BOOKS AND ARTS MARCH 19, 2007
Mindless Eating: Why We Eat More Than We Think
By Brian Wansink
(Bantam, 276 pp., $25)
The idea of "the survival of the fittest" is one of the most powerful organizing principles in all of science. That simple idea, stated by Herbert Spencer on the basis of Charles Darwin's work and later endorsed by Darwin himself, captures the theory of evolution, the process of natural selection, and a host of associated notions. And yet the phrase can produce confusion. Some people think that survival of the fittest implies that species will be ever improving, and that those still around today will be well-designed machines, smart, efficient, and, well, "fit." But Darwin was focused on reproductive success, and nothing in his work suggests that surviving species are particularly smart or, in any global sense, fit.
A visit to Darwin's beloved Galapagos Islands drives this point home. The famous blue-footed booby got its name because of its apparent stupidity. The bird would land on the decks of sailing ships and allow itself to be captured with a lunge from a drunken sailor. (On the islands the birds have essentially no predators and thus no need to be wary.) The giant tortoises are slow-footed, and if they are flipped over onto their backs they have no way of righting themselves. (They never had to cope with teenage boys.) And many of the Galapagos species lean to the chubby side. Bull sea lions have figures resembling NFL nose tackles and can weigh more than eight hundred pounds. The waved albatross can fly enormous distances and withstand very cold water, but it has enough padding that it cannot get airborne without the aid of a wind or a cliff. On a calm day, even a running (er, waddling) start is not enough for it to attain lift-off.
Yet all these animals have survived because they were well-suited to their environment--as long as that environment did not change. The giant tortoises were wiped out on some of the islands because of the introduction of two new species brought by human beings. Goats were faster at eating the grasses favored by the tortoises, and rats (who seem to follow people everywhere) soon acquired a taste for tortoise eggs. Tortoises were as slow and large as ever (they can live for months without food or water), but they were no longer fit enough. In an unchanging environment, natural selection can lead to the survival of the fattest. But if things change, the fat can be in big trouble.
Which brings us to the current state of the American waistline. Over the past few decades, many Americans have grown fat and are growing fatter. More than 27 percent of us are obese. This is a genuine personal and social problem, because obesity is associated with heart disease, diabetes, and many other life-threatening conditions. Why are so many Americans fat? At one level the explanation is rather simple. We gain weight when we consume more calories than we burn. So any theory for why we are gaining weight has to explain why we are eating more, exercising less, or both. A traditional economic argument would be based on rational calculations and technological change. Innovations in packaging, preserving, and cooking have greatly decreased the time it takes to prepare food. And if you lower the cost of something, rational consumers will buy more of it. American consumers eat more, and more often, than they did fifty years ago. Technology has also made it easier for us to avoid exercise. Many of us ride an elevator down to our car, drive to work, and then ride another elevator up to our office. The most strenuous exercise we exert is at the keyboard, or hoisting a glass. According to this view, obesity is a rational choice. When faced with the choice of swapping a Big Mac for an apple or climbing the stairs instead of riding, some people buy a larger pair of pants instead.
Another approach, favored by behavioral economists, stresses issues of self-control. According to this view, our environment has changed in a way that presents us with increasing temptations. When tempted to pop corn in the 1950s, consumers had to get out the pot and start popping. Now the microwave produces popcorn nearly instantly. Fast-food restaurants, convenience stores, and packaged food have all made sticking to a diet more difficult. This argument strengthens and reinforces the standard economic analysis, but it adds the idea that people would, in some sense, prefer to be thinner. The billions of dollars that consumers spend on diet books and self-help programs such as Weight Watchers lend credence to the claim that people are conflicted about their weight gain, whereas in a purely rational world, choosing to be plump would be a simple choice, no more conflicting than choosing to wear a blue hat.
