by Bob Holmes
These flavor neurons in the OFC are also where learning enters the flavor picture. Remember Paul Breslin’s rose–bitter chewing gum, which people eventually learned to treat as a coherent flavor? It turns out that Rolls had tried a very similar experiment in rats, while looking closely at individual neurons in the OFC. And indeed, when he switched up his odor/taste pairings, he saw those neurons gradually switch their responses to reflect the new associations. “You can watch the neurons learn,” says Rolls.
While the neurons do relearn new odor/taste pairings, though, they aren’t speedy about it. Rolls found he had to expose the rats to the new pairing about fifty times before they switched. In contrast, when he did the same experiment, except pairing tastes with visual signals instead of odors, the neurons began to relearn the very first time they saw the new pairing.
Why the difference? Well, says Rolls, it probably has to do with flavor’s role in protecting us from eating the wrong things. “You don’t want to realign your whole flavor system too rapidly.” In the real world, smells tend to be rather reliably paired with the same tastes day after day, while the appearance of things can change quickly. Our brains, it seems, reflect this reality by being unusually conservative about taste/smell pairings, but looser with visual information.
Not that vision isn’t crucial to our perception of flavor. Humans, after all, are a largely visual species, so it’s not surprising that vision creeps into most of our experiences, says Lundström. A simple experiment shows how central vision is, he says. Try it: Imagine the fragrance of a ripe strawberry. Really focus on it. Now, didn’t you also call to mind a mental picture of the fruit itself? “It’s impossible to do that without actually visualizing a strawberry,” says Lundström. “I think that vision is the key component when it comes to memorizing an odor, and you have a strong input from vision when it comes to odor quality.”
To study this effect further, Lundström turned to a technique called transcranial magnetic stimulation—essentially an electromagnetic helmet that can be programmed to stimulate particular regions of the brain and make them work better. TMS of the visual center of the brain, for example, makes people about 10 percent better at discriminating among subtle shades of gray.
Lundström was after something more, though. If vision is linked to flavor processing, he wondered, could you stimulate the brain’s visual system and improve flavor perception in the bargain? If so, that would tie vision even more firmly into the bundle of flavor senses.
So Lundström and his colleagues set up a study to test the idea. They offered volunteers three smell samples to sniff: two of one aroma and one of a different, but somewhat similar aroma—strawberry and raspberry, say, or pineapple and orange. The people had to identify which of the three was the different sample, a task they could do correctly about three-quarters of the time. Lundström had his subjects do the test three times: once without TMS, once with TMS stimulating their visual center, and once with a fake TMS helmet that emitted an impressive, official-sounding hum but caused no actual changes in the brain.
Sure enough, people treated with the real TMS, but not the pretend one, proved to be about 10 percent better at picking which odor was not like the others. In other words, helping people see better helped them to smell more accurately, too. And the effect wasn’t just a general heightening of the senses. TMS of the visual center did nothing to help people tell which of three odor samples was more intense than the others, he found. That makes sense, says Lundström—visualizing the source of an odor is important in identifying it, but irrelevant in deciding how intense it is.
We’ve seen that the orbitofrontal cortex is the birthplace of flavor. It’s also the crossroads for several other key parts of consciousness. All five senses pass through the OFC on their way into the brain, and the OFC also gets input from the brain regions responsible for emotions, reward, and motivation, as well as higher-order thought. The orbitofrontal cortex has been called the sensory packaging center of the brain, the place where all our world experience comes together. That suggests that flavor is not just a filigree on our lives, a little bit of aesthetic fluff. It’s a key part of our interaction with the world.
Chapter 5
FEEDING YOUR HUNGER
Dana Small vividly remembers the first and last time she drank Malibu rum and 7UP. “It was a big party, and it was my first time, I think, having alcohol, and I was probably underage,” she recalls. Small is, well, small, with long, bright-copper hair and a barely detectable soft lisp. “I didn’t know what I was doing, and I probably didn’t have that much, but Malibu and 7UP doesn’t really taste like alcohol, it’s like this really sweet kind of . . . Anyway, so I had a few of those, and felt not so good the next day. That was 20 years ago. In those 20 years, I’ve continued to eat a lot of sweet things, but I particularly avoid Malibu and 7UP.”
