The target article is pitched as an endorsement of genetic causation, but is mostly concerned with discounting it. Start at the beginning: What do we actually know, with dead solid no-philosophy certainty, about the relationship between the human genome and complex behavioral phenotypes? We know there is a correlation between genetic and phenotypic similarity. In humans (as opposed to the farm animals for whom the concept was originally designed) the “heritability” of a trait is estimated as the unstandardized slope of the regression of phenotypic on genotypic similarity. I will refer to this as the “G-P correlation” to steer clear of the briar patch surrounding heritability per se.
Pairs of identical twins are more similar than pairs of fraternal twins, for more or less every behavioral trait. Genetically more similar pairs in a GREML matrix are more similar than less similar pairs, once again for every phenotype. (The magnitude of the GREML G-P correlation is usually much smaller than it is for twins; more about this below.) Although it depends on how you measure it, the G-P correlation is not small. Especially during the twin study era of the previous scientific generation, G-P correlations for the common objects of investigation – intelligence, personality, mental illness, problematic behavior like criminality – were estimated to be from 0.4 to 0.8.
All correlations are caused by something. Thinking about what might cause the G-P correlation of a behavioral phenotype focuses our attention on what we don't know. In the twin study era we did not know which genes were involved; without knowing the genes we could obviously not know anything about the direction, let alone the mechanism, of the genetic effect. Without the direction or the mechanism of the effect, we could know nothing about its potency or its scope. A G-P correlation could be caused by the hard biological consequences of rare genes of large effect, systems of polygenes operating on endophenotypes, violations of the equal environments assumption, uncontrollable correlations between genetic and environmental effects, gene–environment interactions, and so on. The activity of sorting through this hairball of causes, mostly without the benefit of experimental control, is called social science genomics. Social science has its virtues, and twin studies produced a great deal of social science, but twin studies were decidedly unsuccessful at identifying genetic (or for that matter environmental) causes.
My major difference with the authors of the target article involves their contention that this state of affairs was fundamentally changed by the completion of the human genome project and the development of genome-wide association studies (GWASs) and their attendant methods. Modern DNA-based genomics, to be sure, has provided the ability to conduct genetically informed social science in new ways, but these new methods, as well-documented by the target article, have all served to diminish our estimation of genetic causation. First, linkage and association studies showed that there are few big genes down there, at least not for behavioral traits in the normal range. Then GWASs showed there were no non-tiny genes down there, and that G-P correlations estimated among unrelated people were much smaller than twin correlations, closer to 0.2 than to 0.6. Then, in what is undeniably the most interesting social scientific method developed since the multivariate twin study, all of the within-family methods described in the target article reduced the causally relevant part of the G-P correlation for behavioral traits by at least half. The actionable part of the correlation, estimated as a real number in the form of a polygenic score (PGS), is less than that, under 5% for even the most intensely studied traits, with samples in the millions. Might there be some solid gold genetic causes down there somewhere in the remaining 5%? Maybe, but with a few exceptions we still don't know the which, why, how, or where of any of them.
Zuckerman (Reference Zuckerman1987) observed on these pages that “All parents are environmentalists until they have their second child.” I have three children, and although their parents happen to be concordant for educational attainment (EA), there remains some within-family variation. It seems reasonable that 5% of those differences (the metric doesn't quite apply, but you get the idea) are causally related to unknown, but certainly complex, genetic differences. If there is a big list somewhere called “causes of human differences in behavior,” genetic differences deserve to be on it. The nature side of the old debate, however, sometimes lures opponents into defending the absurd null hypothesis that the net causal genetic effect on human behavioral differences is zero. Does anyone actually believe this? I (third from the skeptical left in Fig. 1 of the target article) certainly don't. Block (Reference Block1995, second from left) doesn't deny genetic causation; in fact, just like the target article it is about genetic causation, and the inadequacy of heritability coefficients for quantifying it. Lewontin (Reference Lewontin1974, first on the left) was, after all, a geneticist: “The analysis of causes in human genetics is meant to provide us with the basic knowledge we require for correct schemes of environmental modification and intervention” (p. 525).
We can now see the outcome of the nature–nurture debate as regards human behavioral differences, and it is not what anyone expected. Genetic differences among humans do not determine behavioral differences. Although genetic differences aren't irrelevant to behavioral differences either, genetic causation of human behavior is weak, thin, contingent, gloomy, first-generation, call it what you will. That conclusion comes as no surprise. The surprise is that the crucial discoveries about the limitations of genetic causation were made not by environmentalists, left-leaning scientists, or philosophers of causation, but by the geneticists themselves. Once upon a time the so-called genetic revolution produced a paradigm shift in social science by showing that human genetic and behavioral differences are always correlated. Now it has produced a second shift, a counter-revolution, by showing that while G-P correlations have certain methodological consequences for social scientific practice, the direct causal genetic effects that underlie them are so small and indeterminate as to place few constraints on our individual or collective self-determination.
