The calorie debate

The day after I appeared on the radio show The Takeaway, Scientific American editor Michael Moyer came on to criticize me.  I welcome the debate and my response is below.

Here is a link to The Takeaway’s series on obesity, which has his audio file.  Moyer also writes in his blog:

Unfortunately Chow’s outsider’s perspective on the obesity crisis isn’t really an outsider’s perspective at all: it is the physicist’s perspective. Physicists have a long history of marching into other sciences with grand plans of stripping complex phenomena down to the essentials with the hope of uncovering simple fundamental laws. Occasionally this works. More often, they tend to overlook the very biochemistry at the heart of the process in question.

Chow’s conclusion is not just obvious—it’s a tautology. Because for Chow, a calorie is just a unit of energy. Eat more calories than you burn, and the energy must go somewhere. That somewhere is fat cells. The conclusion is built into the assumptions.

But perhaps a calorie is not just a calorie. Perhaps, as some prominent researchers argue, the body processes calories from sugar in a fundamentally unique and harmful way. According to this hypothesis, we’re not getting fat because we’re eating more. We’re getting fat because of what we’re eating more of. The biochemistry that explains why this would happen is complex—certainly difficult to include in a computer model—but that doesn’t make it wrong.

Ultimately experiments will decide if this hypothesis is true, or if it is not true, or if it is true but just one part of a nuanced understanding of obesity that includes biochemistry, microbiology, neurobiology, politics, economics and much more. The obesity crisis isn’t rocket science. It’s complicated.

Moyer’s criticism of me is ironic in two ways. The first is with regards to his claim that biology is not like physics. I fully agree and have posted on this very topic here. Additionally, while I have been frustrated in the past trying to get biologists to pay attention to my work, this is the one area where I am not a complete outsider and have access to and input from some of the very best clinicians, experimentalists, and public health scientists in the field.

His main physiological criticism is that there is more to obesity than just “calories in, calories out”. I also agree. However, I disagree that this necessarily means that it can’t be modeled.  As someone who has worked on biological problems for close to twenty years, I am well aware of how difficult it is to model biological processes and I have no grand schemes to come in and change the field. That said, body weight change is an area where I think some progress can be made although with a caveat that I will address at the end. The reason it is amenable to mathematical modeling is that there is a huge separation of time scales between the complicated biochemistry and the change in your body.  This separation of time scales is vast.  Biochemistry occurs on the time scale of minutes to days while the time constant for weight change is a year or more. This time scale separation allows us to take the average of the fast complicated effects, which can be measured experimentally and incorporated into the model. The body weight model Kevin and I have developed may be the most experimentally validated thing I’ve ever done.

At the reduced level, there are two important processes that must be accounted for. The first is that energy must be balanced.  The second is that macronutrient flux must be balanced. This second part is where the “Is a calorie a calorie debate” comes in. Energy balance is indisputable. Even Moyer would agree. However, he and others like Gary Taubes feel that just focusing on energy balance misses the point. We actually do try to account for the second part in our model using the best available data and can change the model when and if the data demands it.

In order for you to live and function you must use energy. The energy comes in the form of food. Any excess energy you take in is either excreted or stored. We are generally very efficient and don’t waste much energy. However, as Moyer and Taubes point out, not all food is created equal. In particular there are carbohydrates, protein and fat. The American diet mostly consists of carbohydrates and Taubes thinks this is bad. Your brain burns carbohydrates (in the form of glucose) but does not burn fat, while your body mostly stores fat but has very little stores of glucose in the form of glycogen. Your body tries to keep glucose levels fairly constant in the blood. When glucose levels are  low, the rest of your body doesn’t burn glucose to save it for the brain and switches to burning fat.

Your fat cells also release free fatty acids into the bloodstream and the amount released is in proportion to how much fat you have. When you eat carbs, your body handles this sudden surge in glucose by secreting insulin from your pancreas. The insulin does two things. It allows the non-brain cells to switch from burning fat to glucose and also suppresses the release of fat from fat cells.  Any glucose you don’t burn gets converted into fat in the liver, which is then released into the blood packaged in VLDL and LDL particles. The surge in insulin brings down the glucose level to baseline levels in a few hours and then your body switches back to burning fat. It has been claimed that this drop in insulin makes you hungry although the experimental evidence for this is inconclusive. Over the course of the day, you mostly burn glucose after you eat and fat between meals. You also burn protein but for simplicity, I won’t deal with that in this post.

