Flavor

by Bob Holmes

  The same thing happens when you eat a mouthful of food. The bite delivers a whole suite of sensations—tastes, smells, texture, temperature, perhaps some crunch—all at once. The mind packages them together into a single experience, and assigns that experience to the mouth, where the most prominent physical stimuli occur. No one notices that some of the sensations—notably the food’s aroma and the sound of its crunch—actually come from somewhere else.

  With their penchant for taking things apart to see how they work in detail, neuroscientists have explored this further, of course. One particularly squirm-inducing experiment involved threading volunteers with a series of plastic tubes so that researchers could puff smells either through the nostrils or up the back of the throat, at the same time that they squirted scentless, odorless water into the mouth. The subjects identified the front-of-nose scents as smells coming from the outside world, but called the back-of-nose smells “tastes,” and perceived them in the mouth.

  We’ve just seen how the brain binds together sensations that come in at the same time, treating them as a single, unified flavor that can be greater than the sum of its parts. But as it turns out, not every set of simultaneous sensations gets bound into a unified perception. “In order to combine as a flavor, they need to be viewed as similar, as things that go together,” says Johan Lundström, a sensory researcher at the Monell Chemical Senses Center.

  As an example, Lundström, who’s Swedish, points to an unpleasant experience common in kitchens back home. In Sweden, as in much of Europe, milk is often sold in cardboard cartons that cannot be resealed tightly after they’re opened. When he happens to store an opened carton in the fridge with a leftover half of an onion, the milk picks up oniony odors—and that sensory dissonance makes it impossible for him to drink the rest, even when he knows it’s fresh. “You cannot make yourself drink the milk,” he says. “Your system is screaming at you that there’s something wrong here.”

  That is healthy caution. One of the reasons we smell and taste in the first place is to make sure we don’t eat something we shouldn’t, so most people react with dislike to new or peculiar flavors, especially if they come as a surprise. (Adventurous foodies who know what they’re getting into have other coping mechanisms that can override this aversion, says Lundström.)

  This “go togetherness” turns out to be an important part of sensory integration. Lundström recounts an as-yet unpublished experiment that some of his colleagues at Monell did to test whether the brain is better at integrating flavors that go together—and whether we can teach our brains new flavor combinations.

  Paul Breslin, Pam Dalton, and their collaborators made up special chewing gum that allowed them to give tasters tiny, measured doses of an aroma (rose scent) and either a bitter or a sweet taste. First, they figured out the minimum dose that people could detect of each scent or taste on its own, then dialed back the amount even more so that they ended up with what should have been flavorless gum. Then they paired taste and aroma in either familiar (rose–sweet) or unfamiliar (rose–bitter) combination gums. Sure enough, they found that people could taste the flavor in the gum with the familiar rose–sweet combination, indicating that they could integrate these two components, just like lip-reading at a noisy party. The people still couldn’t taste anything in the rose–bitter gum, though, which shows their brains didn’t know how to combine these discordant flavor stimuli into one integrated perception.

  But then Breslin and Dalton took their experiment one step further: They made a new rose–bitter gum, but this time with enough of each flavoring that chewers could taste it. Volunteers chewed this peculiar gum daily for a month, then returned to the lab to see if their experience had changed their ability to taste the original, low-dose gum. And indeed it had—proof that after a month of practice, their brains had learned to integrate rose and bitter just as they had once automatically combined rose and sweet. “You can definitely train the system that these go together in a relatively short time,” says Lundström.

  All this evidence—from the gimmickry of Hobkinson’s strange dinner to Lundström’s oniony milk to Spence’s Pringles and barnyard oysters—suggests that flavor is not what we usually think it is. Gordon Shepherd puts it best: “A common misconception is that the foods contain the flavors,” he says. “Foods do contain the flavor molecules, but the flavors of those molecules are actually created by our brains.”

