Thursday, August 21, 2014

What diet should I follow?

A tweet of mine regarding the paleo diet got retweeted quite a bit:

Tweet #1: The paleo diet gets the highest compliment I can give: it should be the null hypothesis.

What got much less notoriety was the tweet that I submitted immediately after that one:

Tweet #2: Buth that doesn't make it gospel.

That second tweet puts the first into context but went mostly unseen. I suppose a blog post is in order to better clarify what I meant by both.

Model vs. reality

The philosopher Plato came up with the “allegory of the cave” to compare “...the effect of education and the lack of it on our nature.”
...imagine a cave inhabited by prisoners who have been imprisoned since childhood. These prisoners are chained in such a way that their legs and necks are fixed, forcing them to gaze at the wall in front of them (514a–b). Behind the prisoners is a fire, and between the fire and the prisoners is a raised walkway. Along this walkway is a low wall, behind which people walk carrying objects “...including figures of men and animals made of wood, stone and other materials.” (514c–515a). In this way, the walking people are compared to puppeteers and the low wall to the screen over which puppeteers display their puppets. The people walking are behind the wall on the walkway, so their bodies do not cast shadows on the wall, but the objects they carry do. The prisoners cannot see any of this behind them, and are only able to view the shadows cast upon the wall in front of them. The sounds of the people walking echo off the shadowed wall, the prisoners falsely perceive these sounds to be that of the shadows.

Socrates suggests that, for the prisoners, the shadows of artifacts would constitute reality, because they have not seen the light. They would not realize that what they see are shadows of the artifacts, which are inspired by real humans and animals outside of the cave.1
This allegory describes the disconnect between human observation and reality. When we try to unravel the truth behind complex systems like the human body, our interpretations are limited by what we can sense and therefore deduce. We come up with abstractions and simplifications to help us cope with an incomplete understanding.
Scientific knowledge is like the knowledge embodied in a map: “Follow this route, and you will pass a valley and then a river.” That maps work convinces us that the landscape exists outside of mind or imagination; that science works convinces scientists (even if not all philosophers, let alone social scientists) that there exists a real world that is not the creature of human imagination. But scientific knowledge is not more than a guide to reality; it is not the real thing itself.2
A map can be a reliable representation of reality. Even maps not drawn to scale or lack other details can be incredibly useful.
Maps, like sciences, are not collections of facts; nor are they entirely theoretical. They are some sort of amalgam of fact and theory that is staggeringly reliable without being guaranteed 100 percent forever reliable. The amalgam is always being modified. Places get renamed, different relations are recognized, above all finer and finer details are explored; and one can see no end in sight to further exploration.3
Maps are models, which are abstractions of reality. Models are useful to get things done in a consistent and practical manner. They help to characterize relationships between factors and can hopefully be used to make predictions. The calorie-in/calorie-out (CICO) model is based on the idea that the number of calories ingested vs. expended can be used to predict weight gain and loss. It’s a model that works. Yes, that’s right; I said it. It does a degree. However, models have limits and they can break very easily when pressed.
When one extrapolates from the known into unexplored territory, however, great caution is in order. One cannot predict how far a mountain range stretches; one must explore it. The “revolutionary” episodes in science are of that sort: they do not really disturb the existing knowledge of “Do this, and that happens,” but they reveal it to be so only within a certain range of conditions or terrain outside which the flora and fauna are strikingly different. Newtonian mechanics is just as valid now as it was 300 years ago, but we now know we cannot use it whenever speeds approach that of light (then we must use relativity theory) or whenever “things” get small enough (then we must use quantum theory.)4
The “eat less, exercise more” solution derived from the CICO model works somewhat under certain conditions, but is not the absolute universal solution so many people seem to believe it is. CICO is not the only “map” at our disposal. Other models have better (and worse) levels of precision and accuracy for predicting an outcome.

But enough about CICO... What about the paleo diet? It’s a solution based on an evolutionary model. That is, our diet is shaped by our physiology, which is the product of millions of years of evolution. I happen to agree with this premise (although maybe not the same interpretation of the details as others may have). Regardless, the point of this post is not to argue the validity of evolution nor the paleo diet itself. The paleo diet is a solution based on a model, not reality. This is not a disparagement of the paleo diet. Every common diet plan is also based on a model. A low carb/high fat (LCHF) diet is based on a model of fat storage driven by insulin. The Weight Watchers diet is a proprietary points system based on calories and other factors and includes psychological support through group meetings. The Mediterranean Diet focuses on the nutritional habits of the people of Crete, Greece, and southern Italy. These diets are solutions based on models that are based on observations, experimentation, and hypotheses. Sure, some diets are based less in reality and more on altruism and wishful thinking. (Vegans, I’m looking in your direction and whistling.)

