How a Secret Service agent, CIA officer, mentalist and poker pro read people

The Short Version

Four people whose work depends on reading others describe much the same method, and none of it is a trick for spotting a liar at a glance. Evy Poumpouras, a former Secret Service polygraph examiner, says to learn a person's normal behavior first and then watch whether their words, body and actions stay "in harmony" [1]. Oz Pearlman, a mentalist, says the same thing in his own terms: there is no foolproof sign, only departures from someone's usual pattern [3]. Jonna Mendez, the CIA's former Chief of Disguise, watches for people trying too hard to convince, and Daniel Negreanu, a poker professional, builds a profile of each opponent and gets his best information when they are relaxed [2][4]. They disagree on one popular cue: Poumpouras calls the idea that looking away means lying largely a myth, while Mendez says eye contact is often one of the first things to falter [1][2]. All four treat a read as information for a better decision, not as proof.

Table of contents

What They Said

Evy Poumpouras: baseline first, then harmony

Poumpouras was an interrogator in the Secret Service's polygraph unit, trained at the Department of Defense Polygraph Institute and later in forensic psychology [1]. Mel Robbins asks her why body language matters when someone is lying, and her first answer is unexpected: "Look, it doesn't. It doesn't. The research goes back and forth." [1] Many researchers, she says, will tell you body language is useless for detecting deception. "It's true. And it's not true." [1]

What she does trust is narrower. "Here's what I know. People give off cues." [1] The cues mean something only against the person's own normal. She shows it by describing how Robbins greeted her: how she walked, how she hugged, that she came over at once with open arms. "I start to in the few minutes I get to know Mel, I get her baseline." [1] It does not take long: "You can get somebody's baseline in just a couple of minutes." [1]

The baseline is personal, not a rule about what honest people do. Poumpouras says she would stand with her arms crossed all day if she could, because it is comfortable for her, and she avoids it only because she knows the impression it gives [1]. Crossed arms on her would mean nothing [1].

Once you know the normal, you look for whether three things agree: what the person says, what their body shows, and what they give off. Those, she says, "need to be in harmony" [1]. Reading Robbins, she found the words and the body matched, and judged her comfortable, confident and warm [1].

The interrogation room version is a sequence. Start with easy questions like name and date of birth, to see the person at rest. Then ask them to recall a past event. Then ask an emotional question. Watch how the behavior shifts from stage to stage [1].

On eye contact she is direct. The idea that looking away means lying is, in her view, largely a myth [1]. How a person makes eye contact depends on their history: whether they grew up shy, were belittled or hit, or have ADHD or autism [1]. She breaks eye contact herself when she searches her memory [1]. So she separates two uses. When you are reading someone, "we leave them alone, but we pay attention to what they do" [1]. Do not demand that a stressed child look at you; what they do under stress is their baseline [1]. When you are presenting yourself, eye contact does build trust and show respect [1].

She adds posture. Squared up, shoulder to shoulder, signals engagement; angled away suggests disinterest, nerves or a mind elsewhere [1]. Suspects often sat in what examiners called the fleeing position, turned toward the door, because they wanted to leave [1].

From interviews she passes on verbal patterns, and is careful to say they are not backed by research; they are what examiners noticed again and again [1]. Arriving with a Bible, which they called a "divine prop", or swearing on a grandmother's grave, usually signaled trouble, because the truth is typically simple and does not need selling [1]. Others are stalling by repeating the question, possessive wording, where "my car" hints at more attachment than "the car", and, above all, whether the person actually answered what was asked [1].

She warns about charm. A Canadian detective she interviewed said the overtly charming, ingratiating person was the type he watched most closely [1].

Two rules govern what to do with all this. First, never call people out on what you notice; the cues are for gathering information and making smarter decisions, not for showing how clever you are [1]. If something seems off, do not make it about yourself; ask a simple question, such as how their morning is going, and let them show you [1]. Second, do not wait for a confession. In a fraud case she asked the suspect only what he was worried about; he said his family, then offered to get his wife out of town before a search, and the small admissions made the picture [1]. "You don't need the smoking gun from people. They show you." [1]

Her summary puts behavior over speech: "I really don't care what people say. I look at what they do." [1] She once walked out of a meeting with a talent manager who plainly did not care, reasoning that his lack of interest then showed how he would work for her later [1].

Jonna Mendez: the person trying too hard, and a record of your hunches

Mendez spent three decades at the CIA and ran its disguise program [2]. Her account of lying is short. Liars, she says, often fall apart a little, because they know they are lying and try too hard to convince you [2]. She watches for shifting facial expressions, stories that bounce around, nervousness, and people who will not let a point go [2]. And she says direct eye contact is often one of the first things to falter [2].

