So Daniel sent us this one — he's been thinking about all those episodes we've done on deception and wants to zero in on micro expressions specifically. How brief are they really? Are they actually beneath conscious control? And the practical question: if you're a tenant being shown an apartment by a realtor who just wants the commission, or you're sitting across from a job interviewer who knows the place is toxic but insists it's great — can you actually use micro expression detection to protect yourself? What's the evidence?
The first thing to know is that almost everything pop culture says about micro expressions is either wrong or wildly overstated. "Lie to Me" was a fun show, but it did real damage to public understanding.
Tim Roth squinting at someone for two seconds and declaring them a liar. Very satisfying television.
And about as scientifically accurate as my brother's claim that sloths invented pizza.
That is documented.
It's not. So let's start with what a micro expression actually is. Paul Ekman's original construct — and he's the foundational researcher here — defines them as full-face emotional expressions that last between one twenty-fifth and one-fifth of a second. That's forty to two hundred milliseconds. They happen when someone feels an emotion and tries to suppress or conceal it, and the expression "leaks" before they can stop it.
Forty milliseconds. That's faster than a blink.
A blink is about a hundred to a hundred fifty milliseconds, so yes — some micro expressions are faster than a blink. And this is where the distinction matters. Ekman distinguishes three categories. You've got macro expressions, which are normal-duration emotional displays — half a second to four seconds, the kind you see in everyday conversation. You've got subtle expressions, which are partial-face — maybe just the eyebrows or just the mouth — but at normal duration. And then micro expressions, which are full-face but vanishingly brief. Most coverage blurs all three together, and that blurring is where the myths start.
So the forty-millisecond thing — where does that number actually come from?
Ekman's original work in the nineteen sixties and seventies. He was studying psychiatric patients — specifically, he filmed interviews with thirty-five inpatients and then analyzed the footage frame by frame at twenty-four frames per second. When patients were lying about their emotional state, he found micro expressions in a hundred percent of them. That's the study everyone cites. But here's the problem: that finding has not replicated.
What do you mean, not replicated?
Porter and ten Brinke did a major study in two thousand eight with a hundred and sixteen participants. They had people lie about emotional experiences and then analyzed the video. They found micro expressions in only twenty-one point seven percent of deceptive statements. That means nearly seventy-eight percent of lies contained no micro expression at all.
So the foundational claim — "liars leak micro expressions" — is true less than a quarter of the time.
In that study, yes. And this is the single most important thing to understand about this field. The mechanism is real. The phenomenon exists. But it's not a reliable truth detector, and anyone who tells you otherwise is selling something.
Let's talk about that mechanism, because that's where it gets genuinely interesting. What's actually happening in the brain when a micro expression fires?
So imagine a duel between two brain systems. On one side you've got the limbic system — specifically the amygdala, which processes emotional stimuli in about a hundred milliseconds. Something happens, you feel an emotional response, and your face starts to react. On the other side you've got the neocortex — the thinking brain — which tries to suppress that expression because the situation calls for a poker face. The micro expression is what leaks through in the gap between the amygdala firing and the neocortex clamping down.
So it's literally a race between emotion and inhibition.
And that's why the timing is so specific — forty to two hundred milliseconds. That's the window where the emotional response has been triggered but the suppression hasn't fully engaged yet. It's a neurological gap.
And that gap is what makes the "involuntary" claim plausible. If it's happening before your conscious brain can intervene, you can't really be blamed for it.
Right, but — and this is where the nuance matters — "largely involuntary" is not the same as "completely involuntary." Trained individuals can suppress them. Actors learn to control their faces. People with certain personality profiles — what researchers call high self-monitors — show fewer leaks. Matsumoto and his colleagues demonstrated in two thousand that while there's cultural universality in how these expressions appear — meaning the same emotions produce the same facial configurations across cultures — there's significant individual variation in how much people leak.
So if you're dealing with a practiced liar, someone who's had time to rehearse, you might see nothing at all.
Or you might see manufactured expressions designed to mislead you. The absence of a micro expression doesn't mean truthfulness any more than the presence of one means deception.