You have
Access
The target article is pitched as an endorsement of genetic causation, but is mostly concerned with discounting it. Start at the beginning: What do we actually know, with dead solid no-philosophy certainty, about the relationship between the human genome and complex behavioral phenotypes? We know there is a correlation between genetic and phenotypic similarity. In humans (as opposed to the farm animals for whom the concept was originally designed) the “heritability” of a trait is estimated as the unstandardized slope of the regression of phenotypic on genotypic similarity. I will refer to this as the “G-P correlation” to steer clear of the briar patch surrounding heritability per se.
Pairs of identical twins are more similar than pairs of fraternal twins, for more or less every behavioral trait. Genetically more similar pairs in a GREML matrix are more similar than less similar pairs, once again for every phenotype. (The magnitude of the GREML G-P correlation is usually much smaller than it is for twins; more about this below.) Although it depends on how you measure it, the G-P correlation is not small. Especially during the twin study era of the previous scientific generation, G-P correlations for the common objects of investigation – intelligence, personality, mental illness, problematic behavior like criminality – were estimated to be from 0.4 to 0.8.
All correlations are caused by something. Thinking about what might cause the G-P correlation of a behavioral phenotype focuses our attention on what we don't know. In the twin study era we did not know which genes were involved; without knowing the genes we could obviously not know anything about the direction, let alone the mechanism, of the genetic effect. Without the direction or the mechanism of the effect, we could know nothing about its potency or its scope. A G-P correlation could be caused by the hard biological consequences of rare genes of large effect, systems of polygenes operating on endophenotypes, violations of the equal environments assumption, uncontrollable correlations between genetic and environmental effects, gene–environment interactions, and so on. The activity of sorting through this hairball of causes, mostly without the benefit of experimental control, is called social science genomics. Social science has its virtues, and twin studies produced a great deal of social science, but twin studies were decidedly unsuccessful at identifying genetic (or for that matter environmental) causes.
My major difference with the authors of the target article involves their contention that this state of affairs was fundamentally changed by the completion of the human genome project and the development of genome-wide association studies (GWASs) and their attendant methods. Modern DNA-based genomics, to be sure, has provided the ability to conduct genetically informed social science in new ways, but these new methods, as well-documented by the target article, have all served to diminish our estimation of genetic causation. First, linkage and association studies showed that there are few big genes down there, at least not for behavioral traits in the normal range. Then GWASs showed there were no non-tiny genes down there, and that G-P correlations estimated among unrelated people were much smaller than twin correlations, closer to 0.2 than to 0.6. Then, in what is undeniably the most interesting social scientific method developed since the multivariate twin study, all of the within-family methods described in the target article reduced the causally relevant part of the G-P correlation for behavioral traits by at least half. The actionable part of the correlation, estimated as a real number in the form of a polygenic score (PGS), is less than that, under 5% for even the most intensely studied traits, with samples in the millions. Might there be some solid gold genetic causes down there somewhere in the remaining 5%? Maybe, but with a few exceptions we still don't know the which, why, how, or where of any of them.
Zuckerman (Reference Zuckerman1987) observed on these pages that “All parents are environmentalists until they have their second child.” I have three children, and although their parents happen to be concordant for educational attainment (EA), there remains some within-family variation. It seems reasonable that 5% of those differences (the metric doesn't quite apply, but you get the idea) are causally related to unknown, but certainly complex, genetic differences. If there is a big list somewhere called “causes of human differences in behavior,” genetic differences deserve to be on it. The nature side of the old debate, however, sometimes lures opponents into defending the absurd null hypothesis that the net causal genetic effect on human behavioral differences is zero. Does anyone actually believe this? I (third from the skeptical left in Fig. 1 of the target article) certainly don't. Block (Reference Block1995, second from left) doesn't deny genetic causation; in fact, just like the target article it is about genetic causation, and the inadequacy of heritability coefficients for quantifying it. Lewontin (Reference Lewontin1974, first on the left) was, after all, a geneticist: “The analysis of causes in human genetics is meant to provide us with the basic knowledge we require for correct schemes of environmental modification and intervention” (p. 525).
We can now see the outcome of the nature–nurture debate as regards human behavioral differences, and it is not what anyone expected. Genetic differences among humans do not determine behavioral differences. Although genetic differences aren't irrelevant to behavioral differences either, genetic causation of human behavior is weak, thin, contingent, gloomy, first-generation, call it what you will. That conclusion comes as no surprise. The surprise is that the crucial discoveries about the limitations of genetic causation were made not by environmentalists, left-leaning scientists, or philosophers of causation, but by the geneticists themselves. Once upon a time the so-called genetic revolution produced a paradigm shift in social science by showing that human genetic and behavioral differences are always correlated. Now it has produced a second shift, a counter-revolution, by showing that while G-P correlations have certain methodological consequences for social scientific practice, the direct causal genetic effects that underlie them are so small and indeterminate as to place few constraints on our individual or collective self-determination.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interest
None.