Seven Body Language 'Rules' TikTok Got Wrong. Here Are the Studies That Killed Them.

Open TikTok. Search "body language dating." Inside three swipes you will be told that 93% of human communication is nonverbal, that crossed arms mean you have lost them, that good eye contact reveals an honest partner, that power posing in the bathroom before a date will rewire your testosterone, and that you can spot a liar by a flicker across their face. Every one of these claims is either misquoted from a tiny lab study, scientifically retracted by the original authors, or never replicated outside a single famous paper. The TikTok dating-coach economy is built on debunked science. The studies that did the debunking are sitting in peer-reviewed journals nobody is opening. This article opens seven of them.

This matters in dating because the wrong frame produces the wrong move. If you believe crossed arms mean rejection, you will misread a partner who is cold or self-soothing as a partner who is closing off, and you will pursue or retreat for the wrong reason. If you believe sustained eye contact equals honesty, you will trust a trained liar over a shy honest person. If you believe power posing fixes anxiety, you will skip the actual emotional work and keep wondering why dates still go badly. Bad body language science is not harmless. It is the operating system most of your dating decisions are running on. Time to patch it.

Albert Mehrabian, UCLA psychologist, ran two small studies in the late 1960s. He used 30 women rating the emotional tone of single words like "thanks" and "maybe" when said in matched and mismatched vocal tones. He found that when there was a CONFLICT between the literal word and the tone of voice and the facial expression, listeners weighted the channels at roughly 7% verbal, 38% vocal, 55% facial. That is where the 7-38-55 rule came from. The lab condition that produced it: 30 women, single words, deliberately mismatched signals. The takeaway most often quoted: 93% of all human communication is nonverbal. Those are not the same finding. They are not in the same time zone.

Mehrabian himself has spent decades publicly correcting the misuse. He has gone on record saying the popular interpretation is "absurd" and that the 7-38-55 figures apply ONLY when verbal and nonverbal channels are inconsistent. They tell you which channel listeners weight when those channels disagree, not the global breakdown of communication. If your date is telling you "I had a great time" while their feet are pointed at the exit, the 7-38-55 finding helps you read the conflict. If they are just telling you a story about their dog, 93% of the meaning is in the words. Stop quoting the number. The man who discovered it has been begging you to stop for 55 years.

“

It is absurd to imply or suggest that the verbal portion of all communication constitutes only 7% of the message. Please note that this and other equations regarding relative importance of verbal and nonverbal messages were derived from experiments dealing with communications of feelings and attitudes. Unless a communicator is talking about their feelings or attitudes, these equations are not applicable.

— Albert Mehrabian, the researcher who produced the 7-38-55 figure, in his own words

Amy Cuddy's 2010 TED Talk on power posing (stand in a Wonder Woman stance for two minutes, raise testosterone, lower cortisol, dominate the job interview or first date) has 75 million views. It is one of the most-watched TED Talks ever. It is also, scientifically, in serious trouble. The original 2010 paper had a tiny sample (42 participants) and reported hormonal changes that have failed to replicate every time anyone has tried. In 2016, Dana Carney, the paper's co-author, posted a public statement on her academic website titled "My Position on Power Poses," in which she said she no longer believed the effects were real, would not teach the material, and would not advise anyone to use it.

An 11-study replication project and a separate 24-study meta-analysis both found no reliable behavioral or hormonal effects from power posing. The only effect that did survive: a small bump in self-reported "feeling powerful," which is the same effect you get from any brief positive ritual including taking a deep breath. The actionable update for dating: doing a Wonder Woman pose in the bathroom before your date will not change your testosterone. It will, at best, give you a tiny mood lift identical to drinking a glass of water and reminding yourself you are a person. Spend the two minutes on something with real evidence, like a slow exhale or eye contact with yourself in the mirror, and skip the choreography.

The most-repeated body language "rule" in dating content. The science behind it is roughly zero. Joe Navarro, the former FBI behavioral analyst who is probably the most credentialed working nonverbal expert alive, has written extensively that crossed arms are a self-soothing posture (the same family of behaviors as crossing legs, hugging a pillow, or holding your own elbows). They appear when people are cold, when they are thinking deeply, when they are physically tired, when they feel chilly air conditioning, and yes, sometimes when they are defensive. The signal does not isolate the reason.

There is a study from the University of Michigan finding that subjects stuck with difficult cognitive tasks 30% longer when their arms were crossed. The posture is associated with perseverance, not closure. In a romantic context: if your date crosses their arms when the conversation gets emotional, that is more likely a self-comfort move than a rejection. If they cross their arms AND lean back AND angle their torso away AND check their phone, that is a cluster, and the cluster is the reading. One arm cross by itself is not a verdict. It is barely punctuation.

