There are few entities in the world where the question of how things taste, and what they taste like, looms larger than with the global flavoring giant McCormick. One day, I drove from my Brooklyn home to suburban Baltimore to visit the company’s headquarters and observe a group of people at work who are, in essence, paid to taste.

As I pulled up, I detected a faint whiff of a spice that I could not quite give a name to. When I met Marianne Gillette, the company’s vice president of applied research, she noted that many old Baltimoreans, back when the company was based on the waterfront, “associated the smell of Baltimore with McCormick.”

Everyone “remembers the cinnamon,” perhaps no surprise, because not only is it a best-selling spice but, citing the company’s internal research on food and emotion, Gillette told me that “cinnamon is the most loving spice.” It is a virtual memory pathway—for many people, one of the most potent smells of early childhood (I remember the oblong white McCormick cinnamon tin in a way that I do not remember the company’s oregano). Although many of us may associate the McCormick name with tins or bottles of spices, much of the company’s business now comes in providing “custom flavor solutions” for products higher up the food chain. “We are in every aisle of the supermarket,” Gillette told me, and in any number of “quick-serve casual restaurants.”

To visit the company’s flavor laboratories is to see where the human vagaries of flavor meet the certainties of hard science. As I spied a few test tubes on a workbench, Silvia King, McCormick’s white-coated chief scientist, invited me to take a whiff. It smelled like tomato or, more accurately, the smell that remains on your fingers after you touch tomato leaves. “We had a customer come in who had a processed tomato product,” she told me, “that was really lacking in that fresh tomato profile—where you pull it off the vine.”

And so, looking to add some “top note”—the first thing you taste—McCormick’s researchers have at their disposal several thousands of molecules, each derived from natural compounds, to impart that fresh tomato flavor. These molecules are added in ridiculously trace amounts. “This thiazole here,” she said, pointing to a test tube, “one drop of that into an Olympic-size swimming pool would make that pool smell like a tomato plant.”

Excerpted from You May Also Like by Tom Vanderbilt.Excerpted from You May Also Like by Tom Vanderbilt.Penguin Random House

To identify what is imparting that tomato flavor to tomato, the lab has any number of tools at its disposal. King pointed to a computer screen, across which bobbed jagged peaks. “This is a gas chromatogram; it’s like a recipe to me. Every one of these peaks represents a specific compound.” So you might blend the thiazole with dimethyl sulfide (which smells like creamed corn) or fennel ethyl alcohol (which smells like “roses or beer,” King says) or isovaleric acids (chocolate and cheese). “Once you blend these,” she said, “you can come up with a really wonderful top note,” the very thing that often disappears as a tomato is cooked or ages. In the company’s huge databases lurk the “molecular fingerprints that will distinguish Mexican from Israeli oregano.”

The only problem with this machine knowledge of flavor is that humans are not machines. We do not taste something and provide, the way a refractometer would, a Brix number (the sugar content by percentage of weight in a liquid solution). Humans bring their own sensory and interpretive mechanisms to things. Seemingly minor changes in flavor have all kinds of ripple effects for humans, things machines would miss. Add a small amount of vanilla extract to low-fat milk, and the milk will suddenly seem, to the human taster, not only sweeter but creamier and thicker—even though vanilla extract does not actually change the sweetness, fat content, or viscosity.

Nailing down what a flavor actually means, to us, takes work. A customer will come to McCormick needing, as King noted, an avocado flavor. “They’ll say, ‘I don’t even know where to start with developing this avocado flavor. I don’t know whether I want a guacamole-type avocado or a freshly peeled avocado.’ ” The team will produce an entire range. At the consumer level, Gillette said, it gets even more confusing. “When a consumer says, ‘I want avocado or guacamole,’ they will have a very clear concept in their mind—which actually might not be avocado at all. It could be more like corn or the lime note in guacamole.” Or an “anchor” guacamole they have every weekend at their favorite Mexican joint.

Meet the Human Chromatographs

In a search for clarity, McCormick turns to what Gillette jokingly terms its “human chromatographs”: trained sensory panelists. I joined Jason Ridgway and Tess Aldredge, two of McCormick’s senior sensory analysts, in a small room that faces, via a two-way mirror, a dimly red-lit room with a round table, around which a number of people were slowly nibbling pretzels from small paper cups. “They taste under red lights,” Gillette had told me, “because if they taste two gravies and one is more brown than the other, our senses will say one is meatier and richer than the other.” The absence of light and color makes the tasters’ job harder, Aldredge said. “You don’t get the bias of ‘it’s red, it’s going to taste like strawberry.’ You have to think it through. It’s very taxing psychologically.”

Ridgway flipped a switch, and audio from the other room filtered in, like a transmission from a distant spaceship. The panel’s director was asking the panel about “persistence of crisp,” which Ridgway defined for me as the “time to change in total quality during chewdown.” A “persistence of crisp” scale was provided to the panelists, ranging from cornflakes all the way up to Pringles potato chips. He asked, “Is anybody else getting the burnt-like pieces?”

