Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Cultural Variation in Emotion Concepts and Expression

Abstract. Cultural variation in emotion concerns the ways communities name, organize, evaluate, perceive, express, and regulate affective episodes within systems of language and social practice. This research monograph critically synthesizes classic universality studies, the methodological debate about facial-expression recognition, cross-linguistic semantic networks, research in relatively isolated communities, cultural display rules, ideal affect, social-relational approaches, a 2025 individual participant data meta-analysis, and recent large-scale computational studies. It evaluates what the evidence establishes about shared structure and local diversity, distinguishing expression production from judgment, emotional concepts from bodily episodes, and cultural-level patterns from individual experience. Methodological limits and proposed research designs identify how comparative emotion science can advance through collaborative, ecologically valid and multilingual measurement. The study is a critical narrative research synthesis of published evidence, not a registered systematic review or new experiment.

Keywords: cultural variation in emotion; emotion concepts; emotion semantics; cross-cultural psychology; facial expression; vocal emotion; emotion recognition; display rules; affect valuation; emotion regulation; emic and etic; colexification; cultural evolution; emotion perception; social construction of emotion

The Research Object and Its Explanatory Boundaries

Cultural variation in emotion concepts and expression is the systematic study of how patterns of naming, evaluating, experiencing, displaying, and interpreting affective events differ across historically situated communities. The unit of comparison may be an emotion word, an appraisal pattern, a bodily response, a facial movement, a communicative act, or a social situation. These units answer different questions and require different evidence. A person can experience agitation without categorizing it as anger, display a smile without communicating happiness, or understand a fear word while using it in locally specific circumstances. Cross-cultural findings therefore require explicit specification of the phenomenon being compared. The early integrative framework of Mesquita and Frijda (1992) remains useful because it follows emotion across several phases rather than making a single component stand for the whole process. Mesquita & Frijda (1992)

The central questions concern which properties recur across groups, which differ, how shared social practices maintain those differences, and whether a reported contrast survives changes in translation, task, setting, and sampling. A cultural contrast in adjective ratings offers evidence about reported concepts, while an observed pattern of behavior in a ceremony offers evidence about situated expression. Neither result alone establishes that people inhabit different biological affective systems. Studies of local social expectations, lexicons, and behavior must be connected through identifiable mechanisms instead of compressed into a single scale of cultural difference. The constructionist synthesis of Mesquita, Boiger, and De Leersnyder (2016) emphasizes that the distribution of emotional episodes itself may be shaped by participation in characteristic relationships and institutions. Mesquita et al. (2016)

This dossier treats culture as learned, transmitted, and changing practices, rather than an attribute that every member of a nation shares in identical form. Nationality can be an administrative sampling variable; it is only an approximate indicator of language, social network, norms, and lived experience. Within-group diversity, migration, generation, and institutional context are consequently necessary components of explanation. The scope includes emotion concepts, expression production, perception by observers, display rules, ideal affect, and interpersonal dynamics. Detailed mechanistic accounts of affect generation, clinical syndromes, and the complete domain of emotion regulation belong in adjacent research dossiers. The present synthesis focuses on the processes through which these domains acquire locally patterned meanings and functions. Lindquist et al. (2022)

Emotion Episodes, Concepts, Categories, and Displays

An emotion episode can include a change in perceived stakes, attention to a situation, bodily activation, felt valence, action readiness, expression, and subsequent interpretation. Emotion concepts are learned semantic and practical resources for identifying, communicating, and predicting such episodes. A category such as fear may refer to several kinds of episode related by family resemblance rather than an invariant physiological package. Researchers should distinguish concepts inferred from linguistic judgments from the occurrence of episodes, and distinguish a participant's own categorization from the experimenter's category labels. Jackson and colleagues (2019) show that languages share some large-scale semantic organizing dimensions while varying in which particular emotions are linguistically linked. Jackson et al. (2019)

An expressive behavior is an observable movement or vocal event, including a smile, narrowed eyes, laughter, a gasp, or silence. Its social meaning arises from context, the individual's action, the observers' knowledge, and locally learned conventions. A smile can communicate affiliation, politeness, embarrassment, reassurance, or joy, depending on the interaction. Expression production means what people do; expression judgment means what observers infer from that behavior. Studies that obtain consensus for a label attached to a photograph establish performance under the tested judgment conditions. They do not, by themselves, establish that a given facial configuration always accompanies a particular felt state. This distinction is central to the controversy reviewed by Russell (1994) and the later work of Keltner and colleagues (2019). Russell (1994) Keltner et al. (2019)

Cultural display rules specify when expressions are encouraged, muted, amplified, masked, or otherwise adapted to a social audience. A society can share a broadly intelligible expression while prescribing its restraint during a formal encounter; conversely, two groups can employ different visible cues to pursue similar relational goals. The object of analysis is therefore a conditional mapping among eliciting situation, emotional appraisal, relationship, audience, communicative intention, and behavior. Matsumoto and colleagues (2018) link beliefs about display management to judgments of others' expressive intensity, making cultural expectations part of perception itself. Concept taxonomies, display rules, and interpersonal functions can be empirically separated only with designs that observe each component. Matsumoto et al. (2018)

The Historical Formation of Comparative Emotion Science

Foundational comparative research asked whether emotions possess components that extend across the cultural diversity of humanity. Ekman, Sorenson, and Friesen (1969) presented photographs of posed facial displays to observers from several societies and reported predicted category choices, while noting stronger agreement in literate samples. The study became a key source for a basic-emotion interpretation of cross-cultural expression. Its importance is historical and methodological: it offered a testable proposal linking visible configurations and discrete emotion names. The evidence, however, was generated through a specific set of materials and response formats, matters that later investigators examined directly rather than treating the original task as a transparent measurement of emotion. Ekman et al. (1969)

The debate sharpened in the 1990s. Russell (1994) reviewed the cross-cultural recognition literature and questioned how forced-choice labels, posed photographs, and within-subject comparison influence responses. Ekman (1994) defended the universality interpretation, arguing that evidence persisted despite methodological objections. These articles disagree not only about findings but about the intended conclusion: whether participants associate some facial configurations with emotion labels, whether such associations are widespread, and whether invariant psychological categories can be inferred from the association. Subsequent experiments separated these claims more effectively by manipulating linguistic cues, using less familiar societies, and observing expressions in context. Methodological disagreement is thus part of the cumulative history of a productive research program. Russell (1994) Ekman (1994)

A parallel program examined emotion in anthropological and linguistic settings. Mesquita and Frijda (1992) synthesized cultural differences in eliciting events, appraisals, behavior repertoires, and regulation. Wierzbicka (1999) investigated how the meaning of emotion words depends on language-specific semantic structures and explored possibilities for cross-language comparison without assuming exact lexical equivalents. Mesquita and Walker (2003) then connected cultural models of self and relationships with recurrent patterns of feeling and behavior. These approaches widened the research object from photographs and judgments to the social conditions that bring emotional experiences into being and make them consequential. Their distinct predictions require evidence at interpersonal and community levels. Mesquita & Frijda (1992) Wierzbicka (1999) Mesquita & Walker (2003)

From Isolated Judgments to Socially Situated Evidence

The older cross-cultural paradigm often compared university participants or small samples viewing still images under standardized instructions. Standardization improves internal comparability, yet the intended equivalence may disappear if the image, emotion label, or instruction has a different local meaning. A carefully matched laboratory task can identify an effect of semantic prompting; it does not automatically identify how people understand a neighbor's spontaneous expression. Gendron and colleagues (2014) directly tested the role of conceptual cues in sorting posed faces in the United States and among Himba participants in Namibia. Their findings demonstrate why an instrument's task structure is a substantive causal ingredient in the result. Gendron et al. (2014)