  The notion that flavor lives in the mind, not the food—or even the mouth or nose—is startling enough. But it looks like we can even go a step further: We construct our notion of flavor almost from scratch, building it from the ground up as we experience the world. Sure, a few preferences do appear to be hardwired, such as a liking for sweet tastes and an avoidance of bitter ones. But even those can be overridden by experience, as anyone who drinks gin and tonics can attest. And once we get to more complex flavors, it’s clear that most of our perceptions and preferences are based on experience. To really understand how our brains create our experience of flavor, we need to dig into the details of what happens in the brain as we taste. First, a little background.

  Psychologists generally think of the brain much like a layer cake. The bottom layer is made up of raw sensations—tastes, smells, touch, and so on. On top of that is a layer of synthetic perceptions, where raw sensations are assembled into objects: a series of shapes, colors, and shadings become a face, for example. Crowning the cake are one or more “cognitive” layers—exactly how many is a matter of debate—where higher-order thought takes place. Here, for example, we attach a name to the face and develop expectations of how that person will behave, how important they are to us, and so forth. For flavors, these cognitive layers are responsible for identifying and naming flavors, deciding whether they’re good or bad, and choosing whether to eat something or not.

  In this standard picture, all the information flows upward, with lower levels serving as data for higher processes. If there’s no reverse flow, we’d expect the lower levels, the sensations and perceptions, to be “clean”—that is, driven purely by the sensory inputs themselves, and unaffected by any preexisting cognitive or emotional baggage. But we’ve already seen that’s not exactly true, since experience can modify the way we bind sensations together. So what’s going on here?

  That’s the question Edmund Rolls, a neuroscientist then at Oxford University, set out to answer. Rolls is one of the grand old men of sensory neuroscience, and many research paths in the subject led through his lab at one time or another. Rolls got to thinking about the particularly pungent spoiled-milk product that we know as cheese. Most people from Western nations like the stuff, while many Asians find it disgusting. (The tables are turned, of course, when it comes to Asian delicacies like aged duck eggs or the slimy, rotten-soybean preparation the Japanese call natto.) We know that cultural experience affects our liking for these foods. But, Rolls wondered, could these cognitive-level concepts reach back down and modify the raw perceptions, too?

  To find out, Rolls and his student Ivan de Araujo devised yet another bit of psychology lab trickery. They prepared a synthetic “cheese flavor” and gave it to volunteers to smell. Half the volunteers read a label describing the odor as “Cheddar cheese,” while the other half saw it labeled as “body odor.” If you’ve been following along to this point, you won’t be surprised to learn that the first group liked the smell better than the second group.

  But then Rolls and de Araujo dug one step deeper, using brain scans to peer into the subjects’ brains. There, they did find a surprise: the two groups’ brains lit up differently all the way down to the second layer of the cake, the regions responsible for basic perceptions, even though nothing had changed except the words that described the odor. In other words, higher-level thought processes—and it’s hard to find a level much higher than language—can change not just how we think about flavor perceptions, they can change the perceptions themselves. Thought itself, in other words, is one of our flavor senses. The brain construct
s flavor by piecing together inputs from virtually every one of our sensory channels, plus inputs from thought, language, and a host of other high-level processes like mood, emotion, and expectation. That makes flavor a remarkably complex and changeable concept. It’s a wonder we can talk about it coherently at all.

  Actually, maybe we can’t. Perhaps our flavor perceptions are so individual, so idiosyncratic, so circumstance dependent that we’re fooling ourselves when we think we’re saying anything objective about flavor. That’s certainly the impression you get when you look more closely at wine. Wine should be a perfect test bed for exploring the reliability of our flavor perceptions. No other foodstuff is so thoroughly, obsessively described and quantified. Detailed tasting notes are available—usually from not just one, but several trained, professional tasters—for almost any wine available commercially. Not only that, but those tasters often assign numeric scores to every wine, too, allowing for numeric comparisons of quality. Wine, if you have the right mind-set, is where the world of food meets Big Data.

