by Bob Holmes
The picture gets even more complicated. Wysocki has tried the same experiment with other odorants, such as the sweaty-smelling 3-methyl-2-hexenoic acid, and found no change in detection ability. His colleague, Pam Dalton, showed that repeated exposures to Maraschino-cherry-smelling benzaldehyde do lead to improvement—but only in women of reproductive age, not in men, young girls, or postmenopausal women. Even now, nearly three decades later, Wysocki’s not sure why some people get better at detecting some odors after being exposed to them, while other people do not.
Some of the answer, no doubt, has to do with the odor receptors themselves, and the way they interact with the odorants. But some, too, must depend on the way your brain processes odor information. People who measure olfactory thresholds quickly learn that they don’t stay put. Your detection threshold for a particular odor can vary many thousandfold from one test to the next—and it doesn’t matter whether the tests are separated by thirty minutes or more than a year. That’s probably at least partly because our noses don’t always claim the same share of our conscious attention.
That’s not to say we can’t get better at recognizing and identifying odors. Practice clearly helps. You can prove this yourself by pulling a bunch of spices off your kitchen shelves and trying to identify them with your eyes closed. After a few rounds of trial and error, you’ll get better. The most vivid demonstration of the power of practice, though, comes from wine experts, who are much better than the rest of us at putting names to the aromas rising from their tasting glasses. But whenever scientists have tested such experts (which doesn’t happen often—what wine pundit would run the risk that their nose might be shown up as below average?) they’ve found that their olfactory ability is nothing special. What gives their ordinary noses the ability to perform extraordinary feats of perception is simply experience. That’s encouraging news for anyone who wants to sharpen their flavor senses.
If you’ve booked a table at the best restaurant in town, or plan to open a treasured bottle of wine, there may be other ways to amp up your sense of smell to wring more flavor out of the experience, though you may look a bit odd in the process. Nasal sprays containing sodium citrate or a compound called EDTA bind calcium ions in the mucus layer coating the olfactory epithelium, and this makes olfactory cells more sensitive for a few minutes before things return to normal.
If the thought of spritzing your schnozz every fifteen minutes at the French Laundry restaurant in California is a bit off-putting, here’s another option: Use one of those nasal dilator strips that professional athletes wear across the bridge of the nose to hold their nostrils open. Athletes use them to help inhale more air faster, but as a side effect the dilators also improve airflow to the olfactory epithelium. Tests have shown that this makes it easier to detect and recognize odors.
But even if experience can alter our odor perceptions somewhat, it’s clear that our genetic makeup of odor receptors is driving the flavor-perception bus. It’s not just a question of odor receptors, though. More than a thousand other genes affect what happens in the sensory pathway after an odor binds to its receptor. Differences in these genes undoubtedly mean that some people have a more acute sense of smell overall than others do, just as some people see or hear better than others. Not many researchers have measured how big these differences in general olfactory sensitivity are, alas, so the subject remains a big question mark.
We’re just starting to understand how these differences in genetic makeup affect our experience of flavor. For example, many people—but not all—notice a distinctive asparagus-type odor to their urine shortly after eating that vegetable. Proust noted that asparagus “transforms my chamber pot into a flask of perfume.” For many years, scientists assumed that the smelly pee folks digested asparagus in a way that produced an odorous molecule called methanethiol, while the others—can we call them sweet pees?—did not. But in 1980, researchers fed a pound of canned asparagus to a sweet-pee volunteer, collected his urine afterward, and offered a whiff to unsuspecting volunteers. To the researchers’ surprise, they found that anyone who could smell asparagus in their own urine could also detect it in the supposedly sweet pee of the donor. In other words, the difference between sweet and smelly is not in the digestion of the eater, but in the nose of the smeller. We now know that a particular odor receptor, OR2M7, may be responsible. (Actually, it turns out that there are a few people who really do produce odorless urine, for unknown reasons.)
