Facial Expression

Abstract

A facial expression is a configuration of facial-muscle movements that conveys an emotional or communicative state. It is one of the oldest objects of scientific psychology: Charles Darwin's 1872 monograph argued that human expressions are inherited, continuous with animal displays, and shared across cultures, and Paul Ekman's twentieth-century program turned that thesis into a measurable claim that a small set of basic emotions have universally recognized facial signals. Cross-cultural recognition studies find agreement well above chance everywhere tested, yet methodological critiques, an in-group advantage, and reverse-correlation work showing culturally distinct signals all complicate a strong universality reading. Expression is also partly a cause of feeling, not only an effect, though the textbook facial-feedback demonstration failed a large replication. This article covers how expression is measured, the universality debate, and the contemporary reappraisal of what the moving face reveals.

Keywords: facial expression, basic emotions, universality, facial action coding system, facial feedback hypothesis

What Facial Expression Is

A facial expression is a patterned movement of the facial muscles that signals an internal state to an observer — most centrally an emotion, but also intention, appraisal, and communicative intent. In MeSH it is defined as the voluntary or involuntary motion of the face or its parts in the expression of the internal emotional state of the individual, and it is indexed both as a sign elicited in physical examination and as a form of nonverbal communication. The face has more than forty independently controllable muscles, and their combinations produce a signalling channel of enormous expressive range built on a compact anatomical substrate.

Two features make facial expression a genuine object of cognitive psychology rather than a mere catalogue of grimaces. First, it is dual-natured: the same musculature produces both involuntary displays driven by felt emotion and deliberate, posed, or regulated configurations shaped by social rules, so any given expression is a mix of what a person feels and what they intend to show. Second, it sits at the center of a century-old empirical question with real stakes — whether the mapping from emotion to face, and from face back to perceived emotion, is a human universal or a cultural convention. The answer governs everything from the design of emotion-recognition software to the admissibility of reading a defendant's face in court, which is why a question that sounds academic has repeatedly spilled into application.

Types of Facial Expression

In the MeSH hierarchy, Facial Expression is filed under two parents that reflect its dual status — Physical Examination, where an expression is a clinical sign read off the face, and Nonverbal Communication, where it is a channel of social signalling. Below it, MeSH records a single narrower descriptor, listed in the table. The classification is an indexing scheme for the biomedical literature, not a psychological taxonomy of emotion: it tracks which configurations have acquired their own research literatures large enough to warrant a descriptor, so a type's presence or absence here reflects bibliographic history rather than any claim about how many distinct expressions exist.

These types are not mutually exclusive modes, and the single narrower descriptor does not mean the face produces only one kind of expression. A smile may co-occur with a brow configuration that changes its meaning entirely, and the great majority of expressions the face produces — fear, disgust, the raised-brow of surprise — carry no separate MeSH descriptor at all, being indexed directly under the parent term. The table therefore marks where a dedicated literature exists, not the boundaries of the expressive repertoire.

The Universality Thesis

The scientific study of facial expression begins with Charles Darwin (1872). Against the prevailing view that expressions were learned conventions, he argued that they are inherited, serve or once served a biological function, are continuous with the displays of other animals, and are therefore shared across all human groups. His evidence ranged from infants and the congenitally blind to questionnaires sent to observers on several continents — an early appeal to cross-cultural data to settle a question about human nature.

Ekman, Sorenson, and Friesen (1969) converted Darwin's thesis into a testable modern claim. Showing photographs of posed emotional faces to observers in several literate cultures, and crucially to the preliterate Fore of Papua New Guinea, they reported that a small set of emotions — happiness, sadness, anger, fear, disgust, surprise — were recognized at rates far above chance everywhere, including in a culture with minimal exposure to Western media. Ekman and Friesen (1971) extended the Fore study with a forced-choice story method designed for a preliterate sample, and the convergence became the empirical backbone of the basic emotions program. Ekman (1992) set out that program's criteria — each basic emotion has a distinct universal signal, a characteristic physiology, and an evolved function — and Ekman (1993) synthesized it as a neurocultural theory: universal expression programs are inherited, but display rules, learned culturally, govern when and how strongly they are shown. The neurocultural move is what let the theory absorb obvious cultural variation without abandoning the universal core.

