“Challenging the utility of polygenic scores for social science” is an erudite and compelling critique. Yet despite the analytic rigor, Burt failed to capture the fundamental flaws that render ‘gene-centric' analyses misleading and polygenic scores (PGSs) meaningless outside of highly controlled environments.
The conflation of “inherited” with “genetic”
The functional unit in biology, biological inheritance, and phenotypic development is the cell – not DNA molecules. To be precise, humans develop from a single inherited cell – in which the genome is but one of many components. Moreover, because each cell's idiosyncratic nature and spatiotemporal context determine gene expression, the genome is merely an “organ” (McClintock, Reference McClintock1984, p. 800) or “tool” of the cell (Archer, Reference Archer2015a, Reference Archer2015b, Reference Archer2015c). This fact renders the distinction between non-genetic (cellular) inheritance and the two forms of genetic inheritance (nuclear and mitochondrial) critical to analyses of phenotypic development. Nevertheless, gene-centric analyses ignore the distinction between cellular, nuclear, and mitochondrial inheritance, and the fact that intra- and extra-cellular environments determine both gene products and their ‘effects.'
For example, the fundamental difference between monozygotic (identical) and dizygotic (fraternal) twins is inherent in the nomenclature – identical twins develop from a single cell (a fertilized egg) with a single placenta (usually); whereas fraternal twins develop from two different cells (two fertilized eggs) with two different placentas (always). Thus, fraternal twins differ in cellular and genetic inheritance, and their intrauterine environments, whereas identical twins do not.
Therefore, the greater phenotypic variability of fraternal twins is due to differences in gene expression engendered by different cells (eggs) acting in concert with inter-twin differences in both genotype and prenatal environments. Yet despite the extreme variability in the developmental competence (oocyte quality; e.g., mitochondrial content) of every female's population of eggs (Santos, El Shourbagy, & St John, Reference Santos, El Shourbagy and St John2006; Wang & Moley, Reference Wang and Moley2010; Zhang et al., Reference Zhang, Liu, Chen, Zhang, Zhang, Hao and Miao2020; Zhou et al., Reference Zhou, McQueen, Schufreider, Lee, Uhler and Feinberg2020) and the irreversible impact of the intrauterine environment on development (Archer, Reference Archer2015a, Reference Archer2015b, Reference Archer2015c, Reference Archer2015d; Archer & Lavie, Reference Archer and Lavie2022; Archer, Lavie, Dobersek, & Hill, Reference Archer, Lavie, Dobersek and Hill2023), the functional distinction between cellular, genetic, and environmental (in utero) inheritance is absent in ‘twin-studies,' heritability statistics, and polygenic scores (PGSs).
Thus, gene-centrism obscures the totality of biologic inheritance by conflating “genetic” with “inherited” – and therefore, the complexity of physiologic, psychological, and social phenotypic development remains unexamined.
Nonlinearity and development
Organismal development is a complex process that extends far beyond the linear amino acid sequence determined by the genetic code. For example, there are cellular, organismal, and environmental processes that lead to “one-to-many,” “many-to-one,” and “many-to-many” genotype–protein and genotype–phenotype relations. These processes include maternal and grandmaternal effects, phenotypic accommodation, alternative splicing, RNA editing, chimeric transcripts, protein multifunctionality, epistatic variance, the metabolic regulation of transcription, and post-translational modifications (Archer, Lavie, & Hill, Reference Archer, Lavie and Hill2018).
Therefore, because the genome does not have linear, predictive, or clear causal relations at the molecular level (e.g., protein species and function [Smith et al., Reference Smith, Agar, Chamot-Rooke, Danis, Ge, Loo and Kelleher2021]), it is illogical to posit that it has these relations with physiologic and psychosocial phenotypes. Yet without these relations heritability statistics and PGSs are meaningless numbers.
