A cornerstone of social genomics research is the creation of what are called ‘polygenic scores’. Although the amount of computational power needed to pool and analyse the relevant genomic data is unfathomably large, the basic idea is easy to fathom. First, social scientists and geneticists collaborate to identify hundreds or thousands of genetic variants, across the genome, correlated with a given trait, behaviour or outcome. Although the inferred effect of each of those genetic variants is miniscule by itself, the second step is for researchers to add up those many tiny effects to create a polygenic score. This strategy for making predictions about future traits or outcomes is also the cornerstone of ‘precision medicine’, which aspires to tailor medical treatments to individuals’ genomes.
Some well-informed observers think that this new strategy is just the latest in the history of efforts to analyse complex phenomena at the wrong level. To paraphrase the psychologist Eric Turkheimer, looking to genetic variants for insight about complex behaviours and social outcomes is like looking to the chemical composition of rocks to understand plate tectonics. And even those who are most enthusiastic about the eventual utility of these scores are acutely aware that previous efforts to use insights from molecular genetics have been hugely disappointing.
To their credit, social genomicists have taken the unprecedented and time-intensive step of creating Frequently Asked Questions documents, which accompany their publications and explain, with remarkable frankness, what they have and have not discovered, and what their findings do and do not mean. They are unfailingly clear about the fact that, when they add up the tiny genetic effects, the aggregate is small compared with, say, the total effect of the environment. They are relentless in their rejection of genetic determinism, and vigorous in their reiteration that environments play a huge role in explaining the outcomes they study.Erik Parens, “The genes we’re dealt” at Aeon