In “Challenging the utility of polygenic scores for social science: Environmental confounding, downward causation, and unknown biology,” sociologist and criminologist Burt demonstrates that genome-wide association studies (GWASs) and polygenic scores (PGSs) for such social outcomes as educational attainment do not live up to the promises made by their enthusiasts. GWASs capture only a small portion of potential genetic influences, and PGSs are irremediably confounded by environmental factors. One often-overlooked confound is “downward causation,” which Burt defines as “sociocultural forces that sort and select individuals based on genetically influenced traits, such as skin pigmentation and height, into different environments and exposures that influence social outcomes.” Sociocultural sorting and selection create what Burt describes as “artificial genetic associations,” or “environmental influences masquerading as genetic influences in GWAS.” In this commentary, I expand the concept of downward causation to include any association between genomic variants and a given outcome that is forged through social practices rather than biochemical pathways. This expanded definition does not require discrimination or privilege on the basis of physiological traits; rather, it recognizes that social stratification produces population stratification, or what we might call the social stratification of population. Because of space limitations, the argument is necessarily stylized and speculative, and should be understood as suggestive rather than demonstrative.
Geneticists define population stratification as “the situation that arises when a study population contains two or more ethnic or racial subgroups that have different allele frequencies and, just coincidentally, different levels of a particular phenotype” (Hamer, Reference Hamer2000). Recognizing the possibility that population stratification can produce spurious associations in GWASs, geneticists usually limit GWASs to a single ancestry group, European in the case of educational attainment. However, recent studies have demonstrated that population stratification also occurs within ancestry groups (Haworth et al., Reference Haworth2019). Geneticists typically attribute this population sub-structure to micro-geographic differentials in allele frequencies. Social scientists, however, should recognize that population sub-structure may also result from social stratification: Systems of hierarchy operating within societies in which some people or groups have more status, power, and resources than others.
Although social scientists investigating epigenetics recognize that the social world can shape cellular processes (Massey et al., Reference Massey, Wagner, Donnelly, McLanahan, Brooks-Gunn, Garfinkel and Notterman2018), those using GWASs and PGSs to identify sources of individual differences in educational attainment typically assume that causality operates only in the upward direction: Genetic differences generate social hierarchy, with status, power, and resources accruing to those with genomic variants that make them more capable of success (Harden, Reference Harden2021). To be sure, an individual's DNA is set at birth, and no life experience – short of high levels of toxic exposure – will change it. On the scale of historical time, however, decades of sociological research suggest the possibility of downward causation: That social hierarchy across generations can shape individual DNA. This is because, in European and Europe-descended societies, children are typically born into the same social position occupied by their parents, and a variety of institutions make mobility difficult. One such institution is endogamy – within-group marriage – which members of high-status groups use as a strategy for maintaining social hierarchy and preserving their position within it (Kalmijn, Reference Kalmijn1991; van Leeuwen, Maas, & Miles, Reference van Leeuwen, Maas and Miles2005). The existence of a tightly bounded “marriage market” in the highest strata of European and American societies is so well-established and well-known as to provide the plot for an entire genre of nineteenth-century novels and such recent television shows as Bridgerton and The Gilded Age. Through several generations of endogamy, social stratification likely also contributed to the development of population sub-structure within European and Europe-descended societies.
In her article, Burt cites the now-classic “chopsticks problem” as an example of population stratification. An analogue for the social stratification of population would be the salad-fork problem. A GWAS conducted 100 years ago for the number of forks used per meal, even if limited to individuals of European ancestry, would likely find hits because the social groups that used the most forks per meal as a symbol of their status also had a long history of using endogamy to perpetuate their status. Over generations, founder effects and genetic drift likely produced different allele frequencies between multiple-fork users and single-fork users within societies. The same variants that predicted salad-fork usage 100 years ago might also have predicted the number of servants employed or the value of property owned. Today, those same variants probably correlate with educational attainment. In the United Kingdom, it is well-known that a person's class status directly produces or precludes educational opportunities (Jackson & Marsden, Reference Jackson and Marsden2012). Access to secondary education is more equitable in the United States, but the majority of Americans do not attend college. In the United States, the rate of high school completion grew substantially around the turn of the twentieth century. At the same time, children of the elite increasingly attended college in order to maintain and justify their status (Groeger, Reference Groeger2021). Because of the social stratification of population, this growing group of college graduates was likely genomically distinct from those who did not attend. As higher education expanded further over the course of the twentieth century, college graduates tended to marry one another and the college admissions process tended to privilege those whose parents graduated from college (Domingue, Fletcher, Conley, & Boardman, Reference Domingue, Fletcher, Conley and Boardman2014; Neidhöfer & Stockhausen, Reference Neidhöfer and Stockhausen2019; Schwartz & Mare, Reference Schwartz and Mare2005), further perpetuating the downward causation of single-nucleotide polymorphism (SNP)–education correlations.
Over the past 5 years, researchers in social genomics have realized that GWASs for educational attainment pick up much more than just direct genetic effects. Scientists have begun to use the term “dynastic effects” to refer to the correlation between a parent's genotype and a child's phenotype (Morris, Davies, Hemani, & Smith, Reference Morris, Davies, Hemani and Smith2020). This term, however, remains undertheorized and underexplored, and is often assumed to describe the direct genetic effect of the parents' genotypes on their parenting (Brumpton et al., Reference Brumpton2020). The concept of social stratification of population, however, suggests that the PGS for educational attainment includes dynastic effects in the original sense of the term “dynasty” – a high-status lineage – reflecting the social capital marshaled by an individual's extended family.
