Flavor

by Bob Holmes

  But counting up odor receptors doesn’t tell the whole story of smell, because there’s another whole layer to the way we perceive odors that isn’t there for taste. Our sense of taste is what sensory scientists call analytic—that is, we easily break it down into its component parts. Sweet and sour pork is, well, sweet and sour. Soy sauce is salty and umami. Ketchup is sweet, sour, salty, and umami.

  Our sense of smell doesn’t work that way. Instead, it’s a synthetic sense: Our brains assemble the component parts into a single, unified perception, and we can’t easily pick out the parts separately. That’s easiest to understand if you think about another synthetic sense: vision. When I gaze fondly at my wife, I don’t see lines, curves, and edges, even though that’s what my brain is actually detecting and processing. I just see her face, the synthetic object of my perception. Similarly, the individual odor molecules sensed by our nose can combine in our brain to create a new perception that’s entirely different from its components. If you combine ethyl isobutyrate (a fruity odor), ethyl maltol (caramel-like), and allyl alpha-ionone (violetlike) in the proper proportions, for example, what you smell is not caramel-coated fruit on a bed of violets, but pineapple. Similarly, one part geraniumy 1,5-octadien-3-one to one hundred parts baked-potatoey methional smells fishy—something neither ingredient shows the least hint of alone.

  Neuroscientists like to refer to these new, higher-level perceptions as “odor objects.” Each one is, in effect, a unique pattern of activation involving a subset of the four hundred or so different kinds of odor receptors in your nose. In essence, these odor objects define reality in our olfactory worlds, just like my wife’s face is a visual object that seems more real to me than its component lines and curves.

  And in the same way that you can create an essentially infinite number of faces out of a smallish set of lines and curves, our four hundred odor receptors can give rise to a dizzying number of different odor objects. A few years ago, researchers gave people mixtures of ten to thirty different odor molecules and asked whether they could tell them apart. Based on those results, they calculated that people ought to be able to distinguish at least a trillion different odor objects—a big step up from the fabled ten thousand smells of received wisdom. (By comparison, sensory scientists say our eyes can perceive a few million different colors and our ears maybe half a million pitches.) Since then, other researchers have pointed out that the “one trillion” number should be treated with caution, since it depends on several iffy assumptions. However, the general message—that the universe of smells is a huge one—still stands.

  To understand how the brain processes these odor objects, I sought out Gordon Shepherd, one of the grand old men of olfaction research. Nearly everyone I spoke to at the Association for Chemoreception Sciences meeting in Florida, in fact, made a point of saying, “You should talk to Gordon Shepherd.” Some even suggested that he’s been so important to research on the neuroscience of smell that he deserved a share of the Nobel Prize for his work. He’s also written a terrific book, Neurogastronomy, about the biology of flavor perception.

  When I caught up with Shepherd on the resort patio, I found a courtly, white-haired man in a red wool sweater, who was happy to spend the afternoon talking about olfaction. Odor objects have a physical equivalent in the brain, he told me. Each one of the nose’s four hundred odor receptors delivers its signal to a different part (or parts) of the brain’s olfactory bulb, the first relay station for odor information. If you imagine the olfactory bulb as a switchboard panel with lights corresponding to individual odor receptor types, then each odor object is represented by a distinct pattern of lights—its own olfactory image, in effect. But when your brain comes to process that pattern of lights, it doesn’t know whether they’re the result of a single odorant molecule or many: it just sees the pattern.

  And we’re generally very bad at articulating complex patterns, says Shepherd. Just try to describe the face of someone familiar to you, or the art of Cy Twombly—you’ll probably struggle just as much as most people do in expressing the aroma of a beefsteak tomato or an artichoke. “It’s the same problem,” says Shepherd. “A highly complex image that’s almost impossible to describe in words.”

  That certainly matches most people’s experience of talking about smells—and by extension, about flavor. Putting names to smells is something humans in general are “astonishingly bad at,” says Noam Sobel of the Weizmann Institute of Science in Israel, one of the most creative, and consistently provocative, smell researchers. To prove this incompetence to a skeptical family member, Sobel once asked her to close her eyes, then he pulled a jar of peanut butter out of the fridge, removed the lid, and waved it under her nose. Even though his relative ate peanut butter almost every day of her life, she couldn’t name that familiar smell. You can repeat the test yourself: Close your eyes and have a friend present you with some familiar household odors, and see how many you can identify. You’ll probably find, as Sobel and other researchers did, that you recognize all of them as familiar, but you can’t name even half of them successfully. (I once failed to identify the flavor of coffee, which at the time was my every-morning breakfast drink.) As one of Sobel’s colleagues is fond of pointing out, if you or I did that badly at naming colors or shapes, we’d go straight to a neurologist to see what’s wrong.