It’s likely that differences in odor perception help explain why people like different foods. Take cilantro, for example. Most people love its bold, grassy flavor, but a substantial minority detests the stuff, describing its flavor as soapy or “buglike.” (How do they know that, I wonder?) Scientists at the personal-genomics company 23andMe recently linked this preference to differences in or near the OR6A2 gene.
But on closer examination, there’s a cautionary story here for anyone who’d like to believe that genes are destiny. If every person with one variant of OR6A2—let’s call it variant X—loved cilantro and every person with variant Y hated it, then we’d say that OR6A2 explained 100 percent of the difference in perception. If OR6A2 had no effect at all, of course, it would explain 0 percent of the difference. The closer to 100 percent you get, the stronger the effect. For cilantro preference, OR6A2 turned out to explain less than 9 percent of the difference in perception. In practical terms, OR6A2 isn’t telling us much about cilantro preference at all.
A lot of olfactory genetics turns out to be like that, as I learned when I asked Joel Mainland to profile my own olfactory genes. Since scientists know only a handful of olfactory receptor genes so far that affect perception, this involved identifying my variants of just a few OR genes, rather than doing my whole genome sequence. A few weeks later, I visited his lab and sat through a battery of olfactory tests where I rated the intensity and pleasantness of the odors targeted by those genes.
The results could only be described as disappointing. Take OR11A1, for example. This OR detects an earthy-smelling molecule called 2-ethylfenchol that sometimes appears as an off flavor in beer and soft drinks. Three variants, or alleles, of this OR are common in the human population, one that is good at detecting 2-ethylfenchol and two that are relatively poor at the job. Mainland’s peek at my genome showed that I have two copies of the sensitive allele, which should make me an especially sensitive smeller of 2-EF. And, since people tend to find stronger odors less pleasant, Mainland predicted that I’d rate 2-EF more unpleasant than the average person.
In fact, though, both predictions turned out wrong. On a scale from 0 (undetectable) to 7 (overpoweringly intense), I rated the intensity of 2-EF as 3.4, much less than the 4.8 that Mainland predicted. I gave it a 5.0 for pleasantness (evidently I like the smell of dirt), while Mainland predicted I’d rate it just 3.2, or mildly unpleasant. Other pairs of receptor and odor, such as OR10G4 and smoky-smelling guaiacol, OR11H7 and cheesy/sweaty-smelling isovaleric acid, and OR5A1 and floral-smelling beta-ionone gave similarly confusing results. Occasionally things worked out more clearly. I’ve got one functional and one broken copy of OR7D4, the receptor that detects androstenone, the boar-urine-and-truffle odor that Wysocki studied. That should leave me as a moderate smeller and an enthusiast for truffles—which, in fact, I am. But it doesn’t always turn out that way, says Mainland. “We have a lot of people who have two functional copies and don’t smell it, and we have some people who have two nonfunctional copies and still smell it.”
It’s not surprising that single genes do such a poor job of predicting flavor perceptions, says Mainland. Since most odorants trigger more than one receptor, my response to any given odor probably depends on my genetic makeup at several genes. That muddies the waters a lot. “I’m dividing you based on one receptor, but you also have 399 other receptors, and that’s a lot of noise to push through,” he says. For example, he and his colleagues have found that OR10G4 can detect both guaiacol and vanillin, the key molecule in the aroma of vanilla, but it’s mu
ch more sensitive at detecting the former. When they look more closely, they find that people with a damaged copy of OR10G4 tend to report that guaiacol smells less intense, but report no difference in vanillin—which, presumably, depends largely on another receptor instead. Clearly, linking genetics to flavor perception still has a long, long way to go.
What we’d really like, of course, is to understand olfaction well enough so we can reproduce smell sensations artificially, as we do for sights and sounds. When Luke Skywalker’s X-wing Starfighter destroys Darth Vader’s Death Star, we see Luke in the cockpit, even though what’s really before us is just pixels on a screen. That’s because we know how to make a video image that mixes those pixels in a way that our eyes and brain interpret just the same as if it were really happening. We hear the explosion, even though none really occurred (and no sound waves would travel through the vacuum of space, but that’s a different issue), because we know how to re-create sounds from a string of zeros and ones in a digital file.
