Non-Verbal Communication curriculum

Researchers define non-verbal communication as messages sent without words, through behavior, appearance and setting. One boundary rule: behavior becomes communication when someone else reads it as a message and gives it meaning. Communication also happens with no intent to interact, as when strangers silently pass on a sidewalk. Exact definitions still vary between research contexts.

Kinesics groups gestures, posture, facial expressions and movements of the head, hands, arms, legs and feet into one family, often called body language. That popular label misleads: critics note these movements lack the vocabulary and grammar of a language, so no rule book gives a single motion one meaning that holds across situations.

Eye behavior is often studied under kinesics, the study of body motion, but it also has its own branch, oculesics. Researchers code gaze direction, mutual gaze (two people looking at each other at once), shifts toward or away, and how much time is spent looking. A review of 29 studies of healthy adults in conversation found gaze consistently helps a speaker hand over the turn.

Haptics covers what touch conveys; one scheme sorts it into welcoming, threatening and persuasive touch. Meaning shifts with where the body is touched: one taxonomy separates social, consent, vulnerable and intimate zones. In a review of healthcare touch, one study concluded that without a relationship and consent, even comforting touch can feel distressing.

Proxemics sorts distance into intimate, personal, social and public ranges, and territory into primary, secondary and public spaces. Distance rarely acts alone: gaze, touch, facial displays, body orientation, lean and posture combine with it. Critics add that spacing preferences shift with relationship, mood and setting, so a fixed personal bubble is a misleading picture.

Paralanguage is the vocal but non-verbal part of speech, and vocalics studies it: pitch, volume, rate, vocal quality and verbal fillers. These cues can carry much information about relationships. In one cross-cultural study, listeners hearing conversations with the words digitally altered still judged positive versus negative relationships at 75 to 90 percent of their full-audio accuracy.

Chronemics covers the communicative side of time, which some scholars divide into biological, personal, physical and cultural time; Bruneau suggested that time may be the basis of all non-verbal communication. A nursing review coded spending time with older adults as a chronemic cue, and researchers also measure how closely two people time their movements together.

This channel covers physical features such as height, body shape, hair and skin color, plus artifacts such as clothing, watches and necklaces, and objects in the setting. These cues need no action: they are on display whenever a person is in view. Settings speak too: a visitor's chair directly across the desk from a supervisor, often set lower, signals more power than an adjacent seat.

In its popular form the rule assigns each channel a set percentage of every message's meaning, and media retellings stretched Mehrabian's findings into 90 percent of all information being non-verbal. A 2019 paper calls such claims an extreme overreading of his classic experiments. Journal articles still restate the split as a rule for reading emotion or decoding messages, so appearing in print is no evidence for it.

In Mehrabian and Ferris, attitude judgments about a stranger saying the single word "maybe" were swayed about 1.5 times more by the stranger's facial expression than by tone of voice. In Mehrabian and Wiener, tone outweighed word content when people rated one stranger's attitude toward another. Both measured judgments of attitude, not the information carried by everyday talk.

Mehrabian and Wiener's 1967 paper was titled Decoding of inconsistent communications: the channels were set against each other. When verbal and non-verbal cues about emotion conflict, several studies report that healthy adults lean on the non-verbal side, called non-verbal dominance; replications report individual scores of 55 to 100 percent. Results stay partly contradictory, and external validity is questioned.

In Mehrabian's 1971 book, the 55-38-7 equation described how attitude, meaning feeling or liking, is communicated. How a message is judged can rest more on its content than on the body language around it, as with a tax demand versus a declaration of love. Stretching the split to instructions or facts goes beyond what it measured.

The equation traces to Mehrabian's 1971 book Silent Messages, where it is stated for communicating attitude, meaning feeling or liking. One later paper describes his claim as applying at least to the communication of affect. His claim has still been questioned on the ecological validity of the tasks and on methodological, statistical and contextual confounds in the experimental designs.

Mehrabian's early findings were treated as sensational and often misreported in the media as proof that 90 percent of information is non-verbal. His quantitative claim that only about 7 percent of meaning sits in the words seems to have drawn the most attention. The split is still repeated: recent health and psychology articles restate it, one with the percentages misassigned.

Definite percentages hide how complex communication is, yet non-verbal context clearly shapes how a message is appraised. In one cross-cultural study, listeners using non-verbal sound alone judged positive versus negative relationships at 75 to 90 percent of their full-audio accuracy, but only 45 to 53 percent for eight relationship types. Later studies built to avoid the original method problems give less extreme estimates.

Gesture research calls this redundancy: the hand carries the same information as the speech and so reinforces it, as when a speaker saying left also gestures left. One model of gesture production even assumes speech dominates and iconic gestures are mostly redundant with it. Not every gesture repeats, though: one study found children aged 5 to 10 paired gestures with non-redundant information more often than adults did.

Complementary gestures supply details the speech lacks: a speaker says here and gestures to the left, or mentions a box and sketches its shape in the air. Listeners combine what they see with what they hear. In a study of 74 Turkish-speaking adults, spatial working memory was linked to better grasp of gesture content, especially when that content was missing from the speech.

Beat gestures are simple up-and-down hand flicks timed to speech, and they can act as emphasis markers. In a study of 100 Spanish speakers, listeners tended to hear stress on whichever syllable a beat coincided with, most strongly when the sound itself was ambiguous. Constant beats may wear thin: in another study they did not help native listeners recall a story and lowered learners' scores.

A review of 29 studies of healthy adults found gaze consistently helps a speaker yield the turn, monitor speech and repair breakdowns, while its role in starting a turn was less consistent. Hands regulate too: an open palm extended toward someone can offer them the floor. Because the review covered adults only, it is unclear whether children show the same gaze patterns.

Emblems are gestures with conventional labels, such as the thumbs-up or the peace sign, that can appear without speech or even replace it. Because their meaning rests on convention, the code has to be shared: the A-OK hand shape signals approval in English-speaking settings but is an obscene gesture in Brazil. Other emblems, such as pointing to the eye, vary less.

A gesture can clash with speech, as when someone mimes stirring while saying break. When verbal and non-verbal emotional cues conflict, studies report a general tendency in healthy adults to favor the non-verbal side, perhaps because it seems harder to fake. The tendency is not universal: in one small study, adults with high-functioning autism relied less on non-verbal cues than matched controls.

A leftward gesture repeats the speech when the speaker says left, but complements it when the speaker says here, since only then does the hand supply the direction. The same motion contradicts speech when it clashes with the words. So a cue's function is read from the speech and context around it, not from the movement alone.

A friend asked about the weather scowls: the scowl could mean the weather is bad, the question is silly, she is preoccupied, or she is still upset about yesterday. Gaze, posture, distance or touch might settle it, or might not. Unlike words, non-verbal behaviors lack a fixed vocabulary, which makes them imprecise, and one 2023 critique notes that checking such an inference takes other evidence.

Speech comes in utterances, but non-verbal cues flow whenever others can see or hear a person, even with no intent to interact, as when strangers wait in line or pass on a sidewalk. People also send and receive these cues at the same moment rather than taking turns. Nursing research lists silences among the non-verbal modalities, so falling quiet is not the same as sending nothing.

Language is propositional: it can assert something that is true or false, such as today being your birthday. A smile with prolonged gaze and close approach asserts nothing; observers read it for what it says about the relationship, and that reading is a guess. One review of relational communication found dominance and affection the most central and recurrent dimensions, with trust and composure among the others.