Measuring the Face

A claim about which configurations signal which emotions requires a way to describe configurations without presupposing the emotions. The Facial Action Coding System, developed by Ekman and Friesen, is that tool: it decomposes any facial movement into action units, each an anatomically defined contraction of one muscle or a small group — the inner-brow raiser, the lip-corner puller, the lid tightener. An expression becomes a combination of action units, coded from the visible movement alone, so that the mapping to emotion is a hypothesis tested against the code rather than built into it.

The system's analytic power is clearest in the smile. Ekman, Davidson, and Friesen (1990) distinguished the Duchenne smile — lip-corner pull (the zygomatic major) combined with orbicularis-oculi contraction that raises the cheek and crinkles the eye — from the non-Duchenne social smile, which moves the mouth alone. The Duchenne marker accompanied reported enjoyment and a left-frontal EEG pattern associated with positive affect, while the social smile did not, giving an objective, muscle-level criterion for the old intuition that a felt smile differs from a polite one. The action-unit approach thus supplies both the universality program's measurement instrument and one of its cleanest confirmations: a morphological feature that tracks felt emotion independently of what the person reports.

The Cross-Cultural Debate

The universality evidence has been contested on method and on substance. Russell (1994) mounted the central methodological critique: the classic studies used forced-choice response formats that inflate agreement by removing wrong-but-plausible options, within-subject designs that cue the expected answer, and posed, exaggerated stimuli unlike spontaneous expression. Recognition, he argued, is graded and dimensional — observers agree most on the pleasant/unpleasant and aroused/calm axes — rather than reflecting discrete universal categories, and the apparent universality is partly an artifact of how the question is asked.

Two bodies of work locate a middle ground and then press past it. Elfenbein and Ambady (2002) meta-analyzed the cross-cultural recognition literature and found a robust in-group advantage: emotions are recognized above chance across cultural boundaries, confirming a universal component, but more accurately when expresser and perceiver share a cultural background — a systematic cultural accent on a shared signal. Jack and colleagues (2012) went further, using reverse correlation to reconstruct the facial-movement patterns that East Asian and Western observers spontaneously associate with each emotion. The reconstructed signals differed systematically: Western representations separated the six emotions into distinct action-unit patterns, while East Asian representations overlapped and relied more on eye-region dynamics. The finding challenges universality at its root, suggesting the signal itself, not only the display rules governing it, varies across cultures. The debate thus runs from the same signals, differently regulated (neurocultural) through the same signals, better read by locals (in-group advantage) to partly different signals (reverse correlation) — a spectrum the rest of the field now occupies.

Figure

Interactive Demonstrations

Three demonstrations make the field concrete: an action-unit builder that assembles emotion configurations on a schematic face, a facial-feedback panel contrasting the 1988 report with its 2016 replication, and a cross-cultural recognition model showing the in-group advantage above chance. Each is deterministic and runs entirely in the browser.

Worked Example

Consider how to read a cross-cultural recognition result quantitatively. The classic studies used a forced choice among the six basic emotions, so the chance baseline is one correct answer in six: 1 / 6 = 0.167, or about 16.7%. A recognition rate is only evidence for a shared signal to the extent it clears that floor. Suppose observers identify an expresser's emotion correctly 82% of the time when expresser and perceiver come from the same culture, and 73% of the time when they come from different cultures — values in the range Elfenbein and Ambady (2002) reported.