Work in relatively small-scale societies makes the question sharper because it reduces some forms of familiarity with conventional Western stimuli. Crivelli and colleagues (2016) studied facial interpretation in indigenous societies and questioned an easy mapping from posed faces to fixed emotion labels. Their subsequent investigation of spontaneous expressions in Papua New Guinea asked what local observers make of local behavior rather than relying only on externally composed photographs. These designs have their own constraints, including the availability of translators, locally appropriate elicitation materials, and potentially small samples. The scientific contribution lies in identifying precisely which generalization does and does not survive changes in ecological and cultural context. Crivelli et al. (2016) Crivelli et al. (2017)

Later work introduced large observational and computational corpora. Cowen and colleagues (2021) used machine learning to examine relationships between recognizable facial movements and naturally occurring social contexts across millions of videos. Brooks and colleagues (2024) constructed a multilingual cross-cultural mapping of meaning judgments elicited by participants' produced expressions. These approaches reduce dependence on tiny inventories of stereotyped poses while raising questions about video selection, online population coverage, training annotations, and the interpretation of learned dimensions. Methodological innovation has therefore created a more differentiated evidence landscape: restricted-choice category agreement, situated behavioral co-occurrence, local meaning attribution, and computational prediction each estimate a different relationship. Cowen et al. (2021) Brooks et al. (2024)

Basic Emotion Accounts and Their Comparative Predictions

Basic emotion theories propose that certain emotion families have evolved organizing functions and partially recurrent behavioral signatures. In comparative research they motivate predictions that some expressive patterns, especially when associated with biologically important social events, will be widely recognizable. Ekman and colleagues' early findings supplied evidence consistent with widespread associations between posed facial configurations and predicted labels. Later formulations acknowledge contextual variation, blends, and multiple expressive channels, but the interpretation still depends on what counts as a signature and what probability of occurrence justifies a universal claim. The contemporary synthesis by Keltner, Sauter, Tracy, and Cowen (2019) extends attention to expression beyond a small set of facial prototypes. Ekman et al. (1969) Keltner et al. (2019)

The strongest defensible version of a shared-systems hypothesis predicts recurring constraints across human populations, not a compulsory one-to-one match between every emotional state and one facial posture. Evidence can favor this view when participants understand independently collected expressions at above-chance rates under comparable conditions, or when certain behaviors recur across similar contexts. Elfenbein and Ambady's (2002) meta-analysis reported above-chance cross-cultural recognition as well as an in-group advantage. Recognition above chance and higher accuracy within familiar groups are logically compatible. One indicates the availability of some cross-group information; the other identifies additional cultural calibration, familiarity, or other shared-context benefits. Elfenbein & Ambady (2002)

Critically, an observer's correct choice among a limited set of provided labels may reflect contextual inference, elimination among alternatives, or shared cultural media. A theory claiming universal emotion mechanisms must state how these alternatives are controlled. Naturalistic expression studies are valuable because they sample behavior outside narrowly posed photographs, but behavior-context correspondence does not directly prove an identical felt experience across participants. The resulting program should specify distinct hypotheses for expression morphology, emotion appraisal, communicative intention, social contingency, and recognition. Keltner and colleagues (2019) review growing evidence for rich expressive repertoires; that diversity makes operational precision more important rather than less. Keltner et al. (2019) Russell (1994)

Constructionist, Appraisal, and Relational Accounts

Constructionist approaches emphasize that emotion categories depend on how bodily sensations, prior learning, concepts, and situational predictions are assembled and interpreted. For cross-cultural inquiry, they predict that differences in learned categories and habitual interpretations can change which affective episodes are distinguished and how observers attribute meaning. This prediction is supported by designs in which adding emotion words or changing category prompts alters how expressions are sorted. The Himba and U.S. comparison conducted by Gendron and colleagues (2014) is especially important because it manipulated access to concept cues and found that such cues moved sorting toward the conventional pattern while cultural differences remained. Gendron et al. (2014)

Relational accounts ask how emotions participate in ongoing interactions and reproduce locally valued forms of social life. An episode of pride, shame, anger, or gratitude can regulate distance, restore obligations, challenge a perceived slight, or affirm belonging. These functions depend on cultural models of personhood and relationships, and they can alter the frequency with which emotions are elicited, sanctioned, and talked about. Mesquita and Walker (2003) describe culture-sensitive emotional functioning in terms of prevalent practices and relational expectations, while Mesquita, Boiger, and De Leersnyder (2016) characterize emotion as activity embedded in normative environments. The strongest research combines participants' own descriptions with observations of interpersonal consequences. Mesquita & Walker (2003) Mesquita et al. (2016)

Appraisal approaches focus on evaluations of goal relevance, responsibility, control, certainty, and social significance. A cultural model can change which events matter and which explanations of an event are plausible, leading to different emotion distributions even when appraisal processes have shared properties. Comparing a named emotion without comparing eliciting circumstances can mistake differences in experience opportunities for differences in mechanisms. Mesquita and Frijda (1992) argued that similarities at one stage of emotional processing can coexist with variation at another. The integrative claim is testable: measure the occurrence of events, individual interpretation, expressive action, and relational outcomes in the same study rather than assigning a cultural cause solely from aggregated emotion ratings. Mesquita & Frijda (1992)

Cultural Evolution and Transmission of Emotional Repertoires

Emotion conventions are acquired through repeated encounters with caregivers, peers, institutions, narratives, media, and public rituals. Their persistence depends on social learning, normative feedback, and coordination with the expectations of others. A cultural-evolutionary perspective asks how recurrent emotional categories and display practices emerge, spread, and change as social environments develop. Lindquist and colleagues (2022) describe a framework integrating biological constraints with cultural transmission rather than treating culture as a static category assigned to individuals. Comparative evidence can then be analyzed as a distribution of learned practices with specific pathways of inheritance and modification. Lindquist et al. (2022)

Shared language can coordinate categories and make particular interpretations more accessible, while communication through gesture and prosody carries partially independent conventions. Migration, schooling, bilingualism, and digitized social interaction can introduce competing norms. The socioecological analysis by Niedenthal, Rychlowska, and Wood (2017) identifies historical migration heterogeneity and present social conditions as possible determinants of expressive repertoires. These proposals shift inquiry from national labels toward variables that can vary within countries and across time. Demonstrating a causal pathway, however, requires evidence linking particular social histories to individual learning and present behavior rather than correlation between national-level indices and survey scores. Niedenthal et al. (2017)

The explanatory level matters in historical comparison. A relation between a country's migration history and its average endorsement of expressive behavior is an ecological association, while a filmed interaction shows a situated behavior, and a repeated-measures experiment can identify immediate contextual effects. Rychlowska and colleagues (2015) associated long-history migration heterogeneity with reports of emotional expressivity and smile functions. Such results motivate comparative predictions about social coordination, including whether people who routinely interact with less familiar partners rely on highly legible displays. The possibility remains that measurement equivalence, contemporary demographics, and regional histories explain part of the association. Cultural transmission is a research mechanism that must be measured, not assumed whenever cross-national means diverge. Rychlowska et al. (2015) Lindquist et al. (2022)

Emotion Lexicons, Translation, and Semantic Equivalence

Emotion words differ in their extension, pragmatic uses, evaluative force, and associations with physical states or social situations. A dictionary equivalent such as anger can serve as a useful starting point but is not sufficient evidence of construct equivalence. Researchers must ask whether speakers understand the word as referring to the same prototypical episode, whether it carries comparable moral implications, and whether it is used in the same communicative situations. Wierzbicka (1999) systematically examined the problem of analyzing emotion language across cultures and proposed semantic decomposition as a route to comparison. The central methodological lesson is to preserve local descriptions before imposing a translated coding scheme. Wierzbicka (1999)

Translation can introduce apparent similarity by inserting an investigator's categories into the task. In a forced-choice facial judgment experiment, the labels themselves direct attention to a small set of hypotheses; in an interview, a translated question may suggest that people should report an emotion as a bounded personal possession. A stronger procedure uses bilingual cognitive interviewing, back-translation as one check among several, spontaneous local vocabulary elicitation, and analysis of actual usage in conversations. The test is functional equivalence in context rather than literal similarity of glosses. Gendron and colleagues (2014) illustrate the importance of conceptual cues: adding labels changed the structure of expression sorting. Gendron et al. (2014)

