Humans share sensory systems with a common anatomical blueprint, but individual sensory experience nevertheless varies. In olfaction, it is not known to what degree sensory perception, particularly the perception of odor pleasantness, is founded on universal principles,dictated by culture,or merely a matter of personal taste.To address this, we asked 225 individuals from 9 diverse nonwestern cultures—hunter-gatherer to urban dwelling—to rank the monomolecular odorants from most to least pleasant. Contrary to expectations, culture explained only 6% of the variance in pleasantness rankings, whereas individual variability or personal taste explained 54%. Importantly, there was substantial global consistency, with molecular identity explaining 41% of the variance in odor pleasantness rankings. Critically, these universal rankings were predicted by the physicochemical properties of out-of-sample molecules and out-of-sample pleasantness ratings given by a tenth group of western urban participants. Taken together, this shows human olfactory perception is strongly constrained by universal principles.
Results
15 Ibáñez-Berganza M.
Amico A.
Loreto V. Subjectivity and complexity of facial attractiveness. 16 Scott I.M.
Clark A.P.
Josephson S.C.
Boyette A.H.
Cuthill I.C.
Fried R.L.
Gibson M.A.
Hewlett B.S.
Jamieson M.
Jankowiak W.
et al. Human preferences for sexually dimorphic faces may be evolutionarily novel. 15 Ibáñez-Berganza M.
Amico A.
Loreto V. Subjectivity and complexity of facial attractiveness. 1 Khan R.M.
Luk C.-H.
Flinker A.
Aggarwal A.
Lapid H.
Haddad R.
Sobel N. Predicting odor pleasantness from odorant structure: pleasantness as a reflection of the physical world. , 17 Berglund B.
Berglund U.
Engen T.
Ekman G. Multidimensional analysis of twenty-one odors. , 18 Schiffman S.S.
Robinson D.E.
Erickson R.P. Multidimensional scaling of odorants: examination of psychological and physicochemical dimensions. , 19 Yeshurun Y.
Sobel N. An odor is not worth a thousand words: from multidimensional odors to unidimensional odor objects. , 20 Schiffman S.S. Physicochemical correlates of olfactory quality. 6 Ayabe-Kanamura S.
Schicker I.
Laska M.
Hudson R.
Distel H.
Kobayakawa T.
Saito S. Differences in perception of everyday odors: a Japanese-German cross-cultural study. , 7 Classen C.
Howes D.
Synnott A. Aroma: The Cultural History of Smell. , 8 Distel H.
Ayabe-Kanamura S.
Martínez-Gómez M.
Schicker I.
Kobayakawa T.
Saito S.
Hudson R. Perception of everyday odors—correlation between intensity, familiarity and strength of hedonic judgement. , 9 Maier-Nöth A.
Schaal B.
Leathwood P.
Issanchou S. The lasting influences of early food-related variety experience: a longitudinal study of vegetable acceptance from 5 months to 6 years in two populations. , 10 Pangborn R.M.
Guinard J.-X.
Davis R.G. Regional aroma preferences. , 11 Poncelet J.
Rinck F.
Bourgeat F.
Schaal B.
Rouby C.
Bensafi M.
Hummel T. The effect of early experience on odor perception in humans: psychological and physiological correlates. , 12 Wysocki C.J.
Pierce J.D.
Gilbert A.N. Geographic, cross-cultural, and individual variation in human olfaction. , 13 Yamin-Pasternak S.
Kliskey A.
Alessa L.
Pasternak I.
Schweitzer P. The rotten renaissance in the Bering Strait: loving, loathing, and washing the smell of foods with a (re)acquired taste. 21 Nygaard M.E. Swedish fermented herring as a marker of rural identity: the Alfta surströmmingsskiva. 6 Ayabe-Kanamura S.
Schicker I.
Laska M.
Hudson R.
Distel H.
Kobayakawa T.
Saito S. Differences in perception of everyday odors: a Japanese-German cross-cultural study. , 8 Distel H.
Ayabe-Kanamura S.
Martínez-Gómez M.
Schicker I.
Kobayakawa T.
Saito S.
Hudson R. Perception of everyday odors—correlation between intensity, familiarity and strength of hedonic judgement. , 9 Maier-Nöth A.
Schaal B.
Leathwood P.