If diets are based on imperfect representations of reality, what’s a person to do? One way to approach this is to pick the best model and declare that the winner. That’s what the US government did when it came up with the original food pyramid. Obviously, we should replace that outdated and faulty model with a new one and impose that on the country. Many people believe that to be the correct course of action. (Dr. Lustig, I’m looking in your direction and whistling.) However, that’s not the course of action I would recommend.

All models are inherently imperfect and no model can completely and accurately predict every situation (assuming a sufficiently complex system). Therefore, I propose a reality based diet and I can describe it simply as:

The diet that works best for you.

Hmmm, that doesn’t seem very controversial or profound. I should break this down into the three important ideas that the statement contains.

The diet that works...

The starting point to determine the right diet is to find out what works and this is where finding a good model can be extremely helpful.

Pick just about any diet that deviates from the junk-food laden fare of the typical westerner and I can show you examples of people losing weight and improving their health on it. LCHF, low fat/high carb, lacto-ovovegetarians, frutarian, Medditieranian, raw food, you name it. Most of them “work” to some degree, so which one should you pick?

There are lots of models out there that claim to be the one true representation of reality. Is obesity a result of hormones or calories? Is overeating driven by food palatabity, genetics, toxins, macronutrient ratios, or none of the above? What about all of the above? Maybe all of these things are contributing to the obesity crisis? If so, are there any factors which are predominantly responsible while others are second and third order variables in the equation? What about synergy between the factors? (E.g., the combination of high glycemic carbs and fat.) These are the types of considerations that ought to go into determining a dietary model.

Let’s say an excellent dietary model has been found based on years of research that’s generally accepted as sound. Are we done? Not so fast. Will it work in practice? On humans? Outside of a metabolic chamber? For more than six months? For a lifetime? What are the downsides? Does it improve all health markers? Does it make some worse?

We generally claim a model to be the best not just because of the scientific knowledge that supports the model, but because of its ability to predict an outcome. If the goal is to reduce obesity and improve health markers, you’d logically want the diet that provides the best “bang for the buck,” so to speak. Of course, other considerations may also apply (like animal cruelty, cost, personal enjoyment, etc.) that complicate matters. Nevertheless, it’s ultimately less important why a model works and more important that it does.

This is at the heart of the arguments regarding which diet is best. If a chosen diet plan improved health the most when compared to other diets, but was near impossible for most to adhere, could anyone claim that it “works?” If one believes in evidence that implies calories are all that matter, but a LCHF diet provides better results in practice over the long term than a CICO diet, which one should they recommend?

...for you.

The pre-Socratic Greek philosopher Protagoras is credited with the phrase “man is the measure of all things.” To claim there is an absolutely best model based on objective evidence might be possible. However, human elements and impacts are far less agreeable. What’s more, natural variation can stymy application of a model too rigid to adapt to a more fluid reality.

Let’s suppose that I come up with a 12,000 calorie per day diet plan, roughly broken up into three meals of 4,000 calories each. How would you picture the person who could consume so much food? Maybe someone from the fattest person on the planet category in the Guinness Book of Records? Indeed, that may be so. Donna Simpson consumes that many calories in a day.

Picture of Donna Simpson.
"Stuffed! The 30,000-calorie Christmas feast eaten by the world's fattest mum in ONE two-hour sitting." Mail Online. Associated Newspapers, 26 Dec. 2010. Web. 19 Aug. 2014.

However, there’s another person who consumes as much as Donna, but doesn’t seem to have the same level of obesity: Michael Phelps.

Picture of Michael Phelps.

Picture with description of Michael Phelps' typical diet.

Quite a contrast. This is an extreme example of how different the same kind of diet can have on individuals with different genetics and lifestyle choices. This will also occur on much smaller scales whenever a dietary intervention is sought. Therefore, tailoring to individual needs, within reason, is a hallmark of a successful diet, in my opinion.

The Atkins diet is a good example of a diet based on a model that takes into account individual tolerances. It goes something like this: two weeks on induction (20 grams effective carbohydrates or less) and then slowly increase carbohydrates while weight loss is consistent or weight is maintained depending on the goal. The model for the diet is based on insulin levels and allows a tailored response to the person’s presumed carbohydrate intolerance. It could very well be that the entire premise of the diet is flawed or that the diet’s tailoring is insufficient. The diet may be scientifically sound and work for a majority of people, but an individual may find it too restrictive and thus untenable for the long term. Nevertheless, an attempt is made with the Atkins diet to allow some flexibility while adhering to the model’s framework.

The tail wagging the dog.