Lexical differences can be informative even when direct translations are unavailable. A language may employ one expression for several experiences that another language differentiates, or it may distinguish social meanings that have no dedicated counterpart elsewhere. Such facts show how communicative categories are organized; they do not alone establish the absence of a physiological capacity or emotional experience. Cross-language measurement should collect narratives of eliciting events, descriptions of bodily and action tendencies, speaker judgments of appropriate usage, and convergent behavior. These methods connect linguistic evidence with emotion episodes without treating words as transparent windows into internal states. Wierzbicka's lexical analysis and Mesquita and Frijda's process model offer complementary methodological resources for doing so. Wierzbicka (1999) Mesquita & Frijda (1992)

Worldwide Emotion Semantics and Colexification Networks

Colexification occurs when a single lexical form expresses multiple concepts within a language. If fear and anxiety are named by the same form, they are connected in that language's lexical system; across languages, networks of such relationships can be compared. Jackson and colleagues (2019) examined colexification patterns in 2,474 spoken languages, a scale that moves beyond the historically dominant comparisons among a few wealthy, industrialized populations. They found substantial differences in emotion semantic networks related to linguistic and geographic proximity while identifying broad organization around valence and activation. Their result provides evidence for shared constraints on semantic structure together with major diversity in the mapping of familiar emotion labels. Jackson et al. (2019)

Lexical colexification is an indirect indicator of conceptual relationships. A shared word may signal related use, historical borrowing, polysemy, or limited differentiation in a particular lexical resource; it is not a real-time measurement of affective experience. Language families are also historically related and geographically clustered, so individual language observations cannot be treated as statistically independent national samples. Jackson and colleagues used network comparisons to study these relationships, but causal inference about why a specific culture distinguishes one emotion requires additional evidence from speakers and social contexts. A well-designed follow-up would connect lexical networks to judgments of stories, spontaneous labeling, and behavioral interpretation within the same speech communities. Jackson et al. (2019)

Valence and activation are attractive comparative dimensions because positive versus negative evaluation and energetic intensity arise across many kinds of episodes. Yet an emotion category also conveys responsibility, relationships, bodily sensations, uncertainty, and moral appraisal, which need not align across languages. The shared structure reported by Jackson and colleagues does not imply semantic identity of anger, fear, or sadness in every language. Conversely, lexical diversity does not rule out common physiological constraints. Distinguishing dimensional overlap from category-specific correspondence helps locate what has been generalized and what remains locally learned. This is the appropriate basis for comparing lexical and observational findings without forcing the entire literature into a binary universality debate. Jackson et al. (2019) Lindquist et al. (2022)

Facial Recognition Paradigms and Their Inferential Limits

The classic recognition design places a facial image in front of an observer and asks which emotion is represented. Some variants supply a short list of words; others allow free labeling, grouping by similarity, or description of what the person might do next. The designs create different response spaces. A label chosen from six alternatives reveals discrimination relative to five distractors, while a free description reveals salient interpretations without presupposing a taxonomic inventory. Russell (1994) showed why response format, stimulus selection, and context affect the inferential force of facial-recognition findings. A rigorous comparative study must report these choices as determinants of the construct measured, rather than merely procedural details. Russell (1994)

The meta-analysis of Elfenbein and Ambady (2002) found recognition performance above chance across cultures and an advantage for observers assessing expressions from their own national, ethnic, or regional group. The in-group advantage was smaller in groups with greater contact, consistent with learning and familiarity as possible explanations. These results challenge both the claim that facial expressions convey no cross-cultural information and the claim that all relevant meanings are decoded equally well everywhere. However, a meta-analysis pools studies whose labels, stimuli, and populations vary; the resulting estimate cannot settle whether particular emotions have identical causal or experiential structures. The analysis supports a graded, context-sensitive account of communicated information. Elfenbein & Ambady (2002)

A coherent measurement model separates sensitivity to expressive features, familiarity with a display repertoire, the decision rule induced by the task, and interpretation of the emotion label. Researchers should evaluate what participants are asked to identify: an internal feeling, a social action, a likely next behavior, or a named category. Comparisons become more informative when participants also rate confidence, alternative plausible meanings, and the situational information that would alter their judgment. Repeated measures across forced choice and free response help estimate the role of task support. Claims about universality should name the level of generalization established by each procedure, a principle sharpened by Russell's review and Gendron's later experimental work. Russell (1994) Gendron et al. (2014)

Evidence From Himba Participants and Conceptual Cue Manipulation

Gendron, Roberson, van der Vyver, and Barrett (2014) compared people in the United States with Himba participants from northwestern Namibia using posed facial images that participants sorted into piles. Without explicit cues to conventional emotion categories, Himba sorting did not reproduce the presumed universal categorical pattern, whereas U.S. sorting more closely did. When cues to emotion concepts were supplied, results in both communities moved nearer the expected organization, although cultural differences persisted. This is especially strong evidence that experimental instructions and conceptual supports can causally affect emotion judgments within the task. The study thereby challenges a simple interpretation of posed-face sorting as a direct readout of preexisting universally shared discrete concepts. Gendron et al. (2014)

The result also raises a more precise theoretical possibility. People may use some shared perceptual and affective information while learning different routines for organizing that information into categories. A lack of identical free sorting does not imply that participants cannot notice similar features or use the same visual cues in another social situation. Similarly, a cue-induced increase in familiar categories may reflect task adaptation, genuine concept activation, or the temporary adoption of the experimenter's framework. To identify which mechanism is operating, future designs should independently assess visual discrimination, lexical knowledge, situation knowledge, and the stability of categorization outside the testing encounter. This preserves the interpretive strength of Gendron and colleagues' intervention. Gendron et al. (2014)

The sample was culturally informative because conventional cross-cultural studies often rely on participants exposed to the same globalized media or formal schooling. Yet remote-population research faces challenges of translation and local collaboration. Sample size, recruitment, participant history, appropriateness of image stimuli, and the relation between interviewer and participant affect generalizability. Rather than treating any individual community as the representative of cultural difference, replication should involve multiple locally grounded populations and a broader repertoire of stimuli. Convergent observations from indigenous communities in Papua New Guinea underscore the value of methodological diversity and locally meaningful interpretation tasks. The goal is a replicable account of specific context-dependent judgments, not a universal claim drawn from one counterexample. Gendron et al. (2014) Crivelli et al. (2016)

Melanesian Research and the Meanings of Gasping Faces

Crivelli and colleagues examined how facial configurations familiar from Western experimental inventories are understood in communities in Papua New Guinea. Their 2016 work on facial reading in two indigenous societies questioned the interpretation of conventional expression photographs as fixed emotion signals. Another study found that a gasping facial display conventionally associated with fear could function as a threat display in a Melanesian society. These findings demonstrate that outward resemblance between configurations and photographs does not guarantee agreement about communicative intent. The contrast is conceptually important because a display can be action-oriented, addressing another person, while conventional recognition inventories ask observers to attribute a private internal state. Crivelli et al. (2016) Crivelli et al. (2016)

A separate study investigated recognition of spontaneous expressions produced by residents of Papua New Guinea and interpreted by other residents of the same society. Local expression samples help address concerns that artificial photographs reflect a Western theory built into the stimuli. However, spontaneous behavior is more variable and sometimes more ambiguous than a carefully posed expression; the analysis must retain the event and relationship in which a behavior occurred. A culturally specific interpretation of one gasping face should not be generalized to all expressions of fear, all Melanesian populations, or every encounter within a studied community. Careful explanation identifies the display, audience, situation, and participant judgments to which the conclusion applies. Crivelli et al. (2017) Crivelli et al. (2016)

The broader lesson is methodological symmetry. A posed photograph made in one population is often treated as a neutral research instrument, while an interpretation supplied by another population is treated as cultural deviation. An equitable comparative design recognizes that both production and interpretation are culturally situated; the source population's response is also a hypothesis, not an unquestioned criterion. Multiple groups should produce and judge locally collected expressions, with equivalent opportunities to define meaningful categories. Such a reciprocal design can distinguish differences in what people produce from differences in how they use available evidence to infer meaning. It also makes room for genuine cross-group regularities to emerge independently of a fixed Western label set. Crivelli et al. (2016) Gendron et al. (2014)