Watching one's own face, tone or posture is possible but rarely needed, and two partners can trade leaning forward and back without noticing. Body language has often been assumed to reveal more truth than words, and one account of why people favor non-verbal cues in conflicts is that they seem harder to fake. Yet the behavioral ecology view of faces calls the idea that faces leak hidden emotion unverified.

Words are discrete units with fairly stable meanings, combined by grammar. Most non-verbal behavior has neither a vocabulary nor a syntax, and relational messages such as dominance or composure are expressed along continua through face, head, posture and voice. Emblems are the exception: gestures like the thumbs-up carry conventional labels and work much like words.

A scowl alone is ambiguous, and gaze direction, posture, orientation, distance or touch may clarify it, alone or in combination, though sometimes even that falls short. Distance likewise works together with gaze, touch, facial displays, lean and posture. Setting, culture, gender, personality and goals all shift what a pattern means, so context stays pivotal.

Different people show different cues, and groups differ as well: in earlier research, truthful Surinamese participants naturally showed more behaviors linked to deception than Dutch participants did. Comparing someone with their own baseline is meant to correct for such differences. In lie detection, though, the baseline approach is described as controversial and mostly ineffective.

Encoding means producing expressions others can read; decoding means interpreting other people's. People differ widely in expressive accuracy, how accurately others perceive their personality. In a 2024 set of three studies, skill at posing facial expressions on request was moderately to strongly related to receptive abilities such as recognizing emotions, so the two are distinct but not unrelated.

Meta-scientists list threats such as low statistical power, analysis choices made after seeing the data, hypotheses invented after the results, publication bias and little data sharing. Non-verbal research adds an eager market, with popular books, seminars and videos tying body-language reading to love, wealth and fame, and pundits' readings are often oversimplified and judged in hindsight.

A p value depends on both effect size and sample size, so with a very large sample even a small effect turns significant. Psychology students in one study took a result labeled significant to mean a big effect, while the label barely changed their guesses about sample size. A meta-analysis of 138 facial feedback studies shows the gap: its average effect was clearly above zero, yet small and highly variable.

Many more studies are run than published, and positive, novel results are more likely to appear than null results or replications, so the literature can look stronger than the evidence. How much the bias distorts a given field is less well known: one meta-meta-analysis of psychology and medicine found only weak evidence, suggestive of mild bias, in the subsets it could test.

Preregistration can record the design, outcomes and analysis plan in advance, which targets publication bias and switching outcomes after the fact. Many Labs projects run one protocol in dozens of laboratories to estimate effect sizes precisely and see how much they vary. A multi-lab test with 1,894 participants failed to reproduce the pen-in-mouth facial feedback effect, though it departed from the original in four respects.

Classic face sets use exaggerated, posed expressions that new ratings suggest many viewers do not see as genuinely emotional, and forced-choice tasks hand people labels they might not pick freely. Samples are narrow too: across 2021 to 2023 issues of five British Psychological Society journals, participants from Latin America, Eastern Europe, the Middle East and Africa made up 8.7 percent of samples combined.

Simpson's paradox, where a trend in a whole population reverses within its subgroups, is most likely when inferences jump levels, from populations to subgroups or from subgroups to individuals. Lie detection adds a caution about single judgments: in meta-analyses, accuracy from seeing and hearing someone is about 56 percent, near the 50 percent expected from guessing, and individual differences in skill seem minute.

In one survey, police officers and students alike believed gaze aversion and increased movement signal lying, yet liars generally do not show more of these nervous behaviors than truth tellers. Tracing the evidence exposes the gap: a 2013 US government audit found that the 178 sources behind an airport screening checklist came down to three original studies, which backed only some indicators.

One argument holds that unless a replication has more statistical power or better meets the theory's assumptions, it may be no more reliable than the study it failed to repeat. Facial feedback shows a middle path: after a large multi-lab test did not reproduce the pen-in-mouth result, a meta-analysis still found a small, variable effect, and researchers have proposed, with preliminary support, that an audience matters.

The Facial Action Coding System ties each facial action to the muscles behind it. The zygomaticus major pulls the lip corners up in a smile, the corrugator helps lower the brows in a frown, and the frontalis raises them. One muscle can yield separate actions: parts of the orbicularis oculi either raise the cheeks or tighten the lids. Such coding covers visible movement only, not muscle tone, skin color or tears.

Ekman and Friesen built FACS to tell apart every visible, anatomically based facial movement. Each action unit, such as AU12, the lip corner puller, has a number, a short name and its muscular basis, and coders rate its intensity on a five-point scale from trace to maximum. FACS itself contains no emotion descriptors; emotion labels come from separate add-ons.

FACS records movements, while add-ons such as EMFACS and an interpretive database are used to infer emotions from single action units or combinations. A review of 55 consumer-research studies found inconsistencies both in how FACS was applied and in how emotion was inferred from muscle activity, and 53 percent of the studies did not validate the emotions FACS indicated.

Reverse inference means reasoning from an observed face back to the feeling behind it. In one study, professional actors posing photographs for emotion-rich scenarios varied their faces within each emotion category; only fear, happiness and surprise were portrayed with moderate reliability and specificity. A face can recur with a feeling without belonging to that feeling alone.

A 2025 meta-analysis of 37 articles found that faces and visual context, such as body posture or a scene, each had large effects on how the other was labeled. Effects grew when face and context signaled different emotions, and once clarity and congruence were accounted for, faces were no more influential than body postures. Ratings of actors' photos likewise shifted when the scenario was shown.

One 2025 review argues that because posed faces are produced on instruction rather than from felt emotion, studies using them can measure how people respond to a look, not whether they read the poser's inner state. Classic sets also tend to be exaggerated and prototypical, and new ratings suggest many are not perceived as genuinely emotional. Naturally elicited expressions can produce different findings.

In one validation, a commercial tool classified about 80 percent of standardized posed expressions correctly, while its agreement with human action-unit coding was only moderate. Translating action-unit codes into emotions has been called elusive. A recruitment platform that scored video interviews partly from facial expressions drew criticism over transparency and bias, and said in 2021 it would drop facial analysis.

In a classic test, people holding a pen in their teeth, which engages a smiling muscle, rated cartoons as funnier than people holding it in their lips. A multi-lab replication with 1,894 participants found no such effect, yet a meta-analysis of 138 studies found a small, variable one. Another experiment found feedback mainly while the face was moving, mostly dampening ongoing feelings.

Popularized mostly by Paul Ekman and his mentor Silvan Tomkins from the late 1960s, basic emotion theory holds that certain emotions are evolved and distinct, each with universal signals, its own physiology and a dedicated function. Its theorists agree on at least happiness, sadness, fear, anger and disgust. Early support came from cross-cultural face recognition studies; rival theorists dispute the one-face-per-emotion picture.

Before 2008, only three papers, published 1969 to 1971, formally tested emotion perception in small-scale societies, all in Melanesia and Southeast Asia, where people matched posed faces to emotion words or brief stories. By one 2020 count, eleven later studies across six such societies followed: four reported results consistent with universality, three using the matching method, while the rest found substantial variation.

In choice-from-array tasks people pick from a few labels or faces, which limits room for disagreement, while free labeling lets differences show. Respondents can also work by elimination, picking options unused on earlier trials, or settle on the best available label. Exporting Western word lists without prior fieldwork also assumes each English emotion term has an exact local match.