The Taubes argument then goes like this.  Since carbs increases insulin and insulin suppresses fat release and also converts carbs to fat then eating carbs will cause you to get fat. The argument is not self-evident because when insulin is low, fat is released and you burn fat.  The balance at the end of the day is what matters so you need to measure the net effect over long time scales.  This has not been done carefully enough for us to really know what will happen. Kevin Hall is doing an experiment right now but even these results may not be enough because we need really long-term data. Gary Taubes and Kevin have actually talked about trying to get funding to do this experiment on a long time scale. It might be true that carbs makes you fat but we don’t know yet. On the whole body level, the question reduces to: As you change the macronutrient content of what you eat (i.e. fat to carb ratio) does your body adapt and burn energy in the same proportion? We could thus be in energy balance but not in macronutrient balance. If the  metabolic flexibility of our body is perfect, then we will adjust to any diet and then all calories are equivalent. However, if we cannot match the energy we burn with our fuel intake, then our body composition will change until we do.

Our model tries to incorporate both energy balance and macronutrient balance.  We estimate the macronutrient balance part by looking at cross-sectional body composition data.  It turns out that the ratio of fat to lean in people lies on a curve.  Using this curve, we can then predict how fuel use is partitioned between lean and fat stores.  This makes the assumption that the cross-sectional curve of a population matches the longitudinal curve of an individual and this may not be true.  However, we have validated the model on multiple data sets and it does predict body weight and body composition changes well over short to medium time scales.

Finally, with regard to food supply causing the obesity epidemic.  We took our model and calibrated it to the average American and then fed it the average food available. What we found is that the excess food more than compensated for the weight gain. In fact, we predict that we are progressively wasting more food over time and this is partially validated with EPA data on municipal waste. If we applied the “Taubes effect” then people would either be even fatter or waste more food. In either case, it doesn’t discount the increase in food supply as the cause.

Finally, the caveat. While I think we can model the metabolic or output end, I have not addressed the input end. We really have no idea what controls how much we eat. This involves many complicated neural and environmental processes, which may or may not be amenable to modeling. It could be true that high carb diets make you more hungry and thus eat more. We simply don’t have enough data to know.

For those interested in reading more about this, click here.

I corrected some grammatical errors after posting.


9 thoughts on “The calorie debate

  1. As a retired biochemist I was intrigued by your article in the NYtimes, but could not figure out how to get to the executable simulation via the link provided. What did I miss? Please help!


  2. “The body weight model Kevin and I have developed may be the most experimentally validated thing I’ve ever done.”

    So, you’ve actually done controlled studies in which a significant sample of people were randomized to differing changes in caloric intake, they filled out your online simulator form, followed the prescribed diet, and then you’ve verified over the passage of many months that your simulator predictions provided accurate predictions of weight change? I would love to read that research.

    I wouldn’t bash you guys, I think what you’re trying is great and I’d be doing the same thing if I were in your shoes. I’m also pretty sure it’s completely hopeless to expect you can model anyone’s weight response with much accuracy, Espeically when you can’t even predict what caloric intake would make a given person weight-stable (you’re off by 1000 cals per day in my case, FYI.)

    According to your sim, if I continue eating as I’ve been eating for some months now, my 200+ lb fatbody will be a svelte concentration-camp 81 pounds in only 2 years! I think you’ve got some tweaking to do.


  3. Your experiment would not work because we cannot verify that they actually eat what they think they eat. We matched the model to various experiments where the subjects were sequestered in a clinic and the food intake was completely controlled. There will always be people that fall outside the parameter range in which the model works.


  4. I appreciate what you’re trying to do, but I didn’t find the calculator to be very accurate for me, based on my actual experience.