Vocalizations, Prosody, and Multimodal Emotion Communication

Emotion is communicated by more than faces. Laughter, cries, sighs, exclamations, speech rhythm, posture, gesture, and coordinated activity provide overlapping information about bodily arousal and interpersonal intent. Sauter, Eisner, Ekman, and Scott (2010) studied cross-cultural recognition of nonverbal emotional vocalizations, producing evidence that some vocal signals can be identified across cultural boundaries. Their work expands the empirical basis for testing recurring patterns of expression and makes clear that the generalization target may differ among positive and negative states or among kinds of vocal signal. Recognition success under a matching task remains a result about that task's available options and acoustic materials. Sauter et al. (2010)

A vocalization's acoustic properties can support interpretation even when an emotion word is difficult to translate. Pitch contour, duration, intensity, and temporal organization may indicate urgency or arousal; laughter and crying also have patterned social uses. A culturally comparative study can use sound-only, face-only, and audiovisual conditions, estimating what each channel contributes and whether modalities interact. It should examine production in natural exchanges as well as perception under controlled presentation. Keltner and colleagues (2019) emphasize the breadth of expressive modalities and the value of going beyond a few prototypical facial movements. The result is a multi-channel research design suited to communication rather than a single-trait decoding exercise. Keltner et al. (2019) Sauter et al. (2010)

Everyday communication also involves strategic and relational use of channels. A polite smile may accompany words that conceal frustration; a vocal hesitation may express deference rather than fear; silence may be meaningful when speaking risks challenging authority. Cultural expectations shape what observers attend to and what actions are available to express an emotion. The cultural display-rule literature provides an account of how internal states and outward behavior can diverge in systematic circumstances. As a result, researchers should resist treating a multimodal machine prediction as a direct measure of felt state until its criterion has been independently validated. What a model predicts is usually a label supplied by people, a behavior pattern, or the context in which expression occurs. Matsumoto et al. (2018) Niedenthal et al. (2017)

Large-Scale Computational Findings and Their Scope

Cowen, Keltner, Schroff, Jou, Adam, and Prasad (2021) applied machine-learning methods to approximately six million videos from 144 countries. Their experiments examined associations between 16 types of facial expression and naturalistic contexts, reporting that the pattern of context associations was approximately 70 percent preserved across 12 world regions. This is important evidence that some forms of facial movement have broadly recurring situated associations. The outcome was an expression-context correspondence estimated from video data, not a universal measurement of inner emotional experiences or a demonstration that every instance of each context elicits the same response. The population of uploaded videos also differs from a random sample of human activity. Cowen et al. (2021)

Brooks and colleagues (2024) recruited 5,833 participants across six countries. Participants produced facial expressions based on naturalistic images and rated expressions in their languages across 48 emotions and mental states, yielding 423,193 participant-generated expressions. The computational analysis identified 28 dimensions of facial meaning, 21 with strong evidence for cross-cultural consistency and others with more specific patterns. The design is particularly valuable because it measures gradients of meaning rather than restricting observations to six labels, although its interpretation depends on participant demographics, translated rating inventories, model architecture, and the facial movements sampled. These are substantive dimensions that an independent replication should evaluate. Brooks et al. (2024)

Large corpora improve breadth but introduce new forms of dependency. Online images and videos are selected by platform participation, algorithmic visibility, access to devices, and norms governing what people record. Automated classifiers may inherit labels or datasets drawn disproportionately from a few populations; a high predictive score does not prove the causal generation of an expression or an internal state. Reports should disclose training and testing splits, person-level leakage controls, cultural sampling weights, annotator language, calibration, and the uncertainty of predicted meanings. Brooks and colleagues reported commercial affiliations and interests relevant to the method, which should be considered in independent evaluation without dismissing the data. Combining large-scale patterns with local ethnography and preregistered experimental tests offers a stronger research strategy. Brooks et al. (2024) Cowen et al. (2021)

Display Rules, Audience, and Expressive Management

An emotion display is regulated within a social encounter. A person may intensify a smile when greeting a guest, restrain visible disappointment before a superior, or maintain a neutral face while experiencing excitement. Display rules are learned expectations about appropriate expression in particular relationships and settings. Their scientific importance is that observable behavior is partly determined by who is present, what outcome the actor seeks, and which actions are socially sanctioned. Matsumoto and colleagues (2018) examined relationships among cultural display rules, emotion regulation, and judgments of expression intensity in American and Japanese participants. The research shows why observers' estimates of inner feeling cannot be separated automatically from their own expectations about expressive management. Matsumoto et al. (2018)

Researchers can test display rules through self-reported norms, hypothetical audiences, expressive tasks, and naturalistic interactions. Each method samples a different level of evidence. A questionnaire asking what one should express may index a norm even when actual behavior deviates under pressure. A laboratory expression task may capture compliance with instructions more than habitual conduct. An observational study can document choices but has limited access to subjective intention. Convergence is strongest when researchers manipulate audience presence and relationship while also recording participants' interpretations of the encounter. The broader framework of Mesquita and Frijda (1992) treats regulation and expression as distinct phases whose cultural organization may differ. Mesquita & Frijda (1992)

Display management can modify the apparent relation between spontaneous experience and the cues available to an observer. This matters for interpersonal research, emotion-recognition benchmarks, and automated inference. An algorithm trained to detect smiling movements may correctly recognize a behavior while incorrectly labeling the person's ongoing state. Similarly, an employee who conceals irritation may be acting according to institutional norms rather than experiencing low arousal. Scientific interpretation should specify which behavior is measured, which contextual pressures are present, and whether feeling was assessed independently. Cultural display rules are therefore an explanatory variable, a potential moderator, and a challenge to unqualified claims that expression transparently reveals emotion. Matsumoto et al. (2018) Niedenthal et al. (2017)

Emotion, Self-Construal, and Relational Goals

Cultural models of the self assign different salience to independence, interdependence, status, obligation, and group harmony. These models can alter which events elicit an emotional response, the value assigned to an episode, and its implications for relationships. Kitayama, Markus, and Kurokawa (2000) compared Japanese and American college students and reported that associations between positive feelings and forms of interpersonal engagement differed by setting. The investigation illustrates how generic pleasantness ratings can conceal distinct social meanings. The findings are associations in specific student samples and should be interpreted within the study's recruitment, measures, and period rather than as permanent descriptions of every person in either country. Kitayama et al. (2000)

An emotion category can be embedded in a model of responsibility and interpersonal action. Pride may communicate achievement and agency in one setting but risk violating expectations of modesty in another; anger may mark justified boundary enforcement or inappropriate relational disruption. These possibilities require direct measures of event appraisal, audience response, and anticipated outcomes rather than an assumption that national averages reveal normative causes. Mesquita and Walker (2003) argue that emotions become prevalent when they fit and reinforce cultural patterns of self and relationship. The framework can be tested with within-person episodes across varied partners and through comparisons of individuals who inhabit more than one normative setting. Mesquita & Walker (2003)

Emotional experience and relational consequence are mutually informative. A moment of embarrassment can prompt repair, a public expression of gratitude can confirm dependence and reciprocity, and resentment can motivate changes to an unequal arrangement. A culturally adequate theory predicts which interactions make such responses functional, when those functions fail, and how participants distinguish moral judgment from felt intensity. Mesquita, Boiger, and De Leersnyder (2016) summarize research on emotional experiences as participation in social worlds. Their formulation suggests that the culturally consequential unit is often an episode distributed across participants rather than an individual's isolated feeling score. This does not prevent individual measurement; it specifies the social conditions needed to interpret it. Mesquita et al. (2016)