Causality and non-genetic inheritance
Phenotypic trajectories are sensitive to initial conditions and all subsequent states – from conception to senescence. At each stage of development, extra-cellular environments alter intra-cellular environments – which then alter gene expression in a recursive process. Thus, because humans inherit molecules, cells, and their biologic, physical, and social environments from their parents, no single level of analysis (e.g., molecular, cellular, organismal, geographic, or societal) can be considered ‘causal' unless the phenotypic changes at ‘lower' levels of biologic organization can be shown to persist at ‘higher' levels.
For example, the “egg” (primary oocyte) from which a human develops was initially created in the mother when she was a fetus developing in the grandmother's uterus. In other words, every “egg” that a female has was created prior to her birth. Thus, the physical and social environments in which a grandmother is immersed alter the intrauterine environment in which her offspring and the eggs of her female offspring develop. As such, physiologic, physical, and social environments impact the phenotypic development of at least three generations – the grandmother, her children, and her children's children – independent of the matrilineal genome.
These non-genetic processes of inheritance and evolution are known as “maternal and grandmaternal effects” and are well-established across species (Bateson et al., Reference Bateson, Barker, Clutton-Brock, Deb, D'Udine, Foley and Sultan2004; Gluckman, Hanson, Cooper, & Thornburg, Reference Gluckman, Hanson, Cooper and Thornburg2008; Maestripieri & Mateo, Reference Maestripieri and Mateo2009). For example, stunting and pediatric obesity are caused by adverse intrauterine environments – independent of genotype (Archer, Reference Archer2015a, Reference Archer2015b, Reference Archer2015c, Reference Archer2015d; Archer et al., 2023; Archer & Lavie, Reference Archer and Lavie2022; Archer & McDonald, Reference Archer, McDonald, Patel and Nielsen2017; Archer et al., Reference Archer, Lavie and Hill2018). In other words, starve any pregnant mammal and she will abort, or bear stunted offspring – independent of her genome.
Maternal effects and social outcomes
Importantly, disparities in egg (oocyte) quality and intrauterine environments can be caused by exposure to adverse physical and social environments (McQueen, Schufreider, Lee, Feinberg, & Uhler, Reference McQueen, Schufreider, Lee, Feinberg and Uhler2015; Navot et al., Reference Navot, Bergh, Williams, Garrisi, Guzman, Sandler and Grunfeld1991; Zhou et al., Reference Zhou, McQueen, Schufreider, Lee, Uhler and Feinberg2020). For example, low educational attainment, racism, sexism, and spousal abuse often lead to unremitting stress, poor nutrition, and alcohol, tobacco, or drug use that damage egg quality and the prenatal environments in which eggs develop. Thus, the environments and behaviors of past generations irreversibly alter the physiologic and behavioral phenotypes of current and future generations – independent of genotype.
Yet because the anatomic, physiologic, and psychological effects of prenatal insults are present at birth (inherited), persist into adulthood, and affect multiple generations, they are also inextricably linked to the matrilineal genome. Thus, gene-centric analyses that ignore maternal and grandmaternal effects will be misleading because of strong but demonstrably specious correlations between genotype and phenotype.
Given these facts, disparities in IQ, educational attainment, obesity, diabetes, physical activity, poverty, and criminal behavior in today's children are caused by the biologic, physical, and social environments in which their grandmothers and mothers were immersed from conception to senescence – independent of genomes and current environments. Thus, the adverse environments (and public policies) of yesteryear – not DNA – are causing disparities in health, wealth, and happiness today, and will continue to do so tomorrow.
Summary and conclusion
A great deal of biology – both established and undiscovered – links cellular and genetic inheritance (and the expression of that inheritance) with phenotypic development. Thus, estimates of genetic heritability and PGSs are often meaningless statistical abstractions derived from attempts to impose artificial dichotomies (nature vs. nurture and genes vs. environment) on demonstrably non-dichotomous developmental processes (Archer et al., Reference Archer, Lavie and Hill2018).
In closing, we agree with Burt's compelling critique and argue that an understanding of the etiology of physiologic, psychological, and social phenotypes “will not be found in the genome” (Archer, Reference Archer2015a).