Two quantities matter, and they pull in opposite directions. Both rates sit far above the 16.7% chance floor — 73% − 16.7% = 56.3 percentage points of above-chance out-group accuracy — which is the universal component: even across a cultural boundary, the signal carries. But the in-group advantage is the gap between the two rates: 82% − 73% = 9 percentage points more accurate within one's own culture. That gap is the cultural accent, and it is reliable across the meta-analytic sample. The lesson is that the universality debate is not all-or-nothing arithmetic: the same two numbers simultaneously support a substantial shared signal (the large above-chance floor) and a real cultural modulation (the in-group gap). A headline declaring emotions universally recognized reports only the first number; a headline declaring emotion recognition culturally specific reports only the second. The recognition demonstration below varies the cultural distance so both quantities change at once.

Discussion

Facial expression occupies a peculiar scientific position: it is among the most intuitively obvious of psychological phenomena — everyone reads faces all day — and among the most stubbornly contested in its particulars. The arc from Darwin through Ekman to the present is not a story of a thesis confirmed or refuted but of a thesis progressively qualified. The universal component is real: recognition clears chance everywhere tested, some displays appear in people who could never have seen them, and a muscle-level marker distinguishes felt from posed smiling. The cultural component is equally real: recognition is better within a group, the signals observers carry in their heads differ across cultures, and what is shown is governed by learned rules. The productive question is no longer universal or cultural? but which aspects of expression are which, and how do they combine?

That reframing has methodological teeth. Russell's (1994) critique showed that the answer depends heavily on how expression is elicited and recognition is measured — posed versus spontaneous stimuli, forced-choice versus free labelling, static photographs versus dynamic video. Much of the apparent disagreement in the literature dissolves into a disagreement about method, which is why the strongest modern work (reverse correlation, large naturalistic video corpora) is defined as much by its measurement innovations as by its conclusions. The practical stakes keep the question honest: emotion-recognition systems deployed at scale inherit whatever the science gets wrong, so the gap between a face configuration and the emotion a person actually feels is not a quibble but the whole problem.

Current Directions

The contemporary reappraisal is sharpest in Barrett and colleagues' (2019) review, which marshals the accumulated evidence into a strong claim: facial configurations are too variable — within a person, within an emotion, across contexts — to support reliable inference from face to felt emotion. People scowl when angry only a minority of the time, and scowl for many reasons other than anger, so the common assumption that a given configuration reveals a specific emotion is, they argue, not supported. The review is a direct challenge to the basic-emotions program and to the commercial emotion-recognition industry built on it, and it reframes expression as one context-dependent cue among many rather than a readout of an internal state.

The large-scale naturalistic studies complicate both extremes. Cordaro and colleagues (2018) catalogued 22 emotional expressions across five cultures and found both a universal core of recognizable patterns and systematic cultural variation — a dialect theory in which shared expressions carry local accents. Cowen and colleagues (2021) applied machine learning to roughly six million naturalistic videos from 144 countries and found that sixteen distinct expressions occur in similar social contexts worldwide, evidence for cross-cultural regularity in how expression is used, not just posed in the lab. And Tracy and Matsumoto (2008) showed that congenitally blind athletes produce the same pride and shame displays as sighted ones after winning or losing, extending innate expression beyond the classic six. The open question the field now shares is how to reconcile strong context-dependence at the level of the single face with robust statistical regularity at the level of millions — a question that is pushing the methods from the posed photograph toward the naturalistic corpus.

Common Misconceptions

A given facial expression reliably reveals the emotion a person is feeling.

The inference from face to felt emotion is far weaker than intuition suggests: configurations are highly variable within an emotion and across contexts, and the same configuration occurs for many reasons (Barrett et al., 2019). Expression is a cue, not a readout.

Facial expressions of emotion are either fully universal or entirely cultural.

Neither pole fits the evidence. Recognition is above chance across cultures (a universal component) yet more accurate within a culture (an in-group advantage), and the signals themselves vary somewhat across groups (Elfenbein & Ambady, 2002; Jack et al., 2012).

Smiling reliably makes a person happier.

The facial-feedback hypothesis is real in principle, but its most famous demonstration — the pen-in-teeth study — failed a 17-laboratory registered replication, so the simple smile to feel happy claim is not well supported in that paradigm (Strack et al., 1988; Wagenmakers et al., 2016).