While technological changes and self-control problems are undoubtedly part of the story, Brian Wansink adds an important additional piece of the puzzle. On the basis of dozens of experiments conducted by his team and others, Wansink has concluded that much eating is mindless. Americans are fat not because they have made a rational calculation that French fries are so yummy that they are worth the costs in health and svelte. Nor are French fries essentially irresistible. Often Americans eat because of contextual cues, or "hidden persuaders," to which they are blind, but which greatly affect their behavior. And though Mindless Eating is partly a diet book, it is not just for the chubby any more than Michael Lewis's Moneyball was just for baseball fans. Wansink's real subject is choosing, not eating, so even thin people should read it.
Wansink begins with the premise that we don't know much about what makes us eat. Often we overeat food that we do not even like, simply because of the context in which we find ourselves. All this happens unconsciously. "In the thousands of debriefings we've done for hundreds of studies, nearly every person who was 'tricked' by the words on a label, the size of a package, the lighting in a room, or the size of a latte said, 'I wasn't influenced by that.'" Wansink calculates that people make more than two hundred food-related decisions every day; most of those decisions are made quickly and intuitively, without any kind of deliberate assessment. As he puts it, "stomachs can't count," and "it's simply not in our nature to pause after every bite and contemplate whether we're full."
This argument is not original to Wansink. The psychologists Richard Nisbett and Timothy Wilson convincingly argued more than two decades ago that people often lack much insight into why they do things, such as buy a certain brand of automobile or beer. When asked, "Why did you do that?" people have no trouble providing answers, but their answers often turn out to be wrong, and they have no more validity than the ones that an outside observer could give. (For example, people sometimes say that they chose one product over another because it looked better, when the two products actually were identical.) Wansink shows that people frequently have little or no idea why they eat so much.
The initial problem is that if we see food, we are likely to eat it, even if we aren't hungry. People tend to eat whatever is put in front of them. Wansink demonstrates this point through a series of somewhat mischievous experiments, some of which would have been great material for Candid Camera. A few years ago, moviegoers in Chicago found themselves with a free bucket of popcorn. Unfortunately, the popcorn was stale; it had been popped five days earlier and stored so as to ensure that it would actually squeak when eaten. People were not specifically informed of its staleness, but they didn't love the popcorn. As one moviegoer said, "It was like eating Styrofoam packing peanuts."
As the experiment was designed, about half of the moviegoers received a big bucket of popcorn and half received a medium-sized bucket. After the movie, Wansink asked the recipients of the big bucket whether they might have eaten more because of the size of their bucket. Most denied the possibility, saying, "Things like that don't trick me." But they were wrong. On average, recipients of the big bucket ate about 53 percent more popcorn--even though they didn't really like it.
Another experiment required some special equipment. People sat down to a large bowl of Campbell's tomato soup and were told to eat as much as they wanted. Unbeknown to them, the soup bowls were designed to refill themselves (with empty bottoms connected to machinery beneath the table). No matter how much soup the subjects ate, the bowl never emptied. Many people just kept eating until the experiment was (mercifully) ended.
The general rule seems to be, "Give them a lot, and they eat a lot." Those who receive large bowls of ice cream eat much more than those who get small bowls. If you are given a half-pound bag of M&M's, chances are that you will eat about half as much as you will if you are given a one-pound bag. The reason is simple: packages "suggest a consumption norm--what it is appropriate or normal to use or eat." In fact, most people do not stop eating when they are no longer hungry. They look to whether their glasses or plates are empty. Those who keep a great deal of food in their refrigerator eat far more than those who do not. (An evident trick for dieters is to buy less each time you go to the grocery store, even if the result is that you go to the grocery store more often. Even better, walk to the store!)
Size operates as a hidden persuader, but it is hardly the only one. Names matter a lot. "Traditional Cajun Red Beans With Rice" is far more popular than "Red Beans with Rice." You might well order "Home-Style Chicken Parmesan," even if you would turn up your nose at mere "Chicken Parmesan." People aren't terribly enthusiastic about "Zucchini Cookies," but they will happily eat "Grandma's Zucchini Cookies." Not only are people more likely to ask for, and to eat, appealingly named dishes; they will also rate those dishes as tastier. Those who have had restaurant items with the foregoing names are more likely to describe their meals as "great" or "fantastic." The right names for dishes can even produce more enthusiastic attitudes toward restaurants as a whole, leading people to characterize them as "trendy and up-to-date." Wansink thinks that if restaurants without particularly good food seek to increase sales, they might choose from a range of effective labels, including the geographic ("Kansas City Barbeque"), the nostalgic ("Classic Old-World Manicotti"), and the sensory ("Hearty Sizzling Steaks"). In this way, Wansink helps to systematize what savvy advertisers already know.