Then someone uttered the word “musty,” which caught my ear. What does mustiness have to do with pretzels, and does anyone actually want a musty pretzel? “One thing to keep in mind when you hear terminology like musty or burnt,” Ridgway told me, “is that they’re describing the product. It’s not a negative thing to have a musty or burnt note.” Aldredge added, “Musty shows up in a lot of products. Often they’ll describe these little water samples”—she pointed to one of the palate-cleansing paper cups—“as slightly musty.” “They can usually tell when we need to change the filters on the filtration systems,” said Ridgway. Even though mustiness does actually stem from a chemical compound—alpha-Fenchyl alcohol—Ridgway says a term like “musty”—or “wet dog” or “dirty socks”—is meant to be a bit more “user-friendly.” He cautioned that “everybody has a slightly different interpretation of what musty is; you hear everything from a wet basement to old books.” Aldredge chimed in, “Mine is hose water.”

Words must be chosen carefully: The mere presence of a sensory descriptor (“fresh pineapple”) may be enough to suggest its actual presence in the food (cheddar cheese), leading panelists to look for “phantom” attributes that are not really there, sending testers down a false path—or consumers down a path they may not want to travel. “We’ve done things before where somebody might say, ‘I like that cheese,’ and we go through and describe it, using all the descriptors, and all of a sudden they realize, ‘Hey, this really does smell a lot like baby vomit.’ They may not even use that cheese at home anymore.” Every food is transformed. Onions have a “rubbery note.” Mango exudes sulfur. “A really good papaya,” says Gillette, “has a distinct aroma of garbage to it.” (Note to self: If forced to eat garbage, think of papayas.)

Perhaps unsurprisingly, given the language on offer, the one question you will never hear asked in a sensory panel is the one we would probably first ask upon tasting: Do you like it? One problem is that expressing liking or disliking is liable to change the way sensory analysts experience the product. Also, expert tasters often tend to not like the same things as nonexpert tasters (that is, consumers). The title of one study, in The Australian Journal of Dairy Technology, says it all: “Cheese Grading Versus Consumer Acceptability: An Inevitable Discrepancy.”

Introducing preference introduces noise. As the influential food scientist Harry Lawless puts it, “You would not ask a gas chromatograph or a pH meter whether it liked the product, so why ask your analytical descriptive panel?” If you ask panelists to identify whether product A or B has more salt, and then which one they prefer, what do you do when people guess the wrong way on the salt test? Is their hedonic judgment still valid? Then there is the problem that not everyone likes the same thing or likes it for the same reasons. An early version of McCormick’s spice lexicon had a category for “off” notes, including “soapy” for cilantro. But casting soapy in this pejorative light rather ignores the idea that soapiness is part of the very nature of cilantro. Sensory lexicons go out of their way to avoid words that might connote quality or that may mean different things to different people. A lot of stereotypical “wine talk”—words like “well-rounded” or “chewy”—is not included in wine’s professional sensory lexicon.

Another issue is that sensory testing does not happen in environments where food is actually consumed. Something that seems good enough in a small sample in a lab setting may be less appealing at the dinner table. You can break down the sensory profile of a soft drink in the lab, as Nancy Farace, McCormick’s food insights manager, told me, but released into the wild, into consumers’ hands, it may become a different beast. “How’s the consumer going to drink it? Will they put it over ice? At thirty-two degrees? Will it be in a plastic cup, a glass cup, or in a can, or through a lid with a straw? What might they eat beforehand? Afterwards? It’s not only knowing what they like but how and when they’ll consume it.” Simply asking people if they like something, as the Dutch food researcher E. P. Köster has pointed out, puts them suddenly on alert: They “taste more attentively and judge probably on different criteria than when they just eat it.” On the other hand, asking average consumers to think more analytically may interfere with how they implicitly feel.

In short, in the food industry at least, you generally do not want to know what experts like (because most consumers probably won’t like it), and you do not want to probe too deeply about why consumers like something (because they won’t be able to explain it in useful terms). Take coffee. Bitterness is the essential thing that people like about coffee as a sensory attribute but you would never want to use the word to sell it. But if the ultimate goal is to sell things that most people will actually like, why bother with this rarefied sensory testing by people whose powers of discrimination are pitched at such a higher frequency? As Gillette described it, one reason is to calibrate sensory compounds with consumer taste. Instead of going to panels with fifty vanillas and trying to figure out which ones consumers might like, you can analyze the flavor profiles of the things they actually like and use that as a starting point for making new products.

