American Scientist

Poets and novelists have long lamented our lack of a language to describe smell. Attend a wine tasting, watch a cooking show, or visit a local perfume counter, and you’ll hear visual terms such as “bright,” “shimmering,” or “sparkling.” Compared with our brief olfactory glossary, our visual vocabulary, like our sense of eyesight, is vibrant and specific. We broadly agree on colors and even systematize them by wavelength, Pantone swatch, or Crayola crayon equivalent.

The authors of an olfactory perception study published in the February issue of Scientific Data are hoping to expand our perceptual glossary to include a full-blown dictionary of odors. A better understanding of olfactory perception could help diagnose and treat people who suffer from smell sense reduction, or anosmia, a potential complication during and after COVID-19 infection. A lingua franca of scents could also support research into digital olfaction (see “Artificial Noses,” January–February 2012), which combines sensors and algorithms to detect and analyze smells. Developing such “electronic noses” is an emerging field with potential applications in the health care, food, environmental monitoring, security, consumer electronics, and entertainment industries.

But compiling a comprehensive dictionary of odors is no easy task, because smells are highly nuanced and personal. We smell scents only when molecules become airborne and waft noseward to meet our olfactory receptors—specialized proteins located on the cilia lining our nasal cavities—and reach our brain via the olfactory nerve.

The lead author of the Scientific Data paper, psychologist Antonie Louise Bierling of the Friedrich Schiller University in Jena, Germany, describes the olfactory process as “messy.” “There are all these factors—culture, background, what experience I’ve had with this odor, what genetics I have,” she says. “Genetics determines, for example, what olfactory receptors I express.” In other words, experts only partly understand the link between perception and an odor molecule’s composition and structure. Solving this puzzle, which researchers call the stimulus perception problem in olfaction, could help scientists understand why people react to smells differently.

Take benzyl acetate, a fruity organic ester. Not unlike the fictional Shimmer Floor Wax touted in the classic Saturday Night Live sketch (“It’s a floor wax!” “It’s a dessert topping!”), benzyl acetate is used in fragrances and flavorings as well as in solvents. When study participants sampled the scent of it, they split over whether it smelled like bananas or nail polish remover. “There’s no true answer to the question,” Bierling says. “The smell of the molecule is the smell of benzyl acetate.”

This unsettling lack of a one-to-one link between odor molecules and the associations and experiences they evoke in humans underlines how factors such as culture and experience affect what we smell—or even if we smell anything at all, as Bierling learned from a colleague: “We had one odor that, to our German fellows, just smelled like nothing,” she said. “I gave it to our Chinese colleagues, and they said, ‘Ugh, take it away!’ For them, it smelled like a cheesy, milky, rancid odor.”

Our tendency to react viscerally to some smells likely stems from the survival value of detecting unseen dangers. Unlike other senses, our olfactory sense determines immediately not only whether we recognize a smell, but how we feel about it. And unlike other senses, smell bypasses the thalamus—which usually processes sensory and motor information—and makes a beeline to brain areas governing emotion and memory, such as the amygdala. This emotional response fulfills an important aspect of each of smell’s three chief functions: finding safe, nutritious food; avoiding environmental hazards; and aiding social functions such as mating.

Understanding how such an evocative yet elusive sense works requires researchers to overcome significant obstacles. Language ranks high on that list. After all, how do you study something when you lack a basic lexicon for asking questions, designing experiments, and interpreting results?

Study participants split over whether benzyl acetate smelled like bananas or nail polish remover.

• Facebook
• Twitter

Perfumers rely on their training kits and fragrance wheels; sommeliers turn to their nez du vin boxes of vials and flash cards. Studies of odor have utilized terminology from both fields. But given the jolts of emotion and flares of memory fragrances can trigger, and considering people’s remarkably variable reactions, Bierling and her colleagues wanted to devise something more universal and data rich. “If you think about a wine taster telling you about a good wine, of course they’re not wrong,” Bierling says.“ But if you ask 10 of your friends, if they’re not wine experts, they’ll probably have something different to say than that wine taster. It’s a similar thing in olfaction.” So the team created a test anyone could take—one that would capture not only responses to well-established descriptors in the research literature but also open-ended responses.