It is healthy to eat soy, and you can lose weight by selecting soy products, but efforts to market those products have fared pretty badly. Why? A big reason is that people have a bad association with the word "soy." In one of Wansink's experiments, a number of people were asked to eat a new PowerBar, prominently labeled with the words "Contains 10 grams of protein." Other people, not different from the first group, were asked to eat the same PowerBar, prominently labeled with the words "Contains 10 grams of soy protein." Members of the first group liked what they ate; they described it as chewy, tasty, and chocolaty. Members of the second group were not so pleased. Many spat out the bar. Many others asked, after the first bite, if they could have a drink of water. Those who knew that they had eaten "soy protein" said that the bars had a bad aftertaste and "didn't even taste like chocolate."
Wansink's lesson is that "we taste what we expect we'll taste." To support this claim, he notes that in the dark, people are willing to believe that chocolate yogurt is strawberry yogurt--and apparently to enjoy it just as if it were strawberry. A military chef found himself with a group of sailors who were tired of eating lemon Jell-O and insisted on getting their favorite flavor, which was cherry. Not having any such Jell-O, he colored lemon-flavored Jell-O red--and the sailors ate it happily. Indeed, even many wine connoisseurs cannot tell the difference between red and white wine when the wine is served in dark, opaque stemware.
Social influences also have a large impact. An especially good way to gain weight is to have dinner with other people. On average, those who eat with one other person eat about 35 percent more than they do when they are alone; members of a group of four eat about 75 percent more; those in groups of seven or more eat 96 percent more. We are also greatly influenced by consumption norms within the relevant group. A light eater eats much more in a group of heavy eaters. A heavy eater will show more restraint in a light-eating group. The group average thus exerts a significant influence. But there are gender differences as well. Women often eat less on dates; men tend to eat a lot more, apparently with the belief that women are impressed by a lot of manly eating. (Note to men: they aren't.)
When it is even modestly less convenient to get food, people will eat less of it, and sometimes much less. If the candy in people's offices is made only slightly harder to reach, their candy consumption drops. If people have to open a lid to get to the ice cream in a freezer chest, they will eat a lot less ice cream than they will if the lid is open. If almonds are in shells, overweight people are much less likely to eat them than if they have already been shelled. In this light, it is easy for cafeterias to exploit the effects of seemingly small changes in convenience. When people are asked to go to a separate lunch line to pay for candy and potato chips, they buy a lot less.
Smells, sounds, and sights are important too. It turns out that if oatmeal is served in a bowl manufactured to smell like cinnamon and raisins, people eat a lot more oatmeal. But if the bowl is engineered to smell like macaroni and cheese, oatmeal becomes pretty unappealing. When the background music is pleasant and slow, people stay a lot longer and drink a lot more; in one study, soft music was worth 41 percent more in drink revenues per table. By contrast, fast-food restaurants, seeking rapid turnover, "decorate for speed eating: bright lights, lots of hard surfaces that reflect lots of noise, and a high contrast, high arousal yellow-and-red color scheme." Sensible restaurants decide on the optimal tradeoff between lingering diners, who might well consume more, and faster but more numerous diners. Once they have made their choice, they can offer contextual cues that will lead people in their preferred directions.
