Are There Measurable Clues to an Aroma’s “Roundness”?

We sometimes describe an aroma as “round,” “smooth,” or “bright.” We are not looking at the shapes of molecules. Instead, we are connecting an impression received through the nose with another sense, such as shape, touch, or color. Could aroma characteristics measurable by instruments also be involved in these associations?

In 2023, Ryan J. Ward and colleagues compared people’s responses about shapes, textures, pitches, colors, and other qualities associated with aromas against measurements from a device called an “electronic nose.” Their aim was to investigate how much an aroma’s physical and chemical characteristics contribute to links between senses that have been explained through pleasantness, word meanings, experience, and learning.

Shapes and Colors That 68 People Matched with Aromas

The research team used rating data collected from 68 participants in an earlier study: 45 women and 23 men, with a mean age of 26.75 years. The 10 aromas were black pepper, lavender, lemon, orange, peppermint, caramel, cherry, coffee, freshly cut grass, and pine. Essential oils and fragrance oils from two brands were used.

The sample containers were wrapped in white tape to prevent associations based on the color of the liquid. The aromas were presented in random order, and participants smelled each at least once, with the option to smell it again if needed.

For shape, participants chose a position matching the aroma on a nine-point scale with an angular figure at one end and a rounded figure at the other. For texture, they physically touched rough sandpaper and smooth silk before rating their impression of the aroma on a nine-point scale. The pleasantness or unpleasantness of each aroma was rated separately.

For sound, participants listened to tones and selected a pitch that matched the aroma. For color, they adjusted lightness and hue on a screen. Thus, the “roundness” measured here is a correspondence with a figure selected through an association with an aroma. It is not a measurement of molecular shape or of a phenomenon in which smelling an aroma actually causes a figure to be seen.

An Electronic Nose Reads Sensor Responses, Not Human Impressions

Next, the same 10 fragrance materials were measured with an electronic nose. This device combines several gas sensors and records differences in their responses when exposed to gases. Rather than experiencing an aroma as “lemon-like” or “pleasant” as a person might, it captures a response pattern produced by the aroma mixture.

The measured features included readings from several sensors responsive to volatile substances, as well as temperature, humidity, and air pressure. Each material was placed in the device and recorded for 10 minutes, and this was repeated 10 times per material. There were 100 recordings in total, but only 10 different fragrance materials. The analysis used the average features of each material across its repeated measurements.

Electronic-nose sensors respond to multiple compounds. A relationship between a particular sensor reading and human ratings therefore does not establish that “this one molecule produced a round impression.” Nor does the method identify every compound in an aroma. The original paper also notes limitations, including compounds the device detects poorly and some readings that cannot be quantified.

Do Aromas That Seem Similar to People Also Have Similar Measurements?

The team first constructed a “map” that placed aromas with similar average ratings for shape, texture, color, and other qualities close together, and another map that placed aromas with similar sensor readings close together. There were areas of overlap, such as coffee and caramel falling into the same group on both maps.

After aligning the maps’ orientations and scales, the researchers examined how closely the arrangements of the aromas overlapped. Their metric gave a mean similarity of 49%. However, this number describes similarity between arrangements. It does not mean that “chemistry explained 49% of human associations” or that “unfamiliar aromas were identified with a 49% success rate.”

For example, orange and lemon belonged to the same group in the human ratings but not in the sensor features. Similar human impressions and similar instrument responses therefore showed both overlap and divergence.

Measured Features Explained 6–23% of the Variation in Ratings

Next, the researchers examined individual ratings using statistical models that accounted for differences between participants and fragrance materials. Of the nine measured features, one that overlapped strongly with the others was excluded, leaving eight for use in the models.

Significant relationships with measured features were found for shape, texture, pleasantness, pitch, and color. The measured features alone explained 6–23% of the variation in ratings. For example, the figures were 18% for shape, 6% for texture, 10% for pleasantness, and 9% for pitch. For color, they were 16% for lightness, 17% for the red–green dimension, and 23% for the yellow–blue dimension.

These figures describe how well the models captured differences in ratings within the data collected for these 10 aromas. They do not mean that a given proportion of an individual’s experience is determined by compounds. Nor can these numbers alone establish whether the same accuracy would apply to unmeasured perfumes or another person’s responses.

Aroma Intensity and Quality May Help Shape the Associations

Shape ratings were associated with temperature; texture, pleasantness, and pitch with gas-sensor responses; and color with several features, including temperature and air pressure. The authors suggest that aroma intensity, aroma quality, and the complexity of mixtures may underlie these relationships.

For example, temperature, humidity, and air pressure affect how odorants exist in the gas phase, which might in turn relate to perceived intensity and associations. However, this study did not manipulate temperature and compare the same fragrance material under different conditions. It did not establish a causal relationship such as raising the temperature making an impression rounder.

The original paper also discusses potentially involved chemicals based on the names of particular sensors. However, sensor responses to mixtures alone cannot establish which components or mechanisms generate an association. The use of air-quality sensors is also distinct from investigating the fragrance materials’ effects on health.

The result does not displace explanations based on learning and experience. Alongside experiences of seeing and smelling lemons, or liking or disliking an aroma, the physical and chemical features of the stimulus reaching the nose may offer a partial explanation. The experiment also did not settle whether the associations are innate or acquired through experience.

Predicting Responses to Unfamiliar Aromas Is a Next Step

The authors state that a wider range of aromas, including unfamiliar ones, needs to be tested. They predict that measured features might explain a greater proportion of responses to aromas with fewer associations based on names or experience. This is not a conclusion demonstrated with the 10 aromas in the present study.

They also propose leaving one aroma out of model construction and testing whether its ratings can be predicted, and taking measurements immediately before or after participants smell an aroma. These approaches could test more directly how well the models handle unfamiliar aromas and how the immediate environment affects the results.

The research was partly supported by the UK Engineering and Physical Sciences Research Council (EPSRC). The funder had no role in study design, analysis, or the decision to publish, and the authors declare no conflicts of interest.

Summary

The shapes, textures, sounds, and colors associated with aromas showed small but significant relationships with features measured by an electronic nose. The fact that these features explained 6–23% of the variation in ratings indicates the value of considering physical and chemical aspects of aromas, alongside experience and language, when trying to understand these associations.

An aroma’s “roundness” is not a fixed label attached to a single compound. Yet dismissing it as a fanciful expression also misses shared features. How do stimulus characteristics enter into connections with other senses through perceived intensity and quality? This study provides a basis for investigating that pathway.

References

Ward, R. J., Wuerger, S. M., Ashraf, M., & Marshall, A. (2023). Physicochemical features partially explain olfactory crossmodal correspondences. Scientific Reports, 13, 10590. DOI: 10.1038/s41598-023-37770-1. Full original article. The original article is published under CC BY 4.0.