Her craft was the other side of reading people: controlling what observers read. It shows what she thinks people actually notice. To disguise someone she began by describing them as a watcher would. Of Robbins: long blonde hair, striking glasses, little makeup, not much jewelry, no visible wedding ring, a particular watch. "I would just describe you in such fine detail." [2] "Then, for the disguise, we would change every piece of that that we could." [2] A dark short wig, heavy makeup, no glasses [2]. Teeth could be altered, which changed the gum line and with it the profile [2].

Often the aim was to make someone less visible, so officers were made older: "Nobody looks at old people too much." [2] Men were harder because they rarely change their look, so the team worked on small things: cologne, a gold chain, an extra button undone, a different watch, a shirt the man hated [2]. Each officer got a written sketch of the new character, what he smokes, what he drinks, what he likes, so that he carried the whole person in his mind [2]. Then came the test: lunch in the CIA cafeteria among colleagues and bosses who knew him well. Officers came back saying they had sat next to friends and nobody knew [2].

The lesson for a reader of people is that an impression is built from a handful of details and a manner, and that both can be manufactured [2].

Her stories about behavior under pressure make a related point. At a tightly controlled border, a smuggler with a man and child hidden under his back seat chatted with a guard he knew, and when the guard admired the Billy Joel tape playing, handed it over as a present for the guard's girlfriend and drove through [2]. In a dangerous city, Mendez walked into an alley and met about twenty armed, drunk young men; her training told her she could not turn back, because turning back would signal that something was wrong, so she kept walking and they parted [2]. In both cases safety depended on showing no departure from normal [2].

She also learned which people do not hold up. The agency found that charismatic recruits who needed constant praise did not flourish in real operations, where there was no applause [2].

On intuition, Mendez has a method. She tracks her hunches the way you would follow a newspaper story, to see later whether she was right, and over time that record builds trust in them [2]. Robbins offers a way to tell intuition from anxiety: intuition arrives as a clear statement that pulls you in a direction, while fear sounds like scattered noise and a string of what-ifs [2]. Mendez agrees [2].

Oz Pearlman: there is no foolproof sign

Pearlman is a mentalist, and he begins by saying what he is not. "I am not psychic. I do not have supernatural powers." [3] What he performs are tricks, built on decades of studying how people think [3]. "And I know how to get information from you that you simply don't know how I'm getting." [3]

He shows how small the signals are. With his back turned, Robbins picks a word in her book and counts its letters silently. When she says she is done, he reads the speed: "just by saying K so quickly, it was fast. It was an easy word to count." [3] A long word, or one with double letters, takes longer [3]. From that and from how her eyes moved across the word, he narrows it to four letters beginning with W, and she reveals the word: want [3].

On lying, he says his first instinct has been right often enough to count: "I have seen so many instances where my first instinct was correct. Now, not every time. But enough statistically that it's a statistical advantage." [3] His worry is that people argue themselves out of what they noticed, with their children or on a first date that does not feel right: "There's so much in your instincts that you will talk yourself out of." [3]

He is equally clear that there is no foolproof sign, and he names the eyes as an example of something that is not one [3]. What changes when people lie, in his experience, is their level of detail, their speed and their cadence [3]. Most people add unnecessary detail, like the colleague who calls in sick with a fever of 104 and a long story, when someone who is really ill just says they are not feeling well [3].

The rule is the same as Poumpouras's. Notice someone's normal pattern and watch for departures from it [3]. The direction of the departure depends on the person: someone who always embellishes might give themselves away by being brief [3].

Much of what Pearlman teaches is how to get people to open up in the first place. He learned as a teenage restaurant magician that strangers meet you with a flood of silent questions, so he approached tables at an angle and not head on, set a time limit, and led with what was in it for them [3]. He avoids questions people answer on autopilot and asks ones that open onto a story [3]. His principle is to make the other person feel like the star [3]. He says "We notice patterns when we behave with other people." [3] The patterns he noticed were about what makes people relax and talk [3].

Daniel Negreanu: profiles, comfort and ranges

Negreanu's starting premise is "Everything everyone does at the poker table conveys information." [4] He divides it into conscious sources, such as what someone says, and subconscious ones: posture, eyes, throat, pulse [4].

His way of getting it is to make people comfortable. He contrasts himself with Phil Ivey, whom he considers the greatest player of all time and who projects an aura of watching you that makes people uneasy [4]. Lex Fridman sums it up as Ivey ruling by fear and Negreanu by love, and Negreanu jokes, "I'll get your money, but you're going to enjoy it." [4] Relaxed opponents give off more information [4]. He thinks he is at his own best in the same condition: "I think I'm at my best when I am comfortable like that" and not stiff [4].

