When Two Odors Mix, Which One Shapes Pleasantness?

When two odors are mixed, which one’s pleasantness does the blend resemble? A 2020 study of 198 binary mixtures found that pleasantness usually fell between that of the two components and shifted toward the rating of the component perceived as stronger.

But the stronger component did not always win. Some combinations containing vanillin shifted toward vanillin’s pleasantness even though it was perceived as weaker on its own. Alongside examining these rules and exceptions, the researchers tested whether a blend’s pleasantness could be predicted from its components’ pleasantness and intensity.

Creating 198 combinations from 72 compounds

The researchers selected 72 odor compounds found in natural products, including foods, to cover a broad range of chemical characteristics. They made 198 binary mixtures. Twenty-four pairs were evaluated twice to check reproducibility, yielding a design of 222 trials in all.

Participants were 60 people aged 18–25 recruited at Jiangnan University in China, including 41 women. They were screened using tasks such as distinguishing odor quality and intensity, and practiced using the rating scales. Rather than having each person evaluate every sample, the researchers assigned samples so that each was evaluated by approximately 30 participants.

Each trial involved three samples: component A, component B, and their mixture. Participants rated pleasantness first and intensity afterward. A reference odor helped standardize use of the intensity scale. They were also asked about attributes including the perceived intensity of each component within the mixture.

To avoid reactions from directly mixing the liquids, the two sample solutions were placed on separate sides of a cotton ball, allowing their vapors to mix in the bottle. Preliminary testing sought to bring individual-component intensities close together, but differences remained in the main experiment. The researchers therefore distinguished combinations with and without differences in pleasantness and intensity.

Most mixtures fell in between, with stronger components exerting more influence

Overall, mixture pleasantness fell between the ratings of the two individual components. When the components differed in both intensity and pleasantness, mixture pleasantness tended toward that of the stronger component.

For example, in the pairing of ethyl valerate and p-cymene, ethyl valerate was rated stronger and more pleasant. On a 0–10 scale, their mean pleasantness ratings were 5.39 and 4.06, respectively, while the mixture scored 5.27, close to the more intense ethyl valerate.

Conversely, when the less pleasant component was stronger, some mixtures shifted toward it. A mixture of 1-heptanol and phenylethyl thiol was rated closer to the stronger, less pleasant phenylethyl thiol. The authors also noted a tendency for unpleasant components to have a particularly large influence on mixtures.

When components were similar in intensity but differed in pleasantness, mixture pleasantness often fell near their average or toward the less pleasant component. However, exceptions shifted toward the more pleasant component, so mixing did not follow a simple rule of always producing the arithmetic mean.

Vanillin sometimes led a mixture despite being weaker alone

A clear exception was vanillin paired with diethyl acetal. Vanillin’s intensity rating was 4.03, compared with 5.40 for diethyl acetal, making vanillin the weaker-smelling component. Yet mixture pleasantness was 6.52, close to vanillin alone at 6.48 and farther from diethyl acetal at 4.74.

The authors suggested that vanillin’s odor character may have become prominent in the mixture and suppressed perception of the other component. Although this explanation remains an interpretation, it highlights an important point: intensity measured alone does not always correspond to prominence alongside another odor.

Mixtures more pleasant than either component offer exploratory leads

Some mixtures had mean pleasantness ratings above those of both components. In six pairs, individual statistical tests also reported significantly higher pleasantness than either component. Examples included linalool with 2-octanone and eugenol with 2-octanone.

In all six, the components were classified as similar in pleasantness, and in five they were also similar in intensity. The authors proposed that combining components similar in both properties may make an additional pleasantness benefit more likely.

However, the tests for these six pairs were not corrected for the large number of combinations examined. They are best read as candidates worth retesting. Conversely, among 28 pairs whose mean ratings fell below both components, no statistically clear reduction was confirmed.

Group averages were easier to predict than individual preferences

Another main objective was to predict mixture pleasantness by weighting each component’s pleasantness according to its perceived intensity. This model gives greater influence to the stronger component’s rating.

Using the group-average component pleasantness and intensity ratings from the same trial produced a strong correspondence between predicted and observed values, with R² = 0.857. Using each component’s average across all trials instead reduced R² to 0.732. This does not mean that the model correctly predicted 85.7% of people’s preferences; it describes the correspondence between group-average ratings across mixtures.

Ratings obtained in the context of evaluating a particular pairing corresponded better than treating component properties as a fixed lookup table. The authors suggested that contextual effects, whereby one odor changes how another is perceived, also affect prediction. Pairs with large differences in both intensity and pleasantness remained relatively predictable even when across-trial averages were used.

Predicting individual ratings, however, was less successful than predicting group averages. There remains a gap between explaining average blending tendencies and identifying a fragrance a particular person will like. The evaluation also used component and mixture ratings collected in the same experiment; it did not establish a finished predictor that can be applied unchanged to new people or multicomponent perfumes.

Conclusion

When two odors were mixed, pleasantness usually fell between them and shifted toward the rating of the stronger-smelling component. When intensities were similar, mixtures often fell near the average or toward the less pleasant component, but exceptions such as vanillin showed that a component weak on its own could lead the mixture’s impression.

A model using component pleasantness and perceived intensity explained group-average ratings relatively well, but predicting individual evaluations remained difficult. Thinking about blends requires more than adding two things people like: it also means considering how each is perceived within the mixture.

The study was supported by China’s National Key R&D Program, the National First-class Discipline Program of Light Industry Technology and Engineering, the China Scholarship Council, and the Postgraduate Research & Practice Innovation Program of Jiangsu Province. No separate conflict-of-interest statement was found in the original article.

Reference

Ma Y, Tang K, Thomas-Danguin T, Xu Y. Pleasantness of Binary Odor Mixtures: Rules and Prediction. Chemical Senses. 2020;45(4):303–311. doi:10.1093/chemse/bjaa020