  Bob Hodgson has the right mind-set. An oceanographer by training (now retired), he’s also owned a winery in northern California for forty years. Like any other professional winemaker, he enters his wines in competitions such as the California State Fair, where trained judges taste their way through hundreds of wines and hand out coveted gold medals to the best—medals that can make or break a wine’s salability on store shelves. Sometimes Hodgson’s wines won gold medals, sometimes they didn’t. But unlike most winemakers, he didn’t just shrug at the injustice and carry on. With his scientific turn of mind, Hodgson started to wonder why the very same wine could garner a high score last week and a low one this week. Could you really trust the judges’ scores, he wondered? Hodgson must be a persuasive guy, because somehow, he managed to convince the California State Fair to let him find out.

  Judges at a big competition like the California State Fair taste about 150 wines every day, organized into 4 to 6 “flights” of 30 wines each. The wines within a flight are presented in identical glasses marked with identifying codes, so that no judge knows the identity of any wine he or she is tasting. Each judge individually—no discussion at this stage of the judging—gives each wine a numeric score on a 20-point scale. (Actually, the fair uses a 100-point scale like the ones you sometimes see on the shelves at your local wine shop. But any wine that’s halfway drinkable scores at least 80 points, so for all practical purposes it’s a 20-point scale.)

  With the collaboration of the contest organizers—but unknown to the judges—Hodgson arranged that for one flight per day (usually the second), three of the thirty wines would actually be identical samples, poured from a single bottle of wine but given different code numbers. If judges’ scores are a true reflection of a wine’s quality, then you’d expect these triplicate samples ought to receive identical scores—or at least somewhat similar scores, allowing for a little bit of imprecision in the judges’ ratings.

  The results were shocking. “We did everything we could to make the task easy for the judge: same flight, same bottle. And nobody rated them all the same,” says Hodgson. Only about 10 percent of the judges scored the three samples similarly enough that they awarded the same medal to each. Another 10 percent gave wildly different scores, giving one glass a gold and another a bronze or even no medal at all, and the rest fell somewhere in between. And that wasn’t just because some judges are better than others: judges who were consistent in one year were no more likely to be consistent the next year.

  Hodgson wasn’t done. Next, he compared the results of wines that had been entered not just at the California State Fair but in other major wine competitions as well to see whether wines that aced one competition did well in others, too. As you can probably guess by now, they didn’t. Wines would often win gold in one competition, nothing in another—and not a single wine out of more than twenty-four hundred picked up gold every time. The competitions might as well have handed out gold medals at random, Hodgson calculated.

  So what’s going on? The answer is that people’s perception of a wine changes from moment to moment depending on the circumstances. The wine will taste blander if it follows a robust, fruity wine than if the previous wine was subtle; a particular aroma might have triggered a fond memory (and a higher score) for one glass but not the next one; the judge might have gotten tired as the flight progressed; they might have been distracted by a ray of sunlight or a twinge from an arthritic knee. All of that adds noise to the judge’s rating—so much noise, Hodgson thinks, that it obliterates any real differences in quality. Maybe, in fact, it’s just not humanly possible to judge wines objectively, especially in the crowded, rushed, overwhelming setting of a state fair.

  Hodgson sees this variability at work when he drinks his own wine, too. “Since I have a winery, and I’m cheap, I drink my own wine all the time,” he says. That’s no hardship, because he generally thinks he makes excellent stuff. But even so, he’s not always in the mood. “Sometimes I think, Jesus, I don’t like this wine. But I know not to get upset, because tomorrow is a different day.”

