The story is about why bananas don’t taste like bananas anymore, and it’s the first thing I teach professional beer tasters. We’re usually in a small conference room of some kind, beer samples in front of us, the mood somewhere between grudging participation and the first day of school.
It takes a good yarn to compete with free beer for attention, so I usually start by pointing out that banana-flavored candies like Runts and Circus Peanuts and Laffy Taffy taste like a child’s drawing of the fruit, and nothing like the real thing. People nod in agreement, having known this all their lives, the same way everyone knows that some drinks can taste “purple.” The story gets at their assumptions about flavor, and is useful in reorienting their ideas about how taste actually works.
Another reason I tell it, though—a better reason—is that this particular flavor demonstrates just how complex our tastes are, and just how little we tend to know about them.
Nadia Berenstein, a James Beard award-winning writer and flavor historian, says it’s like this: The chemical compound used to make “fake” banana flavor in candy and other foods actually predates the arrival of bananas in the U.S. by about a decade. It was first sold in Europe as the flavor essence of jargonelle pear, but Americans didn’t know what that was, so to market it here they used the next closest “fruity” thing. Bananas were prized in the late 1800s—food of the wealthy, a tropical bounty that the average person would never touch, carved in bas-relief next to nymphs and demigods. Perfect marketing material, in other words.
“In order to be sold as banana flavor, it didn’t have to conform to anyone’s expectations,” says Berenstein. “Many people probably tasted these flavorings long before they ever tasted the fruit.”
The thing is, the people making fake banana flavorings weren’t too far off. The chemical isoamyl acetate, which is really what they were selling, is found by the boatload in many banana varieties. They just don’t all have it in equal amounts.
The original banana to hit the United States was the Gros Michel. It was a hardy cultivar that could be farmed on large plantations and survive shipping. This was monoculture farming: one strain of one type of plant in isolation, for miles and miles. Americans weren’t getting the best-tasting or the most aromatic bananas in existence—they were getting the ones that could survive a container ship and still look presentable. Still, the fruit’s popularity soared, so much so that the song “Yes! We Have No Bananas” was a 1920s chart-topper.
The history of monoculture farming is a history of collapse. The 1950s saw the Gros Michel nearly wiped out by disease. A new variety resistant to fungus, the Cavendish, was introduced as its replacement, and is now the only kind of banana most Americans know. It also has a lot less isoamyl acetate than the Gros Michel, meaning it is even less similar to your Runts or Circus Peanuts than its predecessor, and is the reason I get to tell people today that “bananas don’t taste like bananas anymore,” like it’s some kind of magic trick.
This is the gateway story to a whole world of history, one of the best examples of where and how our understanding of the flavors we enjoy comes from. The story of bananas in the United States is the story of modern American foodways, of industrialization, of global commerce. It’s also the story of how flavor is viewed as a result of individual choices rather than a network of decisions a community makes.
We ourselves live through these kinds of shifts constantly, often oblivious to them. The forces of industry and history push us slowly forward, mostly invisible in our daily lives. But they have power. They don’t just change what’s in the grocery store, what’s on our plate—they change our perception of the world around us.
The study of flavors and our experience of them falls under the umbrella term of “sensory science.” It’s a relatively new discipline with very old roots, and addresses a singular, central problem: how to describe and compare what may or may not be subjective experiences.
Early attempts at a full classification of sensations, from Aristotle to the early 1900s, were thwarted by limited knowledge of chemistry and biology as well as the sheer complexity of the problem. Many started by trying to determine the elemental building blocks of aromas, such as Henning’s Smell Prism from 1915, a model that showed all smells as a combination of six primary odors: fragrant, spicy, burned, putrid, resinous, and ethereal. All such efforts fell short. It was not until the mid 20th century that the science began to coalesce around a set of verifiable methods that could illuminate how our senses worked.
“The story of bananas in the United States is the story of modern American foodways, of industrialization, of global commerce. It’s also the story of how flavor is viewed as a result of individual choices rather than a network of decisions a community makes.”