Issanchou S. The lasting influences of early food-related variety experience: a longitudinal study of vegetable acceptance from 5 months to 6 years in two populations. , 10 Pangborn R.M.
Guinard J.-X.
Davis R.G. Regional aroma preferences. , 12 Wysocki C.J.
Pierce J.D.
Gilbert A.N. Geographic, cross-cultural, and individual variation in human olfaction. , 14 Logue A.W.
Logue C.M.
Uzzo R.G.
McCarty M.J.
Smith M.E. Food preferences in families. 1 Khan R.M.
Luk C.-H.
Flinker A.
Aggarwal A.
Lapid H.
Haddad R.
Sobel N. Predicting odor pleasantness from odorant structure: pleasantness as a reflection of the physical world. , 2 Haddad R.
Medhanie A.
Roth Y.
Harel D.
Sobel N. Predicting odor pleasantness with an electronic nose. , 3 Keller A.
Gerkin R.C.
Guan Y.
Dhurandhar A.
Turu G.
Szalai B.
Mainland J.D.
Ihara Y.
Yu C.W.
Wolfinger R.
et al. Predicting human olfactory perception from chemical features of odor molecules. , 4 Joussain P.
Chakirian A.
Kermen F.
Rouby C.
Bensafi M. Physicochemical influence on odor hedonics: where does it occur first?. , 5 Kermen F.
Chakirian A.
Sezille C.
Joussain P.
Le Goff G.
Ziessel A.
Chastrette M.
Mandairon N.
Didier A.
Rouby C.
Bensafi M. Molecular complexity determines the number of olfactory notes and the pleasantness of smells. In 1878, Margaret Wolfe Hungerford wrote, “Beauty is in the eye of the beholder,” suggesting what one person finds beautiful, another may not. Consistent with this, we now know that facial preferences vary across individuals.Importantly, however, they are also strongly shaped by cultureand may even have components that are universal.Similar to beauty, the perception of odor pleasantness or valence—the principal dimension by which odors are categorized—is said to vary across cultures.For example, fermented herring is a greatly appreciated delicacy in Sweden, but it also emits a smell described as the “most repulsive in the world.”In addition, people also display individual variability in food preference, even within families.At the same time, more recent studies of urban western participants demonstrate that valence can be objectively predicted from an odorant’s chemical structure,despite the fact that universal odor preferences have been disputed historically. It is unclear how to reconcile these perspectives: is odor preference culturally relative, driven by individual preferences, or universally constrained by molecular structure?
1 Khan R.M.
Luk C.-H.
Flinker A.
Aggarwal A.
Lapid H.
Haddad R.
Sobel N. Predicting odor pleasantness from odorant structure: pleasantness as a reflection of the physical world. , 2 Haddad R.
Medhanie A.
Roth Y.
Harel D.
Sobel N. Predicting odor pleasantness with an electronic nose. , 3 Keller A.
Gerkin R.C.
Guan Y.
Dhurandhar A.
Turu G.
Szalai B.
Mainland J.D.
Ihara Y.
Yu C.W.
Wolfinger R.
et al. Predicting human olfactory perception from chemical features of odor molecules. , 4 Joussain P.
Chakirian A.
Kermen F.
Rouby C.
Bensafi M. Physicochemical influence on odor hedonics: where does it occur first?. , 5 Kermen F.
Chakirian A.
Sezille C.
Joussain P.
Le Goff G.
Ziessel A.
Chastrette M.
Mandairon N.
Didier A.
Rouby C.
Bensafi M. Molecular complexity determines the number of olfactory notes and the pleasantness of smells. 2 Haddad R.
Medhanie A.
Roth Y.
Harel D.
Sobel N. Predicting odor pleasantness with an electronic nose. 22 Majid A.
Burenhult N.
Stensmyr M.C.
de Valk J.M.
Hansson B.S. Olfactory language and abstraction across cultures. In order to address this question, it is necessary to assess all three factors simultaneously, but this has never been done with a diverse sample of cultures. Studies that have used an experimental approach to study the impact of molecular structure on odor preference tend to sample people with similar urban lifestyles and experiences—i.e., literate, educated, and technologically savvy individuals who partake of a common global fragrance and flavor industry(although see Haddad et al.and Majid et al.). This provides only a weak and narrow test of the possible role of culture. To quantify the role that culture may play in odor preference, it is necessary to study diverse cultures, including those of small-scale societies that vary in their subsistence style and geography and where people are minimally influenced by global odor experiences, while at the same time measuring individual variability and the chemical structure of odorants.
