The familiar rule starts to wobble
Cognitive reappraisal is often described as adaptive. Expressive suppression is often described as maladaptive. Those summaries are not invented; decades of research have produced average associations in that direction.
But an average can become a stereotype when it is carried across cultures without testing what changes.
A 2025 Nature Human Behaviour paper by Mark Shuquan Chen and colleagues brought a much larger evidence base to that problem. The study did not simply compare “Western” and “Eastern” samples. It synthesized associations between two heavily studied emotion-regulation strategies and mental-health outcomes across many countries and then tested several cultural and demographic moderators.
Its most important contribution is not a new list of good and bad cultures. It is evidence that the relationship between strategy use and mental health is conditional.
Source card
Title: Emotion regulation and mental health across cultures: a systematic review and meta-analysis
Authors: Mark Shuquan Chen, Qiyue Cai, Deemah Omari, Drishti Enna Sanghvi, Shibo Lyu, and colleagues, including George A. Bonanno
Journal: Nature Human Behaviour
Published: 15 April 2025
Volume/pages: 9, 1176–1200
DOI: https://doi.org/10.1038/s41562-025-02168-8
Design: systematic review and meta-analysis
Registration: PROSPERO CRD42021258190
Evidence object: 249 articles, 150,474 participants, 861 effect sizes, 37 countries or regions, according to the published paper
The research question
The study asked how cognitive reappraisal and expressive suppression relate to two broad categories of mental-health outcome: psychopathology and positive functioning. It then asked whether those relationships vary with cultural dimensions and sample demographics.
This framing matters because earlier cross-cultural discussions often relied on a simplified individualism-versus-collectivism axis or a broad West-versus-East contrast. Chen and colleagues examined several national cultural dimensions, including Hofstede-based indicators and a measure of education, industrialization, richness, and democracy described as EIRDness.
The paper therefore treats culture as multidimensional. That is an improvement over binary regional labeling, even though national indices remain coarse proxies for the actual cultural worlds participants inhabit.
What was analyzed
The meta-analysis focused on the two most-studied strategies in this literature: cognitive reappraisal and expressive suppression. Reappraisal involves altering the interpretation of a situation to change its emotional impact. Suppression involves inhibiting outward emotional expression after a response is underway.
The outcomes were organized around psychopathology and positive functioning rather than one diagnosis or one wellbeing scale. This created a large synthesis but also introduced heterogeneity because studies differed in populations, measures, languages, settings, and outcome definitions.
The authors modeled cultural moderators alongside demographic features such as age, gender composition, and racial-minority representation. Meta-analysis can estimate average patterns and moderator relationships across studies. It cannot recreate the contextual detail missing from the original studies.
Main findings
Across the literature, reappraisal was generally associated with more favorable mental-health outcomes and suppression with less favorable outcomes. The crucial result was that the strength of those associations varied.
The authors reported that greater reappraisal was more adaptive in cultures characterized in their analyses as more short-term oriented, more tolerant of uncertainty, and more competition-driven. Suppression was more maladaptive in more indulgent and more competition-driven cultural contexts.
Sample demographics also moderated effects. The association between reappraisal and better outcomes was stronger in samples with higher proportions of women and racial-minority participants. Suppression appeared more maladaptive in younger samples.
These are moderator effects in a meta-analytic literature, not rules for individuals. They show statistical variation in associations across studies. They do not tell a clinician or educator which strategy a particular person from a particular country should use tomorrow.
Effect estimates and uncertainty
The paper reports meta-regression coefficients for several moderators. For example, the adaptive association of reappraisal was stronger in samples with a higher proportion of women and in samples with greater racial-minority representation. Suppression showed a stronger maladaptive relationship in younger samples.
The exact coefficients are useful for researchers but can be misleading in public summaries if detached from the scales used for moderators and outcomes. The Institute therefore treats the qualitative direction and uncertainty structure as more important for a general Knowledge Library brief than a list of decontextualized numbers.
Meta-regression is also observational at the study level. Moderator coefficients can be influenced by correlations among national indicators, differences in measurement, publication patterns, and the composition of the available literature.
What the paper changes
The paper weakens a common universalist shortcut: “reappraisal good, suppression bad.” It does not reverse the average pattern. It shows that average associations are not equally strong across contexts.
That is a meaningful advance because emotion-regulation advice often moves faster than the evidence. A strategy that is socially costly in one environment may be expected in another. Expressive restraint can carry different relational meanings depending on norms, hierarchy, role, and group expectations. Reappraisal can also differ in usefulness depending on whether the situation calls for adaptation, resistance, action, or acknowledgment of a real problem.
The paper also demonstrates why cultural research benefits from moving beyond a single individualism-collectivism score. No national index captures culture in full, but multidimensional modeling can reveal moderators hidden by binary categories.