A smile is a smile — one expression with one meaning.

Felt and posed smiles are anatomically distinct: the Duchenne smile of genuine enjoyment contracts the muscle around the eye, which the voluntary social smile does not, and the two have different physiological correlates (Ekman et al., 1990).

Glossary

Key Researchers

Lisa Feldman Barrett

(Northeastern University). Neuroscientist whose theory of constructed emotion and 2019 review of the expression literature mount the leading contemporary challenge to the view that discrete emotions can be read reliably from facial movements. ORCID · Wikipedia · Scholar · Wikidata

Charles Darwin

(1809–1882). Naturalist whose 1872 The Expression of the Emotions in Man and Animals founded the evolutionary study of facial expression — the thesis that human expressions are inherited, continuous with animal display, and shared across cultures. Wikipedia · Wikidata

Paul Ekman

(1934–2025). Psychologist whose cross-cultural recognition studies, the Facial Action Coding System, and the neurocultural theory of basic emotions made facial expression a measurable scientific object — the dominant twentieth-century framework the field still argues with. Wikipedia · Scholar · Wikidata

Rachael E. Jack

(University of Glasgow). Psychologist whose reverse-correlation modelling of the mental representations underlying facial expressions produced the strongest modern evidence that the expression signal itself varies across cultures. ORCID · Faculty · Scholar

Dacher Keltner

(University of California, Berkeley). Psychologist whose large-scale cross-cultural and naturalistic studies of emotional expression — the 22-expression catalog and the worldwide-video analysis — anchor the contemporary dialect-theory synthesis of universals and cultural variation. ORCID · Wikipedia · Scholar · Wikidata

James A. Russell

(Boston College). Psychologist whose circumplex model of affect and methodological critique of the universality studies reframed the recognition literature as graded and dimensional rather than a set of discrete universal categories. Wikipedia · Wikidata

Frequently Asked Questions

What is a facial expression?

It is a patterned movement of the facial muscles that signals an internal state, most centrally an emotion, to an observer. In MeSH it is the voluntary or involuntary motion of the face expressing the internal emotional state of the individual, indexed both as a sign in physical examination and as nonverbal communication.

Are facial expressions of emotion universal?

Partly. A universal component is well supported, since recognition clears chance in every culture tested and some displays appear in people who could not have learned them, yet recognition is more accurate within one's own culture, and the signals themselves vary somewhat across groups. The modern consensus is a shared core with cultural accents, not strict universality.

Who were the key figures in the science of facial expression?

Charles Darwin founded the evolutionary account in 1872; Paul Ekman and Wallace Friesen built the twentieth-century universality program and the Facial Action Coding System; James Russell, Hillary Elfenbein, Rachael Jack, Lisa Feldman Barrett, and Dacher Keltner lead the contemporary debate over how universal expression really is.

What is the Facial Action Coding System?

It is a system for describing any facial movement as a combination of anatomically defined action units, individual muscle contractions such as the inner-brow raiser or the lip-corner puller, coded from the visible motion alone, so that the mapping from face to emotion can be tested rather than assumed.

What is the difference between a Duchenne and a social smile?

A Duchenne smile combines the mouth movement with contraction of the muscle around the eye, which raises the cheek and crinkles the eye corner; it accompanies genuinely felt enjoyment. The social smile moves the mouth alone, is usually voluntary, and lacks the eye-muscle marker and the associated positive-affect physiology.

Does smiling make a person happier?

The facial-feedback hypothesis holds that expression can influence feeling, and in principle it can. But the most famous demonstration, holding a pen in the teeth to force a smile, failed a 17-laboratory registered replication, so the simple claim that making oneself smile reliably increases happiness is not well supported by that paradigm.

Can an observer tell what someone is feeling from their face?

