Scan barcode
jasonfurman's review against another edition
5.0
This was exactly the book I was looking for: a rigorous guide to the cutting edge of the emerging genetic-based social science that explains the methodologies scientists are using and also their limitations. The book is both hopeful about the future of research but also appropriately critical of the limitations of the study so far. Could not recommend it more highly—and I would not skip any of the footnotes or appendices which contain a lot of important insights, elaborations, and background. Reading the book left me excited for the future of the field but also frustrated about our ability to overcome the many inevitable limitations that they are so careful in expositing.
Dalton Conley and Jason Fletcher are both sociologists (and Dalton is also got a Ph.D. in genetics). The main point of their book is that genetic analysis is giving social scientists a powerful new tool to better understand the causes of differential educational attainment, incomes, poverty and other social phenomenon. But that we are still having a hard time linking particular genes to social outcomes let alone understanding the biological pathways and there is a huge amount of complex intersection between different genes and genes and the environment. Much of the book is about the social outcome of educational attainment, both because it is important but also because it is easier to study because is often included in the genetic data.
Conley and Fletcher start out by going through the research that establishes the high degree of heredity in many traits, including physical ones like height (80%) but also social ones like educational attainment (40%). They explain the models used to assess heredity, like comparing identical twins and fraternal twins or other well measured genetic distances. They explain in detail the methodology of the ACE model, the assumptions underlying it, why they thought it might be wrong, and how they ended up confirming it in their own research.
They then go on from the overall measures of the hereditary component of different traits to linking these to actual genes. Their take on the attempts to link behavior to single genes is that it was also spurious data mining, caused by the fact that while medical outcomes are well defined social outcomes have numerous measures and you can always find one correlated with the genes. The single gene efforts have given way to genome wide association studies (GWAS) that look at millions of genes and correlate them with outcomes at very high levels of significance to avoid data mining. Taken together, GWAS can produce a “polygenic score” that predicts a particular outcome, like educational attainment. GWAS, however, has three shortcomings: (1) you need to be careful to avoid the “chopsticks problem” (i.e., mistakenly inferring there is a gene for chopsticks when really is just a correlation with East Asian genes); (2) it gives up most any hope of a biological pathway because so many genes; and (3) it suffers from the “missing heredibility” problem of not explaining as much hereditability as know is there from twin studies. This last problem, they suggest, is caused by not analyzing the full genome for cost reasons.
They then go through a series of topics. One is genetic sorting where they examine the arguments in The Bell Curve about the increasing salience of genes and increased marriage based on them and thus locked in genetic stratification. Contrary to this the evidence they produce shows: (1) some traits are more genetically determined now than in the past (like height or weight, because most people have access to enough food so environment matters less) but others are less genetically determined now than in the past (like education, because of compulsory schooling laws); (2) people do mate based on phenotypes but much less on genotypes; and (3) no strong correlation between polygenic score for education and number of children.
They then have a chapter on race where they make the (familiar) point that there is much more genetic distance between most any two African tribes than between Europeans and Asians—which undermines many genetic concepts of race. This is because of the bottleneck of only ~1,000 people leaving Africa for Europe and Asia. They rebut the standard Stephen Jay Gould arguments against important racial differences (e.g., small genetic differences can matter a lot and evolution can happen quickly), but they establish that much of genetic variation is due to random drift not natural selection and that there is no links, and not really much research that could find a link, between race-based genetics and important social outcomes.
The chapter on the genetics of economic growth and war is a good literature review on the non-genetic studies in these areas but thin on the actual genetic studies, really just one for each topic. And finally the book concludes with a discussion of future “designer babies” either through embryo selection or gene editing, raising many concerns including that often “bad” traits are associated with “good” ones and have a benefit for the ecosystem as a whole so we will be taking risks.
Overall, I really appreciated that the book was research-based, did not just list discoveries but explained their methodology, and also that it was critical and skeptical throughout—but used that as an argument for more research not less.