The ecological-fallacy problem
Country-level cultural indicators describe aggregates. Individuals do not literally possess their country's Hofstede score. A person may live within multiple cultural systems at once: family, profession, language, migration history, region, class, religion, generation, online community, and institution.
It would therefore be an ecological fallacy to infer an individual's likely emotional strategy or mental-health response directly from a national average.
This is especially important in Northern Europe, where internationalized urban populations, migration, multilingualism, and strong institutional subcultures make national labels particularly incomplete. A Swedish workplace may contain norms drawn from several countries and professions at once.
Measurement remains part of the culture question
Cross-cultural meta-analysis inherits the validity of the measures used in primary studies. If reappraisal, suppression, symptoms, or positive functioning are not measured equivalently across languages and populations, cultural moderation may partly reflect measurement differences.
Translation quality, response styles, social desirability, and differing norms around disclosure can all affect self-report. The study's size cannot erase those source-level limitations.
This does not make cross-cultural synthesis futile. It means conclusions should be proportional to the measurement architecture underneath them.
What the evidence does not show
The paper does not show that suppression is beneficial in “collectivist cultures,” nor that reappraisal is harmful outside the West. It does not establish that national cultural traits cause the observed strategy-outcome relationships. It does not show that every member of a country shares the same regulatory norms.
It also does not establish intervention effects. Most associations in the underlying literature are observational. Knowing that habitual reappraisal correlates with better outcomes in a set of studies is not the same as proving that training reappraisal will improve mental health in every context.
The paper does not justify ranking cultures by emotional health. Cultural dimensions are moderators used to examine variation, not measures of civilizational maturity.
Relation to the wider evidence
The study sits inside a broader movement from strategy essentialism toward regulatory flexibility and context sensitivity. Earlier work established robust average relationships between certain strategy tendencies and wellbeing. More recent work asks when, for whom, and under what conditions those associations change.
That movement is also visible in research on emotion goals, emotion beliefs, situation selection, and dynamic regulation. The relevant unit is increasingly not “strategy X” alone but person × goal × context × timing × implementation × outcome.
The 2025 meta-analysis provides strong evidence that culture belongs somewhere in that interaction.
Why this matters in Northern Europe
The Nordic relevance is conceptual rather than decorative. Northern European institutions frequently import psychological measures, leadership advice, school programmes, and digital wellbeing tools from international evidence bases. If strategy-outcome associations vary by cultural context, implementation should include local validation rather than assume transportability.
This does not require a uniquely Nordic psychology. It requires ordinary scientific discipline: test whether constructs and relationships travel.
Swedish and Nordic researchers can contribute by using multilingual samples, strong population registries where ethically appropriate, transparent measurement-invariance testing, and designs that distinguish national context from individual values and institutional norms.
Institute synthesis: context belongs inside capacity
The following is Institute synthesis.
For Inner Technology, this paper supports a distinction between repertoire and judgment. A person may know how to reappraise and still need judgment about whether reinterpretation fits the situation. They may be able to suppress expression and still need judgment about whether restraint serves safety, dignity, hierarchy, intimacy, or avoidance.
Capacity is therefore not maximized by teaching one strategy as universally superior. It is strengthened by the ability to perceive context, select deliberately, notice consequences, and revise.
At institutional scale, that principle becomes a warning against exporting psychological programmes as though culture were a translation problem only. The meaning of a behavior can change across settings even when the questionnaire label stays the same.
Open questions
Future research needs stronger within-culture heterogeneity measures, more direct tests of individual values, longitudinal and intervention designs, better representation outside heavily studied populations, and more measurement-invariance work. It also needs to distinguish cultural norms around expression from the internal experience of emotion.
A large meta-analysis can show that context matters. The next challenge is to measure context without turning it into another stereotype.
Why moderator findings are easy to overread
A moderator tells us that an association differs systematically across values of another variable. It does not automatically explain why. If suppression is more strongly associated with poor outcomes in one cultural context, several pathways could produce that pattern: social sanctions, mismatch with norms, relationship consequences, differences in why people suppress, measurement non-equivalence, or correlated economic and institutional factors.
Meta-regression can identify the pattern while leaving the mechanism unresolved. Public summaries often skip that distinction because “culture changes the effect” sounds more satisfying than “culture-linked indicators explain some between-study variation whose mechanism remains open.” The second formulation is less elegant and more accurate.
The same caution applies to demographic moderators. A stronger association in samples with a higher proportion of women does not show that biological sex caused the difference. Gendered socialization, sample composition, diagnosis mix, measurement, and other study characteristics can covary.