Wansink's emphasis on hidden persuaders and on the mindless character of much eating leads him to wonder about the widespread idea that the best response to obesity and poor diet is to provide informational labels. If people eat mindlessly, what exactly will they learn from such labels? To answer that question, Wansink conducted a little study of how much people know about their meals at Subway and at McDonald's. Subway restaurants are pervaded by information; nutritional data can be found on posters, cups, napkins, and tray liners. If you eat at Subway, you will have easy and immediate access to information about the contents of the various options. And with its extensive advertising campaign, Subway has tried to convince people to go there for nutritious, low-calorie meals. But McDonald's is very different. While nutritional information is available, it is not prominently displayed; few people choose McDonald's in the hope of getting a low-calorie meal. So it is not surprising that McDonald's customers do not know much about the nutritional value of their meals. Of 250 patrons at McDonald's, Wansink found that only 57 had even a crude recollection of any nutritional information. On average, the 250 diners consumed nearly 1,100 calories apiece, but they estimated that they had eaten only 876--25 percent less than reality. (This finding is not unusual: people of normal weight think that they eat 20 percent less than they actually do, and obese people are even more optimistic.)
What about the Subway customers who have been bombarded with nutritional information? In some ways the Subway customers seemed far more knowledgeable, as 157 of 250 Subway diners recalled some nutrition information. But most of the Subway group, including those 157, did not seem to pay much attention to that information. In general, they "ignored the low-fat sandwiches and stampeded straight to the high-calorie ones with meatballs, cold cuts, and bacon." They also ate a lot of cheese, mayonnaise, chips, and cookies. True, the average diner at Subway consumed 677 calories, much less than the 1,100 at McDonalds, but more than the 495 they thought they were eating--an underestimate by 34 percent. So by this measure (the percent by which they underestimated their caloric consumption), Subway customers were even more misinformed than the McDonald's customers.
Wansink attributes this odd result to Subway's "halo effect," and he finds that this effect is quite widespread. People who are given low-fat foods tend to eat more of them. Often the increase in the overall volume of eating means that, on balance, they are gaining weight, not losing it. For example, low-fat granola has only 10 percent fewer calories than regular granola. Those who get low-fat granola often feel free to eat a lot of it--more than enough to make up the difference in calories.
Wansink found a similar halo effect in a study involving the consequences of giving people a free glass of wine. In a cafeteria one night, half the diners received wine that was said to be from California; the other half received wine said to be from North Dakota. In fact, the wine was identical cheap California wine. (As far as we know, there is no such thing as North Dakota wine, at least wine made with grapes. We hear there is a so-called wine from North Dakota made with rhubarb, but with respect, we are reluctant to call such a beverage wine.) Almost all the diners drank all of their free glass of wine. What was surprising was that those who thought that they had been given California wine ate much more food and stayed much longer. "Exact same meals, exact same wine"--but once people were given a free glass of California wine, they decided that the meal "was going to be good," and they ate an extra amount.
With an understanding of the role of hidden persuaders, how does Wansink propose to help people to lose weight? His organizing idea is that "the best diet is the one you don't know you're on." He wants people to diet mindlessly--to eat smaller amounts fairly automatically. If you avoid stocking up at the grocery store, you will almost certainly lose weight. A more formal approach is to adopt three simple, easily memorized rules that will lead you to consume 100 fewer calories each day (which can turn into as much as a twenty- to thirty-pound loss in a year). The advantage of such rules is that they can develop into habits and thus be applied without continued thinking. Wansink also points to the potential use of "food tradeoffs," as in the idea that I will have a candy bar only if I am playing tennis today, or have dessert only if I don't snack in the morning.
Wansink can be read either as a diet adviser or as an analyst of decision and choice. As a diet adviser, he is sensible but hardly earth-shattering. His suggestions should help people who want to lose ten or twenty pounds; as he notes, his ideas aren't likely to do a lot for people who need to lose much more. Wansink is much more interesting if his clever studies of food consumption are used as a way of gaining insight into everyday decision-making in many domains. His key lesson is that largely invisible features of the context in which we make choices often move our decisions in one direction or another. In deciding what to eat, we are influenced by factors that we believe to be irrelevant and in fact barely observe. If the point holds for eating, it should hold for many other decisions as well, including those that involve health, safety, friendship, finances, and even romance.
