It is a hotly contested issue whether polygenic scores (PGSs) and genome-wide association studies (GWASs) should play a major role in the social sciences. As described in the target article, what we see is both (over)hype and a staunch opposition, with harsh accusations thrown around, straw man arguments, and ad hominem attacks. All this makes it difficult to not only evaluate the positions, but even to ask important methodological questions about the potential uses of these genetic tools within the social sciences.

Here, Burt offers an elegant methodological target article with the aim of addressing just this problem. In it, Burt objectively criticizes the hype that has often accompanied heritability research, without committing any of the above sins, drawing attention to the methodological limits and challenges of adding genetics research to the social sciences. Although we agree with many of Burt's points, however, we can't help but feel that she ends up overstating her conclusions and overplaying the differences between sociogenomics and traditional research within the social sciences.

In her conclusion, Burt states that “GWASs and PGSs may be powerful tools for identifying genetic associations, but they are not the right tools for understanding complex social traits” (target article, sect. 9, para. 3). Naturally, we wholeheartedly agree. However, our reasons for accepting this claim aren't a belief that these tools cannot at all help us to understand genetic influences or social outcomes, but rather that there is no such thing as the right tools for understanding complex social traits. That is, we do not think that there is some kind of unique or privileged combination of scientific tools for investigation of whatever complex social trait we are interested in, whether that is poverty, educational attainment, or criminal behavior. Let us elaborate.

As philosophers of science (and in particular, philosophers of the social sciences) have long recognized, complex phenomena are not to be understood through the competition of various methods with the aim of finding the ideal one, but rather through use of a broad range of tools that complement each other in various ways (Mitchell, Reference Mitchell2009; Veit, Reference Veit2019, Reference Veit2021; Wimsatt, Reference Wimsatt2007; Ylikoski & Aydinonat, Reference Ylikoski and Aydinonat2014). Although there are often conflicts within scientific disciplines regarding what sets of methods, models, experiments, and the like should be employed, these often appear to be driven by “indoctrination” into the methodology of a lab and ideological disputes over the correct methods. As the saying goes: If all you learn is how to swing a hammer, all problems will start to look like nails. But from a higher-level perspective, it is precisely because of the pluralism of different methods that science has flourished. And this conclusion, we think, likewise applies to the use of GWASs and PGSs.

These methods should not act as a replacement for standard social science tools, nor should they be seen as competitors to randomized controlled trials (RCTs) that investigate environmental factors. Instead, we argue that they can provide us with a useful complement for research into the main targets of the social sciences, that is: complex causal systems with great heterogeneity and no strong generalizations. Just as the study of genome-wide associations bears the danger of falsely attributing causality to observed correlations, so too does standard social science. Burt is right in her criticism of the hype around PGSs: That they are often seen as deterministic, fail to control for a wide range of potential confounds, risk reviving the unfortunate gene-culture war, and so forth. But it is possible to arrive at such a critical stance by highlighting that sociogenomics will of course suffer from all the methodological difficulties of the social sciences – causal indeterminacy, the complexity of the social world, looping effects, and so forth. Within such an alternative picture, however, sociogenomics could still play a valuable role, within its own limited sphere.

Rather than simplifying the complexity of social phenomena, we argue that sociogenomics can help us to highlight how complex and causally interdependent social phenomena truly are. That is, we can buy into the main criticisms of the usefulness of PGSs in the social sciences, without being led to the strong conclusion that sociogenomics is methodologically doomed. Rather than returning to old and unhelpful discussions of social versus genetic causes, we think that sociogenomics might in fact help us toward a recognition of the complexity of our social traits and their myriad bases. This is how one should understand the argument that PGSs may improve RCTs by finding further variables to be controlled for (Harden, Reference Harden2021). It's an embrace of a supplementary and pluralistic stance in the face of complexity. Rather than eliminating sociogenomics, or buying into the mistaken hype that it is going to replace and revolutionize standard social science, we can see its role instead as a complementary method to be added to the vast toolkit of the social sciences. Burt rightly points out that the methods as they are currently used too often fail to appreciate their own limitations, but this can be used as a starting-point, with these careful criticisms forming the basis for refining and strengthening the methods to better fit the contexts of use.

We therefore think that neither the majority of advocates nor the majority of critics of PGSs hold an adequate epistemic stance toward their use in the social sciences. Instead, we have here advocated for something of a mid-level approach, in which proponents of sociogenomics are urged to recognize the methodological difficulties of social science research and familiarize themselves with the philosophy of the social sciences in order to improve their own methods. Once the hype dies down, what remains will be better science, one practiced with adequate attention paid to the current problems and limitations of the methods. At the moment, without knowing exactly how this will unfold, we would like to avoid making any firm predictions regarding the likely payoffs of sociogenomics; however, we hold a (cautiously) optimistic stance regarding its future use.