Suppression is not one behavior
Expressive suppression is usually operationalized as inhibiting outward expression once emotion is active. In daily life, that can include tact, concealment, professionalism, fear, deference, strategic ambiguity, emotional labor, safety behavior, or culturally valued restraint.
Two acts that look identical from outside can have different motives and consequences. Suppressing visible anger during a dangerous encounter is not the same social event as habitually withholding emotion from a trusted partner. A questionnaire average combines across episodes unless the design asks otherwise.
This helps explain why context-sensitive research is not a rejection of strategy research. It is a refinement. The construct remains useful, but the unit of explanation expands.
Reappraisal also has content
Reappraisal is sometimes spoken about as if it means finding a healthier thought. Scientifically, the content of the reinterpretation matters. A person can reappraise an ambiguous comment as harmless, reinterpret a setback as temporary, minimize real discrimination, excuse abuse, or find meaning in loss. All are changes in construal; their consequences differ.
Cross-cultural variation may therefore involve not only how often people reappraise but what kinds of reinterpretation are socially available and valued. Future work can move beyond frequency measures toward content, goal, timing, and outcome.
Implications for measurement transport
International programmes often translate a questionnaire and compare scores across countries. The 2025 paper encourages a more demanding sequence: establish linguistic quality, test structural comparability, examine item functioning, report contextual features, and interpret mean differences with cultural humility.
For policy, that means imported wellbeing indicators should not become benchmarks before their meaning is tested locally. For clinical science, it means treatment mechanism models should be examined across populations rather than assumed universal. For organizational research, it means norms around expression and hierarchy belong in the design.
The strongest contribution of the meta-analysis may therefore be procedural. It turns culture from a paragraph in the limitations section into a variable that can alter the central association.
What replication should look like
The most useful replication is unlikely to be another global average using the same country proxies. Strong follow-up work would measure individual values and norms alongside national context, test item-level invariance, and gather repeated data about actual regulation episodes. That would help distinguish whether a moderator operates through preference, social consequence, perceived usefulness, or measurement.
Longitudinal studies could also ask whether the same strategy predicts later functioning differently across contexts rather than relying on cross-sectional associations. Intervention studies could test whether teaching a strategy has different effects when it fits or conflicts with local norms.
The 2025 meta-analysis makes those studies more necessary because it gives a credible reason to expect heterogeneity. Its scientific value is not only in the moderator coefficients already reported, but in making universal assumptions testable.
The paper is stronger for resisting a binary culture story
One reason this study deserves a dedicated brief is methodological ambition. Cross-cultural psychology has repeatedly been criticized for using nation as a shorthand for culture and for reducing variation to individualism versus collectivism. Chen and colleagues do not eliminate those problems, but they explicitly widen the moderator set. Education, industrialization, economic resources, democracy, uncertainty tolerance, indulgence, time orientation, and competition-related values create a more differentiated map.
That broader map can reveal patterns that a binary East-West analysis would miss. It can also increase researcher degrees of freedom if many correlated moderators are tested. Readers should therefore pay attention to model specification, robustness, and whether findings replicate in independent datasets.
Publication bias and the archive available to meta-analysis
A meta-analysis is only as representative as the literature it can find. Emotion-regulation studies with significant associations may be more likely to appear in journals than null results. English-language indexing and database coverage can also privilege particular regions. Statistical assessments of publication bias help, but they cannot recover studies that were never conducted or never archived.
Cross-cultural conclusions are especially sensitive to this because the least represented populations may also be those whose inclusion would change the apparent moderator pattern. The sample of countries is broad relative to much of psychology, but “37 countries or regions” is not the world.
From cultural moderation to institutional design
The findings are relevant beyond individual counseling. Schools, workplaces, hospitals, and public services teach emotional norms through rules and incentives. An international organization that promotes one preferred regulation style across offices may unknowingly reward behavior differently across local contexts.
A better design question is not “Which culture is right about suppression?” It is “What social function does expression or restraint serve here, and what outcomes follow?” That question can be investigated without essentializing culture.
Research references
Chen, M. S., Cai, Q., Omari, D., Sanghvi, D. E., Lyu, S., et al. (2025). Emotion regulation and mental health across cultures: a systematic review and meta-analysis. Nature Human Behaviour, 9, 1176–1200. https://doi.org/10.1038/s41562-025-02168-8
Gross, J. J., & John, O. P. (2003). Individual differences in two emotion regulation processes: Implications for affect, relationships, and well-being. Journal of Personality and Social Psychology, 85, 348–362. https://doi.org/10.1037/0022-3514.85.2.348
Tamir, M. (2016). Why Do People Regulate Their Emotions? A Taxonomy of Motives in Emotion Regulation. Personality and Social Psychology Review, 20, 199–222. https://doi.org/10.1177/1088868315586325