We’re nowhere near being able to do that for flavor. Sure, you can find a few oddball episodes in cinematic history where people have re-created—or, more precisely, imported—specific odors that fit the scenes of a movie. Take Smell-O-Vision, for example. In 1960, film producer Mike Todd Jr. (Elizabeth Taylor’s stepson) employed a system for mechanically releasing odors into a movie theater during the movie Scent of Mystery. Audiences were supposed to get a whiff of pipe smoke, for example, when one particular character appeared onscreen. The system cost tens of thousands of dollars per theater—a lot of money in 1960—and didn’t work very well. In 2000, readers of Time magazine voted Smell-O-Vision one of the “Top 100 Worst Ideas of All Time.” Still, that hasn’t kept novelty-seeking filmmakers from trying again now and then, albeit usually using scratch-and-sniff cards instead of a forced-air system.
But all of these novelties merely used odors prepared ahead of time. In that sense, they’re the equivalent of showing someone a photocopied picture. The real goal of digital olfaction—being able to make up any smell (and hence, any flavor) you want, to order, by combining elements from a small set of “primary odors”—was nowhere in sight.
Today, several decades later, that goal might just be visible. At the very least, we can estimate the scale of the problem. Every smell on Earth must be encoded by some combination of our four hundred-odd odor receptors. In theory, then, an arsenal of about four hundred primary odorants, each of which tickled a different odor receptor, should allow you to mix and match to recreate any smell. In practice, the task ought to be somewhat simpler than that, because it’s likely that some of our odor receptors are redundant copies. For anyone interested in digitizing only flavor-related odors, the field gets a little narrower yet, since we can ignore all the receptors that are never activated by food odorants. In fact, Mainland thinks it should be possible to get at least a rough sketch of the odors—and therefore the flavors—of the vast majority of foods with many fewer primaries than that. He’s been working with a flavorist at Coca-Cola who claims that with just forty primaries, you can get a recognizable facsimile of 85 percent of all foods.
When I visited Mainland’s lab, he screwed the top off a vial and gave me a whiff of the concoction. “Do you recognize this?” he asked. It certainly smelled familiar, but—as so often happens when we try to identify smells cold, without any prompting—I found myself tongue-tied and unable to put a name to it. Once he told me—strawberry—it all snapped into place: Of course, strawberry! It was indeed a recognizable, though not perfect, imitation. A real strawberry contains hundreds of scented molecules. But with just four of these—cis-3-hexenol (cut grass), gamma-decalactone (waxy), ethyl butyrate (generic fruitiness), and furaneol (caramelized sugar)—Mainland can build a mix that smells recognizably like strawberry. It’s not perfect, he admits—more like a pixelated image than a high-resolution version. However, he says, “We’re okay with eight-bit graphics that gives you a sketch of what’s going on. If we’re making a poor strawberry, but it’s still strawberry and not cherry or banana, we’re happy with that.”
Even if he could match the real thing perfectly, people might not realize it. “The problem we have is that everybody tells us it’s a terrible strawberry,” Mainland says of his facsimile. “But if you smash up a strawberry and put it in an olfactometer, people will also tell us it’s a terrible strawberry.” In our day-to-day lives, it turns out, we don’t generally notice all the components of a familiar odor, so we often don’t have a very good mental image of what the real thing smells like—especially when we lack the visual context. People don’t usually notice the green, vegetative note in strawberry, for example, so its presence in a crushed-up real strawberry can strike them as false, somehow.
So far, all of Mainland’s efforts to mimic strawberry or blueberry or orange aromas use odor components that are naturally found in the target aroma. Ideally, he’d like to go one better someday. “What we would really like to do is make strawberry without anything that’s in a real strawberry,” he says. To that end, he’s intrigued by a molecule known to chemists as ethyl methylphenylglycidate, a mouthful unpronounceable without sounding like Sylvester the cat spluttering at Tweety Bird. To flavorists, the chemical is known as “strawberry aldehyde.” As you might guess from the name, it has a strawberrylike aroma and is often used as an artificial strawberry flavor, even though it doesn’t occur in a real strawberry. (You can’t always trust a name, though—despite its moniker, strawberry aldehyde isn’t actually an aldehyde.) Mainland would love to know whether strawberry aldehyde activates the same odor receptors as the components of real strawberry odor, to see whether that accounts for its mimicry.