Dalton Conley and Jason Fletcher are both sociologists (and Dalton is also got a Ph.D. in genetics). The main point of their book is that genetic analysis is giving social scientists a powerful new tool to better understand the causes of differential educational attainment, incomes, poverty and other social phenomenon. But that we are still having a hard time linking particular genes to social outcomes let alone understanding the biological pathways and there is a huge amount of complex intersection between different genes and genes and the environment. Much of the book is about the social outcome of educational attainment, both because it is important but also because it is easier to study because is often included in the genetic data.
Conley and Fletcher start out by going through the research that establishes the high degree of heredity in many traits, including physical ones like height (80%) but also social ones like educational attainment (40%). They explain the models used to assess heredity, like comparing identical twins and fraternal twins or other well measured genetic distances. They explain in detail the methodology of the ACE model, the assumptions underlying it, why they thought it might be wrong, and how they ended up confirming it in their own research.
They then go on from the overall measures of the hereditary component of different traits to linking these to actual genes. Their take on the attempts to link behavior to single genes is that it was also spurious data mining, caused by the fact that while medical outcomes are well defined social outcomes have numerous measures and you can always find one correlated with the genes. The single gene efforts have given way to genome wide association studies (GWAS) that look at millions of genes and correlate them with outcomes at very high levels of significance to avoid data mining. Taken together, GWAS can produce a “polygenic score” that predicts a particular outcome, like educational attainment. GWAS, however, has three shortcomings: (1) you need to be careful to avoid the “chopsticks problem” (i.e., mistakenly inferring there is a gene for chopsticks when really is just a correlation with East Asian genes); (2) it gives up most any hope of a biological pathway because so many genes; and (3) it suffers from the “missing heredibility” problem of not explaining as much hereditability as know is there from twin studies. This last problem, they suggest, is caused by not analyzing the full genome for cost reasons.
They then go through a series of topics. One is genetic sorting where they examine the arguments in The Bell Curve about the increasing salience of genes and increased marriage based on them and thus locked in genetic stratification. Contrary to this the evidence they produce shows: (1) some traits are more genetically determined now than in the past (like height or weight, because most people have access to enough food so environment matters less) but others are less genetically determined now than in the past (like education, because of compulsory schooling laws); (2) people do mate based on phenotypes but much less on genotypes; and (3) no strong correlation between polygenic score for education and number of children.
They then have a chapter on race where they make the (familiar) point that there is much more genetic distance between most any two African tribes than between Europeans and Asians—which undermines many genetic concepts of race. This is because of the bottleneck of only ~1,000 people leaving Africa for Europe and Asia. They rebut the standard Stephen Jay Gould arguments against important racial differences (e.g., small genetic differences can matter a lot and evolution can happen quickly), but they establish that much of genetic variation is due to random drift not natural selection and that there is no links, and not really much research that could find a link, between race-based genetics and important social outcomes.
The chapter on the genetics of economic growth and war is a good literature review on the non-genetic studies in these areas but thin on the actual genetic studies, really just one for each topic. And finally the book concludes with a discussion of future “designer babies” either through embryo selection or gene editing, raising many concerns including that often “bad” traits are associated with “good” ones and have a benefit for the ecosystem as a whole so we will be taking risks.
Overall, I really appreciated that the book was research-based, did not just list discoveries but explained their methodology, and also that it was critical and skeptical throughout—but used that as an argument for more research not less.
archytas's review against another edition
3.0
Conley's book is clearly articulated, with some excellent explanations of the science (especially in the appendices, which are worth the effort) and an elaboration of the emerging field of sociogenetics. Unfortunately - or fortunately - the clarity of the presentation just aided my growing sense that this discipline is presenting results as meaningful when they are not. Paradoxically, Conley is most persuasive when he is arguing against the use of sociogenetic analysis, and least when he is arguing for it.
Conley starts great, with an introduction explaining that who we are - our cognitive and personality traits if you like - and what happens to us - the way we experience the schooling system, our jobs, our health - are all a result of a complex interplay between our genetics and our environment. However, from here, it is mostly downhill, as instead of focusing on how that interaction works, Conley describes a field which primarily uses statistical techniques to correlate social outcomes to genetic profiles. Environmental factors are either ignored, or "controlled for" in ways which create imaginary realities, in which those measured are largely white and socially uniform, but conclusions are extrapolated to a world which is anything but.