Barrett's theory of constructed emotion starts from the brain's job of regulating the body. It holds that emotions are constructed from past experience, current bodily state and the situation, so an emotion category need not come with a specific face or physiology. One review argues that it explains feeling while basic emotion theory explains emotion, a reading Barrett opposes.

A 2026 review of 21 studies found facial expression production largely universal, with culturally variable nuances. One study of 22 emotions found consistent core action units alongside 104 cultural accents; open-mouthed lip puckers, for example, were more common for positive emotions in India. Cultural differences surfaced mainly in posed expressions, while spontaneous ones looked largely universal.

Asked to name facial expressions, Polish speakers leaned toward mental-state words, while speakers of Maniq, a language of Thailand, named the movements themselves more often, more precisely and with more agreement. In a separate study, mental-state words made up about 70 percent of Hadza participants' labels on average, against 97 percent for US participants. Translating emotion words is not a neutral step.

In one choice task, both US and Hadza participants picked the predicted face more often than chance, and they performed most alike when pleasantness could separate the options. When neither pleasantness nor arousal could, only 58 percent of Hadza participants beat chance, while 94 percent of US participants usually chose the target. A Hadza label used consistently for scowls also fit other faces.

Databases of posed faces are used to train AI systems meant to detect emotions, and US airport officers have been trained to spot body movements thought to predict terrorism. Critics argue such uses inherit the disputed assumption that outward behavior reveals a specific inner state. Ethicists add that automated emotion scoring in hiring may disadvantage people whose expressions differ through disability, neurology or culture.

A Duchenne smile pairs the lip-corner pull of the zygomaticus major with the eye constriction of the orbicularis oculi. Observers rate such smiles as more felt and spontaneous, but people can also produce the eye action on purpose without feeling happy. In one study, weakening that eye muscle with Botox made posed smiles look only subtly less genuine.

Not every smile reports pleasure. One functional account sorts smiles into reward smiles that share enjoyment, affiliative smiles meant to look friendly and polite and to keep bonds, and dominance smiles that assert status. In a study of 190 mostly student viewers, people told actors' versions apart above chance, though affiliative smiles were recognized less well from still photos than from video.

A masking smile is put on to cover a negative feeling. That is hard to do: the smile muscles must be activated while the muscles of the felt emotion are held back, and one hypothesis holds that a strong feeling leaks through as a trace. In one earlier study, observers judged masking smiles not really happy more often than enjoyment smiles, yet named the hidden emotion only 20 to 50 percent of the time.

Display rules are learned norms about which feelings it is proper to show, where and to whom, and they lead people to dampen, intensify, neutralize, mask or fake an expression. In a widely cited but unpublished 1972 study, American and Japanese viewers of distressing films showed no cultural difference when alone, but only the Japanese masked negative expressions with an experimenter present.

Emotional labor is managing feelings and their display to meet a job's expectations. Workers may change only the outward display (surface acting), try to change the feeling itself (deep acting), or show what they naturally feel. Research, including meta-analyses, links surface acting with emotional exhaustion, depressive symptoms and negative affect, while deep acting appears less harmful; these are associations.

People sometimes smile while embarrassed, and smiling without pleasure can serve to conceal a feeling or to appease a partner. In adults the embarrassed smile is often controlled, with lips pressed together, and embarrassment is linked with appeasement signals such as blushing, face touching, nervous laughter and a lowered head. Coy smiles appear in infancy, though their early function is still being studied.

Few people reliably tell a genuine smile from a posed one. Studies put typical accuracy at around 55 to 70 percent, with large individual differences, and the eye-crinkle marker often shows up in deliberately posed smiles too. One study found that directing attention to mouth movement rather than the eyes improved discrimination, but its authors say the result needs further testing.

A micro-expression is a fleeting, usually low-intensity facial movement, often showing only part of a full expression, said to occur when someone conceals a feeling. Ekman's classic window was 1/25 to 1/5 of a second; researchers now generally treat about half a second as the upper limit. The cutoff matters: one team argued that narrow definitions may explain earlier equivocal findings.

The leakage hypothesis says that when people mask or suppress a feeling, the suppression is imperfect, so a brief involuntary expression slips out. The idea grew from flashes of expression seen in slowed-down psychotherapy videotapes. A 2023 review notes micro-expressions were later invoked after the fact to explain null lie-detection results, and the behavioral ecology view it sets out treats leakage as unverified.

In one study, only about 22 percent of participants showed any micro-expression, they made up about 2 percent of all expressions analyzed, and they were no more frequent when people were deceiving. Another study found them in fewer than 20 percent of remorse narratives. Researchers building databases report they are hard to elicit, and reviews note that inferences from them yield both misses and false alarms.

Short computerized courses can raise how well people recognize micro-expressions. In one randomized study of 67 participants, three weekly sessions improved only the trained facet: micro-expression training lifted the micro-expression test but not multimodal recognition or reading patients' emotional cues. Studies cited there found no gain in lie detection, though one reported better job-skill ratings after training.

Seeing a brief flash of fear or anger does not reveal its source: a person under suspicion may be stressed because they fear being accused, not because they are lying. In one study, skill at recognizing micro-expressions was unrelated to accuracy at judging true and false statements, and other work finds emotion judgments are often predicted as well by the situation alone.

Behavior-detection screening applies the idea that cues betray hidden intent. The US airport program SPOT, introduced in 2006, trained about 3,000 officers to watch passengers against a 92-item checklist of behaviors, and one study gave micro-expression training to airport behavior-detection officers. A GAO review concluded that published research did not support using such nonverbal indicators to reliably identify deception.

Brief facial expressions do occur, and one mock-crime study of 77 community volunteers found that short negative-emotion expressions separated liars from truth tellers about a planned act, which its authors offered as first evidence of a possible lie cue. Yet reviews find micro-expressions infrequent, and inferences from them produce both missed lies and false alarms, so they fall well short of a reliable detector.

Where someone looks does several jobs at once. Perhaps most basically, it shows where their attention is directed, and it is also used to regulate interaction, express intimacy and exert social control. Context decides how a look lands: a listener's gaze during a speech can feel comforting, while someone looking at you as you wait for the last bus at midnight can feel very negative.

Eye contact is arousing: one study found partners' skin conductance highest when both were sending and receiving a clear gaze, and implicit measures suggest eye contact triggers a positive reaction. Conversation usually holds it only briefly; museum visitors' preferred length of an actor's direct gaze averaged about 3.3 seconds. Averting gaze is also one way people add psychological distance when someone sits too close.

Looking away has many sources. People tend to avert gaze under heavy cognitive load and while planning what to say; they avoid mutual gaze more when asked very personal questions; and in Japanese culture avoiding eye contact can signal respect or deference. Liars and truth tellers also show similar gaze patterns, so averted eyes by themselves do not say which cause is at work.

Gaze helps people trade turns. Kendon proposed that speakers look away as a turn begins, to plan or hold the floor, and look at the listener as it ends, and an eye-tracking study of pairs playing guessing games found that pattern. A review of 29 studies confirms gaze aids yielding the turn, alongside intonation and gesture, but finds less consistent results for how turns begin.

The visual dominance ratio divides the share of time a person looks at a partner while speaking by the share while listening. People are generally thought to look more while listening, so looking more while talking and less while listening has been hypothesized to signal dominance, and the ratio is a popular leadership measure in automated studies. It is a hypothesis about a pattern, not a reading of one glance.