In “Challenging the utility of polygenic scores for social science: Environmental confounding, downward causation, and unknown biology,” sociologist and criminologist Burt demonstrates that genome-wide association studies (GWASs) and polygenic scores (PGSs) for such social outcomes as educational attainment do not live up to the promises made by their enthusiasts. GWASs capture only a small portion of potential genetic influences, and PGSs are irremediably confounded by environmental factors. One often-overlooked confound is “downward causation,” which Burt defines as “sociocultural forces that sort and select individuals based on genetically influenced traits, such as skin pigmentation and height, into different environments and exposures that influence social outcomes.” Sociocultural sorting and selection create what Burt describes as “artificial genetic associations,” or “environmental influences masquerading as genetic influences in GWAS.” In this commentary, I expand the concept of downward causation to include any association between genomic variants and a given outcome that is forged through social practices rather than biochemical pathways. This expanded definition does not require discrimination or privilege on the basis of physiological traits; rather, it recognizes that social stratification produces population stratification, or what we might call the social stratification of population. Because of space limitations, the argument is necessarily stylized and speculative, and should be understood as suggestive rather than demonstrative.
Geneticists define population stratification as “the situation that arises when a study population contains two or more ethnic or racial subgroups that have different allele frequencies and, just coincidentally, different levels of a particular phenotype” (Hamer, Reference Hamer2000). Recognizing the possibility that population stratification can produce spurious associations in GWASs, geneticists usually limit GWASs to a single ancestry group, European in the case of educational attainment. However, recent studies have demonstrated that population stratification also occurs within ancestry groups (Haworth et al., Reference Haworth2019). Geneticists typically attribute this population sub-structure to micro-geographic differentials in allele frequencies. Social scientists, however, should recognize that population sub-structure may also result from social stratification: Systems of hierarchy operating within societies in which some people or groups have more status, power, and resources than others.
Although social scientists investigating epigenetics recognize that the social world can shape cellular processes (Massey et al., Reference Massey, Wagner, Donnelly, McLanahan, Brooks-Gunn, Garfinkel and Notterman2018), those using GWASs and PGSs to identify sources of individual differences in educational attainment typically assume that causality operates only in the upward direction: Genetic differences generate social hierarchy, with status, power, and resources accruing to those with genomic variants that make them more capable of success (Harden, Reference Harden2021). To be sure, an individual's DNA is set at birth, and no life experience – short of high levels of toxic exposure – will change it. On the scale of historical time, however, decades of sociological research suggest the possibility of downward causation: That social hierarchy across generations can shape individual DNA. This is because, in European and Europe-descended societies, children are typically born into the same social position occupied by their parents, and a variety of institutions make mobility difficult. One such institution is endogamy – within-group marriage – which members of high-status groups use as a strategy for maintaining social hierarchy and preserving their position within it (Kalmijn, Reference Kalmijn1991; van Leeuwen, Maas, & Miles, Reference van Leeuwen, Maas and Miles2005). The existence of a tightly bounded “marriage market” in the highest strata of European and American societies is so well-established and well-known as to provide the plot for an entire genre of nineteenth-century novels and such recent television shows as Bridgerton and The Gilded Age. Through several generations of endogamy, social stratification likely also contributed to the development of population sub-structure within European and Europe-descended societies.
In her article, Burt cites the now-classic “chopsticks problem” as an example of population stratification. An analogue for the social stratification of population would be the salad-fork problem. A GWAS conducted 100 years ago for the number of forks used per meal, even if limited to individuals of European ancestry, would likely find hits because the social groups that used the most forks per meal as a symbol of their status also had a long history of using endogamy to perpetuate their status. Over generations, founder effects and genetic drift likely produced different allele frequencies between multiple-fork users and single-fork users within societies. The same variants that predicted salad-fork usage 100 years ago might also have predicted the number of servants employed or the value of property owned. Today, those same variants probably correlate with educational attainment. In the United Kingdom, it is well-known that a person's class status directly produces or precludes educational opportunities (Jackson & Marsden, Reference Jackson and Marsden2012). Access to secondary education is more equitable in the United States, but the majority of Americans do not attend college. In the United States, the rate of high school completion grew substantially around the turn of the twentieth century. At the same time, children of the elite increasingly attended college in order to maintain and justify their status (Groeger, Reference Groeger2021). Because of the social stratification of population, this growing group of college graduates was likely genomically distinct from those who did not attend. As higher education expanded further over the course of the twentieth century, college graduates tended to marry one another and the college admissions process tended to privilege those whose parents graduated from college (Domingue, Fletcher, Conley, & Boardman, Reference Domingue, Fletcher, Conley and Boardman2014; Neidhöfer & Stockhausen, Reference Neidhöfer and Stockhausen2019; Schwartz & Mare, Reference Schwartz and Mare2005), further perpetuating the downward causation of single-nucleotide polymorphism (SNP)–education correlations.
Over the past 5 years, researchers in social genomics have realized that GWASs for educational attainment pick up much more than just direct genetic effects. Scientists have begun to use the term “dynastic effects” to refer to the correlation between a parent's genotype and a child's phenotype (Morris, Davies, Hemani, & Smith, Reference Morris, Davies, Hemani and Smith2020). This term, however, remains undertheorized and underexplored, and is often assumed to describe the direct genetic effect of the parents' genotypes on their parenting (Brumpton et al., Reference Brumpton2020). The concept of social stratification of population, however, suggests that the PGS for educational attainment includes dynastic effects in the original sense of the term “dynasty” – a high-status lineage – reflecting the social capital marshaled by an individual's extended family.
Competing interest
None.