Underpinning Conley's work is a set of unexamined assumptions about society, which I do not think hold up. This is manifested in a conflating of "success" with educational achievement, and intelligence with salary outcomes. This culminates in a bizarre sequence at the end of the book in which Conley (apocalyptically) imagines an upper-middle-class couple genetically engineering their children for maximum IQ because it will guarantee them a good job, even at the risk of blindness. The assumption that IQ is a better predictor of income than social privilege doesn't hold in our real world. Nor, frankly, does imagining any world in which IQ can be predicted from DNA alone.
So what happened to the dream, outlined at the beginning, of studying the interaction between genes and environment? The frustration is that apparently, sociogenomics doesn't look at interaction at all. Instead, it tries to separate the "environment" and "genetic" components, and estimate a numerical proportional contribution of each to the progress of a person's life. This actually seems to entrench the 'nature' ' nurture' divide, despite a body of research making it less and less tenable to see the world in those terms.
Conley reinforces this frequently through a joking "geneticists vs sociologists" meme which implies sociologists are barracking for social impacts, while geneticists want to disregard them. The times he does refer to interaction, he reinforces a genetic-determinist view, suggesting that someone may be genetically predetermined to choose particular environments.
At no point does Conley define the "environment". Understanding our relationship with bacteria within us, for example, challenges the simplistic idea of genetics vs society. Epigenetics does get a mention, but not in any depth. Recent advances in these fields are hugely significant, suggesting a myriad of subtle interactions shaping our bodies and brains. Cognitive plasticity, the way our brains change as a result of both physical changes and external events, also gets barely a discussion. Environment to Conley seems to be always determined by social policies and parenting techniques. At one point, he suggests that an equitable society could be one in which everything is fully heritable as if the environment is entirely controllable in predictable ways.
Instead, we are primarily in a world of mapping genes (or alleles) to outcomes, and extrapolating causality. Unsurprisingly, as Conley freely admits, it doesn't work very well most of the time. To summarise, before gene sequencing technologies, his sociogeneticist forebears came up with a proportional "heritability" percentage by comparing how similar outcomes between identical twins are to how similar outcomes between non-identical twins are. (I know, terrible sentence, Conley explains better).
Before I go any further, I just want to point out the biggest issue here, the assumption that this is a constant, across different social and natural environments. That is, in a relatively equal society, you would expect outcomes to be less affected by social inequality than in an unequal society. This could also apply to natural environments - if two districts have similar soil properties, the crop production may be more influenced by farmer. But if one district has uneven soil - some farms better than others - and the other is equal, the environment contribution will vary between them. So twin studies carried across differing social conditions are going to produce averages which may not represent the variety within them. Or in other words, aren't very meaningful. Where studies have only measured in constant conditions, those results can't meaningfully be extrapolated to other states.
With the advent of gene sequencing, sociogenomics moved to look first at particular genes or alleles, comparing them to outcomes. This produced very, very small heritability estimates. Nothing like the estimates twin studies had accustomed sociologists to. There are many reasons for this, as Conley freely admits. The most likely relates to the complexity of how genes translate into people. Even for most physical traits, hundreds of genes will contribute to a phenotype (e.g. hair colour). When it comes to cognitive traits - well, we don't even understand how our physical brains produce the result in most cases. It is likely to be thousands of genes contributing to something as simple as memory, for example, making a gene-by-gene approach improbable to provide much that is meaningful. But Conley and his colleagues are not trying to correlate to memory. They are going for things like school attendance and IQ tests. Which relate to a wide range of cognitive traits, each of which probably has thousands of genes. And, of course, we have social impacts and variations (IQ tests vary by government structures, for very basic starters). Every time Conley referred to college education as a "phenotype", I winced. Phenotype is used by biologists to represent a trait resulting from a genotype. School, which is a social outcome dependent on having access and a billion other things, is stretching the definition.