Pupils change with light but also with a person's cognitive and emotional state, and these changes are hard to control or fake. Observers can be affected without noticing: in an early study, men preferred a photo retouched to enlarge a woman's pupils while unaware of the edit. But dilation is not specific to one emotion and also comes with pain or exercise, so it signals arousal, not its cause.

Neuro-Linguistic Programming (NLP) courses teach that looking up to the right signals an imagined event and up to the left a remembered one, and that the rightward look marks a lie. A 2012 paper tested the lie claim three ways: coding liars' and truth tellers' eye movements, teaching observers the rule, and coding high-profile press conferences. None supported it.

In a lab comparison of Finnish and Japanese students, eye contact with a live person drew similar heart-rate and arousal responses in both groups, yet Japanese participants rated faces making eye contact as angrier, less approachable and more unpleasant than Finnish participants did. The authors attribute such differences to display rules and cultural norms. Two student samples cannot speak for every culture.

Emblems are conventional gestures whose meaning members of a group share, so they can stand in for words without speech, like a thumbs-up for agreement in many Western countries. Because the meaning is learned, it varies by place: the A-OK sign means OK in English-speaking settings but is obscene in Brazil, and Italians and Americans mime a telephone differently.

Illustrators are hand and body movements that accompany speech to depict or stress what is said, such as holding the hands apart to show a fish's size or wagging a finger in disagreement. In Ekman and Friesen's scheme they include what McNeill classifies as iconic, metaphoric, pointing and beat gestures. Unlike emblems, they usually occur with speech rather than replacing it.

Regulators steer a conversation, as when a speaker extends an open palm toward someone to give them the turn. In Duncan's account, turning the head toward the listener or stopping hand gestures signals readiness to yield, while continuing to gesture almost always holds the floor. Listeners regulate too, with head nods while listening that act as visual back-channels.

Adaptors are movements directed at oneself or at objects, such as scratching, rubbing hands, lip biting or brushing back hair; longer repetitive ones are what people call fidgeting. They are thought to carry no deliberate message, though some theories propose they help a speaker manage stress. Observers still read them: animated characters given self-adaptors were judged more neurotic.

Affect displays are movements that convey an emotional state, often through the face or posture, as well as gestures such as tense hand movements. Basic emotions theory reads them as expressions of what the person feels. The behavioral ecology view instead treats them as social tools aimed at the viewer: a scowl may push someone to submit whether or not the displayer is angry.

Because emblems are conventions, the same hand shape can carry opposite meanings: the A-OK circle signals OK in English-speaking settings but is an obscene gesture in Brazil. Even a shared idea takes local forms, as when Italians mime a phone with thumb and little finger out while Americans hold a fist by the cheek. Others travel well, such as pointing to the eye to urge alertness.

Observers expect liars to look nervous, and their deception judgments track gaze aversion and fidgeting. Liars do report feeling more nervous, yet a meta-analysis found they make fewer rather than more movements than truth tellers. Vrij and Fisher point partly to cognitive load: lie tellers face more mental demand in interviews, and heavy demand reduces movement, fidgeting included. The cue observers watch is not the one liars show.

Gesture and speech appear to be planned together: speakers gesture overwhelmingly during fluent speech and rarely during stumbles, and congenitally blind speakers gesture even to blind listeners. McNeill's Growth Point Theory treats the two as one system expressing one idea, while the Sketch Model treats them as separate but tightly coordinated routes from one communicative intention.

Gesture seems to help the person producing it. People gesture more when a word is hard to retrieve, and speakers kept from gesturing about spatial content tend to be less fluent. In one dual-task study, people explaining a math problem remembered more letters when allowed to gesture, read as gesture easing memory load. Some studies disagree, and how gesture would aid word retrieval is unclear.

Gestures help listeners too: a meta-analysis cited in one review found that a speaker's gestures aid comprehension, and preschoolers understand complex spoken messages better when representational gestures accompany them. Children can even pick up information a teacher showed only in gesture. The help is not guaranteed, though: the same review notes times when gestures do not promote learning.

Sometimes a child's hands say what the words do not, as when a child says a spread-out row has more checkers while pointing back and forth to match them one to one. Goldin-Meadow's group reports that children producing such gesture-speech mismatches are more likely to benefit from instruction than children whose gestures match their speech, and that the gestures let teachers see it.

Studies reviewed in 2021 report that children learn, retain and generalize math better when a teacher's gestures carry part of the lesson. In one study of 51 Spanish-speaking first graders shown an English-language lesson, adding gesture raised learning in one group from 0 to 50 percent. The benefit varies: in another study, learning with gesture tracked learners' visuospatial working memory.

Pointing is a deictic gesture: it picks out a referent that both people can see, here and now. Reaching appears early in the first year and pointing around 9 months, and pointing helps establish joint attention on a shared object. In one longitudinal study, about three quarters of toddlers' gestures were deictic, though spoken words still far outnumbered word-like gestures.

Sign languages are fully linguistic, and how sign relates to gesture is debated. Goldin-Meadow argues for a sharp break between gesture and sign, while Kendon emphasizes the common ground they share. Pointing shows one difference: a toddler's point picks out something present, whereas words and signs are flexible enough to refer to abstractions.

Immediacy means behavior that signals closeness and approachability. Nonverbal examples include eye contact, facing someone directly, leaning forward, a relaxed posture and smiling. In classrooms, teacher immediacy is defined as cues that cut physical or psychological distance, and it is associated with better student experiences. These cues combine as a system, so a single one is hard to read alone.

An open posture has arms and legs uncrossed; a closed one has them crossed. Some research describes closing up as a sign of feeling threatened or unwilling to interact, and observers do judge by it: in one online study of 167 people, male physicians were rated better when open, but open female physicians tended to be rated lower on social skills. Without more context, one cue stays imprecise.

Where people point their bodies sorts who belongs to a conversation. In Kendon's F-formation, members stand in a circle, horseshoe, L-shape or side by side around a shared space each can easily reach, their backs screening out the outside world. Facing someone directly is counted among involvement cues, but orientation works together with distance, gaze and lean.

Leaning toward someone is counted among behaviors that show attention and involvement. In one randomized trial with 181 participants, a receptionist robot that leaned forward was rated more stimulating than a neutral one, and people looked at it longer and leaned toward it in return. Lean is one part of a system with distance and gaze, and partners may trade forward and backward lean without noticing.

When two people fall into similar postures and movements, that coordination has been linked with rapport: in one coding study of high school pairs, judges' ratings of coordination were strongly related to the rapport partners reported. Evidence that mimicry creates rapport is mixed: in virtual reality, avatars copying people's head and torso movements raised rapport in one study, but a preregistered replication found no effect.

Taking up space with an open, expansive posture has repeatedly been found to go with holding power, in humans and other animals, while constricted postures go with lacking it. That is a link between status and display. Whether adopting a pose changes the body is a separate causal claim: in one study of 82 men, testosterone and cortisol did not differ between repeated expansive and constrictive postures.

How much someone moves is a poor cue on its own. Burgoon's group describes moderate anxiety bringing more fidgeting, but heavier pressure such as an interrogation possibly producing a rigid lock-down, so one state can show as more or less movement. In one study, overall movement tracked arousal across the group, yet within individuals the link held for only about half of them.

Carney, Cuddy and Yap had people hold expansive or contractive poses for two minutes. They reported that expansive posers showed higher testosterone, lower cortisol, stronger feelings of power and riskier choices. Later, larger studies did not reproduce the hormone and risk-taking results, so the report is best read as the claim that started a long replication debate, not as an established effect.