Ideal Affect, Desired Feeling, and Emotional Values

Actual affect denotes the states people report experiencing; ideal affect denotes the states they would like to experience. The constructs can be related yet empirically distinguishable. Tsai, Knutson, and Fung (2006) proposed that cultural influences may be especially prominent for ideal affect. Their work compared valued affective states across groups and examined how desired feeling differs from current feeling. This distinction provides a mechanism by which a cultural norm can influence choices and expressive practices even when everyday physiological tendencies overlap. A community may favor energetic enthusiasm in social display, while another rewards calm composure, without either requiring one uniform actual state across its members. Tsai et al. (2006)

Affect Valuation Theory connects desired feeling to cultural goals and behavioral consequences. Tsai (2007) described contrasts in the valuation of high-arousal positive states, such as excitement, and low-arousal positive states, such as calm, in American and East Asian settings. Tsai and colleagues (2007) further examined influence and adjustment goals as sources of these differences. This research introduces testable pathways from social priorities to affective ideals and subsequently to choices about people, activities, and expressive style. Cross-group effects should not be interpreted as fixed personality attributes: variation in age, social role, and exposure to different cultural institutions can shift ideal affect. Tsai (2007) Tsai et al. (2007)

Affect ideals also affect what counts as an appropriate response to another person's emotion. A highly animated display may be perceived as sincere enthusiasm where visible excitement is valued, yet as excessive in a context favoring restraint. To identify this mechanism, researchers should measure desired and actual states separately and evaluate whether ideal affect mediates behavioral preferences. They should additionally distinguish cultural norms from selection effects, because participants may choose environments and peers that confirm existing preferences. The research of Tsai and colleagues makes this distinction central to the study of cultural emotion differences and provides a useful framework for intervention and institutional design without assuming that one ideal state is universally desirable. Tsai et al. (2006) Tsai (2007)

A substantially larger test was provided by Tsai and colleagues (2025), whose individual participant data meta-analysis aggregated more than 31,000 participants across 124 datasets assembled by independent research teams. The study found that ideal affect and actual affect remain empirically distinct and that cultural differences in desired feeling were larger than differences in reported actual feeling. It also found change over time, including increasing European American valuation of low-arousal positive states, and associations with age, gender, and socioeconomic position. These are aggregate, observational findings rather than random-assignment evidence that culture alone causes affect preferences. The authors' results refine earlier formulations by identifying both recurring cultural patterns and within-group heterogeneity across time and context. Tsai et al. (2025)

Emotion Regulation and the Social Meaning of Control

Emotion regulation comprises processes that influence which emotional episodes arise, their timing, their intensity, and their expression. A cultural comparison requires specifying whether the outcome is felt experience, outward display, social appropriateness, or psychological well-being. Suppression of a visible expression and deliberate reappraisal of an event are different processes with different potential consequences. Ford and Mauss (2015) review evidence that cultural evaluations of emotion influence both regulation and its relation to well-being. The same outward behavior may serve different interpersonal functions depending on whether it expresses self-control, deference, avoidance, or strategic concealment. Ford & Mauss (2015)

Regulation research often imports questionnaires developed in one language and assumes items have the same meaning elsewhere. Yet statements about controlling emotions can refer to controlling bodily feeling, preventing hurtful actions, or managing public display. Researchers should test cognitive equivalence of prompts and examine the difference between endorsement of a norm and actual regulation in episodes. Matsumoto and colleagues (2018) found that beliefs about expressive management were related to how observers inferred emotional experience from faces. This relationship illustrates an interpretive loop: learned regulation expectations shape judgments of people, and those judgments can reinforce the expectations that guide later displays. Matsumoto et al. (2018)

Normative fit can matter for welfare. Mesquita, Boiger, and De Leersnyder (2016) review findings that experiencing culturally normative emotions is associated with better well-being in studied settings, while Ford and Mauss (2015) emphasize variation in the consequences of emotion regulation across cultural contexts. Such associations require careful causal interpretation. Individuals with greater social support may simultaneously report more normative emotions and better functioning; the norm itself need not be the only explanation. Future work should assess within-person regulation processes, relationship outcomes, baseline distress, and changes across time. Cultural sensitivity in the interpretation of emotion is a methodological requirement rather than an assumption that a single regulation strategy benefits all groups. Mesquita et al. (2016) Ford & Mauss (2015)

Historical Migration, Social Ecology, and Expressive Repertoires

A socioecological explanation identifies measurable features of social environments that may shape expression. Historical migration heterogeneity, residential mobility, and frequent interaction among strangers can create distinctive demands for legible communication. Rychlowska and colleagues (2015) analyzed relationships between long-history migration heterogeneity and reports about expressive behavior and the functions of smiles. The account proposes that communication among people who lack shared conventions can favor displays whose interpersonal purpose is broadly understandable. This offers an explanatory pathway linking historical population processes with current communicative norms and complements models built around broad national culture labels. Rychlowska et al. (2015)

The empirical challenge is distinguishing history from correlated contemporary factors. A country with a heterogeneous migration history may also differ in urbanization, economic conditions, contact with global media, and institutional incentives for public interaction. National aggregate analyses can generate hypotheses but cannot establish how an individual acquires a smile repertoire. Niedenthal, Rychlowska, and Wood (2017) describe socioecological variables as tractable directions for explaining variation. Strong tests would measure actual interaction networks, exposure to strangers, community display expectations, and behavioral choices within multiple regions. Experimental manipulation of audience familiarity could provide evidence for immediate mechanisms without pretending to manipulate historical migration itself. Niedenthal et al. (2017)

Ecological variation also cautions against attributing all differences to stable mental dispositions. Two people with similar affective capacities may behave differently because one setting rewards emotional visibility and another penalizes it. Likewise, social and professional organizations can induce new display conventions within a single country. A multilevel account should include individuals, situations, institutions, communities, and historical trajectories, estimating variance at each level instead of relying on an undifferentiated cultural score. The cultural evolutionary synthesis of Lindquist and colleagues (2022) suggests that socially transmitted patterns interact with human learning and biological constraints. Such a model is compatible with the observed possibility of both wide cross-group regularities and local specialization. Lindquist et al. (2022) Rychlowska et al. (2015)

Developmental Learning and Intergenerational Transmission

Cultural meanings of emotion are learned through everyday feedback. Children encounter adult labels for feelings, observe which displays receive approval, and discover how emotion talk affects peer relations and authority. The acquisition of these practices is likely to depend on participation in specific linguistic and social environments rather than nationality alone. The comparative literature reviewed by Mesquita and Frijda (1992) and the cultural-evolutionary approach outlined by Lindquist and colleagues (2022) provide theoretical grounds for studying the development of local emotion categories and expressive contingencies. This chapter treats such developmental pathways as research hypotheses unless a particular age-specific experiment supports a narrower claim. Mesquita & Frijda (1992) Lindquist et al. (2022)

A longitudinal design could examine children's vocabulary for emotions, spontaneous interpretations of social incidents, expressive movements, and parental commentary in repeated observations. Researchers should determine whether apparent age differences result from growing linguistic competence, changing familiarity with the tasks, increased knowledge of local social rules, or changes in affective experiences. Comparing age groups on a translated adult emotion inventory is especially likely to confound these processes. Cross-generational comparisons within migrant or bilingual households could reveal how competing conventions are learned and negotiated. The principle follows from Wierzbicka's emphasis on language-specific meanings and the finding that emotion concepts influence judgment tasks. Wierzbicka (1999) Gendron et al. (2014)

The socially acquired character of emotion repertoires also implies temporal change. Schools, occupational institutions, and digital media can alter communicative practices over a generation; new emotion vocabulary can diffuse rapidly while older interpersonal conventions remain in place. To understand change, investigators should connect historical documentation, participant narratives, and direct observations rather than inferring transitions from a single cross-sectional national comparison. A cultural-evolutionary framework makes transmission channels explicit and allows the researcher to ask when variants spread through imitation, normative sanction, institutional reinforcement, or communicative utility. Longitudinal evidence is needed to distinguish these mechanisms and establish durable cultural transformation. Lindquist et al. (2022) Niedenthal et al. (2017)