I’ve written in the past about how science is refined iteratively through testing against reality.
Science remains the study of nature; rational opinion finds science the more satisfactory the more it properly reflects what nature does. The consensus of the scientific community—the consensus of rational opinion formed as widely as possible, as John Ziman puts it—is exceedingly sensitive to the test of nature. In Richard Burian’s happily chosen phrase, scientists make liberal use of “reality therapy.” Perpetual-motion machines are believed to be impossible because, no matter how ingeniously designed, they have never worked—period; pigs do not fly—period; there is no element of atomic number between that of hydrogen and that of helium—period; and so on and so forth. Our explanations for those truths may have little warrant, and the explanations we use do change from time to time. But that human interpretation of nature is always subject to change does not entail that human knowledge of nature’s phenomena is always fragile: maps can be crude or flawed and yet perfectly reliable in important respects.5
The ultimate arbiter of diet and human health is reality, not a model of reality. If one views the evolutionary model of the paleo diet as a dogma rather than a template, then eventually there will be instances where the diet will fail due to its rigidity. There may be very good reasons to avoid dairy, especially if an individual has negative reactions to those foodstuffs. However, to dismiss dairy products out of hand because they don’t fit with the model is foolish. Ultimately, the pros and cons of foods should be determined on their actual effects, and not on assumptions or preconceived categorizations.

Artificial sweeteners are perhaps the ultimate antithesis to the paleo lifestyle. They’re manmade products designed to affect an unnatural biological response. They are composed of chemicals designed with the sole purpose of appeasing the innate craving for sweet but without the undesired, albeit natural, effects a sweetener should have. What sticks in the craw of some paleoistas is the lack of outrage by the scientific community. Where are the warning labels for these “frankenfoods” with the requisite dire threats? Okay, maybe they won’t cause cancer, but what about other unintended consequences? Perhaps this tricking of the brain’s reward centers leads to overeating, thus negating their effectiveness? That’s possible and certainly a worthy area of scientific exploration. However, there are many who will rail strongly against artificial sweeteners simply because they’re incompatible with the evolutionary model from which the paleo diet is supposedly based.

Optimal vs. natural

There’s something else to consider. When I crafted the reality based diet statement earlier, I tried to define the term “best” but I didn’t question a basic assumption: Is the natural state an absolutely optimal one? Evolutionary pressures take a very long time to gradually optimize a system. However, these optimizations do not necessarily represent the best possible design. (A simple mathematical illustration would be local vs. global minima and maxima.) Take the human eye for example. The slow and gradual progression from simple light sensing cells to a fully formed human eye is almost miraculous. Heredity and random mutation yielded an organ so complex that some consider it irreducibly so. However, from a design perspective, it’s extremely flawed.
Any engineer would naturally assume that the photocells would point towards the light, with their wires leading backwards towards the brain. He would laugh at any suggestion that the photocells might point away, from the light, with their wires departing on the side nearest the light. Yet this is exactly what happens in all vertebrate retinas. Each photocell is, in effect, wired in backwards, with its wire sticking out on the side nearest the light. The wire has to travel over the surface of the retina to a point where it dives through a hole in the retina (the so-called ‘blind spot’) to join the optic nerve. This means that the light, instead of being granted an unrestricted passage to the photocells, has to pass through a forest of connecting wires, presumably suffering at least some attenuation and distortion (actually, probably not much but, still, it is the principle of the thing that would offend any tidy-minded engineer). I don't know the exact explanation for this strange state of affairs. The relevant period of evolution is so long ago.6
It’s ironic that some people who claim to embrace evolution understand it so poorly to think that its “designs” are to be considered sacrosanct. It wouldn’t matter to them how much positive experimental evidence a processed milk product like powdered whey protein would have on health. If it does not fit with their worldview, it’s considered “treif.”

The highest compliment

I believe an evolutionary model provides the best map to human health. I further contend that a dietary model based on evolution should be considered the null hypothesis. Anything that falls outside of what’s known (and that’s subject to debate, of course) should be considered suspect. That is the highest compliment that I can give any diet, but that’s it. That does not absolve the diet (or model) from further inquiry. That does not give us warrant to forever dismiss any foods that are not in line with our ancestral diet. It does mean that those foods should be viewed suspiciously. It may be reasonable to avoid such items in absence of scientific consensus to their safety. However, if we continue to consider things like artificial sweeteners dangerous even after decades of experimentation, mass exposure and chronic use without evidence of human harm, then please, call it a religion and be done with it.


  1. “Allegory of the Cave.” Wikipedia. Wikimedia Foundation, 19 Aug. 2014. Web. 19 Aug. 2014. <>.
  2. Bauer, Henry H. “Other Fables about Science.” Scientific Literacy and the Myth of the Scientific Method. Urbana: University of Illinois Press, 1992. 68. Print.
  3. p. 68.
  4. p. 70.
  5. Bauer, Henry H. “Imperfections of the filter.” 89.
  6. Dawkins, Richard. “Making tracks through animal space.” The Blind Watchmaker: Why the Evidence of Evolution Reveals a Universe without Design. New York: Norton, 1996. 93. Print.

No comments:

Post a Comment