Which brings us to the accuracy question. If I'm sitting across from a realtor and I think I caught something — a flicker of contempt when they said "solid foundation" — how likely am I to be right?
The meta-analyses here are sobering. Bond and DePaulo did the definitive one in two thousand six. Across hundreds of studies, untrained humans detect deception at about fifty-four percent accuracy. That's barely above chance — a coin flip is fifty percent. And here's the kicker: training doesn't help as much as you'd think.
Wait, really? The whole premise of micro expression training is that it improves your accuracy.
It improves your confidence. That's not the same thing. Ekman developed something called METT — the Micro Expression Training Tool — and in lab settings, trained participants can reach seventy to eighty percent accuracy. But that's in controlled conditions with high-resolution video, repeated viewings, and known ground truth about who's lying. In the real world, accuracy drops significantly.
There was a study about this, wasn't there? Where telling people to look for micro expressions actually made them worse?
The basketball study. Porter and colleagues, two thousand nine. They showed participants a video of someone lying and told some of them to watch for micro expressions. The group that was specifically instructed to look for micro expressions performed worse than the group given no instruction at all. They were more confident in their judgments, but less accurate.
That's almost comical.
It's the worst-case scenario for a practical application. You're more certain you're right while being more wrong. Imagine that in a job interview — you decline an offer because you're convinced you spotted disgust when the interviewer described the culture, and you walk away from what was actually a great opportunity.
Or the reverse — you ignore real red flags because no micro expression triggered your alarm.
And this is where we need to talk about the TSA. The SPOT program — Screening Passengers by Observation Techniques — trained agents to spot micro expressions and other behavioral cues to identify potential terrorists. The Government Accountability Office reviewed the program in twenty thirteen. Over a decade, SPOT cost nine hundred million dollars and produced zero confirmed terrorist detections. Nine hundred million dollars, zero results.
That's a staggering failure. And it wasn't some fringe experiment — this was a major federal program.
It was the largest deployment of micro expression training in history, and it failed completely. The GAO report basically said there's no scientific evidence that behavioral indicators can reliably identify threats in an airport setting. The signal-to-noise ratio is just too poor.
So we've got a real neurological phenomenon, we've got a mechanism that makes theoretical sense, but the practical detection rate is terrible and training makes people overconfident. Where does that leave Daniel's question about the realtor and the job interviewer?
It leaves us with a more nuanced approach than "watch for the micro expression and you'll know." Let me lay out what I think the evidence actually supports. First, micro expressions are a valid signal of concealed emotion — that part is real. When you see a flash of contempt or fear or disgust, something is probably going on emotionally that the person isn't fully expressing. The problem is you don't know what that something is.
That's the key distinction. Concealed emotion versus deception.
Right. A realtor might show fear when saying "the inspection was clean" — not because they're lying about the inspection, but because they're afraid the buyer's financing will fall through and they'll lose the commission. The emotion is real, the micro expression is genuine, but it has nothing to do with the truth of the statement they're making. You're reading an emotion, not a lie.
And a job interviewer might show disgust when describing the company culture — not because they're lying about it being a great place to work, but because they just remembered a disgusting interaction with their boss that morning. Same expression, completely different meaning.
So the first rule is: micro expressions tell you someone is experiencing an emotion they're trying to hide. They don't tell you why.
Let's get concrete. Daniel asked about two specific scenarios — the realtor and the job interviewer. If I'm walking through an apartment and I want to actually use this knowledge, what should I be looking for?
Let's take the real estate case first. You're being shown a property, and the realtor says "the foundation is solid as a rock." The micro expressions to watch for would be contempt — that's a unilateral lip raise, one corner of the mouth pulling up — or fear, which is eyebrows raised and drawn together, upper eyelid raised, mouth slightly stretched. But here's what I'd actually recommend: don't try to spot these in real time.
Because of the basketball study problem.