After the first report, scores of new posture studies followed, the idea drew wide media attention, and Cuddy's TED talk on it went viral. A 2016 paper described its appeal: a few minutes of posing that anyone could adopt, whatever their role or rank. Attention of this kind measures interest in an idea, not the strength of its evidence.

A larger replication with 200 participants found higher self-reported power but no significant effect on hormones or risk taking, and other well-powered studies also failed to reproduce the hormone results. A 2019 study had 82 men adopt the poses repeatedly and still found no hormonal difference. Two meta-analyses that corrected for publication bias likewise found no reliable difference in hormones.

Two large preregistered studies, each with about 1,000 participants, suggest that posing raises self-reported feelings of power by a real but small amount, and a meta-analysis of six preregistered studies agrees. Feeling more powerful on a questionnaire is a far weaker claim than changed hormones. Behavior is disputed: two meta-analyses found a difference, but one of them judged it not robust.

Demand effects occur when participants shift their responses toward what they think a study expects. In a meta-analysis of six preregistered studies, posing raised reported power more in people familiar with power posing, and later authors judge demand a likely partial cause. Demand is not always strong, though: a 2026 study found no evidence of demand effects in three unrelated online experiments.

As the replication dispute grew, Carney, first author of the original study, renounced the power posing theory, while coauthor Cuddy kept supporting it and pointed to a growing literature. The debate around them turned heated, with accusations of bullying and malpractice from both sides. A finding's standing rests on its replication record, not on which of its authors still defends it.

In one experiment, about 2,500 participants rated photos of men and women in high-power or low-power poses; high-power posers were rated more competent and more likely to draw admiration, envy and contempt, with no change in perceived warmth. That is a finding about observers. What a pose does to the poser's own hormones or behavior is the disputed part.

Power pose studies that found effects tended to use small samples, while large studies with preregistered analysis plans mostly found none. Publication bias remains a problem, few studies included a neutral pose for comparison, and a 2024 reanalysis found that a popular check for selective reporting gave unreliable answers. Those are the questions to ask of the next headline about the body changing the mind.

Anthropologist Edward Hall proposed four zones: intimate up to about 45 cm, personal to about 1.2 m, social to about 3.65 m, and public beyond that. He drew them from observational case studies alone, and controlled experiments later found a smaller personal space of about 1 m. One later review argues that his zones described where interaction happens, not a bubble around each person.

People let friends closer than strangers and their children closer than friends, and may want more room at work than at a party or around someone in a bad mood. In a paper-and-pencil task across 42 countries, average preferred distance ranged from 77 to 140 cm, and people accept closer distances briefly in crowds. Spacing is negotiated in the moment.

Discomfort climbs immediately once someone enters a person's comfort zone, and more gently when a partner stands too far away. People may respond by moving back or ending the exchange, or by looking away, falling silent, turning the body or showing anger, which adds psychological distance without changing physical distance. A close approach by a friend or partner may instead be returned with more gaze, smiles and touch.

In Argyle and Dean's equilibrium theory, preferred distance is a compromise between forces of approach and avoidance. A related account treats intimacy as built from distance, eye contact, smiling and topic, so a change in one can be offset by another, such as less eye contact when standing closer. The limit: people sometimes return a close approach instead of offsetting it, as with close friends and partners.

Altman described primary territories as one person's exclusive domain, secondary territories as tied to personal connections but not exclusive, and public territories as open to all. People use design, decor and the placement of objects to signal that a space is theirs, as many mammals do with scent. Altman linked territory to privacy, which privacy regulation theory defines as selective control of access to oneself or one's group.

Researchers separate density, a feature of the physical setting, from crowding, a psychological experience usually measured by self-report. In privacy regulation theory, crowding occurs when the privacy a person wants exceeds the privacy they can get. In housing studies, perceived crowding was more strongly linked to aggression than residential density was, a correlation rather than a cause.

In two studies of imagined seat choice in a busy cafe, people preferred seats farther from strangers in one study, and seats with less eye contact or beside a wall in the other, so the preferences did not replicate cleanly. In one study of grade 3 to 5 classrooms, children seated near each other were more likely to form new friendships, though teachers, not chance, assigned the seats.

In neurocognitive research, the zone within arm's reach is called peripersonal space, sometimes described as a protective buffer that prompts defensive action. In a virtual reality study, comfort distance exceeded reaching distance when people stood still, but the two matched when they walked toward a virtual person or object, and both shrank for virtual women compared with men. The authors conclude the spaces share a motor basis.

Touch travels on fast, myelinated A-beta fibers, which respond to many kinds of touch, and slow C-tactile fibers, which seem tuned to slow stroking near skin temperature and fire in step with how pleasant it feels. The usual account splits them into discriminative and social-affective pathways, but a 2023 study found pleasantness ratings nearly vanished when nerve blocks greatly reduced A-fiber function.

Burgoon reported that most people share broad readings of touch: a handshake suggests a formal relationship, an embrace an intimate one. The closer the relationship, the larger the area of the body where touch feels acceptable. Health care research mostly splits touch into instrumental and expressive kinds, usually treating only expressive touch as communicative, though one review argues any touch can communicate.

Hertenstein and colleagues found that people could recognize emotions such as anger, fear, gratitude, love and sympathy from touch alone, and later work showed strangers and romantic partners could do so through touch on the arm. In one study of 27 pairs of Japanese women with matched chance levels, a spoken syllable conveyed negative emotions better than touch, while touch had only a marginal edge for positive ones.

In two experiments with video clips of two men, observers read touch by default as communal, but saw whoever started it as more dominant, especially when the pair was framed as hierarchical. Returning the touch reduced that imbalance when the touch fit the setting. One health care review describes touch as sitting on a fragile line between reassuring presence and control.

One nursing taxonomy maps the body into zones, from hands, arms and shoulders, which studies find are touched most and usually seen as safe, through face and neck, to intimate areas. One study concluded that when a nurse has not built mutual closeness and consent, even comforting touch can feel deeply distressing. Fisher, Rytting and Heslin proposed that touch helps only when it conveys care and matches the intimacy the receiver prefers.

A common scheme sorts cultures into high-contact and low-contact groups, and most research on the benefits of touch comes from low-contact countries. In a UK and Japan comparison, closer bonds allowed more of the body to be touched in both countries, while UK participants found social touch more pleasant. In another study, women in South Africa rated stroking touch more positively than women in the UK.

Crusco and Wetzel named the Midas touch after servers who briefly touched diners while returning change and received higher tips, and field studies have reported more helping and compliance. A small lab study found people complied more with a robot that touched them. Yet three lab experiments found slow stroking did not raise prosocial choices, and another found no significant effect of touch on trust, altruism or risk taking.

A preregistered meta-analysis of 137 studies found that touch interventions, such as massage, had a medium overall benefit for adults, with larger benefits for pain, depression and anxiety, though few single studies could reliably detect it and blinding is impossible. During COVID-19 restrictions, people lacking intimate touch reported more anxiety and loneliness, a correlation rather than proof of cause.

The pitch movement at the end of an utterance, its boundary tone, helps mark a question versus a statement and a finished versus continuing thought. Listeners also often hear raised pitch and rising endings as less confident. That reading can be biased: with one talker, listeners judged a female voice using rising intonation as less confident than a male-pitched voice using the same cue, a gap that vanished with many talkers.

Speakers asked to sound confident talk measurably louder, and listeners rate them as more confident. In a multinational group game, a louder voice went with being seen as dominant. Loudness also has a practical side: in field experiments that set salespeople at different distances from customers, large distances produced louder, shorter and more tense conversations.

