Van Lange et al. formulated a theoretical model in which they proposed climate as a predictor of self-control (and aggressive behavior). We comment on the proposition “that lower temperatures and especially greater seasonal variation in temperature call for individuals and societies to adopt … a greater degree of self-control” (sect. 3, para. 4), which, they argue, is due to a slower life history strategy. In developing their theoretical position, the authors propose distance from the equator as a predictor of self-control. They advocated a “data-driven” approach, allowing one “to derive precise estimates of the variance accounted for by various predictor variables” (sect. 5.3.1, para. 2). In our Human Penguin Project (HPP; available at: https://osf.io/2rm5b/), we collected latitude, self-control, and a variety of important social predictors from 12 countries with varying distances from the equator. These variables allowed us to test the proposed relationship between distance from the equator and self-control. The social predictors included variables such as social network quality and size and variables that are crucial for Life History Theory, like people's attachment styles (Del Giudice Reference Del Giudice2008). Together, we provide the very first test of the authors' proposed model through data-driven analyses (a method called supervised machine learning; for more technical discussions, see Breiman [Reference Breiman2001]; IJzerman et al. [Reference IJzerman, Pollet, Ebersole and Kun2016]; Yarkoni & Westfall [Reference Yarkoni and Westfall2016]) as proposed by the authors, and in a more traditional null hypothesis significance testing confirmatory manner (a mediation analysis).

We analyzed data from 1507 participants from 12 countries on three different continents, with countries at varying levels of distance from the equator. The underlying analytical details (including a detailed explanation of supervised machine learning) are reported online (https://osf.io/gtj38/). Our method is very powerful and robust, as it validates the model internally as it tests the strength of the model and the size of the error. We found distance from the equator to be a significant predictor of self-control (Tangney et al. Reference Tangney, Baumeister and Boone2004), but barely so: It was the 14th predictor in our list and comparable in prediction power to whether participants spoke Serbian or not (the 13th predictor).Footnote ² As we could compare the strength of different predictors, our analyses revealed that the power to predict self-control was much greater for such variables as attachment anxiety, proneness to feeling nostalgia (a complex social emotion), social network size, level of complex social integration, and participants' attachments to their homes. Plotting these variables, controlling for the remainder of the model, further confirmed the stronger relationship of social determinants over distance from the equator (https://osf.io/vzwbe/).

Was distance from the equator then a reliable predictor of climate? We think so: Equator distance correlated strongly with the minimum temperature of that day (r = 0.90, N = 1463). We further explored whether attachment anxiety would mediate the relationship between equator distance and self-control; attachment anxiety (a strong predictor of self-control) could be indicative of differing life history strategies (Del Giudice Reference Del Giudice2008). There was no such relationship, as distance from the equator failed to be related to attachment anxiety (t = 0.02, p = .99), with a nonsignificant mediation onto self-control (95% confidence interval [CI]: 0.0007 to 0.0007). To be sure, we also tested for attachment avoidance, which also was not predicted by equator distance (t = 0.02, p = .85, 95% CI: 0.0003 to 0.0004). We thus find little evidence that distance from the equator matters for predicting self-control and life history strategies, and our analyses support the idea that the social environment is much more important in predicting self-control.

How could this be so? The authors reviewed evidence that seemed supportive of their relationship between climate and self-control. However, our reading of the literature suggests more nuance. Warmer water temperatures are indeed associated with faster growth, earlier death, and higher risk acceptance for some animals (all indicative of a faster life history trajectory [Holt & Jørgensen Reference Holt and Jørgensen2014]). But in contrast, warm-blooded mesopelagic fish (which live in an extremely cold environment) also exhibit a precocious maturation comparable to that of other animals in warmer environments (Miya & Nemoto Reference Miya and Nemoto1986). And life history strategies cannot be easily extended from animals to humans. Humans are more unusual in that both slow and fast life history strategies can involve effortful control and impulsivity (Del Giudice Reference Del Giudice, Gendolla, Tops and Koole2015). Furthermore, warmer temperatures are not just related to aggression; a wealth of evidence in humans indicates that higher temperatures can also relate to prosocial behavior (cf. IJzerman et al. Reference IJzerman, Coan, Wagemans, Missler, van Beest, Lindenberg and Tops2015a).

Furthermore, the authors neglected the fact that self-control via internalized norms is not linked to time horizon, but to a feeling of obligation. It is thus strongly influenced by the relevance of goals (Lindenberg Reference Lindenberg, Wittek, Snijders and Nee2013). For example, Dutch males scoring higher on honor concerns respond more aggressively when insulted, but less aggressively when not insulted (IJzerman et al. Reference IJzerman, van Dijk and Gallucci2007). A recent study also indicated that the relationship between armed conflict and heat waves or droughts is due to the fact that heat waves or droughts exacerbate existing ethnic strife (Schleussner et al. Reference Schleussner, Donges, Donner and Schellnhuber2016). Goal relevance, rather than a lack of self-control, strengthens aggressive responding in these cases.

In short, a broader consideration of the literature speaks against most of the propositions of their theoretical model. Most critically our data, which were suited for testing many of their claims, do not provide empirical support, likely because their theoretical model is underfitted, with the link between climate and self-control being underspecified. It is thus not surprising that the HPP provided evidence against their model, which we take as a strong suggestion for the authors to revisit their proposed link between climate and self-control.

In reformulating their model, we further advise that they give greater weight to established theories on the complex relationships between self-control, life-histories, culture, social organization, and violence, such as the literature mentioned previously, as well as literature on the development of culture (Diamond Reference Diamond1999) and the antecedents of violence (Fiske & Rai Reference Fiske and Rai2015). These theories support our results that the social environment rather than climate predicts self-control, and we suspect it is this social environment that mostly predicts aggressive behaviors. Cultures are not individuals writ large; they invariantly are complex and cannot be reduced to a simple main-effect model.