Nettle et al. provide a long-needed integration of behavioral ecological research on the costs and benefits of fat storage with data on rates of obesity in humans. The logic is convincing, the model helps to specify the exact argument, and the data on birds document that experiencing unreliable access to food shifts the fat-storage set point upward. One study on a mammal is mentioned (Li et al. Reference Li, Cope, Johnson, Smith and Nagy2010), but supporting data are also available from multiple species (Dionne et al. Reference Dionne, Skovsø, Templeman, Clee and Johnson2016; Wilson & Cantor Reference Wilson and Cantor1987). Nettle et al. note the unfortunate neglect of such behavioral ecological models in the social science literature, although there is some relevant work, especially from economists (Bellisari Reference Bellisari2008; Smith Reference Smith2009).

The authors’ meta-analysis supports their insurance hypothesis, but I was surprised that the effect was not larger and more generalizable. Food insecurity increased the risk of obesity by only 21% and only for women in high-income countries. The authors are admirably restrained in their conclusions, emphasizing that factors other than the insurance hypothesis are also important. I wonder, however, if the limited strength of the results might be partly accounted for by measurement problems related to relying heavily on the questionnaire by Radimer et al. (Reference Radimer, Olson and Campbell1990) and its derivatives.

Examination of the specific questions on these questionnaires reveals that they measure only food insecurity resulting from lack of money, not from other causes. The first question is, “I worry that my food will run out before I get money to buy more.” Out of the 13 questions, 12 refer explicitly to not having enough money for food. Not surprisingly, food security on this measure is achieved for 50% of families with incomes over $25,000 but less than 12% of families with household incomes below $20,000 (Kendall et al. Reference Kendall, Olson and Frongillo1995). In short, these instruments measure levels of poverty severe enough to compromise food availability at times; such levels of poverty are, of course, highly correlated with social class, education, race, and sex. For people of middle-class and higher incomes in developed countries, the experience of hunger is rarely from lack of money. Dieting, medical conditions, and schedule constraints are more likely reasons for experiencing episodes of hunger. This makes the usual measures of food insecurity of limited utility, even while they remain valuable for studies of more diverse populations.

This target article has major implications for anorexia nervosa. As the article notes, substantial evidence supports the hypothesis that restrictive dieting causes long-term weight gain (Dulloo et al. Reference Dulloo, Jacquet, Montani and Schutz2015; Mann et al. Reference Mann, Tomiyama, Westling, Lew, Samuels and Chatman2007; Pietiläinen et al. Reference Pietiläinen, Saarni, Kaprio and Rissanen2012). Some of these studies, and those on rodents, also note that caloric restriction induces binge eating. The authors suggest that the insurance hypothesis predicts that anorexia should be more common when food security is high because that makes low body mass safe. That fits the epidemiologic evidence showing the high prevalence of anorexia in upper-class women in wealthy countries, but anorexia is vastly different from merely maintaining a low body mass when food is reliably available. Anorexia is not an adaptation; it is a disorder that arises from evolved eating regulation mechanisms that malfunction in some individuals when exposed to certain aspects of modern environments. People with anorexia nervosa do not maintain a low weight and go about life's business; they are obsessed and often deluded in their monomaniacal dedication to achieving thinness as a primary life goal. They are constantly hungry and preoccupied with food and weight. Many die.

A corollary of the insurance hypothesis offers an alternative explanation (Nesse Reference Nesse, Trevathan, McKenna and Smith1999, p. 363). Almost all cases of anorexia begin with a stringent diet. The reliable result is an episode of out-of-control binge eating within a few days, behavior that would be life-saving in a famine. The experience of out-of-control eating combines with experiencing increased appetite, and observing increased fat storage, to arouse yet more intense fear of obesity, spurring yet more stringent efforts to control caloric intake, creating a vicious cycle that results in bulimia for many and anorexia in a few.

Nettle et al. have provided a valuable service by bringing principles from behavioral ecology together with those from obesity research. Related principles can also explain how exposure to modern media and other incentives to be thin can induce severe dieting that sets off adaptive responses that are useful during a famine, but prone to runaway positive feedback and disorder when aroused by dieting. Severe dietary restriction arouses responses adaptive during starvation – including binge eating, increased appetite, and increased fat storage – that motivate yet more intense efforts to control intake, in a vicious circle that spirals out of control to anorexia or bulimia. If correct, this explanation has obvious utility for prevention and treatment.