Most of us can think of a similar event in our past, where a bad experience has permanently scarred our taste for some particular food or drink. But for Small, a neuroscientist at Yale University, the lesson runs much deeper than “Don’t drink this.” She thinks experiences like hers—and the positive ones on the flip side of the coin—are the whole reason our brains assemble a unified perception of flavor from what could have been left as separate senses of taste, retronasal smell, and texture. “The reason that we have flavor, when we already have taste and smell, is for the purpose of associating foods that we encounter in the environment with their post-ingestive effects, because ultimately that’s what it’s all about. That’s really the role of flavor,” she says. Translation: We remember the flavors of what we ate, and what happened afterward, so that the next time we can seek out the good stuff and avoid the bad. “Flavor perception allows us to have a representation of precisely a particular kind of food. So in the case of Malibu and 7UP, there is specific learning to avoid that item. This is really like no other kind of learning. It’s very strong—one trial—and very long lasting. That makes perfect evolutionary sense: you want to only need one trial.”
For our ancient ancestors, omnivore hunter-gatherers that they were, these eating decisions would have been far more than a minor matter of aesthetics. Their choices of what to eat could literally be a matter of life and death. Pick the wrong root to eat, and you poison yourself and your family. Pass up a nourishing root, and you could all starve. More subtly, being a successful hunter-gatherer hinges on finding the foods that deliver the biggest nutritive bang for your hunting, gathering, and chewing buck. In modern terms, if you’re uncertain you’re going to eat tomorrow, then today you damn sure want to eat potatoes or burgers or ice cream, or something else with a lot of calories, rather than wasting your time munching raw celery.
So you’d expect that evolution would have endowed us with a pretty good system for identifying and remembering potential foods and what happens when we eat them. A little bit of this system, as we’ve seen, is built-in: even newborns have an innate liking for sweet tastes. But mostly, we learn through experience. That’s what flavor does for us—and why our brains assemble all the relevant data of taste, texture, retronasal olfaction, and all the rest into a single, unified flavor perception. Thanks to this synthetic perception, we can learn and remember the flavors that made us sick, and we also learn to like the flavors that nourished us. We don’t generally notice that we’re learning about nourishment, because in our everyday world, flavors and calories are inextricably linked. We don’t usually encounter a baked-potato flavor without also ingesting a big slug of carbs, or salmon flavor without protein and fat. Teasing the flavor apart from its nutritional consequences takes careful experimentation, the sort that’s easier to do with rats than with people.
The classic studies here come from Anthony Sclafani, a researcher at Brooklyn College in New York. Sclafani offered rats one of two water bottles to drink from, one grape flavored, the other cherry flavored. Neither contained any sweetener or other nutrients—just flavoring and water. But he also inserted a stomach tube, so that he co
uld deliver sugar solution straight into the rat’s gut when it drank the cherry flavor, but not the grape. Since the sugar never entered the rat’s mouth, it tasted no sweetness. Yet within a few minutes of encountering the two flavors, the rat learned to drink almost exclusively from the cherry-flavored water bottle. What’s happening, says Sclafani, is that nutrient receptors in the rat’s gut quickly signal to the brain that good stuff is coming in. The brain pairs this with flavor information from the nose and mouth, and the rats learn that cherry means calories, even though they never taste the sweetness. Sclafani also reversed the pairings in other rats, just to be sure there wasn’t something special about cherry flavor. Rats that got sugar infused into their stomachs when they drank the grape-flavored water quickly learned to prefer grape instead. And it’s not just sweetness—Sclafani’s rats learn just as well if the calories delivered through the stomach tube come from proteins or fats. It’s exactly the same learning process that Russian biologist Ivan Pavlov used to train dogs to associate the ringing of a bell with imminent food. After not too long, Pavlov could just ring the bell and his dogs would start salivating in anticipation of the meal to come. Sclafani’s rats hit the cherry-flavored water because they expect calories to follow.