Figure 1 Cross-cultural sample Show full caption Odor preference rankings were collected from nine culturally and geographically diverse populations. These included the three hunter-gatherer groups, Seri from a coastal desert and Maniq and Semaq Beri from tropical rainforest, one shoreline forager, Mah Meri, from a tropical coast; one swidden-horticulturalist, Semelai, from tropical rainforest; one farmer-foraging community, Chachi, from tropical rainforest; one subsistence agriculturalist community, Imbabura Quichua, from temperate highlands; and two urban dwellers from industrial and postindustrial communities of bustling urban settings, Mexican and Thai. The data from these nine communities were then related to available data from a large dataset on odor preference collected from urban dwellers from the USA (New York City). Here, we assess the unique contribution of each of these factors by experimentally testing nine diverse communities. Critically, seven of these groups belonged to small-scale societies—including hunter-gatherers, horticulturalists, and subsistence agriculturalists—with a more traditional lifestyle and who do not experience the same chemical ecology as western and nonwestern urbanites ( Figure 1 STAR Methods ).
23 Keller A.
Vosshall L.B. Olfactory perception of chemically diverse molecules. 24 Hummel T.
Sekinger B.
Wolf S.R.
Pauli E.
Kobal G. ‘Sniffin’ sticks’: olfactory performance assessed by the combined testing of odor identification, odor discrimination and olfactory threshold. Odorants were selected based on a previous study with postindustrial urban dwellers from the United States (New York City) who rated the pleasantness of 476 diverse molecules.We selected ten of these odorants such that the mean ratings would span the valence dimension from unpleasant to pleasant (for more details, see STAR Methods ). Participants from nine communities were, with the help of a network of fieldworkers, presented with ten pen-like odor-dispensing devices,each containing a unique odorant. A rank-order paradigm was chosen to assess odor pleasantness because not all groups had numeracy and use of scales and ratings is not the norm in these communities. The pens were randomly ordered and placed in a line in front of the subject. The participant first smelled all the odors in front of them and then ordered the pens from most pleasant to most unpleasant (from their left to right).
Figure 2 Pleasantness rankings across individuals and cultures Show full caption 23 Keller A.
Vosshall L.B. Olfactory perception of chemically diverse molecules. Between n = 16 and n = 55, individuals from each culture assessed each of 10 odorants. Nine cultures ranked the odorants in order from most (1, blue) to least (10, red) pleasant, whereas the Americans (specifically US Americans residing in New York City metropolitan area; data from Keller and Vosshall) used numerical ratings, converted into ranks here. Each color patch represents the integer ranking that one individual (from the culture indicated at the bottom) gave to one odorant (indicated on the left). The broad column on the far left represents the average ranking for each odorant across all individuals. See Table S2 for more information about the odors, Figure S1 for in-depth analysis odor pleasantness ranking, and Figure S3 for the relationship between odor pleasantness and odor intensity. If odor valence is learned from exposure to cultural traditions, then societies should differ in their rankings of perceived odor pleasantness, with a diverse set of rank orders across cultures. If, however, odor valence is a matter of individual preference, there should be large within group variation. Finally, if perceived odor pleasantness is universal, then all groups should rank odors in the same way. Using the within-culture mean ranking for each odorant, we found that odor valence rankings correlated strongly and positively across all cultures ( Figure 2 ; r = 0.82 ± 0.18), supporting the idea that culture has a relatively small influence overall on odor pleasantness. Pleasantness rankings were also correlated for both the most pleasant and most unpleasant odorants ( Figures S1 A–S1D) and for their ranking consistency across individuals and cultures ( Figures S1 E and S1F).