Modern sensory science was born at the Quartermaster Food and Container Institute of the U.S. Armed Forces. There, staffers were tasked with answering supply problems for the military such as, “Which types of uniform fabric do soldiers prefer?” or “Why do they like one M.R.E. and not another?”
These were thorny issues, because they had to do with the slipperiness of preference or liking. Researchers could easily measure the attributes of these materials and foods, but new challenges arose when subjective elements changed the response of the user, sometimes with life-or-death consequences.
This early research gave rise to a new way of linking personal sensory experience to measurable variables, using methods now familiar to any social scientist: blind trials, questionnaires, focus groups, vocabulary development, and most importantly, statistics. All of these strategies together were designed to arrive at the consensus of a flavor experience. They were meant to average out responses, and eliminate the variability of personal taste.
Once developed, these methods were refined, and rapidly expanded. In 1958, a chemical analysis company-turned-consultancy called Arthur D. Little began using its knowledge of individual chemicals and their flavor attributes in something it called the Flavor Profile Method, and the modern taste panel was born. Typically, small panels of tasters would agree on the defining characteristics of a food or drink and then rate those attributes on a scale of intensity.
The results were supposedly objective, having been analyzed by statistical means to eliminate any major outliers. Trained tasters could identify hundreds of chemical compounds, and advanced chemistry allowed scientists to understand where and how those compounds developed. This approach also legitimized the use of human tasters as dispassionate scientific instruments.
Such methods have since branched out into many proprietary systems for classifying aromas, flavors, and sensations, though their corporate usage has mostly drifted back towards a more marketable focus: consumer preference and acceptance. It is now extremely common for a marketing or product development team to use a sensory panel, chemical analysis, statistical methods such as principal component analysis (PCA), and demographic data to not only tailor products to a particular customer base, but also to analyze where market gaps might exist for new products.
Consumer acceptance has been a driving force behind the industrialization of food around the world. Much sensory research is done in pursuit of making food cheaper, more profitable, and longer-lasting. This is not an inherently bad thing—food has become less expensive and more widely available year-round as a result of improvements to our foodways.
There are also the less noble pursuits. Many companies use sensory methods to make their products into the lowest common denominator possible, just barely recognizable as food, and often devoid of any real nutrition, in a bid for ever-increasing profitability. Anyone upset over the reduced quality of, for example, Breyers ice cream (or, as the label now says, “frozen dairy dessert”), can aim their ire at a sensory scientist.
It is somewhat ironic, then, that the craft beer industry has so latched onto sensory science as a tool to improve its products, to understand them in a deeper way, and especially to discuss them with customers.
Breweries were some of the first industrial food producers to jump on the nascent sensory bandwagon. After all, the industry had a longstanding problem: Brewers aimed for a consistent product, but shifts in ingredients and fermentation made changes inevitable. Also, beer ages. There is no original beer somewhere under glass that a brewer can taste side-by-side with today’s batch to compare. Without tools to control it, flavor will drift.
Sensory science was seen as a way to compensate for these problems, its air of scientific accuracy making it an easy sell for managers and owners. A new quality control tool began spreading through the brewing industry in the 1970s and ’80s: the sensory profile, similar to the now-popularized Flavor Profile Method, which used chemical names and their rated intensities to describe a product.
The use of chemical names is an important strength of the taste panel in a brewery setting. With so much in flux, between changes in malt lots, the water supply, yeast generations, and brewhouse parameters, the ability to pinpoint the exact origins of a flavor compound in the brewing process—and how to change it—is invaluable knowledge.
Beyond its use in describing a variable product, the taste panel had another important strength versus lab instrumentation: Humans can detect some compounds in far smaller amounts than machines. The most important example is probably 2,4,6-Trichloroanisole, or TCA. This is the main compound you think of when you describe something as moldy or musty. Humans have evolved to be extremely sensitive to it, as eating moldy food is generally a bad idea. We are, in fact, sensitive to it on the order of a few parts per trillion, an infinitesimal amount that no machines come close to matching.
