13 min read