Olfactory White: Similar Smells from 30 Different Ingredients

Mixtures can smell similar even when they have no ingredients in common. A 2012 study showed that when odorants with a broad range of characteristics were matched for perceived intensity and combined, mixtures began to smell similar at approximately 30 components or more.

The researchers called this shared percept “olfactory white.” The name draws an analogy with white light and white noise; it does not mean odorless. The phenomenon involves an overall similarity that emerges despite differences between individual components.

It is not just the number: intensity and the range of characteristics matter

The study began with a comparison to vision and hearing. Combining a broad range of wavelengths or frequencies at equal intensity produces white light or white noise. Could comparable conditions be created for smell?

The researchers selected 86 compounds spanning a broad range of odor impressions and molecular properties. Using ratings from 24 people, they diluted the individual compounds to approximately equal perceived intensity. Importantly, they matched how strong the compounds smelled, rather than giving them equal concentrations.

To avoid new chemical reactions in the liquid phase, they also placed the individual components on an absorbent pad so that the vapors mixed in the headspace of the container. A computational method selected combinations without a narrow bias in their odor or physicochemical characteristics.

A total of 208 people aged 21–40 participated across the study, completing different tasks involving intensity matching, similarity, discrimination, classification by name, and comparison with memory. The first similarity experiment is reported as having 56 participants in the Results section and 59 in the Methods section.

More components meant greater similarity, even without shared ingredients

In the similarity experiment, mixtures containing 1, 4, 10, 15, 20, 30, 40, or 43 components were paired, and participants rated the similarity of 191 pairs in total. Apart from control comparisons of a sample with itself, the two mixtures in each pair had no components in common.

Nevertheless, the mixtures smelled more similar as the number of components increased. At around 30 components, mixtures with different compositions were perceived as quite similar.

In a three-alternative discrimination task completed by 20 people, however, accuracy and confidence declined as the number of components increased, but participants could still distinguish 30-component mixtures above chance. Forming a similar overall percept was not the same as becoming completely indistinguishable.

People applied a learned name to unfamiliar mixtures

Next, the researchers asked whether a name could capture this shared overall percept. To avoid associations prompted by the word “white,” they gave the mixtures the meaningless name “Laurax.”

Participants learned their assigned 40-component Laurax by smelling it over three days, then classified new mixtures on the fourth day. The more components a new mixture contained, the more often participants chose the name Laurax. A separate experiment that added an “other” response option produced results in the same direction.

The researchers also tested mixtures with no components in common with the learned Laurax, asking participants to decide from memory alone whether they were Laurax. Mixtures with few components were relatively easy to reject, whereas discrimination of mixtures containing 20–30 components did not exceed chance. Unlike the three-alternative task involving direct comparison, this result came from matching a smell to memory.

Mixing any 30 ingredients is not enough

The researchers matched the intensity of another 58 compounds and conducted an experiment selecting combinations from a total of 144. They compared mixtures with widely distributed characteristics, mixtures biased toward relatively similar characteristics, and mixtures with the same composition but without intensity matching.

Mixtures with a narrower range of characteristics were more likely to be judged different from the learned Laurax. Even with the same composition, failing to match intensity reduced similarity to Laurax. Under these experimental conditions, the breadth of characteristics had a particularly large influence.

The authors suggest that natural coffee and rose do not smell the same despite containing many components because neither the characteristics nor the intensities of their components are evenly distributed. The figure of approximately 30 is a guide obtained with the compounds and adjustment methods used in this study.

What did “white” smell like, and how long was it remembered?

In ratings by another 20 people, olfactory white was broadly intermediate both on a pleasant–unpleasant scale and on a scale of impressions ranging from edible to harmful. This measured impressions from smelling, not actual safety. Because all components had been diluted to match their intensities, the overall smell was not overwhelming either.

Approximately six months later, the researchers retested 20 people who had previously learned Laurax. They assigned the name Laurax to new multicomponent mixtures more often than expected by chance given the number of response options, indicating that the shared impression remained in memory.

This experiment also compared a four-component rose mixture with a mixture containing those same four components plus 26 others. With just the four components, participants answered “rose” in 70% of trials; with the 30-component mixture, “rose” responses fell to 7%, and approximately 60% of responses were Laurax. Although the original components remained, a change in the overall percept made recognition as rose less prominent.

Summary

When compounds spanning a broad range of characteristics were mixed at matched perceived intensities, even mixtures with no shared components became similar. This common impression could be learned under the name Laurax and remained a cue for classification approximately six months later.

Olfactory white is neither odorless nor a single, completely indistinguishable smell. The study showed that an overall fragrance percept arises not simply from matching molecular identities, but also from the breadth of characteristics and the distribution of intensity. The authors interpret the results as supporting the idea that olfaction combines molecular features rather than processing only information corresponding to individual molecular identities. Brain activity itself was not measured in this experiment.

The research received support from the James S. McDonnell Foundation, the European Research Council, and an Israel–Japan joint grant from Israel’s Ministry of Science and Technology, among other sources. At the time of the study, the Weizmann Institute of Science’s technology-transfer office disclosed that it was preparing a patent application relating to applications of olfactory white. One coauthor was affiliated with Via-Scent at the time.

Reference

Weiss T, Snitz K, Yablonka A, et al. Perceptual convergence of multi-component mixtures in olfaction implies an olfactory white. Proceedings of the National Academy of Sciences. 2012;109:19959–19964. Original research paper