Nettle et al. suggest that there are few integrative explanations for disparities in obesity among humans. In response to this apparent gap, they propose the insurance hypothesis, arguing that the human body is adapted to store more fat when faced with cues of uncertainty in food supplies. An empirical keystone of their argument is the well-established observation in high-income countries that women are more likely to be obese in situations of food insecurity. Public health researchers often refer to this phenomenon as the poverty–obesity paradox. A positive association between obesity and food insecurity is consistent with the insurance hypothesis, but it is also consistent with a number of other explanations neglected by the target article.

Some of these neglected explanations for the poverty–obesity paradox share deep similarities with the insurance hypothesis by proposing that certain food environments drive people to store more fat. Different versions of such resource-driven theories argue that deprivation can lead people to choose energy-dense or protein-poor foods that are cheaper per calorie but that are also less satiating and thus more likely to cause people to consume excess calories (Hruschka Reference Hruschka2012). Such explanations are popular in the fields of nutrition, public health, and a number of social sciences. However, as the target article concedes for the insurance hypothesis, they cannot easily account for the absence of the poverty–obesity paradox among certain groups – such as men or children.

By contrast, an alternative explanation based on female-biased anti-fat discrimination in marriage can better account for both the presence of the poverty–obesity paradox among women and its absence among men and children by positing that marital choices sort some obese individuals into poorer households (Hruschka Reference Hruschka, Anderson-Fye and Brewis2017; Han & Hruschka, Reference Han and Hruschkasubmitted). Importantly, this explanation does not require that people pursue and achieve different weight goals. As long as there is variation in obesity and wealth, if wealthy individuals choose thinner partners and thinner individuals choose wealthier partners, correlations between individual obesity and household poverty can arise. Recent studies of heterosexual marital preferences in the United States suggest that men, on average, have stronger preferences for thinner partners, and women have stronger preferences for wealthier partners (Oreffice & Quintana-Domeque Reference Oreffice and Quintana-Domeque2010). In such situations, we should observe heavier women married to poorer men and thus living in poorer households. However, we should not expect heavier men married to women with lower incomes. This gendered prediction is fully consistent with the gender-specific correlations described in the target article. Moreover, if correlations of adult obesity and poverty arise through marriage, then there should be little association of childhood obesity with household income, except perhaps due to propensities inherited from parents.

Resource-driven theories (including the insurance hypothesis) and explanations based on anti-fat discrimination in marriage also make contrasting predictions about individuals who have never married. According to resource-driven theories, there is no a priori reason why the relationships between food insecurity and obesity should differ by marital status. By contrast, anti-fat discrimination in marriage would only predict positive correlations between food insecurity and obesity among those individuals sorted into households through marriage. Conversely, among individuals who have never married, we should observe no such correlation.

Only one study to our knowledge has explicitly tested whether the association of food insecurity and obesity varies by marital status (Hanson et al. Reference Hanson, Sobal and Frongillo2007). As expected from theories of anti-fat discrimination in marriage, the study found an association among women who had entered marriage at least once in their lives, but not among never-married women (Hanson et al. Reference Hanson, Sobal and Frongillo2007). We find similar results when looking at obesity and another proxy of poverty – household income – in nationally representative samples from both the United States and South Korea (Hruschka Reference Hruschka, Anderson-Fye and Brewis2017; Han & Hruschka, Reference Han and Hruschkasubmitted).

The target article also confirms prior work that the poverty–obesity paradox is largely confined to high-income settings, and that for the 80% of contemporary humanity living on less than $10 (USD) per day, increasing income leads to greater weight gain (Hruschka et al. Reference Hruschka, Hadley and Brewis2014). A simple dual process model may be able to account for these varying associations of income and obesity (Hruschka Reference Hruschka, Anderson-Fye and Brewis2017). The first underlying process is simply that people accrue more body fat as they encounter increasing abundance. However, as people become richer, any additional resources have less of an effect until the relationship between resources and body fat becomes effectively flat in the highest-income countries. The second process – sorting of heavier individuals into poorer households through marriage – is weaker and only becomes apparent in higher-income settings where the first process no longer creates a noticeable association between resources and obesity. Importantly, this second process will only take place when there are specific, gendered cultural preferences for thinness and wealth.

In sum, resource-driven theories positing that deprivation or uncertainty lead to greater obesity are inconsistent with findings from men, children, and never-married women, as well as the poorest 80% of contemporary humanity. On the other hand, the theory of anti-fat discrimination in marriage markets is consistent with demographic patterns in a number of high-income countries and is readily integrated into straightforward models of weight gain resulting from increasing resources.

Here we focus on the empirical keystone of the target article's argument – the poverty–obesity paradox in high-income settings. However, the authors also briefly describe additional puzzles that might be consistent with the insurance hypothesis. Some of these empirical patterns – the association between societal inequality and obesity – are not nearly as reliable or well-established as the poverty–obesity paradox. Meanwhile, other puzzles might be explained by the misuse of body mass index as a measure of adiposity across human populations when comparing, for example, Japan and the United States (Hruschka et al. Reference Hruschka, Hadley and Brewis2014; Hruschka & Hadley Reference Hruschka and Hadley2016). Further work that lays out all potential hypotheses on a level playing field and identifies contrasting predictions to investigate with finer-grained data will lead to a better understanding of what best accounts for these puzzles in the distribution of obesity.