Across many studies, listeners rate faster talkers as more confident, and speakers told to sound confident speed up. Rate is not a general advantage, though. In a multinational group game, speaking tempo was not significantly related to perceived dominance, and slower turns at talk went with being trusted. Uncertainty, too, has been linked to slower and sometimes faster speech.

Listeners respond faster to a word that follows a delay, whether the delay is an um, a silence or even an artificial tone, so fillers need not be deliberate signals to help. Silence itself is ambiguous: it can read as threat and dominance in one setting and as submission in another, and in therapy sessions its meaning was shaped by the turns on either side.

Creaky voice, or vocal fry, is a low, rattling phonation, while hoarseness is a rough, breathy quality. In one study, listeners hearing a hoarse voice reported more listening effort and annoyance, yet their memory scores were mostly unchanged, apart from an early dip in one recall task. An earlier study found students remembered less content from a creaky voice than from a typical one.

Accenting a word makes listeners think of its alternatives: in an eye-tracking study, German listeners who heard a contrastive accent on a subject looked more at a contrasting word, such as diver after swimmer. In Mandarin, one sentence about not seeing even one ant led listeners to different inferences depending on its prosody, though context could override the prosody.

Restarts, self-repairs, repetitions and filled pauses are common ways of producing speech in real time, and researchers argue they need not hinder communication and can even help it. Their rate differs by setting, though not in one direction: a Slovenian corpus found more disfluencies in private speech, with different speakers in each sample, while an earlier study of the same speakers found more in classroom talks.

A pioneering study found trivia statements were rated less true when spoken by non-native speakers, which its authors attributed to harder processing. A 2024 study found the rise in rated truth from repetition was similar for foreign- and native-accented voices, even though a separate group rated the foreign-accented speakers less favorably. Accented speakers are also often rated lower in hiring contexts.

The voice carries emotion in two ways: brief nonverbal sounds such as laughs, and prosody, the tone of speech. In a preregistered study of 22 positive emotions, Dutch listeners recognized 14 above chance from nonverbal sounds but only 10 from prosody. A 2026 study of Chinese and Arab listeners found they identified emotions more accurately and faster from nonverbal sounds than from speech.

Emotion researchers describe feelings by arousal, how activated a state is, and valence, how pleasant or unpleasant it is. One review reports that in most species, opposite valence tends to produce different call types, while higher arousal makes calls louder, faster and higher in frequency. In one study, 333 adults told positive from negative infant sounds almost perfectly, yet identified only three of six specific situations above chance.

When English, Swedish and Russian speakers named 132 real nonverbal vocalizations, the same main types emerged: laugh, cry, scream, moan, and possibly roar and sigh. Each type was tied to a limited, sometimes wide, range of emotions, so a laugh is not one feeling. Producing them on purpose may depend partly on hearing: profoundly deaf adults made atypical aggression and pain sounds that listeners identified less accurately.

In a 2023 study, 618 listeners rated very short vowel recordings; agreement on age, sex, health and dominance was already high after 25 ms, while agreement on traits such as trustworthiness and on social traits rose more slowly. The study measured shared impressions, not whether they were true, and other researchers caution that judging strength or condition from pitch is not necessarily accurate.

Among 160 men, lower fundamental frequency, the voice's basic pitch, was linked to higher dominance ratings from male listeners. In mock job interviews, people raised their pitch when facing dominant or prestigious interviewers, especially if they saw themselves as low in dominance. Pitch is not the only cue, though: in one group game, loudness rather than pitch tracked perceived dominance.

Adults in many languages speak to infants in a distinctive register, often called motherese, and it is not limited to mothers. A review of 144 studies linked it to infants' affect, attention and language learning. Its universality is questioned, though: untrained English-speaking raters could pick out child-directed speech in Farsi, but only trained raters managed it for a rural Mayan language.

Chronemics is the study of how people use and perceive time in communication, and Bruneau argued that time may underlie all nonverbal communication. Punctuality can be one such message: in a study at two Swedish schools, the two mothers interviewed described being on time as a way to show respect. Yet staff and students said how a late arrival played out depended on the teacher and the situation.

Anthropologist Edward Hall described monochronic time as a line from past to future, cut into units and scheduled so tasks follow one another. Polychronic time runs several activities at once and gives weight to time-consuming relationships. Hall grouped whole regions under one type despite their differences, and polychronicity is now also measured as a trait that varies between individuals.

Researchers describe waiting as unequal: tied to power, status and resources, it both mirrors and sustains the gap between the powerful and the powerless. A history of British general practice argues that appointment systems let doctors regain control of their time and some of what they saw as lost status. Two people interviewed about waiting for prison health care in England described feeling lower class or second-class.

Conversation researchers suggest people are highly sensitive to timing: a long gap after a question can signal trouble answering yes, or uncertainty. In corpus studies, gaps of 700 ms or more go with responses that resist what the question invited, and in experiments gaps of 600 ms or more led listeners to such unwelcome inferences. The typical gap between speakers is around 200 ms.

A 10-language study of yes-no questions found a mean gap of about 200 ms between turns, with each language's mean within 250 ms of the overall figure. A later 24-language corpus study found an average of just 59 ms, with slight average overlap in 46% of the languages. Since producing a word takes about 600 ms, speakers seem to plan replies before the other person finishes.

In a review of nurses' nonverbal communication with older adults, nurse aides described spending time with residents as a way of showing respect, acknowledgment and interest, and time was reported to benefit nurse practitioner relationships with older patients. Only three of the 22 studies addressed time at all, so this is an observation from care settings, not a general rule.

Time perspective is the largely automatic sorting of experience into past, present and future frames, and these frames are linked to behaviors such as risk taking and health habits. The widely used Zimbardo inventory has five subscales, from past positive and past negative to present hedonistic, present fatalistic and future, but studies across countries have had mixed results, with several failing to recover that structure.

Research reviewed in 2022 shows that one glimpse of a photo is enough for people to form trait impressions such as trustworthiness, and perceived facial trustworthiness has predicted consumer choices, online lending and even death penalty decisions. These impressions have limited accuracy at best, and about half of their variation reflects the individual observer rather than shared agreement.

A meta-analysis of 76 studies found attractiveness strongly shaped ratings of social competence and intelligence but only weakly shaped ratings of moral integrity. In a 2025 paper, US and Polish samples judged highly attractive women, not men, as more moral than moderately attractive ones, and liking explained the link; when liking was varied separately, attractiveness alone no longer raised moral ratings.

A 2023 review proposes that observers use dress to infer social categories, mental states, status and aesthetic taste, and notes earlier findings that uniforms strongly define roles. In one study, 201 people rated photographs, and male dentists in professional attire and male lawyers in formal attire were judged more suitable and capable than those in smart or casual dress.

One definition used in person-perception research treats dress as any modification of, or supplement to, the body, so backpacks, purses and jewelry count, while wheelchairs and prosthetics are borderline. In one survey, lesbians and gay men named clothing style, fit and jewelry as cues that help identify their own group, and dress has long let LGBTQ+ people signal identity to one another without being out publicly.

Under the modification-or-supplement definition of dress, hair itself is not dress but styling it is, and makeup and tattoos count. In a survey of 278 tattooed adults at a free clinic in Tijuana, 31% reported job barriers due to their tattoos. A Hungarian survey of 1,226 students and teachers found private acceptance of tattooed teachers exceeded the acceptance attributed to others, with face and neck tattoos accepted least.