Because actively hunting for micro expressions makes you worse at detecting deception. Instead, record the conversation if you can — many states allow one-party consent recording — and review it later at reduced speed. That's the only evidence-based way to use this technique. Frame by frame, looking for the forty to two hundred millisecond flashes.
And in a job interview? You can't exactly set up a tripod.
Most interviews are on video now anyway — Zoom, Teams, whatever. That actually helps. You've got a recording by default, you can review it afterward, and you can slow it down. But there's a new problem: compression artifacts, latency, and what researchers are calling "Zoom fatigue" — the flattening of affect that happens when people are staring at a screen instead of a person.
So the medium that gives you the recording also degrades the signal you're trying to capture.
It's a tradeoff. And there's another factor: on video, you're only seeing a small portion of the person's body language. You lose posture, hand movements, leg movements — all the other channels that might corroborate or contradict what you're seeing in the face.
What about the ethical dimension here? If I'm using micro expression detection on a realtor or an interviewer, am I crossing a line?
There's a real question about power dynamics. In a job interview, the employer already has vastly more power than the candidate. If the interviewer is trained in micro expression detection and the candidate isn't, you've created what some researchers call a "mind-reading" power imbalance. The EEOC hasn't ruled on this specifically, but using behavioral analysis techniques in hiring could potentially create disparate impact issues — if the technique is unreliable and leads to biased decisions, you're in dangerous territory.
But Daniel's asking from the other side — the person with less power trying to protect themselves.
And I think that's different. If you're the candidate or the tenant, you're not making decisions that affect someone else's livelihood or housing. You're gathering information to protect yourself. The ethical concern is more about overconfidence — if you convince yourself you've spotted deception and make a decision based on that, you might be wrong.
There's a study I want to bring up — ten Brinke and colleagues in twenty fifteen. They showed participants videos of CEOs lying about company performance. The participants who were trained to spot micro expressions were no better at detecting the lies than untrained participants. But they were significantly more confident in their wrong judgments.
That's the finding that should keep everyone humble. Training increases confidence more than accuracy. Every single time.
So what does actually work? If micro expression spotting in real time is worse than useless, what's the alternative?
I want to propose a practical protocol based on what the evidence actually supports. Three steps. Step one: establish a baseline. Before you get to the high-stakes questions, ask neutral ones — "how long have you been in this business," "what's the neighborhood like on weekends" — and watch how the person expresses themselves normally. What's their resting face? How much do they gesture? What's their blink rate? You're not looking for micro expressions yet; you're just calibrating.
So you know what "normal" looks like for this specific person.
Everyone has a different emotional baseline. Some people are naturally expressive, some are flat. If you don't know what normal looks like, you can't spot deviation from normal.
Step two?
Look for clusters, not single signals. A micro expression by itself is noise. A micro expression followed by a verbal qualifier — "honestly," "to be perfectly frank," "I'm not gonna lie to you" — that's a stronger signal. A micro expression paired with a change in vocal pitch or a pause that wasn't there before — that's a cluster. The more channels that deviate from baseline simultaneously, the more likely something is being concealed.
And step three?
Never confront based on micro expressions. Never. Use them to generate follow-up questions instead. You see a flash of contempt when the realtor says the inspection was clean — don't say "you're lying about the inspection." Ask "what specifically did the inspector say about the plumbing?" or "can I see the written report?" The verbal response to a targeted question is far more diagnostic than the micro expression that prompted it.
So the micro expression is not the verdict. It's the clue that tells you where to dig.
That's the single most important reframing. Micro expressions are a question generator, not a lie detector. Their value is in telling you which topics need more probing, not in telling you what's true.
That's actually more useful in practice. If I'm in a real estate negotiation, I don't need to know for certain that the realtor is lying — I just need to know which claims to verify independently.
And that brings me to the -takeaway here. Micro expressions are real, the mechanism is fascinating, but they are a weak signal in an extremely noisy channel. The most practical skill is not spotting them — it's knowing what to do when you think you've seen one. And what you do is ask better questions.
There's something else I want to address — the baseline deviation point. You mentioned it in the protocol, but I think it deserves more emphasis. Sometimes the most informative signal is not a micro expression but the absence of an expected expression.