Migration, Bilingualism, and Emotion in Multiple Cultural Worlds

People frequently participate in more than one linguistic and cultural environment. Migration and bilingualism make this variation observable within individuals and across relationships. A bilingual speaker may possess different lexical distinctions in two languages and different expectations about expressing disappointment to family, colleagues, and strangers. These are hypotheses about context-sensitive performance; they should be tested through repeated episodes and matched tasks rather than inferred from language background alone. The semantic research of Wierzbicka (1999) and the comparative networks of Jackson and colleagues (2019) show why translating one emotion word cannot be presumed to preserve every distinction made available by another language. Wierzbicka (1999) Jackson et al. (2019)

An individual-level design can compare emotion descriptions collected in each language, social judgments under different audience cues, and episodes recorded close to their occurrence. The analysis should measure language proficiency, dominance, social networks, length of residence, and the meaning of the context to each participant. Counterbalancing task language reduces order effects, while interviewing about lexical connotations can identify concepts with only partial overlap. Because individuals may self-select into different social environments, repeated observations and natural experiments may be more informative than simple contrasts between migrant and nonmigrant groups. The core outcome is the flexibility and stability of emotional categorization and expression under identifiable contextual changes. Jackson et al. (2019) Mesquita et al. (2016)

The social consequences of emotional flexibility can include misunderstanding when an expression is interpreted under a different local norm. A signal intended to show respect may be read as disengagement; visible enthusiasm may be judged as competence or as intrusion. These possibilities should be evaluated through actual interaction, ideally using accounts from both participants and measures of subsequent relationship behavior. Ford and Mauss (2015) provide a framework for examining how norms shape emotion regulation and its experienced consequences. The goal is a process-sensitive account of cultural participation that can explain differences within a person as well as similarities between people living in different countries. Ford & Mauss (2015) Mesquita et al. (2016)

Construct Validity, Emic–Etic Designs, and Measurement Invariance

Cross-cultural emotion research requires a defensible account of the construct measured. An etic instrument begins with categories selected for broad comparison; an emic strategy elicits locally meaningful distinctions and uses those materials to interpret behavior. The approaches can be combined rather than treated as incompatible. Researchers might first gather narratives and vernacular concepts, construct locally relevant stimuli, and then test a common dimensional model across populations. Wierzbicka's (1999) semantic analysis and Jackson and colleagues' (2019) large language comparison demonstrate how local lexical organization and comparative structure can inform one another. Measurement validity depends on the adequacy of both levels. Wierzbicka (1999) Jackson et al. (2019)

An instrument has to support the comparison being made. Linguistic translation alone cannot show that a scale has equal factor structure, item interpretation, response thresholds, or relation to behavior across groups. Apparent differences in emotion frequency can emerge from differing concepts of frequency, willingness to disclose, response styles, or the events participants imagine when answering. A rigorous protocol should document translation decisions, cognitive interviews, factor or item-level invariance tests where suitable, and convergent evidence from behavior and contextual reports. Mesquita and Frijda (1992) emphasized that cultural similarities and differences depend partly on the level at which emotion phenomena are described. That insight has direct psychometric consequences. Mesquita & Frijda (1992)

For a facial-expression dataset, construct validity additionally requires specifying whether the criterion is the actor's reported feeling, a judge's emotion attribution, an objectively coded muscular movement, or the social context. Brooks and colleagues' (2024) multidimensional work predicts locally rated meanings; Cowen and colleagues' (2021) naturalistic analysis studies expression–context associations. Their outcomes are related but nonidentical. Combining them without a common criterion would create spurious agreement or conflict. Researchers should publish a construct map connecting each observed variable to the inference they intend to make and disclose situations in which the link is expected to fail. Brooks et al. (2024) Cowen et al. (2021)

Sampling, Statistical Dependencies, and Generalizability

The word culture often stands in for a nation or language, but neither is a statistically independent unit of psychological life. Recruitment by university, online platform, or rural field site selects people with specific histories. Samples within one language family may share historical vocabulary, and neighboring societies may have overlapping expressive practices. Comparative estimates should therefore describe participant recruitment, language competence, age, gender and socioeconomic distribution where relevant, and the degree of exposure to shared media. Jackson and colleagues' (2019) linguistic network analysis makes historical and geographic relatedness analytically visible; cross-cultural experiments likewise need a defensible structure of dependencies. Jackson et al. (2019)

The effective evidentiary population is frequently narrower than the number of countries in a dataset. Cowen and colleagues (2021) analyzed videos from 144 countries, yet the sample consists of videos encountered under specific technological and cultural filters. Brooks and colleagues (2024) studied six countries through a structured production and rating protocol, yielding extensive observations but not an unrestricted sample of all human emotion communication. Participants nested within countries, repeated expressions from the same participant, and shared stimuli create dependencies requiring appropriate estimation. Reported precision should reflect the actual sampling design instead of treating every recorded image as an independent person. Cowen et al. (2021) Brooks et al. (2024)

At least three generalization levels should be distinguished: transfer to another stimulus set, transfer to another social environment in the same community, and transfer to another population. A result can be replicable under a fixed task while failing under open-ended interpretations, and such a failure can inform the mechanism. Elfenbein and Ambady (2002) found evidence of recognition across communities and group-specific advantages, supporting the idea of graded generalizability. Replication programs should preregister which level is being evaluated, collect comparable but locally meaningful stimuli, and analyze uncertainty both within and between groups. This makes the scope of a claim visible rather than hiding it in the adjective universal. Elfenbein & Ambady (2002)

Causal Inference and the Separation of Competing Explanations

Cultural comparison by itself identifies associations between group membership and observed outcomes. Possible mechanisms include language, normative expectations, prior exposure, social structure, appraisal conventions, and the experimental procedure. To distinguish them, investigators must manipulate what can be ethically changed or exploit natural variation that identifies a particular pathway. Gendron and colleagues (2014) provide an example by experimentally varying emotion-concept cues in a sorting task. The intervention demonstrated a causal effect of instructions within the laboratory; it did not experimentally assign participants to cultures. The distinction clarifies the strength of the result while preserving its importance to theory. Gendron et al. (2014)

Other mechanisms invite targeted interventions. Researchers can vary audience familiarity or the perceived social consequences of an expression, present competing situational contexts alongside a face, or manipulate the language used for judging an episode. Such experiments should predefine what result would discriminate accounts. A norm-learning model predicts changes with exposure and audience; a fixed-expression model predicts greater robustness to those manipulations for certain signals; a lexical-construction account predicts cue-sensitive category choices. Theoretical frameworks reviewed by Keltner and colleagues (2019), Mesquita and colleagues (2016), and Lindquist and colleagues (2022) offer distinct hypotheses that become most informative when tested against the same data. Keltner et al. (2019) Mesquita et al. (2016) Lindquist et al. (2022)

Causal analysis also requires avoiding adjustment for consequences of the phenomenon under study. For example, self-reported display rules may both shape and reflect workplace experiences; conditioning on a workplace variable can distort the relation if that workplace was itself selected partly because of expressive style. Longitudinal data and multi-level designs help establish sequences, but they do not guarantee causal identification. The ecological model of Rychlowska and colleagues (2015) makes historical history a predictor of current expressive norms; evaluating that pathway calls for mediation studies, sensitivity analyses, and independent within-community measures. A scientifically responsible conclusion states the level of causal support, not simply whether a cultural contrast is statistically significant. Rychlowska et al. (2015)

Reconciling Apparently Conflicting Findings

The evidence often appears divided between widespread expression patterns and culturally distinctive emotional meanings. Much of this division reflects different estimands. Ekman, Sorenson, and Friesen (1969) studied selections among predicted categories for posed faces; Gendron and colleagues (2014) studied sorting behavior under altered concept cues; Cowen and colleagues (2021) studied expression–context correspondence; and Brooks and colleagues (2024) modeled dimensions of attributed meaning in multiple countries. These studies measure different conditional relations. One can establish recurring behavior–context patterns without establishing identical semantic boundaries, and another can establish semantic variability without denying shared perceptual information. Ekman et al. (1969) Gendron et al. (2014) Cowen et al. (2021) Brooks et al. (2024)