These studies are most successful where populations are relatively genetically homogenous, so differences on the genome are clearly delineated, and socially homogenous, so environmental factors are reduced. It should be unsurprising that Scandinavia, which keeps meticulous records of adoptees, is a significant source of the more successful studies, which Conley describes as a "godsend". However, these studies tell us very little about anything outside of Scandinavia, for precisely those reasons. This, unfortunately, does not stop these studies being cited broadly as universal human outcomes. Conley to his credit does not do this (much), and the clarity of his explanation made me better understand why it is so problematic.
In any case, to move beyond the difficulties of specific studies, genome-wide studies are becoming more common. These crunch vast amounts of data, looking for correlations across the entire genomes of multiple individuals. The size of the datasets makes finding patterns easier, but the methodology is rife with risk. If you go looking for any correlation in a large data set, you will find random relationships. It is for precisely this reason that scientific funding bodies increasingly regulate against studies which don't have a clear hypothesis to test. It is not impossible to use these techniques responsibly, but academic structures, which reward positive novel findings over negative, provide strong incentives to take shortcuts. Conley describes the safeguards in place, but risk remains.
Even with these techniques, the results provide lower heritability percentages than twin studies had predicted.
It isn't that this kind of research is unique to this field. Schizophrenia, for example, is a mental illness diagnosed based on symptoms whose physical causes are not well understood, and almost certainly have a complex web of genetic contributors. Statistical correlation studies are used to try to develop a better understanding of this disease, despite all the same difficulties and problems. The risks here, however, may be better balanced against potential gains. Schizophrenia causes enormous suffering, and identifying possible genetic correlations may help identify the neurological processes associated with it. This is a more interactive process than a crude guess about cause and effect.
These studies have another problem, however, which is that with very large datasets, it becomes difficult to control for known social factors adequately. Conley discusses this in terms of the Chopsticks problem - an actual study which purported to have found a gene correlation to chopsticks use before realising it correlated with south-east populations much more likely to be taught to use chopsticks as children.
Most of the arguments I've put here, Conley covers himself in some form in his chapters on race and global inequality.
Here Conley starts raising for the first time that we do know a lot about the impact of some environmental measures on generating inequality. In the lead up to the first quote above, Conley talks about the impossibility of finding an area to study African-American school outcomes where known factors like poor school funding, health disparities and racism do not obliterate any other differences.
Similarly, Conley deconstructs the various attempts to find a genetic basis for global inequality, pointing and efficiently and effectively to the number of possible interpretations for the data, and the impossibility of narrowing it down. For example,
Conley's deconstruction of racist interpretations is welcome. It is particularly necessary given the resurgence of racist science circles based around arguments for hypothetical genetic differences in intelligence between population groups. It hovers over the entire field, given race is a substantial predictor of exactly the factors Conley equates with success - educational attainment, IQ score and absence of incarceration. One of the biggest dangers in heading into genome-wide studies is that correlations caused by clearly understood social factors - easily summarised as 'racism' (personal and systemic) - become used to justify and perpetuate inequality, rather than dismantle it.
In other areas, however, Conley fails to discuss the specifics of social factors at all. While he pays lip service to their importance, this tends to trivialise or dismiss them. For example, this quote took my breath away a little "Someone who is highly cognitively endowed and able to earn lots of money could choose a stay-at-home spouse who contributes other abilities to the household, such as empathy and care-taking ability." There is a set of assumptions about both how high salaries work, and the reasons that caregivers choose to take on the role, which are not discussed here at all. The impact that having children has on the careers of women, their pay, and their subsequent options for domestic divisions of work, are extensively studied. But to Conley, it seems, smart men earn money, and empathetic women are driven to domesticity.
Conley doesn't discuss the disproportionate exposure to violence that some communities have, and how trauma effects precisely the outcomes he is measuring. Similarly, in discussing spousal choice, Conley fails to identify one of the most significant driving factors - the social stratification between college students and others. It would be laughable if it weren't so serious.
And it is a bit serious. Conley starts to speculate about what this research - which by his own account has yet to produce any persuasive evidence of anything - might mean for policy. What if we start streaming kids by genetics, based on teeny tiny effect sizes? Conley may argue to exclude race from these calculations, but it is hard to see that happening in practice. What if dubious genetic associations are used in recruitment? Even at the more medical end, what if we determine medical treatment based on a genetic association, which is still a generalisation which statistically will be wrong a proportion of the time?