His table talk is profiling. A criminal defense lawyer may be comfortable twisting the truth and more willing to bluff; a Sunday school teacher may regard bluffing as dishonest [4]. He listens for a person's outlook on life, because it predicts how they will act under pressure: "Cynicism is easy to play against, because you can convince them that their cards suck." [4] He keeps a written profile of everyone he faces in his phone, and prepares by watching footage of opponents and noting their tendencies [4].

He dismisses the movie version of a tell. The cookie tell in the film Rounders was chosen to be obvious to an audience; real tells at high levels are much subtler, things like timing, posture or reaching for water in the middle of a hand [4].

The discipline he describes is about not jumping to one conclusion. In his early days players tried to put an opponent on a single hand. Modern players think in ranges: every hand the opponent could hold, narrowed by each action they take [4]. A range should only shrink. If you suddenly think someone could have a hand you had excluded, you misjudged earlier [4].

He also describes the limits. Against the very best, talking gives away more than it gets: "if I'm planning against Phil Ivy, I'm not engaging in anything, because he can read through it" [4]. Online there is nothing to read at all: "You don't have the added bonus of looking across table and getting any sense of whether your opponent is strong or weak" [4]. The most common error amateurs make about professionals is assuming they bluff more than they do [4].

And he separates a good read from a good result. In poker, he says, "making the best possible decision doesn't mean you win. Often you lose." [4] The hardest skill is telling bad luck from bad decisions, and it cuts both ways: "One guy's winning and he thinks he's great. He's really not." [4]

Where They Agree, and Where They Don't

They agree on the baseline. Poumpouras and Pearlman state it as the central rule: learn the normal, then watch for change [1][3]. Negreanu does the same over a longer period with notes and footage [4]. Mendez's detailed description of a person before disguising them is a baseline taken from the other side [2].

They agree that comfort produces information. Poumpouras asks easy questions and never calls people out [1]. Negreanu makes opponents enjoy losing to him [4]. Pearlman defuses a stranger's wariness before he does anything else [3]. Mendez's officers worked to make foreign agents comfortable and confident [2].

They agree that over-selling is a sign. Poumpouras: the truth is simple and does not need a Bible or an oath [1]. Mendez: liars try too hard and will not let a point go [2]. Pearlman: the false sick call has too much detail [3].

They agree on distrusting charm. Poumpouras passes on a detective's wariness of the ingratiating person, and Mendez says charismatic people who needed praise did not hold up [1][2]. There is an irony here that the sources leave unremarked: the skills Pearlman teaches, making people feel like the star and drawing information from them without their knowing how, are the charm the other two say to watch [1][2][3]. It follows that a pleasant, attentive manner is not evidence of honesty or of its absence.

They agree that no single cue settles anything. Poumpouras says the research is mixed; Pearlman says there is no foolproof sign; Negreanu says real tells are subtle and that a read is a range [1][3][4].

They disagree about eye contact. Poumpouras calls the link between looking away and lying largely a myth, because gaze depends on temperament, upbringing and conditions such as ADHD and autism [1]. Pearlman lists the eyes as not a reliable sign [3]. Mendez says direct eye contact is often among the first things to falter in a liar [2]. Negreanu counts the eyes among sources of subconscious information [4]. The positions can be partly reconciled through the baseline: a change in someone's usual eye contact may mean something, while the absence of eye contact in a person who never makes much means nothing [1][3]. But Mendez does present it as a general cue, and Poumpouras does not accept it as one [1][2].

They differ on how much to trust body language at all. Poumpouras, the one with formal training in detecting deception, is the most cautious, and labels her own verbal red flags as unsupported by research [1]. Mendez speaks more confidently of signs [2].

They differ on first instincts. Pearlman urges people not to talk themselves out of a first impression [3]. Negreanu's whole method is to hold many possibilities open and to distrust conclusions drawn from results, since luck can make a wrong read look right [4]. Mendez stands between them: trust the hunch, and check it against what happens [2]. Pearlman's own claim is modest when read closely, an advantage over many trials, not certainty in one [3].

They differ on whether to engage. Poumpouras and Pearlman get people talking [1][3]. Negreanu does too, except against someone who reads better than he does, where he goes quiet [4].

Why It Works

Change carries the signal. A behavior means little without knowing whether it is normal for that person. Crossed arms are Poumpouras's resting posture; a brief answer is suspicious only from a habitual embellisher [1][3]. Comparing a person with themselves removes most of the differences between people that make general rules fail [1].