A 2023 review notes that a visitor chair facing a supervisor across a desk, often set lower, signals more power and dominance than seating side by side. Furniture also limits the distances and angles people can take, and lighting, room color, temperature, windows, sound and scent can affect moods and interaction. The review treats the setting as context that the body-language idea neglects.

In an Israeli study, accounting managers rated their interest in hiring candidates shown in photographs and favored more attractive ones, a premium larger at the five biggest firms. In an Italian study of 221 students, competence read from faces predicted hiring choices for male and female applicants, while for women morality and attractiveness read from faces also mattered. Both used photos, not real hires.

A 2020 systematic review reports that chemosignals in body odor collected during emotional states can affect receivers' social behavior and danger detection, and that even odors of happiness can be perceived. In one experiment, sweat taken from eight people during the fast-acting stress of preparing a speech led 31 receivers to show fear-like facial muscle activity and faster classification of emotional faces.

Hall described low-context communication as carrying most of the information in the explicit code, and high-context communication as leaving most of it in the setting or in what people already share. Research links collectivism with high-context, indirect styles and individualism with low-context, direct ones. All cultures use both styles, though one tends to dominate.

Researchers define display rules as the social rules that set which emotional expressions are appropriate, and report strong cultural differences in them. They have a decoding side: perceivers tend to read expressions their culture discourages less well. In a study in Germany, Israel, Greece and the US, anger signaled a broken norm more strongly than sadness in all four, but how intense expressions seemed varied by country.

Cross-country research treats preferred interpersonal distance as a cultural trait. Comparing Finnish and Japanese participants, one study found the Japanese rated a neutral face making eye contact as angrier, less approachable and more unpleasant. Touch shows sameness and difference: UK and Japanese samples both allowed more touch in closer relationships, but UK participants found social touch more pleasant.

Emblems such as the thumbs-up have conventional meanings and can stand in for words, but the A-OK sign that means okay in English-speaking settings is obscene in Brazil. When American raters judged Italian emblems, some meant nothing to them and others took a different meaning: the Italian gesture for quietly was read as stop. Some emblems were understood by both groups.

This in-group advantage is well established for faces and voices. In one study, Dutch listeners recognized emotion in Dutch vocal sounds better than in Namibian ones but no better than in British ones, though they could not tell where a voice came from; the authors favor cultural dialects over motivation. In a diverse Toronto sample, race alone, without cultural difference, produced no in-group advantage.

Even where most members of a culture share values, researchers stress substantial individual variation, so not everyone from a culture responds the same way. Attribution style varies too: in an East London study of Bangladeshi immigrants, their UK-raised children and non-migrants, model comparisons suggested that dispositional explanations, typical of Western societies, spread mainly through wider society rather than family.

National stereotypes are shared beliefs about the personality of typical citizens, and one might expect them to reflect real average differences. A 2023 review reports they have not converged with measured traits: Italians share a stereotype of warm Southerners and organized Northerners, yet direct ratings cannot tell the two groups apart. Real cultural tendencies still leave wide individual variation.

In a survey of WeChat users in China, middle-aged and older users tended to read a goodbye emoji as a simple farewell, while younger users leaned toward speechlessness or the end of a friendship. In East London, second-generation British Bangladeshis scored between their parents' immigrant generation and non-migrants on collectivism, a pattern the researchers call partial acculturation.

A meta-analysis of 206 documents and 24,483 judges found people sort lies from truths about 54% of the time, where guessing gives 50%, and they classify truths correctly more often than lies. When observers can only see and not hear the speaker, accuracy is about 52%. A federal GAO review of four meta-analyses covering over 400 studies found detection from behavioral indicators at or slightly above chance.

A review by Vrij put law enforcement accuracy near 56%, only marginally above the roughly 54% average. Professionals and laypeople hold similar beliefs about cues to lying, and a meta-analysis found only a small effect of nonverbal cue training. In two studies, officers who cited more manual-taught nervous cues, or observers trained on them, did worse.

A meta-analysis by DePaulo and colleagues found a correlation of just 0.04 between how confident people felt in a lie judgment and whether they were right. In two online studies, people on average rated their own lie detection ability above chance, and higher self-ratings were predicted by lower trust in out-groups and by a stronger tendency toward social desirability.

In a survey across 58 countries, 64% of people named gaze aversion as a sign of lying, the top belief in 51 of them, and 73% of British police officers in another study named it too. Yet a meta-analysis found liars and truth tellers show similar gaze patterns, and liars tend to move less rather than more, a pattern that held in high-stakes settings as well.

Meta-analyses find verbal cues more diagnostic than nonverbal ones, and observers who can only hear a speaker reach about 63% accuracy. Under the Verifiability Approach, liars report fewer checkable details, such as activities with named witnesses or digital traces; the gap holds on average and becomes large when people are asked to include details the investigator can check.

The name comes from the story of Othello, who disbelieved his wife Desdemona; growing desperate, she wept and so behaved as a guilty wife might, and he took it as proof. Researchers note that truth tellers can fear not being believed, and that an accusation can make liars and truth tellers alike look nervous, which an interviewer may then misread as deception.

Levine calls the overall impression made by clusters of cues a demeanor, such as seeming confident or sincere, and argues many people's demeanor does not match their honesty; an honest autistic person, for example, may seem deceptive. One study found honest-looking truth tellers and dishonest-looking liars were judged more accurately than mismatched ones, and accuracy depends more on the sender than the judge.

Because lying is thought to take more mental effort, these methods add load with unexpected questions about minor details, surprise drawings, or recalling events in reverse order. In one study, pairs who lied about lunching together were identified beyond chance from inconsistencies in their answers and sketches. A 2018 review says the methods need validation and that rehearsal may undermine them.

After 9/11, the TSA trained officers to watch passengers for cues such as avoiding eye contact. A 2013 GAO report found no validated evidence that such indicators identify threats, most officers interviewed called some indicators subjective, and referral rates varied widely across airports; DHS disagreed. Because honest people also get nervous, such flags can land on them.

Researchers treat rapport as visible only in an interaction, with the pair as the smallest unit, and describe it as built from mutual attentiveness, a back-and-forth of positive cues and, they argue most importantly, coordinated nonverbal behavior. In one study, German and Gulf Arab observers watching animated figures driven by real movement agreed closely on how much rapport each pair had.

Synchrony means partners adapt to each other's rhythms, like people dancing together; it does not require imitation, since what matters is the timing of responses rather than their form. Researchers usually measure it by correlating partners' recorded movements over time. Therapy research reports that patient and therapist tend to synchronize pitch, movement and even physiology spontaneously.

Evidence summarized in a 2016 article shows the link runs both ways: people copy models they like, and being copied raises liking for the copier. The effect is not guaranteed. In virtual reality, avatars that mimicked participants' head and torso movements raised rapport in a first study of 50 people, but a preregistered second study found no effect of mimicry on rapport or trust.

Research summarized in a 2018 study describes behavioral matching as unintentional and most prosocial when people do not notice it, while noticing that one is being mimicked can harm liking. One theory holds that mimicry signals a genuine disposition when it is truly spontaneous, so observers attend to cues such as its timing, fidelity and selectivity.

Thin slices are short excerpts from a longer stream of behavior. In one study, personality judgments beat chance after just 5 seconds for every trait tested except agreeableness, and going from 60 to 300 seconds added little. Traits differ: extraversion may be easy to judge from slices while agreeableness is harder, and some settings restrict expression enough to hide traits.