“What I think the beer industry hasn’t caught up with is just how good a well-trained, highly performing taste panel can be,” says Bill Simpson, founder of Cara Technology, a U.K.-based brewery consultancy. With a good sensory panel, human tasters can’t just detect tiny amounts of the compounds most important to us, but they can use a shared language that pinpoints the chemical source of the sensation.
Sensory methods are designed for quality control, and are almost entirely preoccupied with flaws. Most major breweries still use a version of the Check All That Apply format: a list of potential off-flavors, with a box to tick for their presence, followed by an intensity rating. There are a few slots for positive attributes a brewery may think are important, but for the most part, focus rests on the negatives.
Most brewers today use some variety of this language to describe their products. Terms like diacetyl, acetaldehyde, and yes, isoamyl acetate get thrown around, not just in the brewhouse but by hardcore hobbyists as well. Homebrewers gift each other off-flavor tasting kits. The Flavor Profile Method and its offshoots are, today, ingrained in how we talk about beer. “Even the term ‘Flavor Profile’ itself has morphed into a general term and become divorced from its origin as a method,” says Berenstein.
The proliferation of the sensory lexicon is due in part to excitement. Customers want details of every step of the brewing process, and in the spirit of meeting that demand (and getting to geek out with fellow beer nerds), those making beer have happily obliged.
By comparison, the wine world has been less willing to show these methods publicly, largely because their marketing revolves around region, climate, soil, and grape varietal as selling points. Vintages are expected to be slightly different every year, and wine’s mythos of terroir clashes with chemistry: There is magic there that science kills. Though sensory science is used heavily behind the scenes for wine, the beer industry has, for better or worse, made it part of the customer-facing culture.
While all this terminology can expand our understanding and experience of beer, it is also limiting. A well-trained panel can reliably identify somewhere between 20 and 50 attributes. Some of the best professionals can identify hundreds of compounds. To put that in perspective, hops alone may have over 1,000. While they don’t all contribute in an appreciable way to aroma or flavor, a thorough accounting of them is simply not possible with the tool of a taste panel.
In some cases, technical terms can be a crutch drinkers use when finding the right words for flavor. “Being able to describe things is a skill. Some people can come up with those metaphors very easily, but it’s understandable why people reach for the science terms, because it’s hard,” says food writer Rachel Wharton. “For those of us who do this professionally, it’s a question of how to maintain your inner beer child and keep exploring the language.”
In the cases when a taster encounters an attribute they don’t recognize, it can be difficult to come up with words to describe it. They will often resort to naming the closest compound they’ve been trained on instead. The phenomenon is so common that sensory scientists have a term for it: “dumping.” This can be problematic, especially in the hands of consumers, because most basic sensory training is so focused on flaws.
“Being able to describe things is a skill. Some people can come up with those metaphors very easily, but it’s understandable why people reach for the science terms, because it’s hard. For those of us who do this professionally, it’s a question of how to maintain your inner beer child and keep exploring the language. ”
Take, for example, butyric acid. Included in most tasting kits, it is a common compound for beginners to learn about. Tasted on its own, butyric is absolutely disgusting: vomit and bile are its main descriptors. For a customer to say a beer tastes like butyric acid would amount to a serious complaint. Yet not only is butyric acid present in small quantities in most sour beers, and a necessary precursor to ethyl butyrate (the wonderful smell of pineapple), it is actually used in many dairy products to enhance the perception of creaminess, including in most milk chocolate. Both lactose and sour beers can have butyric elements to them that a beginner may pick up on and flag as quality issues. But if they hadn’t known the term, they would arguably have enjoyed the beer just fine, oblivious to the supposed flaw.
Diacetyl is another good example of this phenomenon. Dreaded in most circles, scourge of amateur brewers, the butter-like compound is often the first off-flavor would-be tasters are exposed to. Professional brewers pull faces at the slightest hint of it in a beer. Diacetyl can be a flaw, true, because it is often unintentional and can cover up other intentional aromas. Still, it is considered to-style in certain cases—Czech Pale Lagers, for instance—and is not inherently a bad flavor. The proof? Powdered diacetyl is the “butter” flavoring on your popcorn. It is added to many foods to enhance their buttery quality. I guarantee that at some point you’ve licked it off your fingers, unable to get enough of it.