Thinking takes time and effort, and it shows. Pearlman reads how long Robbins takes to count letters [3]. Poumpouras notes that she looks away when searching memory [1]. Negreanu watches timing [4]. Invented accounts and hard decisions take longer or come out differently from recalled ones, which is why her interview sequence moves from easy facts to recall to emotion [1].

People who doubt they will be believed add support. The oath, the prop, the surplus of detail and the refusal to drop a point are all attempts to persuade, where a true statement is usually just stated [1][2][3].

Relaxed people leak more. Negreanu's comfort, Poumpouras's simple questions and Pearlman's openers all lower a person's guard, so that their ordinary behavior appears and departures from it become visible [1][3][4]. Calling someone out does the opposite [1].

Behavior over time is the strongest evidence. Poumpouras's rule of watching what people do, and her point that small admissions build a picture, treat reading as accumulation [1]. Negreanu's shrinking range is the same idea in formal dress: each action removes possibilities [4].

Intuition is pattern recognition that needs scoring. Mendez tracks whether her hunches were right [2]. Robbins describes intuition as pattern recognition that improves by acting on it and learning from both right and wrong calls [3]. Negreanu warns what happens without scoring: results fool people in both directions [4].

Impressions are assembled from few details. Mendez could make a colleague unrecognizable to friends by changing hair, glasses, teeth, a shirt and a manner [2]. That is why a reader should hold first impressions loosely and why Poumpouras says people forget they are half of the exchange and are being read too [1][2].

Where It Applies, and Where It Doesn't

It applies face to face, with a little time. A baseline takes a couple of minutes at least [1]. Negreanu says that online, with no one to look at, the reading disappears and only fundamentals remain [4].

It does not make anyone a lie detector. Poumpouras says the research goes back and forth, and some of her own cues are examiners' lore [1]. Pearlman claims a statistical edge, not accuracy in a given case [3].

It fails on people unlike your assumptions. Shyness, a history of being belittled, ADHD and autism all change eye contact and manner, which is why the comparison must be with the person's own normal [1].

It fails against the trained and the prepared. Mendez's officers passed among friends in disguise and kept their behavior ordinary at a border [2]. Negreanu will not talk to Ivey [4]. A skilled performer can produce comfort and attention on demand [3].

It is for decisions, not accusations. Poumpouras's instruction is never to call people out, and she often responds to a mismatch by quietly making more space, letting calls go to voicemail, without a confrontation [1].

It can damage close relationships if used as judgment. Negreanu says the judging habit that makes him good at poker was poison in his marriage, and that what he had to learn there was acceptance [4].

A good read can still lose. Negreanu's separation of decisions from results applies to any judgment about a person: you can read correctly and be unlucky, or guess wrong and be flattered by the outcome [4].

What To Do With It

1. Take a baseline before judging anything. Spend the first minutes on easy, neutral talk and notice how the person normally stands, speaks and looks at you [1].

2. Watch for change, in either direction. More detail than usual, less than usual, a different speed or cadence, a shift in posture when the subject changes [1][3].

3. Check for harmony. Ask whether the words, the body and what the person actually does agree, and give most weight to what they do [1].

4. Listen for selling. Oaths, props, surplus detail and a point that will not be dropped suggest someone who expects not to be believed [1][2][3].

5. Notice whether the question was answered. Poumpouras puts this above every other verbal cue [1].

6. Do not use eye contact as a test. Leave people's gaze alone when reading them; use your own eye contact to show respect when it is your turn [1].

7. Make people comfortable, and do not announce what you see. Ask simple open questions, stay easy yourself, and keep your observations for your own decisions [1][3][4].

8. Think in ranges. Hold several explanations open and let each new piece of behavior remove some, instead of fixing on one story [4].

9. Score your hunches. Write down what you suspected and check later whether you were right [2].

10. Remember you are being read. Square up, slow down, and consider what you are giving off; and be wary of the person who seems to be working hardest to charm you [1][2].

Sources

1. Former Secret Service Agent Evy Poumpouras on Reading People, Spotting Lies and Building Steady Confidence · The Mel Robbins Podcast
2. Former CIA Chief of Disguise Jonna Mendez on Courage, Reinvention, Reading People and Disappearing in Plain Sight · The Mel Robbins Podcast
3. Mentalist Oz Pearlman on Reading People, Memorable First Impressions and Beating the Fear of Rejection · The Mel Robbins Podcast
4. Daniel Negreanu on Poker Strategy, Game Theory, Resilience and the Greatest of All Time · Lex Fridman Podcast