In theory, this shouldn't happen because the science doesn't justify it. In practice, however, it is likely for precisely the reason that sociogenomics is attracting much more Foundation and philanthropic funding than more proven research approaches to the impact of inequality. That is that power, who has it and who wants to retain it, is not threatened by science which can identify people's failures in their genes, while it is by research which indicates that we need to redistribute what we have.
To be clear here, I am not a reader who believes genetics have no impact on our cognition or motivations. In fact, I am fascinated by the complexity that makes us *us*, that produces such an amazing array of people, no two of whom are that much alike. I want this science to be better, to tell us things. But the more I engage with it, the more convinced I become that it is going nowhere useful. Nothing here is very meaningful, and it seems that the researchers themselves want a simpler reality than the one we live in. More dangerously, this research seems to be coming at the expense of examining well-defined social analysis, and it is doing so as we see a resurgence in white supremacy, including among governments looking for ways to justify closed borders and ongoing inequality. It scares me far more than a hypothetical future IQ/blindness trade-off.
Conley starts great, with an introduction explaining that who we are - our cognitive and personality traits if you like - and what happens to us - the way we experience the schooling system, our jobs, our health - are all a result of a complex interplay between our genetics and our environment. However, from here, it is mostly downhill, as instead of focusing on how that interaction works, Conley describes a field which primarily uses statistical techniques to correlate social outcomes to genetic profiles. Environmental factors are either ignored, or "controlled for" in ways which create imaginary realities, in which those measured are largely white and socially uniform, but conclusions are extrapolated to a world which is anything but.
Underpinning Conley's work is a set of unexamined assumptions about society, which I do not think hold up. This is manifested in a conflating of "success" with educational achievement, and intelligence with salary outcomes. This culminates in a bizarre sequence at the end of the book in which Conley (apocalyptically) imagines an upper-middle-class couple genetically engineering their children for maximum IQ because it will guarantee them a good job, even at the risk of blindness. The assumption that IQ is a better predictor of income than social privilege doesn't hold in our real world. Nor, frankly, does imagining any world in which IQ can be predicted from DNA alone.
So what happened to the dream, outlined at the beginning, of studying the interaction between genes and environment? The frustration is that apparently, sociogenomics doesn't look at interaction at all. Instead, it tries to separate the "environment" and "genetic" components, and estimate a numerical proportional contribution of each to the progress of a person's life. This actually seems to entrench the 'nature' ' nurture' divide, despite a body of research making it less and less tenable to see the world in those terms.
Conley reinforces this frequently through a joking "geneticists vs sociologists" meme which implies sociologists are barracking for social impacts, while geneticists want to disregard them. The times he does refer to interaction, he reinforces a genetic-determinist view, suggesting that someone may be genetically predetermined to choose particular environments.
At no point does Conley define the "environment". Understanding our relationship with bacteria within us, for example, challenges the simplistic idea of genetics vs society. Epigenetics does get a mention, but not in any depth. Recent advances in these fields are hugely significant, suggesting a myriad of subtle interactions shaping our bodies and brains. Cognitive plasticity, the way our brains change as a result of both physical changes and external events, also gets barely a discussion. Environment to Conley seems to be always determined by social policies and parenting techniques. At one point, he suggests that an equitable society could be one in which everything is fully heritable as if the environment is entirely controllable in predictable ways.
Instead, we are primarily in a world of mapping genes (or alleles) to outcomes, and extrapolating causality. Unsurprisingly, as Conley freely admits, it doesn't work very well most of the time. To summarise, before gene sequencing technologies, his sociogeneticist forebears came up with a proportional "heritability" percentage by comparing how similar outcomes between identical twins are to how similar outcomes between non-identical twins are. (I know, terrible sentence, Conley explains better).
Before I go any further, I just want to point out the biggest issue here, the assumption that this is a constant, across different social and natural environments. That is, in a relatively equal society, you would expect outcomes to be less affected by social inequality than in an unequal society. This could also apply to natural environments - if two districts have similar soil properties, the crop production may be more influenced by farmer. But if one district has uneven soil - some farms better than others - and the other is equal, the environment contribution will vary between them. So twin studies carried across differing social conditions are going to produce averages which may not represent the variety within them. Or in other words, aren't very meaningful. Where studies have only measured in constant conditions, those results can't meaningfully be extrapolated to other states.