  Another big reason we’re so bad at naming smells is that our brains process odor information—one of our most ancient senses—much differently than they handle newer senses like sight and hearing. Sights and sounds take an express route to the thalamus, the part of the brain that acts as the gatekeeper of consciousness. We’re wired to pay conscious attention to them. That direct line also means that sights and sounds have rapid access to the newer, more powerful brain regions that handle speech and language. In contrast, olfactory signals go first to the ancient, preconscious brain regions that control emotion and memory, the amygdala and hippocampus—which helps explain why smells are so powerfully evocative—and don’t pass through the gateway to consciousness and language until several stops later.

  But there’s a second reason for our difficulty. In English—and most other Western languages—we pretty much lack a distinct vocabulary for describing odors. We describe smells, if we can describe them at all, by saying what they’re like: a New Zealand sauvignon blanc smells grassy, we say, or a furniture polish smells lemony, and that’s about the best we can do. Here’s an English-speaking American trying to put a name to the smell of cinnamon: ‘‘I don’t know how to say that, sweet, yeah; I have tasted that gum like Big Red or something tastes like, what do I want to say? I can’t get the word. Jesus, it’s like that gum smell like something like Big Red. Can I say that? Okay. Big Red. Big Red gum.” You’ve probably flailed about in a similar way trying to describe a smell—I certainly have. But we don’t do that for colors, for which we do have a specialized vocabulary. We don’t have to describe the colors of the Swedish flag, say, as lemonlike and skylike—we can call them yellow and blue.

  And as it turns out, some cultures do that for smells, too. For a startling example of just how much better we could be at recognizing, identifying, and talking about smells, consider the Jahai, a small tribe of nomadic hunter-gatherers in the mountains of southern Thailand, near its border with Malaysia. The Jahai language has more than a dozen words to describe smells, none of which relate to the smell of any particular object. The Jahai might say, in their language, that something smells “edible” or “fragrant,” or, my favorite, “attractive to tigers.” Some of the actual concepts they’re expressing make no sense at all to an outsider—the word for “edible,” which sounds a bit like “knus,” is applied to gasoline, smoke, bat droppings, some millipedes, and the wood of wild mango trees, none of which strike me as particularly edible; “fragrant” includes several flowers and fruit, some other kinds of wood, and a species of civet cat.

  However bizarre it seems to us, that specialized vocabulary makes it much easier for the Jahai to think and
talk about smells. When researchers gave a standard smell-identification test to ten Jahai men, they found that the Jahai tended to be quick and consistent in describing the smells, even though most of the actual odors used in the test were unfamiliar to them. In fact, the Jahai proved to be just as comfortable describing odors as they were at describing colors. By comparison, ten English-speaking Texans were quick and precise at describing colors, but all over the map when it came to the smells. (One of those Texans was the source of the hopelessly inarticulate description of cinnamon quoted above.)

  Fortunately, vocabulary is something we can learn with a little effort. Even we Westerners have specialized vocabularies for smells within certain domains. Just listen, for example, to a professional perfumer pick apart the olfactory spectrum of a fragrance, nimbly identifying the floral top notes, musky base notes, and the like. An experienced wine buff can do the same thing with what’s in her glass. In fact, tests show that wine experts’ noses are no better than yours or mine—they’ve just had more practice at noticing and putting into words what they’re smelling. Almost anyone can improve their nose for wine, no matter how hopeless they feel. As long as you can recognize that one wine is different from another, you’ve got the basic perceptual tools you need. All it takes is a little effort to nail down the vocabulary.

  But there’s a limit to how well even the professionals can deconstruct the aromas of a glass of wine or a whiff of perfume. Way back in the 1980s, Australian psychologist David Laing presented volunteers with familiar odors like cloves, spearmint, orange, and almond singly or in combinations of up to five. He provided a list of seven possible odors and asked the volunteers to check off all the odors that were present. People did okay at single odors or mixtures of two, but their performance fell off dramatically at three or more. Not a single person correctly identified all the elements of a five-odor mixture. Later studies have confirmed this result—even professional flavorists and perfumers just don’t seem to be able to correctly identify more than three or four odors in a mixture, probably at least partly because the odors interfere with one another in our nose or brain. With this in mind, I’m inclined to look skeptically at wine tasting notes that claim to identify six or eight aromas.

  Are there ways to help identify odors? That is, can we somehow sort odors into categories to make it easier to understand them? We do that for taste, after all: There’s sweet, sour, salty, bitter, umami, and maybe a few others. Color and sound are also simple to sort: It’s all about the wavelength of light or the frequency of a sound vibration. But odors are caused by thousands to billions of unique molecules, each with a different shape and each, apparently, activating a different set of odor receptors. How to make sense of all this?

  Of course, people have tried, beginning long before they knew anything about molecules. Carl Linnaeus, most famous for his method of classifying all living beings, had a go at odors, too. All odors, he thought, fall into seven categories: fragrant, spicy, musky, garlicky, goaty, repulsive, and nauseating. A contemporary, Albrecht von Haller, had an even simpler system, sorting all odors somewhere on a spectrum between ambrosial and stench. And as we’ve seen, nearly two centuries later Crocker and Henderson—the ten-thousand-odors duo—thought there might be four dimensions: fragrant, acid, burnt, and goaty.

