The cover story in today’s N.Y. Times Magazine is on food science and the “age of nutritionism.” Michael Pollan, a professor of journalism at Berkeley, argues that the last thirty years of food science, ostensibly aimed at making people healthier, have actually made matters worse. He argues that the effort to reduce foods to their component parts — namely, nutrients — has left Americans focused on a never-ending rotation of different nutrients instead of on the importance of the foods themselves. And I think his argument has something to say, as well, about the general methodology of scientific empiricism as applied in any discipline, including the law.
Pollan argues that foods are impossibly complex. He argues that whole foods, such as whole grains, fruits, vegetables, and leafy greens, are the best foods for human consumption, based on a broad-based perspective of human history and environment. But according to Pollan, food science has spent the last thirty years trying to isolate the exact nutrients in these foods that make us healthier. This scientific effort has resulted in an ever-changing series of findings, as scientists proclaim the value of a particular nutrient only to find its effects much less dramatic than initial findings suggested. Pollan argues that this is a result, in part, of an effort to isolate nutrients without the actual ability to do so; foods are so complex that it is impossible to reduce them to simply a list of nutrients. But scientists continue to do so, with predictably incomplete and ultimately erroneous results.
Pollan’s critique of food science has lessons for other forms of empiricism. Here are a few:
Limits to Empirical Methods. Pollan spends the bulk of the article explaining the flaws in various nutritional studies over the last thirty years. He notes that for each time the benefits of a new “nutrient” are discovered, it is later revealed that the nutrient itself does not really create those benefits. Instead, it is the nutrient working within its particular environment — namely, within a certain type of food — that may create the nutritional benefit. Pollan goes through a series of nutritional fads — low-fat foods, beta carotene, omega-3 oils, low-carb foods — to show that each has a kernel of truth but is woefully incomplete on its own. Pollan chalks these failures up to the effort to oversimplify something that cannot be simplified.
Quantitative empirical studies share these same limitations. They depend on the researcher’s ability to isolate a single variable and control all other “factors” that might influence the decision. Certainly, all empiricists would recognize the inherent difficulty in doing this. And better empiricism does a better job at actually controlling extraneous variables. But as Pollan suggests, there is some degree of hubris in even attempting this. From his perspective, we are nowhere near the day when scientists will actually be able to explain how foods actually work. While he pays some respect to the continuing scientific effort, he conveys a skepticism that it will ever actually be able to tell us what we need to know.
Overgeneralizing from the Results. Pollan would not have a problem with food science were its findings not so dramatically announced by the media and so extensively coopted by the food industry. For it is the conclusiveness of the studies and the real-world changes that such studies prompt that really cause the trouble. For example, Pollan describes how the low-fat trend in the 1980s actually prompted folks to eat more carbohydrates than they had been before. This made diets worse, not better. Similarly, the recent study finding that low-fat diets did not reduce health risks was weak science, according to Pollan. But the real problem is that the media’s trumpeting of the study encouraged the average person to pick up a quarter pounder with cheese, despite the study’s questionable and inconclusive results.
This, too, is a problem for all empiricists: how to acknowledge that their results are simply one small piece of data in a ongoing process of data collection and interpretation, while persuading their peers that their study represents a critical and important step forward for the discipline. And with law in particular, there is the temptation to argue that a particular empirical result inexorably leads to a particular policy prescription. After all, if law review articles with no empirical support can make such claims, why can’t demonstrable scientific facts?
Ultimately, I think Pollan swings too far the other way. Although food science has inherent limitations, that does not mean that its effort to isolate discrete nutrients is ultimately fruitless. The fact that Vitamin C prevents scurvy is an important and useful bit of information, and eating oranges is not the only way to get the benefits. Findings like this help prevent a wide array of diseases. Ultimately, food science may lead us to understand a lot more about food, and that understanding will help us in our everyday diets and in times of food crisis. But I agree with a more moderate version of Pollan’s thesis: empiricism is important, but we cannot focus on short-term findings as the new answer to all our problems. Putting food science in its context, and using a broader, more comprehensive vision in coming up with our actual diets, is a wiser course.
Posted by Matt Bodie on January 28, 2007 at 12:36 PM
Comments
hog roasting
Posted by: hog roasting | Mar 25, 2007 3:51:24 PM
Interesting article, and interesting take on it. I haven’t read it yet, but Pollan’s take on food science reminds me of Theodoric of York, Medeival Doctor:
“You know, medicine is not an exact science, but we are learning all the time. Why, just fifty years ago, they thought a disease like your daughter’s was caused by demonic possession or witchcraft. But nowadays we know that Isabelle is suffering from an imbalance of bodily humors, perhaps caused by a toad or a small dwarf living in her stomach.”
Posted by: Scott Moss | Jan 28, 2007 9:18:21 PM
Pollan’s article is useful as a case study about the limits of empiricism. But despite his rhetoric, Pollan does use the nutrient-based approach when it suits one of his arguments. In order to substantiate his theory that people should eat more leaves and fewer seeds, Pollan notes how one shouldn’t just increase intake of omega-3 fatty acids, but also should decrease intake of omega-6 fatty acids — he even states that he is borrowing “the nutritionist’s reductionist vocabulary” to make the point.
It seems that Pollan’s concerns about empiricism are mainly an instrument to present his beef (pun intended) with the way in which scientific findings are publicized. Specifically, the studies that are guided by “nutritionism” (and promulgated through media accounts as well as food manufacturers’ advertisements) lead people towards artificial supplements and/or processed food. People start looking to add beta carotene to their diets artificially, rather than start eating those naturally occurring foods rich in beta carotene. To use your scurvy example, Pollan would likely be displeased if people take a vitamin C pill instead of eating fruit (whether they be oranges, or limes, or grapefruit, or fill-in-the-blank).
I have a feeling that Pollan wouldn’t mind nutritionism so much if it led people to the produce aisle of the supermarket rather than to multivitamins and processed food advertised as “low-fat” or “now with omega-3 fatty acids.”
Posted by: J.R. | Jan 28, 2007 7:18:11 PM
On the first point, I’d suggest that some of the insights of complexity theory may provide some reasons why attempting to isolate individual variables doesn’t seem to get us very far in many cases. Namely, systems are not linear; the prevailing feedback loop models highlight the idea that system behavior in aggregate is the product of myriad interactions between all different components and attractors of the system. Attempting to isolate one attractor and understand its effect on aggregate behavior is in some sense limited in capacity by the notion that system behavior is really a product of the interactions between attractors.
This is obviously not to suggest there is nothing worthwhile in attempting to isolate single variables; only that in dynamic systems, doing so cannot capture some of the important aggregate dynamics of system behavior. JMO.
Posted by: Daniel Goldberg | Jan 28, 2007 2:17:03 PM