With the advent of gene sequencing, sociogenomics moved to look first at particular genes or alleles, comparing them to outcomes. This produced very, very small heritability estimates. Nothing like the estimates twin studies had accustomed sociologists to. There are many reasons for this, as Conley freely admits. The most likely relates to the complexity of how genes translate into people. Even for most physical traits, hundreds of genes will contribute to a phenotype (e.g. hair colour). When it comes to cognitive traits - well, we don't even understand how our physical brains produce the result in most cases. It is likely to be thousands of genes contributing to something as simple as memory, for example, making a gene-by-gene approach improbable to provide much that is meaningful. But Conley and his colleagues are not trying to correlate to memory. They are going for things like school attendance and IQ tests. Which relate to a wide range of cognitive traits, each of which probably has thousands of genes. And, of course, we have social impacts and variations (IQ tests vary by government structures, for very basic starters). Every time Conley referred to college education as a "phenotype", I winced. Phenotype is used by biologists to represent a trait resulting from a genotype. School, which is a social outcome dependent on having access and a billion other things, is stretching the definition.
These studies are most successful where populations are relatively genetically homogenous, so differences on the genome are clearly delineated, and socially homogenous, so environmental factors are reduced. It should be unsurprising that Scandinavia, which keeps meticulous records of adoptees, is a significant source of the more successful studies, which Conley describes as a "godsend". However, these studies tell us very little about anything outside of Scandinavia, for precisely those reasons. This, unfortunately, does not stop these studies being cited broadly as universal human outcomes. Conley to his credit does not do this (much), and the clarity of his explanation made me better understand why it is so problematic.
In any case, to move beyond the difficulties of specific studies, genome-wide studies are becoming more common. These crunch vast amounts of data, looking for correlations across the entire genomes of multiple individuals. The size of the datasets makes finding patterns easier, but the methodology is rife with risk. If you go looking for any correlation in a large data set, you will find random relationships. It is for precisely this reason that scientific funding bodies increasingly regulate against studies which don't have a clear hypothesis to test. It is not impossible to use these techniques responsibly, but academic structures, which reward positive novel findings over negative, provide strong incentives to take shortcuts. Conley describes the safeguards in place, but risk remains.
Even with these techniques, the results provide lower heritability percentages than twin studies had predicted.
It isn't that this kind of research is unique to this field. Schizophrenia, for example, is a mental illness diagnosed based on symptoms whose physical causes are not well understood, and almost certainly have a complex web of genetic contributors. Statistical correlation studies are used to try to develop a better understanding of this disease, despite all the same difficulties and problems. The risks here, however, may be better balanced against potential gains. Schizophrenia causes enormous suffering, and identifying possible genetic correlations may help identify the neurological processes associated with it. This is a more interactive process than a crude guess about cause and effect.
These studies have another problem, however, which is that with very large datasets, it becomes difficult to control for known social factors adequately. Conley discusses this in terms of the Chopsticks problem - an actual study which purported to have found a gene correlation to chopsticks use before realising it correlated with south-east populations much more likely to be taught to use chopsticks as children.
Most of the arguments I've put here, Conley covers himself in some form in his chapters on race and global inequality.
"But our sibling difference methods—or any other approach so far developed—can not separate the more context-independent (i.e., nonsocially mediated) biological effects from genetic effects that interact with the social system, such as when lighter skin is rewarded. That is, it could be that cognitive differences are genetically based, but the mechanism linking genes to IQ acts through social pathways (i.e., response to skin tone) rather than biological ones (i.e., brain structure). ... The near impossibility of a definitive, scientific approach as outlined above stands in stark contrast to the loose claims of pundits or scholars who assert that there is a genetic explanation for the black-white test score gap. In walking through the logic of genetic methods, we believe this discussion provides a cautionary tale for how scientists should proceed (or not) with investigations that combine questions of race, genetics, and socioeconomic or cognitive outcomes. With the outpouring of genetic data we are witnessing in society today, there will no doubt be further ventures in this direction. Clear efforts are therefore needed to sharpen the scientific questions that can be answered and also to guard against repeating past instances of pseudoscientific racism relying on ideologically motivated inferences from inadequate evidence."
