Monday, May 19, 2014

The misguided ambition of nutrition science

I’ve written previously about the myth of the scientific method and experimental evidence vs. quantitative theory. In the latter article, I identified what I believe is a problem with contemporary nutrition science: transitioning from experimental evidence to quantitative analysis. If nutrition were like physics and theories could be accurately expressed in mathematical terms, then perhaps this would be appropriate. The gravitational force of a distant planet, which no human has ever set foot, can be confidently determined by Newton’s law of universal gravitation. Conversely, it’s for all intents and purposes impossible to accurately predict the weight loss or gain of an individual even when all known aspects of its food intake and energy expenditure are accounted. Instead of formulas, modern nutrition science relies on statistics that beget correlations that get elevated to the status of “natural laws” within the field.

Part of the reason for this shift is due to the misconception that physics is the pinnacle of scientific purity and every other branch should strive to be more like it.
So long as science is viewed as monolithic, founded on the scientific method, it is possible (and irresistibly tempting) to label some sciences, or bits of science, as “more scientific” than others, according to the degree of which the method has been or can be successfully deployed. Hence, in the classical epitome of science, quantification and mathematics rule the roost, because hypothesis can obviously be framed and tested with real precision only when numbers are used. Immediately, then, biology and geology come to be seen as somehow less scientific than chemistry, which in turn is less scientific than physics, the wholeheartedly scientific science.1
xkcd comic: Purity
Munroe, Randall. “Purity.” xkcd.
Physics is as a whole the most mature of the major sciences, and it seems plausible that this is a cause underlying the fact that physics also has the greatest degree of unanimity within its ranks: hardly anyone within the discipline questions relativity or quantum mechanics, or that the salient task for the discipline is to achieve the theoretical Grand Unification of the Four Forces. There is little ferment over what should be taught in physics courses, or how. The professional journals stand in well-recognized differentiation of function and hierarchy of status. The professional societies work quietly and without fuss (except, occasionally, over matters external to physics itself…)2
The goal of medicine to operate more like physics predates the 20th century. On the Wikipedia entry for Dr. Claude Bernard: “Although the application of mathematics to every aspect of science is its ultimate goal, biology is still too complex and poorly understood. Therefore, for now the goal of medical science should be to discover all the new facts possible. Qualitative analysis must always precede quantitative analysis.” Denis Noble studied Dr. Bernard’s work and echoed his thoughts on physiology one day becoming more theoretical:
The control of the milieu intérieur meant not that the individual molecules did anything different from what they would do in non-living systems, but rather that the ensemble behaves in a controlled way, the controls being those that maintain the constancy of the internal environment. How could that be formalized? Could there be a theoretical physiology? Physical scientists had long since used mathematics to formalize their theories. Could that also be done in physiology? Bernard’s answer to this question was ‘yes, but not yet.’ He cautioned, ‘The most useful path for physiology and medicine to follow now is to seek to discover new facts instead of trying to reduce to equations the facts which science already possesses.’ I believe that this view has been in part responsible for the broadly antitheoretical stance of British and American Physiology. It is important, therefore, to recognize that it represents only half of Bernard’s views on the matter. For the emphasis in that statement should be on the word now. He also wrote that it was necessary to ‘fix numerically the relations’ between the components. He continued: ‘This application of mathematics to natural phenomena is the aim of all science, because the expression of the laws of phenomena should always be mathematical.’ His caution, therefore, was purely practical and temporal. In 1865 he saw, correctly of course, that physiology simply did not have enough data to make much mathematical application worthwhile at that time. But he clearly foresaw that the day would come when there would be sufficient data and that mathematical analysis would then become necessary.3
The rise of observational epidemiology has fulfilled Bernard’s wish for introducing mathematics to medicine. Unfortunately, it only provides the illusion of theory. Observational studies can only show correlation, not causation. Even then, it’s difficult to untangle the confounders.
The science of epidemiology was invented to study infectious diseases, which come on suddenly and can usually be traced back to a source, such as the water supply. Chronic diseases, by contrast, evolve over a much longer period of time, and it’s just about impossible to measure the many thousands of factors over the course of a person’s life that might contribute to a condition decades later.4
Statistical evidence can be a powerful tool to identify areas for further research. Perhaps when the effects of the subject being examined are so overwhelmingly strong, observational data may well be damning enough. The correlation between cancer and smoking is a rare example where the death rate from cancer of the lung in cigarette smokers is nine to ten times the rate in non-smokers and the rate in heavy cigarette smokers is twenty to thirty times as great. To prescribe a cure or behavioral change based on statistical evidence with weak risk ratios, such as processed meats, is a misuse of this kind of research tool. David Fineman discusses attributes of useful (and useless) observational studies in his article: “Observations Generate hypotheses. Observational Studies Test Hypotheses. Association Implies Causality (Sometimes).

