By Michelle N. Meyer, Patrick Turley, and Daniel J. Benjamin
The following is the full version of our Letter to the Editor of the Wall Street Journal in response to Charles Murray’s op-ed, “Genetics Will Revolutionize Social Science” (Jan. 27, 2020). A shorter version was published in the Journal on February 3, 2020.
As researchers with expertise in the creation and communication of polygenic scores, we are concerned that Charles Murray’s attempt to explain how “Genetics Will Revolutionize Social Science” (op-ed, Jan. 27) will cause confusion about the nature and potential of polygenic scores. In our limited space, we confine ourselves to one of the most misleading passages, one that sets the stage for the remainder of the essay: Mr. Murray’s claim that “[p]olygenic scores . . . [are] impervious to racism and other forms of prejudice” and therefore “offer social scientists . . . a secure place to stand in assessing what is innate and what is added by the environment.”
It is true that one’s genome — the sequence of nitrogen bases abbreviated as As, Cs, Ts, and Gs — is fixed at conception and is generally not altered by the environment. However, the implications of that sequence for an individual’s phenotypes can change dramatically depending on the environment. Let us explain.
Mr. Murray references our genetic study of cognitive phenotypes, including educational attainment and IQ, in which we did indeed find many associated genetic variants. However, as clearly explained by sociologist Christopher Jencks in 1980 and expanded on in FAQs we issued along with our paper (see especially FAQ 3.3), those associations very likely operate through environmental channels. In other words, genetic association studies like ours turn up variants that we can only be sure are relevant in the studied environment. As a result, a polygenic score that results from such a study reflects (and only necessarily holds true for) the particular group of people studied and the particular time and place in which they lived. Importantly, this means that polygenic scores can and do reflect racism, sexism, or other prejudices, as well as more benign environmental factors. And an individual with a high polygenic score based in one kind of environment might have a much lower polygenic score for the same phenotype when that score is based in a different environment.
Consider IQ. IQ is not a fixed attribute of individuals and can be affected — for better and worse — by the environment in myriad ways. For example, in a society where people of color are denied access to childhood enrichment programs or adequate nutrition, a polygenic score for IQ might reflect genetic variants associated with skin pigmentation. Relatedly, in a sexist society, variants on the X and Y chromosomes, which determine biological sex, might be related to a variety of socio-economic phenotypes. Such polygenic scores would indeed moderately predict the IQ of people on average, but — and this is key — much of that predictive power would simply reflect social choices, not innate or immutable biology. (Note that polygenic scores for IQ are actually poor predictors of any one individual’s phenotype, as Mr. Murray acknowledges.)
For the same reasons, it is not exactly true that polygenic scores “don’t drop because” of environmental circumstances. Because polygenic scores are unique to a particular time and place, an individual with a high polygenic score based on one kind of environment might have a much lower polygenic score for the same phenotype when that score is based on a different environment. For example, a polygenic score for educational attainment that is created by studying people whose educational environment emphasizes rote memorization or formal writing might be completely different from a score that is based on an educational environment that emphasizes, say, the “flipped classroom” or experiential learning. This implies that the same person with the same genome might be predicted to have a very different educational phenotype based on that genome if their environment changed.
For this reason, the notion of “genetic potential” is deeply problematic, especially when applied to polygenic scores. Polygenic scores do not distinguish between causal pathways that one might consider social versus biological. Furthermore, even with genetic variants that operate primarily through non-environmental mechanisms that are themselves difficult to modify (such as variants that directly influence the formation of neurons in the brain and the biochemical interactions among them), this does not limit the potential for environmental interventions to be powerful. The economist Arthur Goldberger famously gave the example of prescribing eyeglasses to eliminate genetic disparities in vision.
None of this is to say that polygenic scores are useless or scientifically invalid. They can help shed light on the nature of environmental influences, as statistical geneticist Augustine Kong and colleagues recently showed, and the effectiveness of environmental inventions. For example, controlling for polygenic scores enables researchers to measure the effects of the environment more precisely.
However, something as scientifically complex, and as historically fraught, as using genetics to predict human phenotypes deserves to be communicated with exquisite care. Genes influence countless (indeed, virtually all) human phenotypes. But that does not mean that those phenotypes are driven purely (or even primarily) by “biology” hermetically sealed off from environmental influence. Nor does it mean that these phenotypes are determined at birth or immutable, as too many think. Indeed, increasing evidence suggests the opposite. As we learn more about how genes influence phenotypes, we will also need to learn how to better communicate these subtleties and complexities.
Michelle N. Meyer, Ph.D., J.D.
Geisinger Health System
Patrick Turley, Ph.D.
Massachusetts General Hospital
Daniel J. Benjamin, Ph.D. (uncredited in the WSJ due to two-author limit)
University of Southern California
Los Angeles, CA