  The list goes on and on, with many classification systems appearing bizarre when viewed from the outside. The Suya of Brazil regard odors as bland, strong, or pungent. Sounds sensible—but oddly, adult men smell bland, while women smell strong and the elderly smell pungent. The Serer-Ndut of Senegal have five categories: urinous, rotten, milky/fishy, acrid, and fragrant. Monkeys, cats, and Europeans have a urinous odor. Rotten-smelling things include cadavers (obviously), mushrooms (understandably), and ducks (um . . . ); acidic smells include those of tomatoes and spiritual beings. (Prizes for anyone who can explain what tomatoes and ghosts have in common.) The Serer-Ndut themselves have a fragrant odor, the most positive of the five categories—but then again, so do onions.

  Any classification system that uses words (and their underlying concepts) is bound to suffer from cultural blinkers. You name what’s important to you—and that, overwhelmingly, is what’s in front of your nose from day to day. “We” always smell good, and “they” smell bad. You can’t understand goaty odors if you’ve never encountered a goat. As we’ll see, professional flavorists sort odors into categories like fruity, floral, and spicy: basically, the kinds of ingredients they deal with in their daily work.

  Is there any way out of this cultural trap, any way to sort odors into dimensions without having to resort to language? Andreas Keller of Rockefeller University thinks so. A big bear of a man with a soft German accent, Keller works the boundary between sensory science and philosophy, making significant contributions to each. To test the dimensionality of smell from first principles, Keller sets out three vials with different odor molecules and asks people to group them by similarity. If everyone picks the same pair of odorants as most similar, he knows that those two sit together along some dimension—they’re both fruity, say. If no pair is put together more commonly than the others, on the other hand, then all three odorants must be equidistant from one another, like the points of an equilateral triangle. That means there must be at least two dimensions. Four equidistant odors require three dimensions, and so on. The concept is straightforward, even though the math gets more than a little bit hairy as the number of dimensions grows.

  Keller’s hope is that sooner or later, adding more odors no longer requires new dimensions. The big question is whether that happens after just a few dimensions—in which case odors really do fall into meaningful categories—or many. The worst-case scenario would be that there’s a separate dimension for each of our four hundred or so odor receptors, which would effectively mean that there is no underlying structure, no effective way to group smells into perceptual categories. “I think everything up to about twenty or thirty dimensions would be interesting,” says Keller. His experiments are still ongoing as I write, but he’s less and less optimistic that he’ll end up with a manageable number of dimensions.

  Our poor performance at naming and sorting smells is, no doubt, part of the reason why most people think humans are olfactory incompetents, with noses that are good for little more than keeping our glasses from falling off. But in fact, we’re too hard on ourselves. Our noses are a much more powerful tool than most of us realize—more sensitive, in many cases, than the most expensive piece of laboratory equipment.

  Case in point: If you had happened to cross the University of California at Berkeley campus in the early 2000s, you might have noticed an undergraduate—blindfolded, earplugged, and wearing coveralls, knee pads, and heavy gloves—crawling across the lawn with nose to ground, zigzagging slightly back and forth. Was he rolling a peanut across the campus with his nose as punishment for some arbitrary offense during a fraternity initiation? Was he groveling before more senior fraternity brothers? No. He was following a scent trail laid down by a chocolate-soaked string—and doing it almost perfectly.

  This rather odd spectacle was another of Noam Sobel’s slightly skewed brainchildren. (At the time, Sobel was a junior professor at Berkeley, though he’s now at Israel’s Weizmann Institute of Science.) For the chocolate-tracking experiment, Sobel and his students tested a total of thirty-two people and found that twenty-one of them could find and follow the chocolate track by nose alone, with all their other senses blocked. Better yet, when Sobel gave four of the volunteers a chance to practice repeatedly, every one of them got better at following the trail, moving faster and casting about less. When the trackers tried again while wearing a nose clip, every one of them failed to find the trail—clear proof that they weren’t navigating by some other cue that the experimenters had overlooked.

  And it’s not just that we’re less worse than we thought: our noses actually compare favorably with those of other animals—even ones renowned for their sense of smell. Matthias Laska, a psychologist a
t Linköping University in Sweden, has been measuring the acuity of animals’ noses for decades, since long before Sobel’s chocolate study. The gold standard for this sort of thing is to measure the olfactory threshold, the lowest concentration of an odorant that can be detected—exactly what Doty’s machine tried to measure for my nose. Since you can’t just ask a monkey or an elephant whether it can smell something, Laska does the next best thing: He teaches the animal to associate the odor with a food reward—a yummy carrot for the elephant, for example, or a peanut for a squirrel monkey. Then he lets the animal choose one of two boxes: one unscented and empty, and the other bearing the tell-tale scent and containing the treat. If the animal picks the treat consistently, it must be able to smell the signal, and Laska repeats the test with a lower concentration of the odorant. When he gets to the point where the animal can’t tell which box has the treat, he knows the odorant signal has fallen below the olfactory threshold.