Hypothetical Apostasy on Nutrition

As many of you know, I am a major proponent of paleo/primal/ancestral/etc type diets. At this point the term “paleo” has come to be applied to many very different diets, but for the record my own personal beliefs coincide most strongly with the Perfect Health Diet. Whatever you want to call it, it is certainly outside of conventional wisdom and mainstream scientific/medical opinion. This has been a point of contention between me and others who put more stock in mainstream opinion. I have spent many hundreds, or maybe thousands, of hours doing research into human metabolism, and as a result my ideas are starting to get sticky.

Periodically I like to subject my beliefs to one of my all-time favorite techniques, the hypothetical apostasy by philosopher Nick Bostrom. The basic idea is to produce a good faith effort at destroying your currently held position. This process has helped me improve my thinking on a number of topics, including the original mind-killer itself: politics. Given my particularly strong beliefs about diet, it is long past due for me to try this exercise. Below the fold is my best attempt to undermine the paleo position:

First of all, let me begin by saying that the differences between a paleo diet and a conventional healthy diet are relatively small in comparison to the Standard American Diet. It would be foolhardy to make a case that eating candy and fast food and soda would constitute a healthy diet or would be a remotely good idea. There is substantial overlap in terms of things like eating freshly prepared food, high vegetable intake, low sugar intake, etc. (There is also major overlap in terms of getting exercise – with disagreement about cardio versus HIIT – avoiding smoking and excessive alcohol, avoiding stress, having good social support, etc. These factors may end up mattering more than diet!) The biggest differences boil down to paleo diets including more meat and fat, and avoiding whole grains and legumes and vegetable 0il (and to a lesser degree fruit and nuts). That said, the proponents of each believe that their own diets are significantly superior to each other, so this dispute still deserves attention even if both of them handily beat the SAD. In the coming paragraphs I will address both the methodological points, as well as specific differences between the diets themselves.

The paleo diet relies almost entirely on anthropological and biochemical evidence, along with evolutionary reasoning, to make its case. In particular, paleo proponents tend to ignore or dismiss the large-scale observational studies that we have, including the very populations we ourselves are drawn from. Correlation is not causation, true, but correlation is correlated with causation. This does indeed count as evidence. In fact, it is more parsimonious to assume that the correlation is in fact causal, otherwise you have to appeal to random chance or a complicated third factor (or something even more elaborate). If the paleo view were correct, we should expect to see more support – really any support – showing up in prospective cohort studies. The healthy user bias is often cited here, that people doing the conventionally healthy things are just more health conscious in general and take care of themselves better. Consider this from a different angle: for whatever reason, these kinds of people are objectively, empirically healthier, and thus we would do well to copy their behavior in all regards if we want the same outcomes. By deciding to go down a different road, we are ignoring this strong outside view evidence in favor of more flimsy justifications. Call this the healthy user heuristic.

There is also the question of what biomarkers to look at, and what those mean exactly. Once again, we can observe correlations between biomarkers and health outcomes. We also have a ton of short-term RCTs that seem to move these biomarkers in one direction or another. Doing any true, long-term, double-blind, RCTs with eating different foods is simply impractical. It will never happen. This represents not just the best evidence we have, but also the best evidence we are ultimately going to get for a long time. (We can test causal theories with this degree of rigor in animal models or cell cultures, which gets back to the biochemistry argument, addressed below.) Perhaps more data will start rolling in once we have large populations eating differently for a long time, with an observed breakdown in the correlations, or if we create drugs that target these biomarkers with no effect on morbidity or mortality, but these processes take time. We have correlational evidence and we need to use it.

Much is made of science itself being a diseased discipline – there is ample evidence that much of published science is false for instance. First of all, even if true, this does not mean we should throw the baby out with the bathwater. What is the alternative to science? Is that process any more likely to converge on truth? (And let’s be serious about the value of self-selected anecdotes from personal experimentation – every side has plenty of those.) Proponents of paleo often argue that government recommendations influence government funding, a major source of supposedly unbiased studies. But what is the alternative unbiased source of funding? An argument along the lines of group think is usually made, that if you find something in opposition to the mainstream opinion you will lose your funding and status forever. But is this based in real experience, or a hypothetical? Overturning a major finding can itself be a source of status. The paleo proponents need to make a better case that the scientific establishment will not converge on true beliefs over time.

