Here, I intend to clarify the core claims of the palatability theory of obesity outlined by Stephan J. Guyenet in his book The Hungry Brain (reviewed by Scott Alexander). Previously, I had written a comment defending his theory against a recently popular alternative, proposed by Slime Mold Time Mold (SMTM), which states that environmental contaminants are responsible for the ongoing obesity epidemic.

As with many other topics, I often find public debates about obesity frustrating because it's sometimes unclear to me what people mean when they champion a theory. To address this concern, I will briefly summarize (but not necessarily defend) what I see as the key elements of Guyenet's theory, including how it can be empirically falsified and distinguished from its competition.

The theory in a nutshell

The palatability theory is a theory of why obesity has increased over time. It posits that the primary cause is an increase in reward signals in the modern environment from processed food, mediated by the motivation system in the brain, causing us to overeat.

This sounds very simple, but there are many ways to be confused about what the theory is actually claiming, and indeed, I think many people are at present confused about it. Therefore, I'm going to go through the main parts of the theory, step by step, in the hopes that this clears up some confusion.

The datum to be explained

The palatability theory of obesity is mainly trying to explain why humans, especially those in rich nations like the United States, on average weigh more and have higher body fat percentages than their recent ancestors at similar ages.

Note that this is different from saying that the theory is trying to explain variation in individual obesity, at any given time, within a society. The palatability theory does not provide any direct explanation for why some people are more obese than others, whether and why some people's metabolisms are more efficient than others, why some people have more difficult time dieting than others, or why some people prefer hyperpalatable food relative to others.

To be sure, the palatability theory can shed light on each of these questions, and it can certainly be empirically tested using data collected across space rather than across time. For example, if we compared two contemporaneous communities, one whose members ate a modern "cafeteria diet" and another whose members ate only traditional staples, like rice, quinoa, and lentils, then the palatability theory predicts that the second community would have lower levels of obesity. 

The theory does not say that the second community would have no obese people at all, just a lot fewer. I think a charitable interpretation of Guyenet recognizes the potential for other significant environmental factors, including parasite load and infection, prescription medication (in fact, he recently wrote about this), and to some degree, environmental contaminants. I think he's just saying that these are comparatively minor factors for explaining the obesity epidemic, especially in rich nations. (And of course, obese people existed even in the ancient world).

In regards to individual variation in obesity in the modern world, Guyenet does provide his own speculation, pointing to genetics. From his book,

In modern affluent nations like the United States, genetic differences account for about 70 percent of the difference in body weight between individuals. They also play a prominent role in many of the details of our eating behavior, such as how much food we eat at a sitting, how responsive we are to the sensation of fullness, and how much impact food reward has on our food intake. In other words, whether a person is lean or fat in today’s world has less to do with willpower and gluttony and more to do with genetic roulette. If you want to be lean, the most effective strategy is to choose your parents wisely.

Genes also explain that friend of yours who seems to eat a lot of food, never exercises, and yet remains lean. Claude Bouchard, a genetics researcher at the Pennington Biomedical Research Center in Baton Rouge, Louisiana, has shown that some people are intrinsically resistant to gaining weight even when they overeat, and that this trait is genetically influenced. Bouchard’s team recruited twelve pairs of identical twins and overfed each person by 1,000 Calories per day above his calorie needs, for one hundred days. In other words, each person overate the same food by the same amount, under controlled conditions, for the duration of the study. 

If overeating affects everyone the same, then they should all have gained the same amount of weight. Yet Bouchard observed that weight gain ranged from nine to twenty-nine pounds! Identical twins tended to gain the same amount of weight and fat as each other, while unrelated subjects had more divergent responses. Furthermore, not only did twins gain a similar amount of fat, they even gained it in the same places. If one person stored the excess fat inside his abdominal cavity—the most dangerous place to gain fat—his twin usually did the same. Not only do some people have more of a tendency to overeat than others, but some people are intrinsically more resistant to gaining fat even if they do overeat.

I think this quotation shows that Guyenet is not merely claiming that fat people have less willpower, or that they're less virtuous. Yet on Facebook, Eliezer Yudkowsky wrote of Guyenet's theory (and recently said, "Nothing I've learned since then has changed my probabilities here."),

I am a bit personally offended by Stephan Guyenet's talk of how his hyperpalatability theory doesn't _really_ amount to blaming fat people for eating too many tasty things, because the poor dears can't help their hungry brains. Yes we can, dear, we have eaten things far worse than bland foods for months on end. But I shouldn't blame Guyenet - theories that let you take a superior perspective and say that other people can't help themselves (unlike him, who did manage to help himself), can be very "tasty". So his brain probably can't help itself. Unlike mine, which is better.