A more informative synthesis treats the evidence as a set of constraints. Human faces and voices have physical properties; social interaction creates recurring communicative problems; local learning establishes conventional interpretations; and different tasks reveal distinct parts of the system. The cross-cultural recognition meta-analysis reported both above-chance performance and an in-group advantage, while the 2,474-language semantic study found both common higher-level organization and differences in emotion word relationships. These findings support a layered explanation in which partial regularities coexist with local specialization. To evaluate whether the layers are causally independent, the same participants need to complete perceptual, semantic, and interactional measures. Elfenbein & Ambady (2002) Jackson et al. (2019)

Certain conflicts remain substantive. Whether a specific facial configuration conveys a particular emotional feeling without contextual support cannot be resolved merely by statistical agreement with a list of words. Research in small-scale societies challenges fixed interpretations of some conventional photographs, while large computational studies document recurring mappings in online and controlled samples. Neither design exhausts the expressive repertoire of humanity. The remaining question is where the cross-group correspondence resides: facial movement, social function, inferred meaning, learned category, or felt episode. A cumulative field should report those distinctions explicitly and allow empirical results to revise the boundaries of its categories. Crivelli et al. (2016) Crivelli et al. (2017) Brooks et al. (2024)

Cross-Cultural Emotion Science in Education and Human Services

Teaching about emotion can benefit from distinguishing the biological and social conditions of affect from the local conventions used to name and display it. Students may encounter claims that certain facial configurations reveal fear or anger automatically. The research examined here supports a more precise account: expressive cues can carry recognizable information, yet their meanings and frequency depend on task, audience, and culture. Educational materials should compare multiple modalities and show how differing contextual prompts alter interpretations. Russell's methodological review and Elfenbein and Ambady's meta-analysis provide complementary examples of why recognition findings require careful interpretation. Russell (1994) Elfenbein & Ambady (2002)

In counseling, education, and organizational interaction, an individual's emotion words and expressive behavior require contextual inquiry. A clinician or educator may need to ask what a locally meaningful term refers to, which event preceded an episode, and what expression was appropriate in the relationship. Mesquita and Walker (2003) outlined why emotional disturbance assessments benefit from cultural context, while Ford and Mauss (2015) reviewed the relation between cultural norms, regulation, and well-being. These sources justify interpretive sensitivity but do not establish that any particular treatment technique is effective for a given population. Clinical decisions require separate diagnostic evidence, validated interventions, and qualified professional judgment. Mesquita & Walker (2003) Ford & Mauss (2015)

Intercultural misunderstandings can also affect workplace and institutional evaluations. A judge who interprets restrained facial behavior as low motivation, or animated expression as disrespect, may be applying a familiar norm to someone using another convention. Research should test such biases with scenarios that manipulate context and with samples whose own interpretive frameworks are documented. Training that merely teaches a different list of stereotypes could reproduce the same measurement error. The display-rule findings of Matsumoto and colleagues (2018) and socioecological work by Niedenthal and colleagues (2017) instead suggest attention to situational cues, audience, and alternative meanings. Application should be evidence-sensitive and context-specific. Matsumoto et al. (2018) Niedenthal et al. (2017)

Artificial Systems, Emotion Inference, and the Artificial Era

The Artificial Era raises a concrete methodological question: what can an artificial system infer from culturally variable human expressions? Automated systems can classify facial movements, acoustic patterns, text, or the contexts in which behavior occurs. They may perform well on a benchmark yet remain uncertain about the person's intended meaning or experienced state. Cowen and colleagues (2021) demonstrate the scale at which algorithms can identify patterned expression–context associations; Brooks and colleagues (2024) show how models can recover dimensions of meaning attributed by participants from several countries. Those achievements concern predictive mappings to defined labels and observations. They do not constitute an independent measure of subjective experience. Cowen et al. (2021) Brooks et al. (2024)

Cross-cultural model evaluation should document the cultural and linguistic origins of training labels, class prevalence, demographic balance, performance calibration, and error distributions. A model trained on emotion words from one language may impose category boundaries absent from another; a system trained on posed smiles may misread strategic politeness. Better evaluation includes multilingual elicitation, multiple possible interpretations, uncertainty reporting, and locally participatory assessment of harms. The distinct findings of Jackson and colleagues (2019) and Gendron and colleagues (2014) explain why category translation and prompt design are especially consequential. Cross-cultural validity must be tested before an algorithm is used to make high-stakes judgments about a person's intentions or welfare. Jackson et al. (2019) Gendron et al. (2014)

Human–artificial interaction adds further questions about how people adapt their emotional communication to a system that produces plausible social responses. An artificial agent may elicit greater self-disclosure or emotional expressiveness in one context than another, yet those outcomes require targeted measurement and cannot be inferred from ordinary face-recognition studies. The broader conceptual framework of the Artificial Era can organize prospective work on communication between human cultural systems and nonbiological information-processing systems. The present evidence supports cautious functional hypotheses about interpretation and interaction rather than empirical claims that artificial agents possess specific internal emotions. The central research need is transparent operational definition of each system's behavior and the social effect of its output. Lindquist et al. (2022) Keltner et al. (2019)

Research Priorities, Replication, and Open Problems

A strong next generation of comparative research needs reciprocal stimulus production and recognition. Participants in each community should create locally meaningful facial, vocal, verbal, and contextual materials, and groups should then interpret both familiar and unfamiliar examples. This design can reveal whether difficulty lies in an expression's production, the observer's interpretation, or the imposed category list. It should combine free descriptions with standardized ratings, retain the original-language responses, and test the effects of cues as experimentally controlled conditions. Gendron and colleagues' task manipulation and the Papua New Guinea investigations supply methodological reasons for this research design. Gendron et al. (2014) Crivelli et al. (2016)

Another priority is integrating language, episodes, and social outcomes within the same longitudinal sample. Emotion lexicons can be mapped using semantic elicitation, while experience sampling can record situations and feelings close to occurrence and observations can document interpersonal behavior. Investigators can then test whether a category difference predicts distinct appraisals, practices, or reactions after controlling for exposure and demographic factors. Jackson and colleagues (2019) provide global hypotheses about language structure; Mesquita and colleagues (2016) provide a socially embedded conception of emotional episodes. Their combination becomes productive when theoretical distinctions are operationalized and assessed through independent methods rather than treated as mutually exclusive slogans. Jackson et al. (2019) Mesquita et al. (2016)

Reproducibility requires accessible materials, preregistered confirmatory contrasts, justified exclusions, independent validation samples, and a transparent record of translation and recruitment. Large computational findings should be replicated on other platforms and in offline settings; field studies should be repeated across several communities and investigators; historical migration hypotheses need contemporary micro-level mediation measures. Questions that remain open include the causal role of emotion vocabulary, the social transmission of specific display norms, when cross-group recognition is resilient to semantic changes, and how ideal affect affects actual relational outcomes. These are questions about distinct mechanisms and identifiable boundary conditions. The best scientific progress will come from designs that allow competing accounts to make divergent predictions on shared outcomes. Brooks et al. (2024) Rychlowska et al. (2015) Tsai et al. (2006)

Conclusion

The research establishes a complex and increasingly measurable relationship between culture and emotion. Human communities exhibit regularities in expressive behavior and affective organization, and they also differ in lexical distinctions, preferred feelings, audience-dependent display practices, and the meanings attributed to visible and vocal behavior. Early cross-cultural recognition studies, meta-analysis, free-sorting experiments, indigenous-community investigations, large semantic networks, and computational corpora are jointly informative once the constructs each observes are distinguished. The evidence supports scientific questions at several levels: bodily and expressive repertoires, conceptual structure, social purpose, and culturally learned interpretation. None of these levels can stand as an unqualified substitute for the others. Elfenbein & Ambady (2002) Jackson et al. (2019) Brooks et al. (2024)