    I’m not a scientist, but I have successfully lost about 70 pounds from my highest weight (57 pounds in the past 18 months, 26 pounds in the past 12 months).

    I’ve kept fairly careful track of my caloric intake for quite a while. If I plug my numbers from the past, it either indicates that I should weigh about 40 pounds less than I do now, or it indicates that I could have achieved my present weight loss by eating about 800 calories per day more than I actually did.

    My general sense is that my body is hanging on to darn near every calorie I ingest, while the somebody whose weight has always been normal will discard food and use it inefficiently. The model does adjust for the change in weight loss, but the adjustment only amounts to about 100 calories per day or so. The actual adjustment should probably be much higher – 20% to 30% or so of daily caloric intake.

    Another disconnect between the actual world of weight loss and the scientists who study weight loss is the fixation on macronutrient ratios, rather than the actual amount of macronutrients.

    Let’s say I determine that my body needs 100 grams of protein per day to replace the protein that my body uses. That’s 400 calories per day from protein.

    If I’m eating 1200 calories per day, then 33% of my calories are from protein, and 100 g per day is a “high protein diet”.

    If I’m eating 2000 calories per day, then 20% of my calories are from protein and 100 g per day is a “moderate protein diet”.

    If I’m eating 3000 calories per day, then 13% of my calories are from protein and 100 g per day is a “low protein diet”.

    In all three instances, I’m eating the exact same amount of protein. Yet, supposedly, my protein intake might be described as dangerously high on a low calorie diet, and dangerously low on a high calorie diet.

    It’s the same way with all macronutrients. In the days when I was unsuccessfully dieting, I would actually add unnecessary carbohydrates to a meal just to reduce the “percentage of calories from fat”. I had eaten enough food. I was already full. But adding a scoop of rice or a couple slices of white bread or a glass of fruit juice would magically transform my meal from a “high fat” meal to a “moderate fat” meal.

    Now I’m more successful with a bottom-up approach. I make sure I’m actually hungry before eating, and I make sure that each meal contains some protein, some fat and some fiber. Once I reach an adequate amount, I don’t purposely add protein, fat and fiber to the meal simply to add more food. I make sure there’s enough food (based on years of experience) to shut off my hunger until the next meal. Then I count how many calories were in the meal.

    As it turns out, some meals get 60% of their calories from fat. Some meals get 60% of their calories from carbs. Some meals get 60% of their calories from protein. But they’re filling enough to last to the next meal without binging or snacking, even though my actual food intake is fairly sparse (1300 to 1600 calories per day).


  5. Hmmm – not sure why physics and applied mathmatics are not appropriate science or tools to study human physiology – even weight management. I fail to see how making the statement that – an abundance of cheap food has led us to eat too much which has a causal effect on the obesity epidemic – discredits in any way that “what” we are eating also matters. As a family doctor – I welcome all perspectives in addressing the most significant public health threat to Americans – obesity and I find scientific turf wars very boring and counterproductive. The clinicians front line world is one in which we need to take very complex principles and apply them in very simple ways that the average obese or overweight American can not only understand but apply to their every day life. To make this science really matter we have to get better at simplifying the actionable instruction at the bed side. Most patients don’t count calories and could give a hoot about figuring out composition – they need some very simple, easy to understand instructions that they can apply in a real world environment. We see thousands of employees that are fantastically confused about what to do with their nutrition. The most common mistake is missing the forest for the trees – if your BMI is over 30, the exact precise specific balance between macronutrients means nothing if you are overeating by a thousand or more calories a day. That’s not to say that composition is not important – it is – but I commend Dr. Chow’s work on using math and physics to take something that is very complicated and boil it down to something that is easier to digest – not perfect – but applicable. Next step – team up and lets get a national message that is easy for the “every man” to understand and apply in their daily life that includes the “we eat too much” (caloric control) message and the “we eat too much sugar” (sound composition) message. My favorite book cover is Michael Pollan’s book “In Defense of Food” – the simple message is: Eat Food, Not Too Much, Mostly Plants. That is brilliant when you consider application of science to the average man – it is a meaningful message, with a scientific basis that is easy to understand.
    Carry on.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s