  All this points to an uncomfortable conclusion: If trained judges and experienced winemakers don’t consistently prefer one wine over another, then maybe there’s no real basis for calling some wines great and others merely good. And that may be how it really is, though it’s hard to find many wine people who will agree. “I would like to think that Mouton Rothschild is a better wine than Gallo Hearty Burgundy,” says Hodgson. “You and I may agree that one is better—but we may not agree.” Other studies, he notes, have found that ordinary wine drinkers, those of us without special training, tend to prefer cheaper wines over more expensive ones—but only if no one tells us what the price is. If you know the price, on the other hand, that high-level knowledge has a powerful influence on how you perceive the wine’s flavor. Almost everyone tends to think a more expensive bottle of wine tastes better than cheap stuff—even when all that’s changed is the price tag. That sounds like self-delusion—but there’s more to it than that, as one team of researchers learned a few years ago.

  A brain scan is not the ideal setting for savoring wine. For one thing, you have to hold your head perfectly still, which precludes all the sniffing, swirling, and other ceremony that usually accompanies a sip of wine. Instead, you get a tiny dollop—a single milliliter, about a quarter teaspoon—of wine dripped straight into your mouth through a polyethylene tube as you lie in the scanner. But at least researchers can see exactly what the wine’s doing to your brain. For the experiment we’re interested in here, the scannees got sips of what they thought were five different wines of varying price, but in fact four of the wines were paired duplicates: a five-buck plonk was also presented as a $45 bottle, and an exquisite $90 Napa cabernet also appeared under the guise of an everyday $10 wine. Sure enough, the tasters liked the wines better when they appeared with a higher price tag. But the brain scans showed that they weren’t just saying so—the “higher-priced” wines activated the brain’s reward circuitry more than the same wines presented at a lower price. In other words, a higher price tag genuinely led to greater pleasure! As one observer noted wryly, this means that if you’re hosting a dinner party, you can maximize your guests’ pleasure by serving them a cheap wine (which most drinkers prefer in a blind tasting) and telling them it’s expensive.

  As Rolls and other neuroscientists trace the flow of flavor through the brain, their attention comes back again and again to one particular spot, right behind the eyes at the front of the brain. Neuroanatomists have a daunting catalog of tongue-twisting names for parts of the brain, most of which only an expert would need to know. But this little region, known as the orbitofrontal cortex, or OFC, deserves to be more widely known to anyone with an interest in flavor. The OFC, researchers are learning, is one of the key areas where the brain knits together the independent threads of taste, smell, texture, sight, and sound—together with our expectations—into the common cloth of a flavor perception. It
’s not stretching the facts to call the orbitofrontal cortex the birthplace of flavor.

  (As is almost always the case with the brain the reality may be more complex. Another nearby brain region called the frontal operculum could also be a candidate for Flavor Central. In one recent study, researchers monitored brain activity while giving volunteers the odor or taste of orange juice either separately or together. The frontal operculum, but not the OFC, lit up far more strongly to the combined flavor stimulus than you’d predict from its response to smell or taste alone, suggesting that the frontal operculum may be another key area where flavor is constructed.)

  If the OFC is where flavors are born, then it may also be the place to turn if you want to know what a flavor looks like in the brain. And, in fact, that’s just what Rolls and his colleagues have done. By recording the electrical activity of individual nerve cells, or neurons, within the OFC of rats, they’ve found that each neuron there responds to a different set of inputs. One might light up in response to a sweet taste, a pepperlike aroma, and the mouthfeel of capsaicin, the molecule that makes chili peppers hot; another might turn on to sweet taste, a vanilla aroma, and the mouthfeel of fat. You could call the first cell a “chili pepper flavor” neuron and the second an “ice cream flavor” neuron.

  This mapping of particular flavors onto individual neurons helps explain why the first bite of, say, ice cream tastes so much better than the twentieth bite, and why we can eat our fill of stew yet still have room for pie. In essence, Rolls says, a particular flavor neuron gets tired after responding to its flavor over and over, a fatigue he calls “sensory-specific satiety.” He’s actually measured exactly this in monkey brains, showing how repeated doses of a particular flavor combination provoke smaller and smaller responses from its specific flavor neuron.