This is the worst myth in the catalogue because the truth is the exact reverse. Multiple deception-detection studies (going back to the 1980s and replicated repeatedly) find that trained liars produce MORE deliberate eye contact than truth-tellers, precisely because liars know that the cultural script equates eye contact with honesty. The result: the people most likely to lie to you on a first date are also the people most likely to give you the steady, locked-on gaze that you have been trained to read as trustworthy. The polite, shy, autistic, or anxious person who breaks eye contact when telling you something true is the one the rule punishes.

Paul Ekman built a career on microexpressions, brief involuntary facial flickers that allegedly reveal hidden emotion. The TV show Lie to Me popularized the technique. The FBI, TSA, and other agencies adopted training based on it. Then the meta-analyses came in. Across the deception-detection literature, the baseline lie-detection accuracy of untrained humans sits at roughly 54%, barely above chance. With microexpression training, accuracy moves to roughly 30 to 40% in many studies (sometimes WORSE than baseline, because trained observers gain false confidence). The U.S. National Academy of Sciences reviewed the field and concluded there is no reliable evidence that microexpression analysis works as a standalone lie-detection tool.

Translated for dating: you cannot tell if your date is lying about their job, their relationship status, their intentions, or their feelings by watching their face. Nobody can. Not trained interrogators. Not customs agents. Not the woman on TikTok with 4.2 million followers telling you to watch for "duping delight." The science of detection is bad enough that asking the question directly and watching for the pattern of their answers over time will outperform any face-reading exercise. Your gut is not psychic. Their face is not a polygraph. Stop trying.

Eckhard Hess's 1965 pupillometry studies are where this idea was born. Hess found that men rated drawings of women as more attractive when the drawings featured larger pupils, and that pupils tended to dilate when people viewed images they found interesting. The popular dating frame became "if their pupils dilate when they look at you, they are into you." The actual science is much messier.

A 2021 replication project failed to reproduce Hess's gender and mimicry claims. Pupil dilation does occur in response to interest, but it also dilates in response to caffeine, low light, fear, surprise, cognitive load, drug use, and roughly two dozen other inputs. The cue exists. It is weak, context-dependent, and almost impossible to isolate in a bar with dim lighting and four espresso martinis on the table. You cannot trust your read of someone's pupils across a candle-lit dinner. The lighting alone invalidates the signal. Use the rest of the face and the body. Leave the pupils to the optometrists.

The classic teaching: a genuine smile (Duchenne smile) engages the orbicularis oculi muscle around the eyes (crow's feet). A fake smile uses only the zygomatic muscles around the mouth. Therefore, look for the eye crinkle and you can sort sincere from performative. Sarah Gunnery's research at Northeastern, published in the British Journal of Psychology, found that 71% of people can voluntarily produce a convincing Duchenne smile when asked to. The eye crinkle is not the involuntary tell most coaches claim it to be. It can be performed.

The replication that DID hold up is more interesting. A 2024 PNAS paper (Aligning smiles of dating dyads causally increases attraction) ran a controlled experiment in which paired daters were instructed to synchronize the onset of their smiles. The matched-onset condition increased mutual attraction ratings significantly. The takeaway: it is not whether the smile is real, it is whether the timing is aligned. Two smiles arriving in the same half-second create attraction. Two smiles separated by three seconds register as awkward. Stop watching for eye crinkles. Start watching for who smiles at the same moment you do.

After the demolitions, here is the smaller, sharper list of nonverbal cues that have survived replication and meta-analysis. These are the ones to actually pay attention to.

If you cut every disproven myth from the field, what survives is a small handful of cues with strong replication records. Use these. Forget the rest.

That is the entire scientifically defensible list. Three cues. Eye contact, synchrony, voice. Everything else in the TikTok dating-coach economy is either weak, retracted, or never proven in the first place. The size of the actual evidence base is humbling. The size of the lies built on top of it is bigger than the actual science by several orders of magnitude. Knowing the difference is the entire game.

done reading. your situation is next.

The cues that actually replicate (mutual gaze, synchrony, voice) are hard to read in real time and impossible to read after the date is over. The digital equivalents (response cadence, message-length synchrony, initiation balance) leave receipts. Upload a chat. Delulu Check decodes the texting body language using the cues that survived peer review, not the ones TikTok recycled.

GET THE REAL READ

or pocket it.