By Gary Taubes, author of Nobel Dreams (1987), Bad Science (1993), Good Calories, Bad Calories (2007), and Why We Get Fat (2011). Taubes is a former staff member at DISCOVER. He has won the Science in Society Award of the National Association of Science Writers three times and was awarded an MIT Knight Science Journalism Fellowship for 1996-97. A modified version of this post appeared on Taubes’ blog.
The last couple of weeks have witnessed a slightly-greater-than-usual outbreak of extremely newsworthy nutrition stories that could be described as bad journalism feasting on bad science. The first was a report out of the Harvard School of Public Health that meat-eating apparently causes premature death and disease (here’s how the New York Times covered it), and the second out of UC San Diego suggesting that chocolate is a food we should all be eating to lose weight (the Times again).
Both of these studies were classic examples of what is known technically as observational epidemiology, a field of research I discussed at great length back in 2007 in a cover article for in the New York Times Magazine. The article was called “Do We Really Know What Makes Us Healthy?” and I made the argument that this particular pursuit is closer to a pseudoscience than a real science.
As a case study, I used a collaboration of researchers from the Harvard School of Public Health, led by Walter Willett, who runs the Nurses’ Health Study. And I pointed out that every time that these Harvard researchers had claimed that an association observed in their observational trials was a causal relationship—that food or drug X caused disease or health benefit Y—and that this supposed causal relationship had then been tested in experiment, the experiment had failed to confirm the causal interpretation—i.e., the folks from Harvard got it wrong. Not most times, but every time.
Now it’s these very same Harvard researchers—Walter Willett and his colleagues—who have authored the article from two weeks ago claiming that red meat and processed meat consumption is deadly; that eating it regularly raises our risk of dying prematurely and contracting a host of chronic diseases. Zoe Harcombe has done a wonderful job dissecting the paper at her site. I want to talk about the bigger picture (in a less concise way).
This is an issue about science itself and the quality of research done in nutrition. Science is ultimately about establishing cause and effect. It’s not about guessing. You come up with a hypothesis—force x causes observation y—and then you do your best to prove that it’s wrong. If you can’t, you tentatively accept the possibility that your hypothesis might be right. In the words of Karl Popper, a leading philosopher of science, “The method of science is the method of bold conjectures and ingenious and severe attempts to refute them.” The bold conjectures, the hypotheses, making the observations that lead to your conjectures… that’s the easy part. The ingenious and severe attempts to refute your conjectures is the hard part. Anyone can make a bold conjecture. (Here’s one: space aliens cause heart disease.) Testing hypotheses ingeniously and severely is the single most important part of doing science.
The problem with observational studies like the ones from Harvard and UCSD that gave us the bad news about meat and the good news about chocolate, is that the researchers do little of this. The hard part of science is left out, and they skip straight to the endpoint, insisting that their causal interpretation of the association is the correct one and we should probably all change our diets accordingly.
by Richard Wrangham, as told to Discover’s Veronique Greenwood. Wrangham is the chair of biological anthropology at Harvard University, where he studies the cultural similarities between humans and chimpanzees—including our unique tendencies to form murderous alliances and engage in recreational sexual activity. He is the author of Catching Fire: How Cooking Made Us Human.
When I was studying the feeding behavior of wild chimpanzees in the early 1970s, I tried surviving on chimpanzee foods for a day at a time. I learned that nothing that chimpanzees ate (at Gombe, in Tanzania, at least) was so poisonous that it would make you ill, but nothing was so palatable that one could easily fill one’s stomach. Having eaten nothing but chimpanzee foods all day, I fell upon regular cooked food in the evenings with relief and delight.
About 25 years later, it occurred to me that my experience in Gombe of being unable to thrive on wild foods likely reflected a general problem for humans that was somehow overcome at some point, possibly through the development of cooking. (Various of our ancestors would have eaten more roots and meat than chimpanzees do, but I had plenty of experience of seeing chimpanzees working very hard to chew their way through tough raw meat—and had even myself tried chewing monkeys killed and discarded by chimpanzees.) In 1999, I published a paper [pdf] with colleagues that argued that the advent of cooking would have marked a turning point in how much energy our ancestors were able to reap from food.
To my surprise, some of the peer commentaries were dismissive of the idea that cooked food provides more energy than raw. The amazing fact is that no experiments had been published directly testing the effects of cooking on net energy gained. It was remarkable, given the abiding interest in calories, that there was a pronounced lack of studies of the effects of cooking on energy gain, even though there were thousands of studies on the effects of cooking on vitamin concentration, and a fair number on its effects on the physical properties of food such as tenderness. But more than a decade later, thanks particularly to the work of Rachel Carmody, a grad student in my lab, we now have a series of experiments that provide a solid base of evidence showing that the skeptics were wrong.
Whether we are talking about plants or meat, eating cooked food provides more calories than eating the same food raw. And that means that the calorie counts we’ve grown so used to consulting are routinely wrong. Read More