But why do we often keep eating even when we are not hungry? Wansink rightly starts with the observation that "stomachs cannot count." He also stresses that the signals we receive from our stomach that we are full come with a lag of at least twenty minutes. So fast eaters can consume a lot of food before their brain receives the message from the stomach announcing "no more room." But, we might wonder, why are human beings wired this way? We usually feel pain immediately; why does it take us the length of a sitcom to realize we have stuffed ourselves on potato chips and don't feel so good?
A simple evolutionary account provides some clues. For most of human history, the simple rule "if food is present, you should eat it" was probably a good one. Learning to monitor how much we eat, in order to avoid becoming obese, was something most human beings did not have to worry much about, and a little extra padding might come in handy if the larder became bare. In a famine, bet on Jack Sprat's wife to outlive him.
These ancient instincts can be strongly reinforced by cultural norms. What is put in front of you provides a signal; it suggests what you should eat. When the plate is large and packed, it conveys information about what is taken to be the ordinary or appropriate meal--and so people will eat a lot. When a plate is small, or when it has a small portion, it also conveys such information--and so people will eat much less. The conclusion is that social contexts offer a range of subtle, sometimes deliberate, sometimes inadvertent signals, by which people are influenced even if they are unaware of that fact. And this conclusion extends well beyond eating.
Consider an area that seems far afield: the appropriate design of savings plans. In most workplaces, eligible employees receive a notification and are told, "If you want to join the plan, fill out these forms." If employees do nothing, they are not enrolled. But an increasing number of firms use what is called "automatic enrollment." Under automatic enrollment, eligible employees are told, "you are now eligible to join the plan, and unless you fill in the enclosed form you will be enrolled." The choice of automatic enrollment has a large impact. Employees are much more likely to join the savings plan under automatic enrollment, and they join sooner as well.
Social scientists, above all economists, are often puzzled to find that the default option makes such a difference. After all, people can easily select the plan that they prefer. If they want to save money, they can join the relevant plan and opt in; if they don't, they can easily opt out of automatic enrollment. Filling out the form takes just a few minutes. Why does the default have such an impact? Two reasons are important. First, as illustrated by many of Wansink's experiments, people have a strong tendency just to "go with the flow." Call it the "yeah, whatever" heuristic. This tendency creates enormous inertia. People who enroll in the plan almost never drop out, but those who fail to sign up when first given the opportunity to do so can take years to get around to it, even in firms in which the employer is providing a strong monetary incentive to join via a company match.
But failing to join is not just a product of inertia. Another reason is that the chosen default conveys what at least has the appearance of information: it provides a strong signal. If your employer automatically defaults you into a savings plan, you might think that a sensible person has decided that the plan is actually in your interest, and that you should stick with it. Defaults matter because people are often mindless choosers, and not just in what they eat.
Return in this light to Wansink's finding that social influences have a large impact on food intake. One reason is that the behavior of others conveys information about what it is reasonable to do. If you are in a group of people who are eating a great deal, you have reason to believe that the food is good and that it makes sense to eat a lot of it. But peer pressure is also important. If you find yourself in a group of people who are eating only modest amounts, perhaps group members will think ill of you if you eat a lot. To the extent that light eaters consume more in heavy-eating groups and heavy eaters eat less in light-eating groups, information and peer pressure are the principal reasons. Social influences have a powerful influence in many domains; they greatly affect decisions about how much to drink, whether to stay in school, to bring lawsuits, to take precautions against natural disasters, to support a particular political campaign, to make charitable contributions, and to commit crimes.
In sum, many of Wansink's findings are best explained as a product of the cues subtly or inadvertently conveyed by one or another feature of the context. If a restaurant says that it is giving you "North Dakota wine," you have a signal of the quality of the restaurant, and that signal may greatly affect your judgment about the food. If you are eating low-calorie granola, it is not exactly senseless to think that large portions will not be particularly fattening. If Jell-O is labeled "strawberry," and is also red, many people will taste it as strawberry Jell-O even if it is lemon, especially because strawberry Jell-O does not taste much different from lemon Jell-O. Very few of us are always willing and able to make fine-grained decisions about how much to eat or about the nutritional value of what we are eating. Instead, we rely on contextual cues. Much of the time, it is sensible to use those cues. The problem is that Wansink, advertisers, restaurants, and other savvy people can manufacture cues to manipulate us in their preferred directions.