By moving the location of the rats’ stomach tube, Sclafani was able to show that the nutrient receptors responsible for the learning are located right at the beginning of the small intestine, just past the stomach. This is the region that surgeons remove when they do gastric bypass surgery on morbidly obese patients. No one knows exactly why gastric bypass surgery works so well, but one reason may be that it eliminates these nutrient receptors and thus prevents the pairing of flavors to their nutritive consequences. Since the flavors of a meal are no longer associated with incoming nutrients, people would gradually lose interest in the flavors and feel less drive to eat them, Sclafani suggests.
This flavor learning—technically known as flavor-nutrient conditioning—is dead easy to demonstrate in rats, as Sclafani’s work shows and others have verified. But it’s a lot harder to prove that the same kind of learning happens in people. For one thing, the whole stomach-tube business is a nonstarter. People also have an annoying habit of eating whenever they feel like it, so that experimenters have a much more difficult time controlling their food intake or ensuring that they haven’t already formed associations with, say, grape and cherry flavors. As a result, studies of flavor-nutrient conditioning in humans have had mixed results. Sometimes it looks like it happens, and other times it doesn’t.
Probably the best evidence that we really do learn to pair flavors with their nutritive rewards comes from Dana Small’s lab at Yale. Small flipped through a flavor-supply catalog to come up with ten really obscure flavors that no normal person had much chance of encountering in daily life. “They’re novel, and you don’t know what they are,” Small told me when I asked her to describe the flavors. One, for example, was called “aloe,” though it tasted nothing like aloe vera. Not surprisingly, people tended not to like the flavors when they first encountered them—our built-in neophobia raising its head again.
Small and her colleagues picked two flavors and used them to make artificially sweetened soft drinks. One of the flavors also got a dose of maltodextrin, a sugar that delivers a full load of calories—it turns into glucose almost immediately when it reaches the stomach—but is devoid of flavor. (A triangle test—which of these three is not like the others?—confirmed that people couldn’t tell the difference between soft drinks with and without the maltodextrin.) Volunteers consumed each drink several times over the course of a few days, using only one kind of drink each day to keep the postingestive consequences separate. And then Small brought them back into her lab to see how they responded to the two flavors. To be sure she was looking at the effects of learning, and not real-time perception, this time neither flavor was spiked with maltodextrin. Sure enough, the people showed a slight tendency to like the high-calorie flavor better than the low-calorie one. In other words, they had learned which flavor delivered the nutritive goods, and they liked it a little better—but not all that much better. The big difference showed up when Small put them in a brain scanner.
Being in Small’s brain scanner is not exactly a fine dining experience. Just like any hospital MRI, you’re flat on your back inside a giant magnet, with your head immobilized. To get a good image of the effect of each flavor on brain activity, she needs to average over multiple sips: on again, off again. She needs to know exactly when the flavor arrives, and she needs to keep stray odors from lingering and confusing the test. “What that means,” Small says, “is that you’ve got a nasal mask, and then this teflon thing that liquids are dripping off onto your tongue.” Charming.
Even in that utterly strange context, the results were dramatic. When people drank the flavor they’d learned to associate with calories, a part of the brain called the nucleus accumbens lit up like a Christmas tree. The nucleus accumbens is a part of what’s often described as the “reward pathway,” the part of the brain where good things begin to feel good, so that you want to do them again. The reward pathway plays a role in making you want more of things like sex, drugs, and rock and roll (literally—music activates the nucleus accumbens). An old study from the 1950s hooked rats up so they could stimulate their nucleus accumbens by pressing a lever; the rats just kept pressing the lever, over and over and over again, not even pausing to eat or drink.