For indeed, my brain has no problem with swallowing the untasty fact that, after becoming acquainted with many heaps and mounds of studies and theories, I still have basically no clue why people's metabolisms do what they do. And neither does Guyenet, to all appearances. It looks to me like he just took a simplistic theory that sounds a lot like the good old folk "virtue theory of metabolism", while pretending that he isn't really blaming the obese he's blaming their brains; and slapped that on top of a huge mound of studies as his conclusion.

Personally, I think this is an unfair characterization of the palatability theory, and more importantly what's it's even trying to address in the first place. We can easily imagine theories that try to explain why a variable increases over time, such carbon dioxide and the global temperature, without the theory saying much about why the variable differs across space. And while Guyenet does offer a lot of speculation about why obesity differs between people, his speculation on this question is often quite distinct from the virtue theory that Yudkowsky attributes to him in this passage.

The main claims of the theory

In my opinion, Guyenet's core claims should be clearly distinguished from some of his side speculation, such as his theory that hyperpalatable food causes leptin spikes, and over time, leptin resistence, which is supposed to explain why our weight set point increases.

If the leptin mechanism turns out to be false, it would definitely mean that Guyenet was wrong about the specifics, but afterwards his general theory might still be intact. That's because I think his general theory of obesity only makes the following claims,

The core mechanism is overeating

Guyenet asserts that the reason why we're getting more obese over time is because we're eating more. Notably, he doesn't make a big deal about exercise,

I’ve graphed calorie intake estimates from all three methods in figure 3. As you can see, the methods yield different estimates, but they all agree that our calorie intake has increased substantially over the period of time that we gained weight the most rapidly (218–367 additional Calories per day between 1978 and 2006). The third method, pioneered by National Institutes of Health researcher Kevin Hall and highlighted in black on the graph, probably comes the closest to capturing the true increase in daily calorie intake over the course of the obesity epidemic: 218 Calories. Remarkably, this increase is singlehandedly sufficient to explain the obesity epidemic that developed over the same period of time, without having to invoke changes in physical activity or anything else.

How could Guyenet be wrong here? Well, he'd be wrong if people either weren't eating a lot more calories than we were in 1978, or if we we are currently exercising a lot less, and that explains the data instead. In general, he'd be wrong if some other mechanism other than calories in, calories out (CICO, by which I mean the difference between calories consumed and calories expended) explained the rise in obesity.

I should note that I do not think that SMTM actually disagrees with this mechanism. In one of their posts, they argue against the CICO theory, but in a follow-up post, they clarify that by CICO they merely meant the folk theory that willpower is responsible for obesity (which neither Guyenet nor I believe!).

Still, I think it's worth mentioning this mechanism, because it's the central way in which Guyenet's theory is supposed to work. If overeating doesn't explain why we're more obese than in the past, then the palatability theory would fail right out of the starting gate.

It has something to do with our food

Guyenet claims that the reason why we're overeating is related to food and, to a lesser extent, the sugary drinks we consume. This part of his theory might sound too trivial to mention, but it's important to point out because of how we might use this fact to distinguish the palatability theory from SMTM's theory of obesity

SMTM, by contrast, does not derive a necessary connection to food. In general, the contamination hypothesis is compatible with the idea that we're overeating because of what's in our food, but it's not a necessary ingredient of the theory.

In fact, SMTM's leading hypothesis is that it has to do with what's in our water supply. They state,

Obesity is less common at high altitudes because of the watershed. Environmental contaminants build up as water flows downhill and are in much higher concentrations as you approach sea level.

This divergence can be used to distinguish the empirical predictions of the palatability theory compared to SMTM's theory.

There are many causes

Guyenet does not single out one part of our food diet. Instead, he blames almost everything in the modern diet, and the various ways these things are processed, and combined. For instance, he writes that,

In popular media, there is a perennial debate over whether sugar or fat is responsible for the obesity epidemic. This has led some people to view obesity research as a team sport rather than a scientific discipline. Allow me to end the debate by stating what most researchers find quite obvious: It’s both.

Furthermore, he lays blame to the convenience of getting food in the modern environment, 

[Edit: Actually, maybe Guyenet doesn't believe this part of his theory anymore, due to allegations against Brian Wansink's research. Perhaps Guyenet has more than one line of evidence for this claim, but I have yet to check. See this comment chain.]