“Challenging the utility of polygenic scores for social science” is an erudite and compelling critique. Yet despite the analytic rigor, Burt failed to capture the fundamental flaws that render ‘gene-centric' analyses misleading and polygenic scores (PGSs) meaningless outside of highly controlled environments.
The conflation of “inherited” with “genetic”
The functional unit in biology, biological inheritance, and phenotypic development is the cell – not DNA molecules. To be precise, humans develop from a single inherited cell – in which the genome is but one of many components. Moreover, because each cell's idiosyncratic nature and spatiotemporal context determine gene expression, the genome is merely an “organ” (McClintock, Reference McClintock1984, p. 800) or “tool” of the cell (Archer, Reference Archer2015a, Reference Archer2015b, Reference Archer2015c). This fact renders the distinction between non-genetic (cellular) inheritance and the two forms of genetic inheritance (nuclear and mitochondrial) critical to analyses of phenotypic development. Nevertheless, gene-centric analyses ignore the distinction between cellular, nuclear, and mitochondrial inheritance, and the fact that intra- and extra-cellular environments determine both gene products and their ‘effects.'
For example, the fundamental difference between monozygotic (identical) and dizygotic (fraternal) twins is inherent in the nomenclature – identical twins develop from a single cell (a fertilized egg) with a single placenta (usually); whereas fraternal twins develop from two different cells (two fertilized eggs) with two different placentas (always). Thus, fraternal twins differ in cellular and genetic inheritance, and their intrauterine environments, whereas identical twins do not.
Therefore, the greater phenotypic variability of fraternal twins is due to differences in gene expression engendered by different cells (eggs) acting in concert with inter-twin differences in both genotype and prenatal environments. Yet despite the extreme variability in the developmental competence (oocyte quality; e.g., mitochondrial content) of every female's population of eggs (Santos, El Shourbagy, & St John, Reference Santos, El Shourbagy and St John2006; Wang & Moley, Reference Wang and Moley2010; Zhang et al., Reference Zhang, Liu, Chen, Zhang, Zhang, Hao and Miao2020; Zhou et al., Reference Zhou, McQueen, Schufreider, Lee, Uhler and Feinberg2020) and the irreversible impact of the intrauterine environment on development (Archer, Reference Archer2015a, Reference Archer2015b, Reference Archer2015c, Reference Archer2015d; Archer & Lavie, Reference Archer and Lavie2022; Archer, Lavie, Dobersek, & Hill, Reference Archer, Lavie, Dobersek and Hill2023), the functional distinction between cellular, genetic, and environmental (in utero) inheritance is absent in ‘twin-studies,' heritability statistics, and polygenic scores (PGSs).
Thus, gene-centrism obscures the totality of biologic inheritance by conflating “genetic” with “inherited” – and therefore, the complexity of physiologic, psychological, and social phenotypic development remains unexamined.
Nonlinearity and development
Organismal development is a complex process that extends far beyond the linear amino acid sequence determined by the genetic code. For example, there are cellular, organismal, and environmental processes that lead to “one-to-many,” “many-to-one,” and “many-to-many” genotype–protein and genotype–phenotype relations. These processes include maternal and grandmaternal effects, phenotypic accommodation, alternative splicing, RNA editing, chimeric transcripts, protein multifunctionality, epistatic variance, the metabolic regulation of transcription, and post-translational modifications (Archer, Lavie, & Hill, Reference Archer, Lavie and Hill2018).
Therefore, because the genome does not have linear, predictive, or clear causal relations at the molecular level (e.g., protein species and function [Smith et al., Reference Smith, Agar, Chamot-Rooke, Danis, Ge, Loo and Kelleher2021]), it is illogical to posit that it has these relations with physiologic and psychosocial phenotypes. Yet without these relations heritability statistics and PGSs are meaningless numbers.