Crucially, the learned flavor-nutrient link swayed the response of people’s reward pathways much more strongly than it affected their conscious liking of the two flavors. Let’s pause for a moment to underscore that point: When Small asked her subjects which flavor they preferred, she didn’t find all that much difference. That might explain why previous studies of flavor-nutrient conditioning in humans haven’t been very convincing. But Small didn’t stop there. Instead, she also let the subjects’ brains tell her which flavor they valued more—and their brains spoke loud and clear. All the real work, it turned out, was happening under the surface, in the unconscious.
Small points to another recent study that reinforces the point. Researchers at her alma mater—McGill University in Montreal—wanted to separate our conscious and unconscious valuations of food items to see how they differed. To do this, they showed pictures of food to hungry volunteers, and asked them to estimate their calorie content. (That’s the conscious valuation.) At the same time, a brain scanner measured the activity in a region of the brain called the ventromedial prefrontal cortex, another area involved in valuation and appetite. (That’s the unconscious valuation.) To top it off, the subjects were also given five dollars and asked how much of it they’d pay to have that food item to eat right now. Remember, these were hungry college students, who presumably cared about getting some calories at the time.
People turned out to be pretty lousy at consciously guessing how many calories the food items contained. Their unconscious brains, however, did much better: Their brain activity matched the real caloric content of the foods, not their estimate of the calories. The interesting result, though, showed up in people’s willingness to pay for the food. You’d think that when people are consciously deciding how much to pay for a snack, they’d base their decision on their conscious estimate of calorie count. But in fact, the amount they paid was a much closer match to the actual calories—the information accurately assessed by their unconscious.
At this point, you might be wondering why people persist in drinking Diet Coke, or continue to put sugar substitutes in their coffee. You’d think that their bodies would learn that those flavors don’t deliver calories and thus aren’t worth craving. One reason we don’t learn to ignore those flavors is that they deliver a jolt of caffeine, which also feels good. Our bodies learn to like the flavors associated with that kick—and with the buzz of alcohol, too. That’s why so many of us so easily develop a predilection for what are, objectively speaking, nasty, bitter, burning flavors.
There’s another point t
o consider when it comes to fooling the flavor system. You might be a dedicated Diet Coke drinker, but you probably encounter some of the same sweet, citrusy, caramely flavors in other foods, too, where they’re accompanied by real calories. That variability—sometimes sweet citrus means calories, sometimes not—might interfere with flavor-nutrient conditioning and make it harder for our internal calorie counter to keep track of how much we’ve eaten and when to stop. We might even be making matters worse, because we turn the flavors into a caloric slot machine that sometimes pays off and sometimes doesn’t. This sort of “intermittent reinforcement,” to use the technical lingo, is especially good at snaring our reward pathway. (Just look at all the zoned-out people sitting in front of actual slot machines in your nearest casino.) If so, artificial sweeteners might actually increase our attraction toward sweetness and the other flavors that accompany it. This may help explain why artificial sweeteners haven’t exactly been a weight-loss panacea.
It makes good evolutionary sense that all of this sophisticated learning takes place below the threshold of consciousness. Long before human beings ever walked the planet, and even before the first primates picked their way through the trees looking for fruit, our primitive mammalian ancestors would have needed to identify which foods were most nutritious. In short, they would have needed flavor-nutrient conditioning. And they probably had little or no conscious thought to help them with the task. “These circuits evolved so long ago,” says Small. “They were working perfectly well before we had consciousness.” As good mammals, then, we have evolved to want the flavor of calories. Or, to put the matter more precisely, we want the flavors that we’ve learned are accompanied by a dose of calories, while we ignore the flavors that aren’t. And this happens mostly without our conscious awareness.
But modern humans, with very few exceptions, no longer live on the African savannas, digging up roots and picking fruits and running down the occasional gazelle. We’re surrounded by an abundance of foods, and many of them are calorie rich in a way our ancestors rarely experienced. In this new context, our evolved instincts let us down. We no longer benefit by being attracted to high-calorie flavors when they’re always there—and the increased caloric density kicks our flavor-nutrient conditioning into overdrive, making those foods even more attractive. We want those flavors even when getting them is bad for us.