The central methodological result is that a cultural contrast must be described at the level at which it was obtained. Translation choices, forced-choice labels, posed photographs, sampling histories, audiences, and computational annotations can change what an experiment measures. Shared dimensions such as valence and activation coexist with culturally patterned semantic organization; recurring expression-context associations coexist with cases of divergent local meaning. The resulting synthesis favors carefully specified, layered hypotheses over broad labels. It identifies processes of learning, regulation, social interaction, and cultural transmission as promising explanations whose specific causal contributions must be demonstrated with designs that separate them. Gendron et al. (2014) Cowen et al. (2021) Lindquist et al. (2022)

A mature science of cultural emotion variation should follow emotional episodes across language, action, relationships, and time. It should study populations as participants in changing cultural worlds, test the portability of instruments, and ask when similarity is robust across independent methods. The practical significance lies in better comparative theory, fairer assessment, more informed intercultural communication, and better-calibrated technologies of emotion inference. Within Psychology by Bogdanova, this dossier supplies the cross-cultural research bridge between Emotion and Motivation and Cultural and Cross-Cultural Psychology, while neighboring dossiers examine general emotion mechanisms, cultural cognition, language, and regulation in depth. The scientific program remains open to future evidence and independent replication. Mesquita & Frijda (1992) Wierzbicka (1999) Ford & Mauss (2015)

References

  • Brooks, J. A., Kim, L., Opara, M., Keltner, D., Fang, X., Monroy, M., Corona, R., Tzirakis, P., Baird, A., Metrick, J., Taddesse, N., Zegeye, K., & Cowen, A. S. (2024). Deep learning reveals what facial expressions mean to people in different cultures. iScience, 27(3), 109175. https://doi.org/10.1016/j.isci.2024.109175
  • Cowen, A. S., Keltner, D., Schroff, F., Jou, B., Adam, H., & Prasad, G. (2021). Sixteen facial expressions occur in similar contexts worldwide. Nature, 589, 251–257. https://doi.org/10.1038/s41586-020-3037-7
  • Crivelli, C., Jarillo, S., Russell, J. A., & Fernández-Dols, J.-M. (2016). Reading emotions from faces in two indigenous societies. Journal of Experimental Psychology: General, 145(7), 830–843. https://doi.org/10.1037/xge0000172
  • Crivelli, C., Russell, J. A., Jarillo, S., & Fernández-Dols, J.-M. (2016). The fear gasping face as a threat display in a Melanesian society. Proceedings of the National Academy of Sciences, 113(44), 12403–12407. https://doi.org/10.1073/pnas.1611622113
  • Crivelli, C., Russell, J. A., Jarillo, S., & Fernández-Dols, J.-M. (2017). Recognizing spontaneous facial expressions of emotion in a small-scale society of Papua New Guinea. Emotion, 17(2), 337–347. https://doi.org/10.1037/emo0000236
  • Ekman, P. (1994). Strong evidence for universals in facial expressions: A reply to Russell's mistaken critique. Psychological Bulletin, 115(2), 268–287. https://doi.org/10.1037/0033-2909.115.2.268
  • 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
  • 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
  • Ford, B. Q., & Mauss, I. B. (2015). Culture and emotion regulation. Current Opinion in Psychology, 3, 1–5. https://doi.org/10.1016/j.copsyc.2014.12.004
  • Gendron, M., Roberson, D., van der Vyver, J. M., & Barrett, L. F. (2014). Perceptions of emotion from facial expressions are not culturally universal: Evidence from a remote culture. Emotion, 14(2), 251–262. https://doi.org/10.1037/a0036052
  • Jackson, J. C., Watts, J., Henry, T. R., List, J.-M., Forkel, R., Mucha, P. J., Greenhill, S. J., Gray, R. D., & Lindquist, K. A. (2019). Emotion semantics show both cultural variation and universal structure. Science, 366(6472), 1517–1522. https://doi.org/10.1126/science.aaw8160
  • Keltner, D., Sauter, D., Tracy, J., & Cowen, A. (2019). Emotional expression: Advances in basic emotion theory. Journal of Nonverbal Behavior, 43, 133–160. https://doi.org/10.1007/s10919-019-00293-3
  • Kitayama, S., Markus, H. R., & Kurokawa, M. (2000). Culture, emotion, and well-being: Good feelings in Japan and the United States. Cognition and Emotion, 14(1), 93–124. https://doi.org/10.1080/026999300379003
  • Lindquist, K. A., Jackson, J. C., Leshin, J., Satpute, A. B., & Gendron, M. (2022). The cultural evolution of emotion. Nature Reviews Psychology, 1, 669–681. https://doi.org/10.1038/s44159-022-00105-4
  • Matsumoto, D., Choi, J., Hirayama, S., Domae, A., & Yamaguchi, S. (2018). Culture, display rules, and emotion judgments. Psychology: The Journal of the Hellenic Psychological Society, 23(1), 7–23. https://doi.org/10.12681/psy_hps.23014
  • Mesquita, B., & Frijda, N. H. (1992). Cultural variations in emotions: A review. Psychological Bulletin, 112(2), 179–204. https://doi.org/10.1037/0033-2909.112.2.179
  • Mesquita, B., & Walker, R. (2003). Cultural differences in emotions: A context for interpreting emotional experiences. Behaviour Research and Therapy, 41(7), 777–793. https://doi.org/10.1016/S0005-7967(02)00189-4
  • Mesquita, B., Boiger, M., & De Leersnyder, J. (2016). The cultural construction of emotions. Current Opinion in Psychology, 8, 31–36. https://doi.org/10.1016/j.copsyc.2015.09.015
  • Niedenthal, P. M., Rychlowska, M., & Wood, A. (2017). Feelings and contexts: Socioecological influences on the nonverbal expression of emotion. Current Opinion in Psychology, 17, 170–175. https://doi.org/10.1016/j.copsyc.2017.07.025
  • 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
  • Rychlowska, M., Miyamoto, Y., Matsumoto, D., Hess, U., Gilboa-Schechtman, E., Kamble, S., Muluk, H., Masuda, T., & Niedenthal, P. M. (2015). Heterogeneity of long-history migration explains cultural differences in reports of emotional expressivity and the functions of smiles. Proceedings of the National Academy of Sciences, 112(19), E2429–E2436. https://doi.org/10.1073/pnas.1413661112
  • Sauter, D. A., Eisner, F., Ekman, P., & Scott, S. K. (2010). Cross-cultural recognition of basic emotions through nonverbal emotional vocalizations. Proceedings of the National Academy of Sciences, 107(6), 2408–2412. https://doi.org/10.1073/pnas.0908239106
  • Tsai, J. L. (2007). Ideal affect: Cultural causes and behavioral consequences. Perspectives on Psychological Science, 2(3), 242–259. https://doi.org/10.1111/j.1745-6916.2007.00043.x
  • Tsai, J. L., Chen, D. S., Yang, A. M., Cachia, J. Y. A., Blevins, E., Ko, M., Mathur, M. B., Aragón, O. R., Arens, E. A., Bencharit, L. Z., Chen, S. H., Chen, Y.-C., Chentsova Dutton, Y., Cheung, B. Y., Chim, L., Chow, P. I., Clobert, M., Costello, A. M., de Almeida, I., . . . Zhou, X. (2025). A meta-analytic review of cultural variation in affect valuation. Psychological Bulletin, 151(12), 1486–1524. https://doi.org/10.1037/bul0000499
  • Tsai, J. L., Knutson, B., & Fung, H. H. (2006). Cultural variation in affect valuation. Journal of Personality and Social Psychology, 90(2), 288–307. https://doi.org/10.1037/0022-3514.90.2.288
  • Tsai, J. L., Miao, F. F., Seppala, E., Fung, H. H., & Yeung, D. Y. (2007). Influence and adjustment goals: Sources of cultural differences in ideal affect. Journal of Personality and Social Psychology, 92(6), 1102–1117. https://doi.org/10.1037/0022-3514.92.6.1102
  • Wierzbicka, A. (1999). Emotions across languages and cultures: Diversity and universals. Cambridge University Press. https://doi.org/10.1017/CBO9780511521256