Even when present, supposed “flaws” can act as positive aspects of our favorite beers. There are many stories of homebrewers and who try to copy their favorite beers over and over, unable to get the last elusive bit of depth and complexity. It is only years later that they realize what they were missing: the wet cardboard and boot leather of oxidation. Their favorite beers had sat on a hot truck or container ship or store shelf for months, whereas their fresh brews hadn’t had enough time for the oxygen to do its damage and render them (im)perfect.
There is, simply put, no such thing as a good or bad flavor. There is only context.
This is why some newer sensory companies focus on using more everyday descriptors, rather than chemical names and the language of flaws. The same statistical methods are employed to ensure valid results, but it’s becoming increasingly common to work with panels trained to speak of “guava” and “peat” rather than butyric acid and guaiacol.
“It’s a good thing that beer is so focused on quality control—that makes it pretty unique,” says Lindsay Barr, a former sensory specialist for New Belgium Brewing Company and founder of DraughtLab, which makes sensory software for small breweries. “We’ve just taken that stringent QC approach and brought it into our sensory world, sometimes at the expense of getting a full picture of all the positives in our beverage. It’s kept sensory out of reach for many breweries.”
“[Knowing technical terms] has a lot of potential for good,” says Simpson. “It can make you enjoy a beer more if you know more about it, which is a bias one way. You can manipulate people’s enjoyment of a beer with words—best used for good.”
When training my own panels, I remind them day after day that there is no such thing as a bad flavor. We taste diacetyl-laced popcorn, leathery ports, and butyric-heavy tropical fruits. We work on both identifying compounds and exploring how we experience them in context. This is something any brewery is capable of doing with both staff and customers. You don’t need a technical education in something to know what you enjoy about it.
Scientific and technical language has always been accused of obscuring as much as it illuminates. Sensory science, with its focus on measuring the human experience, may be especially vulnerable to this argument.
To be viable, sensory science must stay at a remove from the individual. Averages and other statistics make the whole thing tick: They are there to disregard “incorrect” outliers. The use of chemical terms is also supposed to give an implied certainty.
The craft beer industry has adopted these tools and tried to use them in a strange way: We apply terminology meant for consistency and statistical overview to the ungovernable experience of the individual. In doing so, we accidentally gloss over a whole world of human strangeness.
“There are many stories of homebrewers who try to copy their favorite beers over and over, unable to get the last elusive bit of depth and complexity. It is only years later that they realize what they were missing: the wet cardboard and boot leather of oxidation. Their favorite beers had sat on a hot truck or container ship or store shelf for months, whereas their fresh brews hadn’t had enough time for the oxygen to do its damage and render them (im)perfect.”
Each of our senses is actually a patchwork of systems. Take touch. What we talk about as a single sense is actually multiple avenues of sensation bundled together: tactile experience, cold and warmth sensing, pain. These inputs are all sending signals through your nervous system to your thalamus, a region of the brain that sorts the incoming information. The thalamus cannot actually process everything your nervous system takes in, and instead looks for major changes: a sudden temperature shift, or a new increase or release of pressure.
Should the pattern stay the same, the thalamus weeds out the sensory input as unimportant, and simply throws it away. Your big toe isn’t sending any different signals from a moment ago, and so your thalamus was discarding the information as irrelevant. Your thalamus can be influenced by your conscious attention, though, which is why thinking about your toe just made you able to feel it again (neat trick, that). The same is true with tasting—focus is often at the forefront of any training, because the results will change along with it.
All of your senses pass through the thalamus except for one: aroma. Smell utilizes a region called the olfactory bulb, which bypasses the thalamus entirely, plugging directly into the brain without going through the processing that all the others do. The vivid memories conjured by smells are partly due to this special, privileged position. That is not to say that attention isn’t a factor in olfaction—it just means other parts of your brain screw with it.