Here Conley starts raising for the first time that we do know a lot about the impact of some environmental measures on generating inequality. In the lead up to the first quote above, Conley talks about the impossibility of finding an area to study African-American school outcomes where known factors like poor school funding, health disparities and racism do not obliterate any other differences.
Similarly, Conley deconstructs the various attempts to find a genetic basis for global inequality, pointing and efficiently and effectively to the number of possible interpretations for the data, and the impossibility of narrowing it down. For example,
"Even if we assume that the associations between genetic diversity and economic development show a cause-and-effect relationship, the interpretation of the finding is far from obvious. Although the authors have their preferred interpretation, it is not possible to discard many alternative interpretations. For example, that genetic diversity is correlated with country-level racial and ethnic composition. The potential for these correlations means it is difficult to know whether the links between genetic diversity and country success are tied to Ashraf and Galor’s theoretical ideas or are capturing other processes—such as histories of colonization, war, natural resource exploitation, and so on—that are tied up with race and racism.
Conley's deconstruction of racist interpretations is welcome. It is particularly necessary given the resurgence of racist science circles based around arguments for hypothetical genetic differences in intelligence between population groups. It hovers over the entire field, given race is a substantial predictor of exactly the factors Conley equates with success - educational attainment, IQ score and absence of incarceration. One of the biggest dangers in heading into genome-wide studies is that correlations caused by clearly understood social factors - easily summarised as 'racism' (personal and systemic) - become used to justify and perpetuate inequality, rather than dismantle it.
In other areas, however, Conley fails to discuss the specifics of social factors at all. While he pays lip service to their importance, this tends to trivialise or dismiss them. For example, this quote took my breath away a little "Someone who is highly cognitively endowed and able to earn lots of money could choose a stay-at-home spouse who contributes other abilities to the household, such as empathy and care-taking ability." There is a set of assumptions about both how high salaries work, and the reasons that caregivers choose to take on the role, which are not discussed here at all. The impact that having children has on the careers of women, their pay, and their subsequent options for domestic divisions of work, are extensively studied. But to Conley, it seems, smart men earn money, and empathetic women are driven to domesticity.
Conley doesn't discuss the disproportionate exposure to violence that some communities have, and how trauma effects precisely the outcomes he is measuring. Similarly, in discussing spousal choice, Conley fails to identify one of the most significant driving factors - the social stratification between college students and others. It would be laughable if it weren't so serious.
And it is a bit serious. Conley starts to speculate about what this research - which by his own account has yet to produce any persuasive evidence of anything - might mean for policy. What if we start streaming kids by genetics, based on teeny tiny effect sizes? Conley may argue to exclude race from these calculations, but it is hard to see that happening in practice. What if dubious genetic associations are used in recruitment? Even at the more medical end, what if we determine medical treatment based on a genetic association, which is still a generalisation which statistically will be wrong a proportion of the time?
In theory, this shouldn't happen because the science doesn't justify it. In practice, however, it is likely for precisely the reason that sociogenomics is attracting much more Foundation and philanthropic funding than more proven research approaches to the impact of inequality. That is that power, who has it and who wants to retain it, is not threatened by science which can identify people's failures in their genes, while it is by research which indicates that we need to redistribute what we have.
To be clear here, I am not a reader who believes genetics have no impact on our cognition or motivations. In fact, I am fascinated by the complexity that makes us *us*, that produces such an amazing array of people, no two of whom are that much alike. I want this science to be better, to tell us things. But the more I engage with it, the more convinced I become that it is going nowhere useful. Nothing here is very meaningful, and it seems that the researchers themselves want a simpler reality than the one we live in. More dangerously, this research seems to be coming at the expense of examining well-defined social analysis, and it is doing so as we see a resurgence in white supremacy, including among governments looking for ways to justify closed borders and ongoing inequality. It scares me far more than a hypothetical future IQ/blindness trade-off.
More...