Causality and non-genetic inheritance
Phenotypic trajectories are sensitive to initial conditions and all subsequent states – from conception to senescence. At each stage of development, extra-cellular environments alter intra-cellular environments – which then alter gene expression in a recursive process. Thus, because humans inherit molecules, cells, and their biologic, physical, and social environments from their parents, no single level of analysis (e.g., molecular, cellular, organismal, geographic, or societal) can be considered ‘causal' unless the phenotypic changes at ‘lower' levels of biologic organization can be shown to persist at ‘higher' levels.
For example, the “egg” (primary oocyte) from which a human develops was initially created in the mother when she was a fetus developing in the grandmother's uterus. In other words, every “egg” that a female has was created prior to her birth. Thus, the physical and social environments in which a grandmother is immersed alter the intrauterine environment in which her offspring and the eggs of her female offspring develop. As such, physiologic, physical, and social environments impact the phenotypic development of at least three generations – the grandmother, her children, and her children's children – independent of the matrilineal genome.
These non-genetic processes of inheritance and evolution are known as “maternal and grandmaternal effects” and are well-established across species (Bateson et al., Reference Bateson, Barker, Clutton-Brock, Deb, D'Udine, Foley and Sultan2004; Gluckman, Hanson, Cooper, & Thornburg, Reference Gluckman, Hanson, Cooper and Thornburg2008; Maestripieri & Mateo, Reference Maestripieri and Mateo2009). For example, stunting and pediatric obesity are caused by adverse intrauterine environments – independent of genotype (Archer, Reference Archer2015a, Reference Archer2015b, Reference Archer2015c, Reference Archer2015d; Archer et al., 2023; Archer & Lavie, Reference Archer and Lavie2022; Archer & McDonald, Reference Archer, McDonald, Patel and Nielsen2017; Archer et al., Reference Archer, Lavie and Hill2018). In other words, starve any pregnant mammal and she will abort, or bear stunted offspring – independent of her genome.
Maternal effects and social outcomes
Importantly, disparities in egg (oocyte) quality and intrauterine environments can be caused by exposure to adverse physical and social environments (McQueen, Schufreider, Lee, Feinberg, & Uhler, Reference McQueen, Schufreider, Lee, Feinberg and Uhler2015; Navot et al., Reference Navot, Bergh, Williams, Garrisi, Guzman, Sandler and Grunfeld1991; Zhou et al., Reference Zhou, McQueen, Schufreider, Lee, Uhler and Feinberg2020). For example, low educational attainment, racism, sexism, and spousal abuse often lead to unremitting stress, poor nutrition, and alcohol, tobacco, or drug use that damage egg quality and the prenatal environments in which eggs develop. Thus, the environments and behaviors of past generations irreversibly alter the physiologic and behavioral phenotypes of current and future generations – independent of genotype.
Yet because the anatomic, physiologic, and psychological effects of prenatal insults are present at birth (inherited), persist into adulthood, and affect multiple generations, they are also inextricably linked to the matrilineal genome. Thus, gene-centric analyses that ignore maternal and grandmaternal effects will be misleading because of strong but demonstrably specious correlations between genotype and phenotype.
Given these facts, disparities in IQ, educational attainment, obesity, diabetes, physical activity, poverty, and criminal behavior in today's children are caused by the biologic, physical, and social environments in which their grandmothers and mothers were immersed from conception to senescence – independent of genomes and current environments. Thus, the adverse environments (and public policies) of yesteryear – not DNA – are causing disparities in health, wealth, and happiness today, and will continue to do so tomorrow.
Summary and conclusion
A great deal of biology – both established and undiscovered – links cellular and genetic inheritance (and the expression of that inheritance) with phenotypic development. Thus, estimates of genetic heritability and PGSs are often meaningless statistical abstractions derived from attempts to impose artificial dichotomies (nature vs. nurture and genes vs. environment) on demonstrably non-dichotomous developmental processes (Archer et al., Reference Archer, Lavie and Hill2018).
In closing, we agree with Burt's compelling critique and argue that an understanding of the etiology of physiologic, psychological, and social phenotypes “will not be found in the genome” (Archer, Reference Archer2015a).
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interest
None.