The anthropology literature is as rife with confounding factors and bad studies as any other discipline. Loren Cordain used his own study of anthropology to determine his version of the paleo diet (indeed the Paleo Diet), a version which is now widely rejected as being too influenced by mainstream scientific opinion on saturated fat. There are also concerns that some of the original data collected (back when real hunter-gatherer tribes still existed) is faulty: the !Kung may not really eat only nuts, the Inuit do in fact obtain plant material to supplement their diet, the Tuoli data from the China Study came from an unusual feasting day, etc. This particular form of science should not be given undue weight. Even then, it seems like relatively high intakes of honey and/or fruit do occur in seemingly healthy populations, so this does not clearly show any danger from sugar intake.

Biochemistry is a tricky beast. Let’s ignore for a moment that animals and cell cultures are not human bodies, which should cast all of these findings under doubt. The biggest problem here is not that biochemistry is inadmissible as evidence, but that there are enough studies with enough different variables that either side can make a plausible case by selecting the right studies. Both sides are highly biased in this debate, and with enough motivated cognition anything can be proven from the available data. The paleosphere consists of a number of influential and motivated bloggers, who are scouring PubMed for studies to support their viewpoint, which then get endlessly discussed in the echo chamber that is the internet. These commentators are usually not respectable scientists doing research in their own right. In the face of these issues, it is better to throw out biochemistry almost entirely, and focus on real, living, healthy human beings.

In terms of particular food groups, high meat intake is among one of the most common correlations with poor health outcomes. High saturated fat intake is associated with negative changes in the relevant biomarkers, and those biomarkers improve with more consumption of polyunsaturated fats. Whole grains and legumes and fruits and nuts are correlated with good health outcomes. That should be enough to convince you!

The Result

This did not result in as much of a shift as I expected beforehand, which is somewhat disappointing. At this point I have heard rebuttals to most of the above arguments, so it seems hard to consider them seriously without raising mental objections. I am definitely wary of motivated cognition in the paleosphere, and I think the intellectual honesty of its proponents varies widely. I already discount the echo chamber effect pretty highly, those voices are not independent evidence. I am slightly more skeptical of the biochemical claims of paleo proponents, and would like to do a more in-depth investigation in the literature, particularly looking for contradictory findings. I do think that the outside view argument is the strongest one, but at what point in our travels do we finally tip the scales in favor of the inside view? That question is huge, and warrants a post of its own. For me, the convergence of anthropology, biochemistry, and personal experimentation is a strong one. It would require shaking one of those pillars to make me significantly revise my views at this point.

  • Romeo Stevens

    I reject the premise that PHD supports the generally accepted paleo diet. PHD are, AFAIK, the only “paleo-ish” people saying to eat a pound of fruit, a pound of potatoes, and that dairy in moderation is fine.

    • WilliamEden

      Where do you get a pound of fruit and a pound of potatoes? They think you should keep carbs at 20% of calories…

      • Romeo Stevens

        “Perfect health Diet in Brief:

        About one pound per day—roughly , four fist-sized servings—of “ safe starches ”: white rice, potatoes , sweet potatoes , taro, winter squashes , and a few others . Add up to another pound of sugary plants—fruits , berries , beets , carrots , and such

        by calories , a low-to-moderate-carbohydrate (20 to 35 percent)” -Pg 13

        • WilliamEden

          Touche! =D

          At first a pound seemed absurd to me as well, but then I thought about how much water content is actually in these foods. A pound of apples only has a bit over 200 calories. A pound of potatoes has 350 calories or so. That comes out to roughly a third of calories if you’re at 1500 calories, or just over a quarter at 2000 calories, consistent with their macronutrient recommendations.

          They do stress going closer to the 20% figure if you want a longevity bias, and closer to the 35% figure if you want a performance bias. I’m a huge fan of the longevity side of things myself…

          • Romeo Stevens

            Maybe I just encounter too many extreme low-carbers.