Gillette had cautioned me that sensory experts were people whom “you didn’t want to go to lunch with.” And yet there we were, in a McCormick canteen, eating udon noodles with katzu and oregano, among other flavor-forward delicacies. As a trained panelist, she said, “you’re really sensitive to aging and oil. An oil starts to go bad after a little bit; the consumer will be happily eating something that we would consider rancid, and wouldn’t even notice it.” Aldredge told me she will suddenly stop eating during a meal. “Your friends are like, ‘Why did you stop eating?’ You say, ‘I really don’t want to talk about it.’ And then they’re really interested.”

What Does Dr Pepper Really Taste Like?

My attention somehow drifted to a can of Dr Pepper, one of a number of drinks offered on a nearby table. I realized that I did not have a good sense of what the flavor of Dr Pepper was, nor, admittedly, had I invested much thought in it. My thought was that it tastes “like Dr Pepper.” How would I describe its various qualities to someone who had never had one? Clearly, the company uses this epistemological murk to its advantage, prominently advertising “23 flavors” right on the can. This invokes an appetizing mystery: What could those flavors be? Surely 23 is better than 11!

That mystery indeed informs the heritage of the brand. In the 1960s, the perception of Dr Pepper, notes Joseph Plummer, was riddled with misconceptions: It was medicinal or made from prune juice. But the company was able to turn the eccentricities of this brown not-tasting-of-cola drink into strengths. By the early 1970s, it was the country’s fourth most popular soft drink. Not being able to identify a precise flavor can actually be a strength. As Howard Moskowitz had suggested to me, part of the popularity of Coca-Cola versus, say, an orange soda is its more complex flavor blend. Consumers tire of it less quickly than of an orange soda, which has a simpler, more recognizable profile (which might be “easier to like” on the first go-round). The more you can identify any one flavor, Moskowitz said, the more it sits in memory and thus is easier to remember.

As it happens, I am agnostic on Dr Pepper. It is not something I generally seek out, nor is it something I would reflexively avoid. Whatever my level of liking, it is a feeling for the thing as a whole, instead of an analytical distribution of various sensory and trigeminal attributes. My feelings may be partially informed by exposure. Dr Pepper has a southern regional identity, and not having been raised in the South, I have not had as many chances to consume it. But could it be a lack of appreciation as well? If I knew more about Dr Pepper, would I like it more?

It occurred to me: What better time than sitting with a bunch of sensory experts to have a taste test? “Let’s train Tom on the aromatics of Dr Pepper,” Gillette announced. I bring the glass to my nose. “What does it smell like to you?” Ridgway asked. “If you can’t describe it, what does it remind you of?” There was something, but it evaded me. I could almost feel some frustrated synapse waiting to be fired that would connect my sensory machinery to my memory. Gillette, sensing my struggle, put her nose to the glass. “I smell something that isn’t a beverage. It’s something that I love to eat for dessert.” An image shimmered at the edge of my mind. “It’s always tough when you don’t have the language first,” Ridgway said, consoling me. Gillette gently asked me if she could help. “It reminds me of burgundy cherry ice cream. The vanillin, the creamy note, the black cherries.”

It was as if a door had been opened. I smelled it again, and there it was, hanging like a sign right in front of me; how could I have missed it? I clearly knew the smell. Did I have some memory of what I thought it was, and did it take that terminology to bring forth the memory? Odor is famously talked about as a strongly evocative triggering mechanism for memory (particularly when the smell is unpleasant). But what triggers the memory for odors?

Science is rather divided on whether words (that is, “semantic mediation”) are essential in triggering odor memories or whether smell memory fundamentally works on its own. Regardless, it struck me as curious that I could be having this clear sensation, smelling Dr Pepper, knowing that it was not Coca-Cola or 7UP but not really knowing what it was. How much of life itself comprises this sensory sleepwalking, these subconscious perceptions? How different is this sensation from hearing a piece of music from an unrecognizable genre or not being able to make out something in the distance with our eyes?

Paying too much attention could drive you crazy, I thought as we compared tasting notes. “I thought it was pruney,” Aldredge said. Someone else countered, “I went nonfood—mulch.” “Ah!” Gillette said, eyebrows raised. “There are some earthy notes,” Aldredge said. “It’s woody.” A bit sheepishly, I proffered, “Some clove?” “Maybe,” Ridgway responded evenly. In any case, no one in the room was going to nail all twenty-three flavors; remember the finding that people begin to peak after identifying three compounds. We were talking about flavor, using language to unlock our senses, forging new memories—and thus future tastes—in the process. As Gillette told me, “You will never experience Dr Pepper the same way again.”

Excerpted from the book YOU MAY ALSO LIKE: Taste in an Age of Endless Choice by Tom Vanderbilt, copyright 2016 by Tom Vanderbilt. Published by arrangement with Alfred A. Knopf, an imprint of The Knopf Doubleday Publishing Group, a division of Penguin Random House LLC.

See more here: 

Inside the Lab That’s Inventing Your Next Favorite Flavor