But there is some good news here. As Wansink shows, awareness of our vulnerability to manipulation can provide a degree of inoculation against it. If we want to lose weight, we can insist on small portions, split a main course with our dining partner, aim to eat less than our friends at social occasions, and feel bemused, rather than seduced, by attractively named foods or unnecessarily large portions. If we want to save for our retirements, we can take steps, at least once a year, to see whether it might make sense to put more money into savings plans, or, better yet, sign up for a plan that increases our contributions automatically. To improve our health, we might commit to exercising three times a week at a specific date. But the opportunities go well beyond self-help. In a brief but intriguing discussion, Wansink explores the role of "nutritional gatekeepers"--those in the family who select the meals and "have a huge day-by-day influence on their family's nutrition." It turns out that on average, gatekeepers control no less than 72 percent of what their families eat. Wansink contends that gatekeepers can play a large role in nudging their families toward more nutritious and less fattening diets, in part through choosing variety, in part through clever "marketing" to spouses and children, in part through using small serving sizes.
Many other private and public institutions operate as "decision gatekeepers," or what might be better called decision architects, nudging people in particular directions. Restaurants are an obvious example, but the same point holds for hospitals, schools, and computer companies, which offer default packages of various sorts, imposing powerful contextual influences that move people in their preferred directions. Employers are decision architects, too, offering initial packages including one or another combination of take-home wages, savings plans, vacation time, health care, parking privileges, and more. If Wansink is right, these influences will sometimes be simultaneously invisible and strong.
The same point holds for government programs involving Social Security, prescription drugs, discrimination, health care, poverty relief, the environment, and much more. Many of these programs ask citizens to make choices while also offering confusing, distressing, or bad signals. The Medicare Plan D prescription-drug plan could have been much improved with a better understanding of how people choose; many seniors have been overwhelmed by the wide range of options and the sheer complexity of the decisions they are being asked to make. If changes were made to help seniors select the plan that best suits their preferences and drug requirements, then the value of the program to the participants would increase without raising the costs to the government or the insurers.
Armed with an awareness of the power of contextual cues, many people have been exploring new forms of paternalism, sometimes called "weak" or "thin" or (our preferred term) "libertarian." The central idea behind libertarian paternalism is that it is often possible for decision architects to steer people toward better decisions (as judged by the decision-makers themselves, not the architects) without restricting freedom of choice. Automatic enrollment is a good example. The plan designer has to choose something as the default, either opt-in or opt-out. Many more people join under opt-out, and only a tiny number of people quit the plan after having been automatically enrolled, which implies that most were content with the suggestion.
Wansink's own work suggests a nice possibility for libertarian paternalism in the domain of food consumption: public cafeterias, serving children or even adults, could easily enlist contextual cues to lead people to select meals that are both less fattening and more nutritious. If displaying the fruit more prominently than the desserts in a school cafeteria line produces healthier people, isn't that a good outcome? And such measures are far less intrusive than more coercive actions, such as flat bans on junk food from cafeterias.
To those who object to the effort to enlist such cues, Wansink's answer is powerful: such cues are always present anyway, and the real question is their effects. For would-be dieters and gatekeepers, this point offers a clear warning, but it also signals an opportunity, in the form of promising approaches that work by implementing, rather than blocking, people's reflective judgments. If Wansink is right, the same opportunity can be found in many other domains as well. And even in his favored domain, gatekeepers can do more to improve the survival rates of the fattest members of our society, and make all of us a bit more fit.
Cass R. Sunstein is a contributing editor. Richard H. Thaler is a professor at the University of Chicago and the author of The Winner's Curse: Paradoxes and Anomalies of Economic Life (Free Press).
By Cass R. Sunstein and Richard H. Thaler
Cass R. Sunstein is a contributing editor. Richard H. Thaler is aprofessor at the University of Chicago and the author of TheWinner's Curse: Paradoxes and Anomalies of Economic Life (FreePress).