Brian Wansink, director of the Food and Brand Lab at Cornell University, has conducted clever experiments that illustrate the outsized influence of effort cost on our eating behavior. In one study, he recruited administrative assistants and placed candy dishes containing Hershey’s Kisses in one of three different locations in each of their offices: on the desk, in the top drawer of the desk, or in a filing cabinet six feet away. Eating a Kiss on the desk required only a small arm motion, while Kisses in the drawer required a larger arm motion, and those in the filing cabinet required getting up and walking across the room. Each additional effort barrier, although small, made the Kisses a less attractive deal.

Remarkably, these seemingly trivial differences in effort cost resulted in large differences in candy intake. Participants with candy bowls on their desks ate an average of nine Kisses per day. Those with candy bowls in their desk drawers ate six Kisses per day, and those who had to hike all the way across the room only ate four Kisses. As Wansink puts it, “It’s not worth the effort for an Eskimo to locate and overeat mangos.” Would we be leaner as a nation if we all had to walk three miles and climb a tree each time we wanted a hamburger and fries (or ice cream, or pizza)? Almost certainly. Yet our food today is more convenient than it has ever been in human history.

Finally, Guyenet highlights food variety,

Several independent researchers using various methods have confirmed that we tend to eat more total food—and gain weight—when we’re presented with a large variety of foods. This goes a long way toward explaining what researchers call the buffet effect. We tend to overeat spectacularly at buffets, despite the fact that the food isn’t always the crème de la crème.

The complexity of Guyenet's theory should earn it a prior probability penalty. Each of these causes needs to be substantiated before we accept his theory as true.

However, the complexity of this theory is also a strength in one sense, in that it allows us to distinguish the theory from monocausal alternatives. If there were a single reason why humans are more obese than in the past, then that leaves open the question of why no one's figured it out by now, and moreover, why no major nation has ever been successful at reducing obesity within their borders.

If, as SMTM suspects, the underlying reason for the obesity epidemic is lithium in our water, then it's a little surprising that, not only has no one identified this cause, but literally almost every single nation in the world has unwittingly been contributing in the exact same way, adding lithium to their water, producing the large rise in obesity we see across the world. This is not a perfect heuristic—it's totally conceivable we live in such a world—but still.

It's mediated by the brain

Really, what makes Guyenet's theory stand out among some competing popular theories is the way he talks about the brain, and how our overeating is governed by the motivation system in the brain. He says: it's not a chemical imbalance, a problem with our metabolism, or digestion failure. We just actually want to eat more food these days.

Naively, this is what makes the palatability theory most intuitive to me. I think it's pretty clear to me that chocolate milkshakes are more appetizing than water, that cheeseburgers are more appealing than eating raw meat, and restaurant-prepared burritos taste much better than merely eating each of the unprocessed ingredients alone. In each of these cases, I find myself often overconsuming, and for precisely the reason that Guyenet proposes. In other words, the palatability theory powerfully matches my personal experience—modern food is pretty addicting—and I'm surprised when other people seem not to agree on this point.

It's also worth pointing out that Guyenet thinks this mechanism is not easy to reverse. When I was 15 years old, I decided to become a strict vegan, which lead me to eat food that was quite a bit less palatable compared to what I was used to. More than one year afterwards, however, I still had an intense craving for cheese; it was clear that my brain had not yet forgotten some of its old patterns of thinking about food.

Guyenet suspects that our brain's weight set point might never go down dramatically after living long enough in the modern world, even if we eventually stop eating palatable food altogether. If true, this would make his theory harder to test, and again, his theory would earn a penalty for being more unfalsifiable, but at the same time, we should be clear about what observations his theory strongly predicts, and rapid weight loss on unpalatable diets is just not one of them.

Putting these claims together

In order for a theory of the obesity epidemic to be meaningfully empirically different from the palatability theory, I think it must diverge on at least one of the claims above. Either the root cause must not be overeating, or it must not have something to do with our food, or there must be a single (or a very small set) of proximate causes, or the main mechanism must be unrelated to the motivation system in the brain.

On a personal note, I often find it very difficult to stay thin, and so maybe Guyenet is subtly saying I have a bad brain, as Yudkowsky alluded to. At the same time, I find it shocking when other rationalists think Guyenet is advancing a moral theory, as opposed to an empirical one. Presumably, it would not be common to interpret him as saying that much if he wrote about drug addiction or procrastination, which are both also problems that have likely gotten worse as a result of modernization. I'd prefer, for the sake of rationality, we restrict our discussion to his actual arguments and factual claims.

At the moment I consider Guyenet's theory the current best contender for why obesity has gotten worse over time. My hope is that, if you disagree, you can be more precise in your disagreement by objecting to one of the core claims of the theory that I have now provided.

