In a recent editorial in the New York Times titled “Are You a Carboholic? Why Cutting Carbs Is So Tough”, Gary Taubes describes a phenomenon that is familiar to many of us (1):
The holidays and family vacations are a particular problem. Desserts and sweets, it seems, will appear after every lunch and dinner, and I’m not particularly good at saying no when everyone else is partaking. The more sweets I eat, the more we eat as a family, the longer it takes upon returning home before that expectation of a daily treat fades away.
What I’ve realized is that eating a little of a tasty dessert or a little pasta or bread fails to satisfy me. Rather it ignites a fierce craving for more, to eat it all and then some. I find it easier to avoid sugar, grains and starches entirely, rather than to try to eat them in moderation.
Like Taubes, I have a hard time controlling my intake of certain foods when they’re in front of me, and occasionally, I may continue to think about them for days after I eat them (chocolate-covered ice cream bars, I’m looking at you). Taubes goes on to provide his explanation: these foods elevate insulin levels, and that in turn increases cravings for carbohydrate-rich foods. Par for the course, he provides no evidence for this model other than a few quotes from Drs. Robert Lustig and David Ludwig, which are nothing more than personal speculation. I’m not aware of any research suggesting that high circulating insulin levels promote cravings for carbohydrate, sugar, or dessert foods, and I’m fairly certain those data don’t exist.
What Taubes may not be aware of is that there is no need to propose a speculative new explanation for why we crave desserts, because researchers have already provided an explanation that’s quite compelling. And it doesn’t involve insulin.
It starts with drugs
To understand why we crave food, sex, gambling, winning, catching fish, driving fast cars, playing video games, and everything else we crave in our lives, it’s helpful to start with drugs of abuse. This is because the way in which drugs create cravings is very simple, and we can consider it without burdening ourselves with all of our preconceived notions and feelings about food.
Habit-forming drugs, including drugs of abuse, are habit-forming because they stimulate the dopamine signaling pathway in the brain (2)*. This pathway determines your gut level of motivation for things– what we often call a craving. It evolved to motivate us to get food, water, sex, shelter, social status, and other things that were important for the reproductive success of our distant ancestors. So when a drug increases the level of dopamine in key structures of your brain, you will become more motivated to take the drug– you will crave it. This is called reinforcement.
Some drugs, like amphetamine and injected or smoked (crack) cocaine, are very good at increasing dopamine levels in the brain (3). This means they are also very good at increasing your motivation to take them. If you take the drug repeatedly and your level of motivation/craving crosses a certain threshold, your brain begins to implicitly prioritize drug-seeking behaviors over constructive behaviors like keeping a job and obeying the law. We call that addiction.
When a person with an addiction doesn’t use a drug for a long time, his craving for it will slowly subside. This is why a person who quits smoking will initially feel very strong cravings, but these will diminish to a low level over a period of months to years. Yet when that person exposes himself to the drug again, or cues associated with it– for example, by taking a puff of a cigarette– it often re-awakens those latent brain pathways and triggers a relapse into addictive behavior (4). This can leave a person struggling with strong cravings once again. Note that this abstinence -> cue -> relapse cycle is precisely what Taubes describes in his article, only for sweets instead of cigarettes.
Drug reinforcement and relapse behavior, of course, are not mediated by insulin. The drug goes straight to the brain and spikes dopamine– there is no pit stop in the pancreas. High background levels of circulating insulin are not known to exacerbate addictive behavior. I don’t think anyone believes that cravings and addictions to alcohol, gambling, sex, or video games are due to elevated insulin levels. Is food reinforcement the one exception?
We actually have a pretty good idea of how food reinforcement works, thanks in large part to the research of Anthony Sclafani, which I describe in more detail in my book. As with drugs, food reinforcement works by spiking dopamine in the brain, a fact that Taubes has come around to (5). The question is, how does food increase dopamine levels in the brain?
Research by Sclafani and others shows that food reinforcement primarily occurs when the upper small intestine detects carbohydrate, fat, and protein, causing a spike in brain dopamine levels (taste receptors in the mouth that detect sugars also contribute, to a lesser extent)** (6, 6b). Researchers can also spike brain dopamine and cause reinforcement by injecting glucose into the liver’s portal vein, but this finding has not been consistently replicated (7, 8). After performing multiple experiments and reviewing the evidence on the role of insulin in reinforcement, Sclafani’s group concluded that “these data do not suggest a primary role for insulin in glucose-based [reinforcement]” (9). Part of the reason why they came to this conclusion is evidence suggesting that “insulin may act in the brain to reduce sweet taste signaling and sugar activation of brain reward systems” (emphasis mine)(10, 11).
Furthermore, if insulin is responsible for spiking dopamine and ultimately food reinforcement, then dietary fat shouldn’t have the same effect because it causes less insulin release. Yet fat is highly reinforcing– a key fact that Taubes neglects to share with readers (12, 13, 14).
Essentially, the brain is wired to be motivated by the food properties that kept our distant ancestors alive and fertile***. When you eat food, the brain analyzes its composition via receptors in your mouth, upper small intestine, and perhaps liver that measure the amount of fat, sugar, starch, protein, and salt it contains. This all happens on a nonconscious level. If the brain “likes” what you just ate, it will release dopamine. The more fat, sugar, starch, protein, and salt your food contains (up to the “bliss point”), the more dopamine your brain releases. The more dopamine your brain releases, the more your cravings increase. This is why we crave cookies and bacon but not lentils and plain celery.