2 for each of these factors was used as the primary analysis measure ( Figure 3 Proportion variance explained by factors determining odor pleasantness Show full caption (A) Culture (6%) plays a negligible role in explaining variance in the observed odor pleasantness rankings, whereas individual variability (54%) and odorant identity (41%) explain more. (B) Positive control: odorant ranks were shuffled so that individuals within the same culture received a common shuffle, but individuals across cultures received different shuffles, simulating a “strong culturally determined odor pleasantness” scenario; this demonstrates that if the data did have a strong cultural component, it could have been detected with this method. (C) Negative control: culture labels were shuffled across individuals. This removes culture-specific information in each individual’s set of ranks, yet the contributions of each culture closely resemble the observed data. Bayesian analyses can be found in Figure S1 , with further information in Figure S2 and Table S1 We next conducted a two-way ANOVA using odorant identity and cultural grouping to determine the observed ranks for each individual + odorant pair. The first factor was “odorant identity.” The second factor of specific cultures has no net effect because all individuals must give the same ten ranks, but the interaction between these factors, odorant × cultural grouping, we call “culture” since only this term establishes whether odor rankings vary across groups. The remaining variance not accounted for by these factors and their interaction we term “individual,” which represents some combination of individual preference (not mediated by culture and odorant) and perceptual or task-related noise. The explained variance ηfor each of these factors was used as the primary analysis measure ( Table S1 ). We found culture only explained 6% of the variance, whereas 54% was due to individual variability ( Figure 3 ; see Table S1 for details of statistical analysis and Figure S2 A for the partition between individual preference and perceptual noise). Critically, odorant identity explained 41% of variance in rankings.
As a positive control, we simulated a case where culture drives odor preference by shuffling odorant labels in a manner that was consistent for each member within a culture but varied across cultures. Under these conditions, 41% of the variance was explained by culture, with the rest explained by individual variability. This positive control demonstrates that our method is sensitive enough to measure cultural variability should it exist. As a negative control for a possible effect of culture, we next shuffled individuals between cultures. Under these conditions, culture explained only 2% of the variance, not much smaller than the value observed in the unshuffled data ( Figure 3 ). The analogous Bayesian model comparisons reached the same conclusions ( Figures S1 G and S1H). Consistent with only a small contribution for culture, direct assessment of interindividual ranking similarity using Kendall’s τ showed the mean rank similarity for pairs of individuals within the same culture (τ = 0.32 ± 0.14) was only slightly higher than for pairs of individuals in different cultures (τ = 0.28 ± 0.11). In addition, a follow-up intensity ranking task showed that pleasantness ranking was not explained by the perceived intensity of odorants ( Figure S3 ).
2 = 0.056; subsistence, η2 = 0.015; continent, η2 = 0.021). Next, we used the same subsistence and continent groupings of individual cultures to ask if either explained the (small) differences in odor preferences between cultures better than random groupings. Specifically, we asked if a clustering metric, compactness (measured by the distance of cultures to cluster centers), was lower for either of these groupings than random groupings. We found weak evidence for continent as an organizing force (more compact than 96% of random clusterings) and less for subsistence (more compact than 77%). These analyses suggest that cultural preferences for odors are in large part locally determined. Taken together with the previous analyses, we find only a weak contribution of culture to odor pleasantness rankings. Perhaps, there is another shared factor that could explain odor pleasantness preferences. We considered two such factors: (1) subsistence type—hunter-gatherer (Semaq Beri, Maniq, Seri), subsistence horticulturalist (Semelai, Chachi, Quichuan, Mah Meri), and (post-)industrial urban dwelling (US American, Mexican, Thai)—and (2) continent—North American (US American, Mexican, Seri), South American (Chachi, Quichuan), and Asian (Semaq Beri, Maniq, Semelai, Mah Meri, Thai). We recalculated the ANOVA in Figure 3 to ask whether a factor corresponding to either subsistence or continent explained more variance than specific cultures (as identified in STAR Methods ). We did not observe increase in variance explained; in fact, variance decreased (culture, η= 0.056; subsistence, η= 0.015; continent, η= 0.021). Next, we used the same subsistence and continent groupings of individual cultures to ask if either explained the (small) differences in odor preferences between cultures better than random groupings. Specifically, we asked if a clustering metric, compactness (measured by the distance of cultures to cluster centers), was lower for either of these groupings than random groupings. We found weak evidence for continent as an organizing force (more compact than 96% of random clusterings) and less for subsistence (more compact than 77%). These analyses suggest that cultural preferences for odors are in large part locally determined. Taken together with the previous analyses, we find only a weak contribution of culture to odor pleasantness rankings.