New Comment
19 comments, sorted by Click to highlight new comments since: Today at 10:27 AM

A theory that appeals to me although I am not an obesity researcher is the torpor/omega-ratio theory of obesity. It states that obesity results from an activation of hibernation metabolism (torpor) by an elevated ratio of omega-6 to omega-3 PUFAs in the diet. Humans do not hibernate but our mammalian ancestors did, and the theory is that those metabolic pathways are latent in our biochemistry.

The activation of hibernation metabolism and fat mass gain by high omega-6 to omega-3 ratios is documented in the literature for hibernating mammals, and the ratios of omega-6 to omega-3 PUFAs in Western diets have skyrocketed in the past several decades due to vegetable oils. I think I first heard of this theory from this blog. It could certainly be coupled with the Hyperpalatable Food Theory.

Some additional information on this: 

  • PUFA as a proportion of body fat rose from ~10% in 1950 to ~20% in 2010. At least one hunter gatherer population has PUFA levels of 4%. We introduced high PUFA oils into our diets in the early 1900's. 
  • Human body temperature has dropped 0.6C/1.0F over the last 150-200 years. It is not believed to be measurement error. Measurement of a hunter-gatherer tribe showed decreasing body temperatures over time as they became more connected with civilization. It has been theorized the increased intake/body proportion of PUFAs may be causing this. 
  • Reasonably good video on the topic: 

There is a good case for completely avoiding seed/vegetable oils as they are a new and structurally unique food source we are unadapted to. 

If anyone knows any way to get body fat PUFA proportions tested could you please let me know? I've cut seed oils from my diet for a few months now, but I'm interested in knowing how much effect it's having. 

I agree, with a caveat. There's overeating in terms of food volume (bigger portions, eating past fullness, however you'd like to look at it), and there's "eating the same volume as you did before, except much of the food is more calorically dense."

As I commented on the Hyperpalatable Food Hypothesis post, you can actually compare recipes from then and now to see what's going on:

My Betty Crocker cookbook from 1969 (where I get most of my dessert recipes) has a brownie recipe that calls for 2 cups sugar, 4 oz chocolate, 2/3 cup butter; it's meant to bake in a 13x9 pan and yield 32 brownies.

The brownie recipe on Betty Crocker's website (that is, "today's brownie recipe") calls for 1 3/4 cups sugar, 5 oz chocolate, 2/3 cup butter, but is meant to bake in a 9x9 pan and yield 16 brownies.

At 201 grams of sugar per cup, you get 22 grams of sugar per brownie in today's recipe vs. 12.5 grams of sugar per brownie in the 1969 recipe. 

Today's brownie recipe yields brownies that are 5 square inches and the 1969 recipe yields brownies that are 3.65 square inches, but even if you cut today's brownies to the size of yesterday's brownies you'd still get 16 grams of sugar per brownie.

Yes, excellent point. Guyenet claims we’re overeating calories. He doesn’t claim we’re necessarily overeating raw mass, and in fact talks about caloric density in the book a few times.

I'm still confused about the connection with altitude, under both this theory and the Lithium/contaminant theory. [ETA: in the comments below, I thought to check if altitude is known to affect palatability, and was surprised to find that it does make people's sense of taste less sensitive.]

If you look at a map of obesity rates in the US, it's clear that the most significant factors are time, ethnicity, and altitude - in that order (from here):

On the other hand, presence of Lithium mines doesn't seem relevant, which you would expect to matter even if the accumulation in groundwater was also significant (from here):

The only theory piece that seems reasonably solid to me is that it has to do with damage/malfunction of the set-point regulation mechanism in the VMN. There doesn't seem to be enough available evidence for anything more than that.

These maps don't adjust for socioeconomic status, which has a huge correlation with obesity and health in general. West Virginia, Kentucky, and the Black Belt of the South are some of the poorest areas, while Colorado is one of the richest and best-educated. 

http://proximityone.com/graphics/mhi_stcty_17b.gif

https://www.washingtonpost.com/blogs/govbeat/files/2014/04/tumblr_n4jrdrOOC41rasnq9o1_1280.jpg

If the causality was:

obesity <- socioeconomic status -> altitude

then I would expect cities to stand out more (relative to altitude) on the obesity maps like they do on the maps you linked.

But I don't see that (there is a noticeable correlation, but it looks smaller than the altitude one to me), so I think it's more likely the causality is better approximated with a graph which has an arrow from altitude to obesity than not.

Wait, rather than some cuckoo idea that this is related to altitude or marginally less cuckoo lithium theory, what if it's about regional cuisine and eating habits?