Foods that are skillful combinations of fat, sugar, starch, salt, and delightful flavors likely cause your brain to release high levels of dopamine, sometimes provoking addiction in susceptible people. And when a person excludes a problem food for a long time, then suddenly has it in front of them at the dinner table, those latent craving pathways are reactivated– just as they are for drugs. No insulin required.
I’m not aware of any evidence that insulin is involved in this process, or that chronically high insulin levels accentuate it, and available evidence suggests that insulin is probably not involved. Why propose this unsupported mechanism when we already have an evidence-based mechanism that makes far more sense?
What about fatoholics?
Taubes’s story breaks down further when we consider that many of the most common food cravings aren’t for sweet foods. Rob Markus and colleagues recently published a paper in which they identified the food types that are most commonly associated with addiction-like behavior (15). Here’s what they found:
The majority of respondents experienced these problems for combined high-fat savoury (30%) and high-fat sweet (25%) foods, whereas only a minority experienced such problems for low-fat/savoury (2%) and mainly sugar-containing foods (5%). Overweight correlated only with addictive-like problems for high-fat savoury and high-fat sweet foods (P < 0.0001), while this was not found for foods mainly containing sugar.
Wait, what? The primary foods that trigger addiction-like behaviors are “high-fat savory” foods that aren’t even sweet? And only one person out of twenty reported addiction-like behavior for “mainly sugar-containing” foods? It appears that fatoholics are more common than carboholics.
This lines up with my personal experience, and I suspect, the experience of most readers. I don’t know about you, but to me, hard candy just isn’t that appealing. I have no desire to eat fat-free frozen yogurt. Candy corn and Tootsie Rolls are barely better than starvation. Throw some fat into the mix, as in chocolate, ice cream, or brownies, and now you’re talking! Savory foods like potato chips, bacon, pizza, and ribs are also high on my list. Taubes doesn’t mention that the “desserts and sweets” that provoke his addiction-like behavior are probably also high in fat, tacitly assuming that only the sugar is relevant.
One thing that’s clear from reading Markus’s study is that different people find different foods “addictive”. Most people have a hard time controlling their eating behavior around certain high-fat savory and/or high-fat sweet foods. Yet a small minority have a hard time with low-fat savory foods like pasta or low-fat sweet foods like hard candy.
We’re all built differently, and we each have our own problem foods that drive us to overeat. It’s entirely possible that Taubes is wired to have an unusually strong reinforcement response to foods like bread, pasta, and fatty sweet desserts that are typical on American dinner tables, but not to high-fat savory foods. To him, the common denominator may seem like it’s carbohydrate, which is why the insulin explanation is so irresistible that even a total lack of evidence cannot restrain him from it.
So why are some people carboholics and fatoholics?
The reason is simple. The human brain is hard-wired to be motivated by calorie-dense foods rich in fat, sugar, starch, protein, and salt, because these things kept our distant ancestors alive and fertile in a difficult world. In the modern world, the food industry and home cooks are better at pushing our brain’s reinforcement buttons than ever before in history. Our distant ancestors didn’t have ice cream or pizza, and the brains we inherited from them may not be designed to constructively handle that kind of powerful reinforcement stimulus. We’re so good at pushing our own buttons that some of our modern foods probably provoke very high levels of dopamine release in the brain. Like drugs of abuse, those high levels of dopamine make us crave in ways that don’t always support our health– and sometimes lead to addiction. That’s why we crave dessert.
Due to individual differences in how we’re wired, some of us become “carboholics”, others become “fatoholics”, others become both, and still others become neither. We have much work to do before understanding these individual differences. Yet regardless of whether or not we have addiction-like eating behaviors, nearly all of us eat too much when we encounter foods that are highly reinforcing.
I’m not surprised Taubes wrote this article, but I am surprised the New York Times published it. It might seem like fun and games to publish provocative speculation written by people who seem incapable of doing a Google Scholar search, but pieces like this can harm the public understanding of health science, with potentially serious consequences. How hard is it to run a piece like this by a few experts– people who are truly knowledgeable about the subject? Here are names: Anthony Sclafani, Karen Ackroff, Leann Birch, Roy Wise, Paul Kenny, Nora Volkow, Hisham Ziauddeen, Leonard Epstein, Jennifer Temple, Ashley Gearhardt, Stephen Benoit, Dianne Figlewicz, and Kent Berridge. As a reader and a scientist, can I please see some quality control?
* Typically by increasing dopamine levels, but also sometimes by acting on the same downstream pathway dopamine activates. For example, caffeine blocks the adenosine receptor, which increases dopamine’s action on downstream reward pathways.
** We don’t know what mechanism gets the signal from the intestine to the brain yet. Sclafani and Gary Schwartz found that cutting the nerves connecting to the small intestine doesn’t block reinforcement, suggesting that the signal may not be carried directly by nerves to the brain. Instead, it may be a factor released by the intestine into the bloodstream, but that remains to be demonstrated. This is a very important mechanism to understand because it’s so relevant to obesity. I suspect it won’t be too long before we get a toehold on it.
*** Sugar was obviously not the only nutrient that kept our distant ancestors alive, and from an engineering standpoint it would make no sense to design a system that is only motivated by sugar. That’s why our brains are strongly motivated by all concentrated sources of calories– including sugar.