Yes. This is under-discussed. If someone normally smiles when talking about their work, and then you ask about company culture and their face goes completely neutral — no micro expression, just flat — that deviation from baseline is potentially more informative than any forty-millisecond flash.
Because you don't need frame-by-frame analysis to notice that someone who was animated suddenly went still.
And this is something you can actually use in real time without recording and reviewing. It's within the bandwidth of normal human perception. You're not trying to catch a one-twenty-fifth-of-a-second flicker — you're noticing that the emotional music stopped.
Let's talk about the arms race dimension of this. As micro expression research becomes more widely known — and it has, "Lie to Me" was a global hit — the people who are motivated to deceive are also learning about it.
This is the unsettling part. The more we publicize micro expression detection, the more skilled deceivers will learn to manufacture them. We're already seeing this in high-stakes contexts. People who know they're being recorded for job interviews will practice their expressions. They'll deliberately inject "sincere" micro expressions at the right moments.
So the signal degrades further as awareness spreads.
It's a classic arms race. And the next frontier is even stranger. With AI-generated video avatars — Sora, Google VLOGGER, all the generative video tools — we're approaching a world where the face you're reading doesn't have a limbic system at all. There's no amygdala, no emotional leakage, because there's no emotion. What do you do with micro expression detection when the face is synthetic?
You'd be scanning for signals from a system that doesn't generate them. It's like using a metal detector on plastic.
And yet the synthetic faces will be designed to produce exactly the expressions we've been trained to look for. The AI will learn to simulate the micro expressions of sincerity, of trustworthiness, of honest concern. It'll be better at leaking on cue than any human ever was.
So the future of deception detection might not be about catching micro expressions — it might be about catching their absence. The uncanny valley problem. A face that's too perfect, too appropriately expressive, with none of the noise and randomness of a real human emotional system.
The best deception detection tool might end up being not a trained eye but a well-structured question that forces the liar to do more cognitive work than the truth-teller. When you ask a specific, unexpected follow-up — "what did the inspector say about the plumbing specifically?" — a truth-teller just retrieves the memory. A liar has to construct a plausible answer while maintaining consistency with everything they've already said. That cognitive load is harder to hide than a micro expression.
And you don't need forty-millisecond perception to ask a good follow-up question.
You just need to pay attention and not be afraid to dig. That's the real skill.
So to pull this together for Daniel's question — micro expressions are real, they're fascinating, the mechanism is a genuine window into how emotion and cognition interact. But as a practical tool in a real estate office or a job interview, they're vastly overrated. The evidence says: record if you can, review later at reduced speed, focus on baseline deviation rather than hunting for specific expressions, and most importantly, use whatever you think you saw to generate questions, not conclusions.
And stay humble about your accuracy. If the research shows that trained professionals in lab conditions top out at seventy to eighty percent — and untrained people are at fifty-four — you are not going to be the exception. You're going to be wrong a lot. The goal is to be wrong less often than you would be without the technique, and to use it in ways where being wrong doesn't cause catastrophic harm.
Which means never making a life decision based on a micro expression alone.
Never. Verify independently, every time. The micro expression is a hint, not a verdict.
And now: Hilbert's daily fun fact.
Hilbert: In the nineteen twenties, a French colonial administrator in Niger claimed to have discovered a Tang dynasty bureaucratic manual buried in a Tuareg chieftain's library, which he said proved the Chinese had established an administrative outpost in West Africa eight centuries before the Europeans arrived. It was later corrected — the manuscript was a nineteenth-century forgery produced by a Cairo antiquities dealer who had been selling "ancient Chinese" documents to gullible colonial officials across North and West Africa for nearly a decade.
A whole decade. That's a good run for a forgery operation.
The colonial administrator probably didn't want to admit he'd been had by a Cairo antiquities dealer. Easier to believe in the Tang dynasty in Niger.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you enjoyed this episode, leave us a review wherever you listen — it helps. We'll be back soon.