Most people outside of the field of science do not understand that associations can only be used to fuel speculation and establish hypotheses, but nothing more. The context of terms like significant, linked, associated, relative risk, absolute risk, etc. are often misinterpreted or unknown to the layperson. That’s expected to some degree. We’re supposed to trust in science and scientists implicitly. The scientific method is supposed to be a path to truth. It’s supposed to filter out erroneous conclusions and biases, right?
…the common view of science as a unitary, monolithic enterprise fails to recognize how varied are the people who do it. Scientists are supposedly trained to judiciousness, objectivity, patience, and careful experimentation and observation; scant attention has been paid to how the practice of science is influenced by the fact that scientists, like all other human beings, vary in ability, competence, dedication, and honesty.5
Let’s consider the paleo and “whole foods” movements. An observation of ancestral diets yields a hypothesis: the obesity epidemic and diseases of modernity are attributed to a deviation from the diet of our ancestors. The initial attempt was to match macronutrient ratios, which lead to a generally low carbohydrate interpretation. But, perhaps that’s not enough? If the macronutrient ratios match, but the diet is high in so-called processed foods, then can it still be referred to as paleo? By definition, it could not. Does that matter? Why is it a certainty that any deviation from an ancestral diet is inferior? If artificial sweeteners or processed foods are inherently bad for us, then there needs to be proof, not supposition simply based on those items falling outside the bounds of the hypothesis. Is the goal of the people who subscribe to the paleo diet to discover an optimal human diet or is it just an act of faith based on an appeal to nature and/or antiquity?

My personal stance is that a low carbohydrate diet is optimal for a great many people. I say that not because the existence of insulin demands it. I say it because the evidence demonstrates it. I also understand that certain individuals, say elite athletes that require quick bursts of energy (sprinters rather than marathon runners), may be better off with a much higher carbohydrate macronutrient ratio. Again, this is not primarily because it fits with what’s written in a biochemistry textbook; it’s because that’s what works for those individuals. We were told for years that lactic acid was the culprit of second-day sore muscles. Even today, some coaches tell their athletes to concentrate on cleansing their muscles of lactic acid after a hard workout. Then, very recently, this mechanism of human biology was shown to be false. Does this mean that trainers have to change their workout and recovery regimens? No, because the regimens were based on what worked, not what fit the model. Unlike the coyote in a roadrunner cartoon running off a cliff, the laws of nature aren’t suspended until we become aware of their existence.

My primary goal for understanding nutrition is to find out what works best. The “why” and “how” are important but secondary concerns. Understanding the mechanics may allow us to predict and implement further refinements. However, a hypothesis should never overrule a working implementation. As the physicist Richard Feynman said:
In general, we look for a new law by the following process: First we guess it. ... Then we compute the consequences of the guess to see what, if this is right, if this law that we guessed is right, to see what it would imply. And then we compare the computation results to nature, or we say compare to experiment or experience, compare it directly with observations to see if it works. If it disagrees with experiment, it’s wrong. In that simple statement is the key to science. It doesn’t make any difference how beautiful your guess is, it doesn’t make any difference how smart you are, who made the guess, or what his name is. If it disagrees with experiment, it’s wrong. That’s all there is to it.6


  1. Bauer, Henry H. “The So-called Scientific Method.” Scientific Literacy and the Myth of the Scientific Method. Urbana: University of Illinois Press, 1992. 36. Print.
  2. p. 28.
  3. Noble, Denis. “Claude Bernard, the first systems biologist, and the future of physiology.Experimental Physiology 93.1 (2007): 16-26. Experimental Physiology. Web. 7 May 2014.
  4. Teicholz, Nina. “Why We Think Saturated Fat Is Unhealthy.” The Big Fat Surprise: Why Meat, Butter, and Cheese Belong in a Healthy Diet. New York: Simon & Schuster, 2014. 47. Print.
  5. Bauer, Henry H., 32.
  6. The Character of Physical Law. Dir. Richard P. Feynman. Perf. Richard Feynman. 1964. Education Development Center, 1990. Film.

No comments:

Post a Comment