But what if you wanted not just eight-bit graphics but a high-resolution image that reproduces the real flavor precisely? So far, the closest approach to this ultimate goal comes from a recent German study led by Thomas Hofmann at the Technical University of Munich. In what can only be described as a heroic assault on the university library, Hofmann and his colleagues (including the delightfully named Dietmar Krautwurst, who was clearly destined for a career in food science) read through more than sixty-five hundred scientific books and papers that analyzed the flavor molecules present in particular foods. They winnowed these down, selecting only the best and most detailed studies, until they ended up with more than two hundred food items—everything from mushrooms to taco shells, Scotch whisky to donuts—for which the key odorant molecules had been identified. Most of the papers even took things one step further by showing that a mix of those key odorants smelled indistinguishable from the real item.
The surprising thing is that the aromas, and hence the flavors, of all these diverse foods could be re-created using a palette of just 226 key odorants. That’s remarkably encouraging, given the thousands of smelly molecules present in that range of foods. Some of these key odorants are what they call “generalists” that turn up over and over again. The cooked-potato-smelling methional, for example, figures in the odor of more than half the foods, while green-grassy hexanal and fruity-fresh acetaldehyde play a role in 40 percent and 29 percent, respectively. Many other odorants contributed a distinctive note to just a few food items, such as garlic’s diallyl disulfide and grapefruit’s 1-p-menthene-8-thiol.
Sometimes, they found, it takes only a handful of key odorants to replicate a food’s flavor. Cultured butter, for example, needs only three: the buttery-smelling generalist butan-2,3-dione, coconutlike delta-decalactone, and sweaty-smelling butanoic acid. Other foods, like beer and cognac, required eighteen and thirty-six key odorants, respectively, to mimic their bouquet precisely—a lot, but still just 10–15 percent of the total set of primaries.
Of course, trying to build a digital-flavor unit with 226 primaries is still a huge technical challenge. But if you could do it—even if it took trained technicians and an expensive, well-stocked lab—then the sense of smell (and, by extension, much of flavor itself) would finally free itself from subjectivity and be on a truly objecti
ve footing. We could take an olfactory “snapshot” of a ripe Georgia peach or a tomato fresh from the garden in the heat of August and reproduce it exactly. We could save a famous chef’s signature dish and archive it in a museum. And we could collect flavor memories of our travels and revisit them at home, just as we now do with photographs.
There’s a lot of work still to be done before those fantasies can become realities. And not just on the olfactory front, either. As it turns out, there’s more to flavor than just taste and retronasal smell. The physical sensations of touch—texture, temperature, and the like—also play a huge role in it.
Chapter 3
THE PURSUIT OF PAIN
I’ve been procrastinating. On my dining room table I have lined up three hot peppers: one habanero, flame-orange and lantern-shaped; one skinny little Thai bird’s eye chili; and one relatively innocuous jalapeño, looking by comparison like a big green zeppelin. My mission, should I choose to accept, is to eat them. For you, dear readers.
In ordinary life, I’m at least moderately fond of hot peppers. My fridge has three kinds of salsa, a bottle of Sriracha, and a jar of Szechuan hot bean paste, all of which I use regularly. But I’m not extreme: I pick the whole peppers out of my Thai curries and set them aside uneaten. And I’m a habanero virgin. Its reputation as the hottest pepper you can easily find in the grocery store has me a bit spooked, so I’ve never cooked with one, let alone eaten it neat. (In fact, the first habanero I bought went moldy in my fridge while I was working up my courage.) Still, if I’m going to write a chapter that’s largely about hot peppers, I ought to have firsthand experience at the high end of the range. Plus, I’m curious, in a vaguely spectator-at-my-own-car-crash way.