Volynets, Glerean, Hietanen, Hari & Nummenmaa (2020) — Bodily maps of emotions are culturally universal
The scaling-up of emBODY from Nummenmaa et al.’s (2014) two-country design (Finland, Taiwan) to 3954 respondents in 101 countries. Subjects saw two body silhouettes and an emotion word, and coloured the regions whose activity they felt increasing (left body) and decreasing (right body) for each of 13 emotions plus a neutral state. The resulting bodily sensation maps (BSMs) were tested for cultural, sex, and age dependence.
What was found
Concordance dominates. t-SNE projection of the country × emotion effect-size maps clustered by emotion, not by country: “Love” maps from all 15 well-sampled countries sit together, and apart from both a near neighbour (Happiness) and a distant one (Depression). Correlation-based similarity confirmed it — every emotion’s mean inter-nation similarity exceeded rs = 0.82. Sex differences were similarly minor (r > 0.80 concordance), though real: women reported more gut activation during anger, jealousy, anxiety and shame and more throat activation during anxiety, shame, fear, contempt and sadness; men reported more genital activation during love.
Ridge regression settles the variance question. With emotion, 15 countries, 9 Huntington civilizations, 8 language families, education, age, BMI and sex all in one penalised model (λ = 10⁷), emotion category carried the variance and the cultural/anthropometric regressors did not. No relationship at all with longitude, absolute latitude, height, weight or BMI.
Ageing dampens the maps. The one substantial background effect. Age correlated negatively with map intensity for eleven emotions and positively for depression and sadness — which, because those two are represented by deactivation (negative values), also means dampening. The authors read this alongside Hietanen et al. (2015), where bodily maps become more discrete across childhood, to propose a lifespan arc: sensations sharpen into adulthood, then fade. See age-related-interoceptive-decline.
The methodological argument worth keeping
The paper’s most transferable contribution is not a finding but a critique of how cross-cultural claims get tested (see cultural-universality-of-emotion). Null-hypothesis testing makes “no difference between cultures” the null, which is unprovable by Fisherian statistics; and because cross-cultural work needs large samples, trivially small differences reach significance. The field is therefore biased “towards seeing small differences across cultures even when consistencies would dominate.”
Their alternative: quantify the de facto degree of similarity and judge it against a pre-defined criterion. Their benchmark is deliberately provocative — the observed cross-cultural consistency (r = 0.82) is more than double the typical effect size of medications considered effective in general practice (levodopa, antipsychotics, antidepressants; r < 0.40, Leucht et al. 2015). They also concede the corollary honestly: universality should be treated as a continuum “ranging from universal homeostatic and defensive responses to flexible and malleable social processes,” not an all-or-nothing property, and this calls for Bayesian rather than Fisherian reasoning.
What this does and does not license
The paper concludes that the results “lend support to the basic emotion view (Ekman, 1992)” and show emotions are “experienced in the body in predominantly culture-independent, discrete fashion.” Two boundaries on that conclusion matter for the rest of this wiki, and the first is drawn by the authors themselves.
It is not evidence for autonomic-specificity. The authors are explicit: interoceptive accuracy is imperfect (Critchley et al. 2004) and bodily changes are only loosely coupled to subjective feeling (Mauss et al. 2005), so “the present data do not reflect a detailed description of emotion-specific physiological changes but the net consciously felt changes occurring in the body during emotions, and the BMs are thus subjective feeling maps.” Distinct felt maps are compatible with the absence of distinct autonomic signatures — which is what Scarantino (2018) and Barrett both, for different reasons, think the physiological record shows. Volynets et al. cite Siegel et al. (2018) in the introduction as leaving that question open and never claim to close it.
The constructionist reading is available and unaddressed. Every subject was cued with an emotion word, in English, and asked where they would feel it. Concordant answers are what shared conceptual knowledge of emotion categories predicts just as much as what shared bodily experience predicts — core affect plus conceptualization, in Barrett’s terms, with language doing the cross-cultural work. The paper’s defence is indirect (Nummenmaa et al. 2014 found word-cued maps concordant with maps from nonverbal emotion induction), and it is not engaged in this paper. Flagged as an open tension on cultural-universality-of-emotion rather than resolved here.
An internal strain
The claim in the title is “culturally universal”; the limitation section concedes the sample “lacked sufficient number of, for example, East Asian and African respondents.” The 15 analysable countries are Australia, Brazil, Canada, Croatia, Finland, Germany, India, Italy, Mexico, Philippines, Poland, Romania, Turkey, UK, USA — and 3069 of 3954 subjects were Western. The authors anticipate the objection and answer it conceptually rather than empirically: universality means degree of consistency, not exceptionlessness, so a future African discrepancy would move a continuum rather than falsify a claim. That is a coherent position, but it also means the title’s claim is not the claim the data test.