I'm currently persuaded that it's probably some combination of hypoxia (as suggested here), demographics, and traditional cuisine being more common at higher altitudes. Maybe it's a little bit of groundwater contamination? I'd really want to see a post that actually thoroughly investigated each of these hypotheses, rather than just sort of pointing to a factor on a graph and saying "yep it's probably that".

Hypoxia seems by far the most plausible explanation of those ideas, there just don't seem to be significant enough demographic or cuisine differences by altitude in the US.

Though I just thought to check, and apparently increased air pressure does seem to negatively affect taste directly, which is some points for the hyperpalatability theory.

Under the lithium theory, the amount of lithium in water/food accumulates as water sources go downhill, thus explaining the altitude connection, so just an altitude connection is strictly less "cuckoo".

From the obesity map and my experiences living across the country, I would be surprised if regional cuisine/habits ends up mattering more than altitude, but might be worth testing more rigorously.

You quote Guyenet quoting a Brian Wansink study. Wansink studies probably shouldn't be taken as evidence for much, as he resigned in 2018 after a scientific misconduct investigation against him. I'm just sort of skimming this post and this one but apparently people found 150 mistakes across just 4 of his papers? And some of his published effect sizes are absurd, e.g. "You have a messy kitchen, a cluttered desk, you end up eating 44 percent more snacks than if the same kitchen is clear", and "[Women who have visible cereal boxes in their kitchen are] about 21 pounds heavier than the neighbor next door that doesn’t have any cereal visible at all.”

The quoted effect sizes sound waaay too big, so seeing them along with the misconduct investigation and resignation makes me feel like the Hershey's Kiss study of his (that Guyenet quotes) would not replicate.

That makes sense. I don't actually know whether Guyenet still believes in the convenience element of the theory, given what you said. He might have other evidence. But maybe a disclaimer is in order for that part of the post.

“we should be clear about what observations his theory strongly predicts, and rapid weight loss on unpalatable diets is just not one of them.”

It is worth noting that Guyenet sells a diet plan, which includes a bland food diet as a weight loss strategy, as well as exercise (which he has claimed is ineffective/unrelated to weight loss): https://www.humanos.me/programs/ideal-weight-program

It's also worth pointing out that Guyenet thinks this mechanism is not easy to reverse. When I was 15 years old, I decided to become a strict vegan, which lead me to eat food that was quite a bit less palatable compared to what I was used to. More than one year afterwards, however, I still had an intense craving for cheese; it was clear that my brain had not yet forgotten some of its old patterns of thinking about food.

Guyenet suspects that our brain's weight set point might never go down dramatically after living long enough in the modern world, even if we eventually stop eating palatable food altogether. If true, this would make his theory harder to test, and again, his theory would earn a penalty for being more unfalsifiable, but at the same time, we should be clear about what observations his theory strongly predicts, and rapid weight loss on unpalatable diets is just not one of them.

I feel like one should distinguish between "the cravings are hard to reverse on an unpalatable diet" with "the weight is hard to reverse on an unpalatable diet". At least in my experience, I ate less when I was on an unpalatable diet, so I think it works for weight loss. The continued cravings seem more mediated by an information-based thing to me, like essentially being the knowledge that much better food is available. Of course this is not easily reversible.

Guyenet suspects that our brain's weight set point might never go down dramatically after living long enough in the modern world, even if we eventually stop eating palatable food altogether. If true, this would make his theory harder to test, and again, his theory would earn a penalty for being more unfalsifiable, but at the same time, we should be clear about what observations his theory strongly predicts, and rapid weight loss on unpalatable diets is just not one of them.

I don't understand how CICO can coexist with the idea of a weight set point. If the mechanism of gaining weight is CICO via overeating because food is so palatable, then it seems natural than on unpalatable food you would eat less, and thus I would expect rapid weight loss on unpalatable diets as a prediction of the theory.

to clarify, I don't understand why positive CICO can increase your weight set point but negative CICO can't decrease it.

The story I have heard is that fat cells, when they grow too large (from more energy in than out), create new fat cells, and when they grow smaller (because more energy out than in), shrink and send signals that say "please feed me" instead of disappearing.

If true, then it seems reasonable that you could have the lipostat mostly work as a ratchet, where it's easier to grow accustomed to higher weights than lower weights.

[I think there's some evidence against this story--you would expect liposuction to be more effective if this was the primary mechanism--but I think it's more the case that lots of things are going on, and so nothing is only thing you have to think about.]

Not an explanation, but type 2 diabetes is an example of a system failing(/adapting) in one direction and then not being reversible.