The design was deceptively simple: In Germany, 1,200 young adults each received a kit containing 10 small vials—eight holding different scents selected at random from a set of 74 odors commonly used in industry and research, and two containing “anchor odors” placed in every odor set. The latter were benzyl acetate and 4-decanolide, which emits a fruity peach or apricot fragrance. These two odors, which research shows are detectable even in people with a reduced sense of smell, provided a baseline for comparing groups and guarding against potential systematic problems, such as if one kit contained nothing but unpleasant odors. Bierling says testing all 74 odors on all participants would have taken an impractical amount of time and money. “But I still wanted to have something to compare, kind of a baseline,” she says.

For consistency, the researchers balanced quantities against concentrations so that each smell would pack a similar punch. To simplify analysis, they chose monomolecular fragrances, which require only a single, isolated chemical compound to produce a scent. They also randomized the vials so people who received the same group of 10 did not sample them in the same order.

By the end, and following a few evaluations, at least 120 of the participants had sniffed each odorant. Around 70 percent performed the experiment at home, whereas 20 percent did so in the lab. Another 5 percent participated in a retest study involving six random odors one week later. For comparison, the team also recruited 120 people who had reduced olfactory performance.

Participants ranked the smells from “not at all” to “very” on eight provided terms such as “pleasant,” “irritating,” and “disgusting.” To improve comparability with past research, the authors also asked subjects to choose “yes” or “no” from a list of 16 qualitative labels created by neuroscientists Andreas Keller and Leslie B. Vosshall of The Rockefeller University in New York City, in their 2016 study of nonexpert olfactory perception. These included terms such as “fruity,” “musky,” “fish,” and “grassy.”

First, though, participants were asked to describe the odors in their own words, which the researchers then standardized by isolating and grouping descriptors. They also retained more specific phrases (“old people perfume”) and brand names (“Colgate toothpaste”) as needed. Then came the challenge of translating the standardization from German into other languages. “This is not trivial,” Bierling says. “In German, we have five or six terms that all kind of translate to ‘pungent,’ but with different qualities. We try to map them to ‘stinging’ or ‘biting,’ but it’s hard. Sometimes it doesn’t map one-on-one. You don’t want to lose a degree of resolution.”

Unlike other senses, smell bypasses the thalamus and makes a beeline to brain areas governing emotion and memory.

• Facebook
• Twitter

After all, their study sought not to uncover some physiological mechanism but rather to sketch the rough outlines of an olfactory Rosetta Stone. Consequently, the paper’s chief findings focused on ensuring the results were internally valid and consistent with prior research. For example, the authors found a strong resemblance between their findings and the patterns reported by Keller and Vosshall regarding how participants rated a set of shared smells in terms of pleasantness, familiarity, edibility, and intensity. However, respondents differed when rating substances for warmness—an inconsistency Bierling also saw among participants who took retests. “That makes sense,” she says. “Most participants didn’t really know what ‘warmness’ in an odor is. We tested it because it’s a trigeminal perception—something that might be from the trigeminal nerve.”

The trigeminal nerve contributes to somatosensation, an overlapping subsystem of our sensory nervous system that helps us relate to our environment. Whereas the olfactory nerve provides pure smell information, such as the scent of a flower, the trigeminal nerve detects the burning, cooling, or tingling sensations from chemical irritants such as ammonia or menthol (see “Perfume or Noxious Fume,” September–October 2024).

The participants who retested the same smells a week or so apart were highly consistent when assigning terms such as pleasantness, edibility, and disgust; they were only moderately less consistent in rating warmness, coldness, and intensity. The two chemicals that saw people change their minds the most during the retest were the anchor scent 4-decanolide and allyl caproate, which puts off a fruity pineapple scent.

But to get to reliable digital olfaction—using sensors, signal processing, and machine learning to electronically mimic human olfaction—Bierling says her colleagues need to pay more attention to the wide variation of smell perception among humans instead of eliminating such variation as statistical noise. “This is actually the data we need. If we want electronic noses and digitization of olfaction at some point, then we need to know if 40 percent say this smells like banana and 60 percent say it smells like nail varnish remover, not what it smells like to a panel of 10 trained individuals who have a list of 20 categories we gave them in advance. That doesn’t get us to a digitization of olfaction.”