Generic backchannels like uh huh or yeah show understanding and attention, and are called continuers because they invite the speaker to keep going; specific ones like really respond to the content. When readers supplied the next turn of written dialogues, generic ones drew new information and specific ones drew elaboration. Backchannels and nods can also be used to fake listening.

In two preregistered studies of 646 people talking with strangers, follow-up questions and verbal validation predicted social connection in deep conversations, and overall listening behavior predicted it in small talk, though verbal markers were less consistent there. Nods and uh huhs are seen as weaker evidence of listening, since people can use them while only pretending to listen.

A common account has perceivers sort others by salient categories such as race or gender and then apply beliefs about those categories. A 2023 review argues that motivation shapes even the earlier step, steering which dimensions a perceiver attends to and sorts by, and that the process is not purely sequential. On this view, bias can enter before any stereotype is applied.

In social impressions, the primacy effect means information from the start of an interaction biases the whole impression, partly because later information that fits the first valence gets more weight, so a negative start can color what follows. In perceptual and value-based averaging tasks, one study found people showed a strong recency bias and only a modest primacy bias.

The attractiveness halo, crediting attractive people with other desirable traits, is described as an over-generalization effect. Warmth works similarly in person-perception research: high warmth creates a halo for competence, while low warmth can reduce the benefit of high competence. A 2024 article on clinical debriefing warns that both effects can skew how instructors judge learners.

People explaining others' successes and failures tend to overweight disposition and underweight situation. In one set of studies, experienced professionals favored candidates from lenient grading or forgiving work settings, reading high results as ability. The companion idea, that actors credit situations for their own behavior, is shakier: a later meta-analysis found average effects near zero, with differences only in some conditions.

Perceivers use stereotypes and prior knowledge to predict others. In two brain-imaging experiments, information consistent with social expectations activated reward regions, and in two behavioral experiments people gave up money to meet expectation-consistent targets. Yet a 2026 study of impressions updated in daily life found a contrast bias: people overreacted to new inconsistent information.

Perceivers lean on stereotypes and other prior knowledge to simplify the hard work of predicting how people will respond. Face research shows conceptual stereotypes shaping impressions formed from faces, and overgeneralizing from social groups appears to influence those impressions too. Physical facial cues also matter, since data-driven models predict much of the variance from them.

Nonverbal behavior often does not explain itself: a scowl can mean several different things, so an inference from it needs other evidence before it is trusted. A 2024 article on clinical debriefing describes a case where an instructor assumed a learner did not know to start chest compressions, when a faulty simulator may have shown a pulse; asking for the learner's view could have surfaced the real reason.

Self-fulfilling prophecy research holds that a teacher who wrongly blames low ability may expect less, read later work as confirming it, and treat the student in ways that draw poorer performance. Attribution theory suggests that emotional expression, body language and feedback can carry teachers' causal beliefs to students. A 2018 review found such teachers reported lower expectations, and students seen as low ability expected less success.

Walther's social information processing theory holds that when nonverbal cues are missing, people adapt by shifting that work onto words and other strategies. An analysis of 2,004 emails and text messages collected from college students in 2010 found writers used deliberate style choices, such as extra letters, all capitals, abbreviations and emoticons, to carry tone.

Expressive punctuation, capital letters and emoji arose as substitutes for cues lost in texting. In two experiments, an emoji helped readers grasp a negative, indirect reply faster, though not when the question was a request for action. In another study, replies with emoji seemed more responsive. Meanings also vary by age: younger WeChat users read a goodbye emoji as more than a farewell.

With a webcam on top of the monitor, the angle between the camera and the other person's eyes on screen makes someone trying to make eye contact seem to look at your chin. Careful positioning can shrink that angle to about 2 degrees. In a simulated teletherapy study of 42 clients, restoring eye contact this way did not raise rated empathy; clients seemed to rely on other cues.

Bailenson proposed that gaze on video calls can cause nonverbal overload; long stretches of close eye contact and grids of faces looking at you are unusual offline, and constant self-view is often blamed too. In a small study of German psychology students, turning off self-view reduced fatigue and cognitive load and active participation went with less fatigue, while grid versus speaker view made no significant difference.

Bailenson likens constant self-view to a colleague carrying a mirror of your face all day, and a meta-analysis found a small link between seeing oneself in a mirror and negative mood. A survey of 2,448 US workers linked dissatisfaction with one's own facial appearance to video meeting fatigue. Turning self-view off is often suggested, but many users prefer to watch themselves.

When faces were shown over virtual backgrounds, plants and bookcases earned the highest trust and competence ratings and home and novelty scenes the lowest. Smiling faces were rated higher overall, but smiling appeared to reliably offset the weak home and novelty backdrops only for trust. The study used still photos, and earlier work found backgrounds did not change interview outcomes.

In a WhatsApp experiment with 160 young adults, a partner who replied at once rather than after seven minutes was perceived as more socially present, meaning more accessible, though not as more in tune with the participant. Empathic content and emoji raised both kinds of presence. The authors note that a seven-minute delay may not feel slow in real messaging.

In a preregistered experiment, pairs in virtual reality moved in synchrony beyond what random pairings showed, whether they appeared as humanlike avatars or as abstract cubes. In a separate study of word-description games, pairs described more words when both avatars moved with their users than when both stood still, and did worse when the listener avatar played a prerecorded animation instead of real movement.

In tests such as the Geneva Emotion Recognition Test, people watch short clips and say what the person feels, and accuracy is the match with the correct answer; others include the DANVA, which uses voices, and the PONS. In one study such tests correlated with one another, suggesting a shared skill, and tracked crystallized intelligence rather than self-rated social functioning.

Self-reported emotional intelligence typically shows no or only small correlations with measured emotion recognition. In a study of 227 adults, test scores clustered together and correlated with crystallized intelligence but not with self-rated recognition skill. In a small study, musicians rated their own empathy higher than non-musicians did, yet showed no significant difference on an empathic accuracy task.

In a randomized study of 67 people, three weekly sessions of multimodal or micro-expression training improved only the trained skill, neither group beat the control on recognizing patients' emotion cues, and lower starting scores went with larger gains. A visual-search study found gains carried over to an untrained expression but not from shapes to faces, and gains were small for briefly flashed faces.

Expressive accuracy, the tendency to be perceived accurately, varies widely between people. In 72 small online groups of strangers, more facially expressive members were rated warmer and more cooperative and sat more centrally in liking networks, while attractiveness did not predict standing. In lie detection too, accuracy depends more on the person judged than on the judge.

Studies find autistic adults produce and interpret nonverbal cues differently from non-autistic adults, and vary among themselves. The double empathy problem describes misunderstanding that runs both ways between the two groups, and researchers warn against judging neurodivergent behavior on neurotypical terms. Culture, personality and trauma can also make expression more restrained.

Nonverbal cues lack the fixed meanings of words, so the same smile, gaze and closeness can signal liking in one setting and a sales pitch in another; confirming an inference takes other evidence. A 2025 review warns that automated emotion systems may misread people who express emotion in restrained or unusual ways as flat, deceptive or insincere, errors it calls ethical liabilities.

Police manuals have listed poor eye contact and restless movement as signs of lying, airport screeners were taught similar nervousness cues, and jurors are told to notice witnesses' nonverbal behavior. Police in one study, like students, believed in such non-diagnostic cues. Those at risk include nervous truth tellers and people whose demeanor does not match their honesty.