You cannot pay attention to what you cannot sense, though, and it turns out all of our senses have their own versions of colorblindness. We are all randomly anosmic—unable to smell—certain compounds, a product of our genetics. Some smells even have anosmia rates as high as 50% in the general population, like β-ionone, responsible for the unique aroma of violets. Only half the population will ever smell it—the other half will smell all the other things that violets give off, and each half will talk to the other as if they are experiencing the same thing.
There can be temporary anosmia too, a reaction of your brain shutting out certain overwhelming smells, or ones it considers background noise. It’s the reason that you don’t smell your own home until you leave it and come back. Sulfur compounds in particular are prone to this treatment: Dimethyl sulfide in beer is almost impossible to smell up close, and must be smelled from farther away—hence the snobby-looking wafting sniffs that trained tasters will take before diving into a beer.
External forces modulate your perceptions in ways you don’t notice, either, and that are still poorly understood. Oxford University’s Crossmodal Research Laboratory, in a study for British Airways, showed that passengers’ dislike of airplane food stemmed not from the dry, recycled air, as was commonly believed, but from the loud, low-frequency roar of the engines, which temporarily dampened their ability to taste sweetness and accentuated perceived bitterness. The sonic environment actually changed the flavors passengers experienced. It seems that our senses are more intertwined than we truly understand, each affecting the other in synesthetic waves of overlap.
The language of sensory science suddenly seems like an impoverished way to describe these possibilities. With all of these factors changing how a person might taste a beer, isn’t it just a little presumptuous of us to say that some knowledge of chemistry gives us total insight into their real lived experience?
This brings us to another set of questions that sensory science may be ill-equipped to explore. Why, exactly, do we have any of these systems to begin with? What is it that makes our bodies and brains believe that these chemicals are food? We did not evolve a whole bunch of random sensing mechanisms and then see which ones were useful—each one has grown over time to respond to stimuli already present in our environment.
Isoamyl acetate is not just found in bananas. It is a primary aromatic compound produced in many fermentations. Drink a Belgian beer, and the banana notes, usually alongside clove, are driving characteristics, though neither banana nor clove are anywhere in the recipe. The aromas are created entirely by yeast through complex mechanisms that likely developed millennia before our species did. We have in a sense grown up with these life-derived flavors as older siblings, driving what we even think of as food.
Fermentation, it turns out, is where much of our flavor experience is written. Forget even the microorganisms that create the food—our own bacteria play a crucial role in what we enjoy. We tend to think of flavor as a direct result of our food itself, but the community of microbes in our mouths actually breaks apart and frees aromatic compounds as we eat. We don’t just live with these bacteria—we commune with them, modulating our own behaviors for them as much as they change for us. The ones in our gut are necessary for our very survival, doing some of the work that our body can’t, and the balance of them changes our lives. They are like a shard of primordia embedded within us, the shrapnel of evolution. We, in our oddness, are like trees that have grown around an obstacle until it’s part of us.
There is even a debate as to whether fruit has any “fruity” flavor on its own, or whether most of what we expect to smell and taste is fermentation-derived. This may sound counterintuitive. You would think an apple should have all the characteristics of an apple or we would call it something else—there must, somewhere inside it, exist the Platonic ideal of an apple. There it sits in an orchard. This ideal one is perfectly round and ripe. It smells of waves and waves of rind, flesh, juice, the sweet and sharp cider nose of acid sting and red-green. It is enough to make one believe in divinity.
Now turn to your supermarket. See the abundance: whole orchards of produce, and it smells like …
Nothing.
The argument goes like this: Fruit on its own doesn’t really smell like anything until it starts to ripen. Maybe a little like grass, or sap, or wood—some other generic vegetative matter—but an unripe pear is hard to tell from an unripe apple by smell alone. The process of ripening is our miracle moment that turns something inedible and unappetizing into the fruits that make us swoon.
The supermarket shelves are full of unripened fruits, depriving us of that Platonic orchard moment. It’s for a practical reason. Shopkeepers will tell you that people don’t want to buy ripe fruit that needs to be eaten that day. Ripe fruit is going quick, developing blemishes and soft spots and getting mushy, because, and here’s the kicker: Ripening is really just the first stage of rotting.