Less reliably than intuition suggests. A recent major review argues that facial configurations are too variable within an emotion and across contexts to support confident inference from face to felt emotion; the same scowl occurs for many reasons besides anger. Expression is one context-dependent cue, not a direct readout of an internal state.

Why does the universality question matter outside the laboratory?

Because emotion-recognition systems are now deployed at scale, in hiring, security, and marketing, and in legal settings faces are read for credibility. Any systematic error in the science about how reliably the face reveals emotion is inherited by those applications, which is why the gap between a configuration and a felt emotion has practical stakes.

References

Barrett, L. F., Adolphs, R., Marsella, S., Martinez, A. M., & Pollak, S. D. (2019). Emotional expressions reconsidered: Challenges to inferring emotion from human facial movements. Psychological Science in the Public Interest, 20(1), 1–68. https://doi.org/10.1177/1529100619832930

Cordaro, D. T., Sun, R., Keltner, D., Kamble, S., Huddar, N., & McNeil, G. (2018). Universals and cultural variations in 22 emotional expressions across five cultures. Emotion, 18(1), 75–93. https://doi.org/10.1037/emo0000302

Cowen, A. S., Keltner, D., Schroff, F., Jou, B., Adam, H., & Prasad, G. (2021). Sixteen facial expressions occur in similar contexts worldwide. Nature, 589(7841), 251–257. https://doi.org/10.1038/s41586-020-3037-7

Darwin, C. (1872). The expression of the emotions in man and animals. John Murray.

Ekman, P., Sorenson, E. R., & Friesen, W. V. (1969). Pan-cultural elements in facial displays of emotion. Science, 164(3875), 86–88. https://doi.org/10.1126/science.164.3875.86

Ekman, P., & Friesen, W. V. (1971). Constants across cultures in the face and emotion. Journal of Personality and Social Psychology, 17(2), 124–129. https://doi.org/10.1037/h0030377

Ekman, P. (1992). An argument for basic emotions. Cognition & Emotion, 6(3–4), 169–200. https://doi.org/10.1080/02699939208411068

Ekman, P. (1993). Facial expression and emotion. American Psychologist, 48(4), 384–392. https://doi.org/10.1037/0003-066x.48.4.384

Ekman, P., Davidson, R. J., & Friesen, W. V. (1990). The Duchenne smile: Emotional expression and brain physiology II. Journal of Personality and Social Psychology, 58(2), 342–353. https://doi.org/10.1037/0022-3514.58.2.342

Elfenbein, H. A., & Ambady, N. (2002). On the universality and cultural specificity of emotion recognition: A meta-analysis. Psychological Bulletin, 128(2), 203–235. https://doi.org/10.1037/0033-2909.128.2.203

Jack, R. E., Garrod, O. G. B., Yu, H., Caldara, R., & Schyns, P. G. (2012). Facial expressions of emotion are not culturally universal. Proceedings of the National Academy of Sciences, 109(19), 7241–7244. https://doi.org/10.1073/pnas.1200155109

Russell, J. A. (1994). Is there universal recognition of emotion from facial expression? A review of the cross-cultural studies. Psychological Bulletin, 115(1), 102–141. https://doi.org/10.1037/0033-2909.115.1.102

Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facilitating conditions of the human smile: A nonobtrusive test of the facial feedback hypothesis. Journal of Personality and Social Psychology, 54(5), 768–777. https://doi.org/10.1037/0022-3514.54.5.768

Tracy, J. L., & Matsumoto, D. (2008). The spontaneous expression of pride and shame: Evidence for biologically innate nonverbal displays. Proceedings of the National Academy of Sciences, 105(33), 11655–11660. https://doi.org/10.1073/pnas.0802686105

Wagenmakers, E.-J., Beek, T., Dijkhoff, L., Gronau, Q. F., Acosta, A., Adams, R. B., Jr., et al. (2016). Registered Replication Report: Strack, Martin, & Stepper (1988). Perspectives on Psychological Science, 11(6), 917–928. https://doi.org/10.1177/1745691616674458