Some ripening is chemical in nature. Oxidation and other breakdown reactions occur, causing cell wall damage and browning. The real action, though, is the massive, invisible work of bacteria and yeast, which not only generate chemical—and thus physical—changes, but which produce most of the aromas we associate with the fruit itself.
Standard brewer’s yeast gives off apple notes, including a compound called acetaldehyde. Ales and Belgian styles have more than Lagers, but even your Light Lagers have it in droves. Next time you pick up a Sapporo, think “red apple” and you’ll be shocked you never noticed it before.
An even better example is Torulaspora, a family of yeast often found on apple skins and fundamental to the spontaneous fermentation of some ciders. When I worked at Brooklyn Brewery, we once used a strain of it to make a beer, which didn’t play out well. Malt sugars and the environment of beer are very different from apple sugars and cider, and the project was scrapped. We eventually had to dump some, and as we did, the brewery filled with the smell of an apple cart on the way to market.
We were left wondering: Do apples smell like that because they’re apples, or only because Torulaspora live on them? Is an apple just a pale shadow of itself without the microbiological community it supports?
“People always talk about varietal character in wine grapes, but there is a crucial part they usually miss: Wine grapes don’t actually taste like wine,” says Garrett Oliver, brewmaster of Brooklyn Brewery. “And if wine tasted like grape juice, you wouldn’t drink it. If you eat Nebbiolo grapes, they don’t taste like Barolo. Wine tastes like fermentation—or at least it tastes like what fermentation does to grape juice.”
There is still no better way to make the things we enjoy drinking than by letting microbes do their work. Industrial fermentation is simply the selective use of these organisms to make the flavors we already want: a microbial party with a guest list. In a broad sense it’s a decision about what belongs in our ecological niche. We keep making fermented drinks over and over again because they’re supposed to exist where we do. Guided by flavor, we build up communities around us, microbial and otherwise, with a deeply intuitive sense of what does and doesn’t belong.
“People always talk about varietal character in wine grapes, but there is a crucial part they usually miss: Wine grapes don’t actually taste like wine. And if wine tasted like grape juice, you wouldn’t drink it. If you eat Nebbiolo grapes, they don’t taste like Barolo. Wine tastes like fermentation—or at least it tastes like what fermentation does to grape juice.”
Sensory science, for all its contribution to our understanding of our food and ourselves, can miss this mark quite a bit. The statistics can tell you a lot, and the chemicals can be in the right place, but all the testing in the world can’t fully encapsulate the rich complexity of the best fermentations, just as it can’t make a beer the perfect one for a moment with friends—the one that makes us feel we belong. There is the science of a drink, and then there is still the art and magic that makes everything click.
“The varieties of pleasure we derive from food or drink always escape attempts to control or define it,” says Berenstein, “but that elusiveness pushes the science forward, too.”
Sensory scientists are not blind to any of this. They are often highly trained specialists, vigilant and careful about what they can and cannot prove, and continually finding ingenious ways to plumb the depths of all these phenomena. A strict awareness of its limits is ingrained in the discipline. Even so, the tools have wound up in many places beyond the quality lab, in response to pressure from management to prove the things they want proven, and from consumers demanding more ways to connect. In the beer industry in particular, the result has been a flattening of experience, and an unintentional blindness to the breadth and interconnection of our sensory world.
If you think about the ecology of flavor, all those microbes coming together with us in a complex dance, it can look like a mirror image of our own coming together at the dinner table. Community is the end result of flavor. We all crave that novel experience, be it a banana or Barolo, but we don’t just want to have it by ourselves. The barbecues, the dinner parties, the nights of pasta and wine—we bother less with flavor when we’re alone than when our home is full of friends. We share our tables in the knowledge that we are all finding something we need there, all finding our way by the sonar pulses of our nerves and wholly reliant on our microbiome crew.
We’re deeply marked by these connections, all the way down to the roots. The training of sensory science, valuable as it may be to the industry, can pull us away from something human we already understand—flavor is about community, and we all know it when we taste it.
Words, Scott SimpsonIllustrations, Ben Chlapek
Language 
