This essay is part of Legacies of Eugenics, a series of essays by leading thinkers devoted to exploring the history of eugenics, and the ways in which it lives on in our habits of thought, and in aspects of the sciences and medicine.
¤
IN THE Operation Varsity Blues scandal of 2019, 50 wealthy parents were charged with trying to get their children into elite universities through fraudulent means. The story dramatically demonstrated the lengths to which some parents will go to ensure their children’s acceptance into places like Stanford, Yale, Georgetown, and USC. Actress Lori Loughlin and her husband, fashion designer Mossimo Giannulli, bribed athletic coaches to recruit their children for sports they did not play. Actress Felicity Huffman and private equity mogul William E. McGlashan Jr. were among the parents who paid to falsify their children’s SAT and ACT test scores. Those who were caught faced criminal charges, yet the scandal also shed light on the perfectly legal tactics used by wealthy parents to get their children into elite institutions, such as endowing buildings or hiring expensive consultants.
The Pennsylvania couple Malcolm and Simone Collins have taken a different approach. For their two daughters, Titan Invictus and Industry Americus, the Collinses used eugenics. Titan and Industry are both under three years old, so it is too early to tell whether the experiment worked, but they were conceived through in vitro fertilization (IVF) using a process known as polygenic embryo screening (PES). Unlike preimplantation genetic diagnosis (PGD), which IVF clinics have long used to test embryos for chromosomal and single-gene disorders known to run in the families of prospective parents, PES examines single-nucleotide polymorphisms from across the genome to calculate lifetime risk of complex diseases and propensities for particular social outcomes, including expected educational attainment. A study published in 2023 found that 38 percent of Americans would strongly consider using PES to increase the chances of their child being admitted to a top-100 college or university if they were doing IVF anyway and if PES were available for free, which it currently is not: fees begin around $1,000, plus $400 per embryo (on top of the cost of IVF itself).
In addition to their four children (Titan and Industry have two older brothers, Octavian George and Torsten Savage), the Collinses have 34 embryos on ice. After selection through a similar PES process, they plan to implant one every two or three years until Simone’s uterus literally ruptures and needs to be removed. Malcolm and Simone are part of a new pronatalist movement that encourages elite couples to have as many children as possible to stave off what they and others—including, most prominently, Elon Musk—see as an impending population collapse. According to Musk, “If people don’t have more children, civilisation is going to crumble,” going out “with a whimper in adult diapers.” Or, in the hyperbolic words of Malcolm Collins, “If [humanity] was an animal species it would be called endangered.” Yet pronatalists seem just as concerned about who is having babies. After all, birth rates are still high in some of the world’s poorer countries. For the Collinses, pronatalism is an alternative to immigration. Indeed, some of today’s most ardent pronatalists are also staunch anti-immigrationists.
While critics have paid considerable attention to the overt and covert racism of today’s pronatalism, they have focused much less on the growing enthusiasm for genetically engineered children, not just among pronatalists but among the tech elite more broadly. OpenAI CEO Sam Altman is an investor in Genomic Prediction, the PES company used by the Collinses. Noor Siddiqui, the founder and CEO of another PES company, Orchid Health, has boasted of her company’s numerous Silicon Valley clients. Many of them, including Siddiqui herself, use IVF not due to any fertility issues but in order to control the DNA of their future children.
Even among advocates of PES, debate rages about whether selecting a child on the basis of its predicted educational attainment is actually eugenics. According to Malcolm Collins, “it’s completely different.” He defines eugenics as “state-sponsored selective breeding to influence the dominance of certain genes,” in contrast to PES, which “us[es] technology to give parents the choice over which traits they value most.” Yet according to the evolutionary psychologist Diana Fleischman, a fierce advocate for PES who shared a stage with the Collinses at the 2023 Natal Conference in Austin, Texas, it’s all eugenics. Fleischman contends that selecting an embryo according to its predicted educational attainment is no different from selecting one according to its number of chromosomes. “We are all eugenicists,” she says, and we might as well embrace it.
Americans today typically conflate eugenics with racism, genocide, coercive sterilization, and ableism. This popular image, however, misses two key features of eugenics as it has existed in the United States since its importation from England at the turn of the 20th century. The first is that intelligence—not race—has always been at the center of American eugenics. Historically, eugenics and racism have operated in tandem, but neither is reducible to the other. Eugenics attributes socioeconomic inequality—both within and between racially defined groups—to varying levels of intelligence, which it defines as a biological quality shaped largely by our DNA. The second key feature is that, although the eugenics movement was responsible for the legalization of involuntary sterilization in a majority of US states, eugenics has depended primarily on individuals policing the gene pools of their own families, not on government intervention.
But eugenics does not work by breeding smarter humans. No genetic intervention, not even PES, has been shown to do this. Rather, eugenics works by naturalizing socioeconomic inequality and generating support for policies that enhance the life chances of those at the top of the social hierarchy and reducing the life chances of those at the bottom.
¤
Early in the 20th century, American eugenicists attributed poverty and its various outcomes to the presumed low intelligence of the poor. There was, however, no consensus about what exactly intelligence was. Francis Galton, coiner of the term eugenics, considered socioeconomic status a proxy for intelligence, believing that people who were intelligent would be good at socially valuable activities and would thus be rewarded monetarily. American psychologists followed his lead, calibrating intelligence tests to correlate with educational success and socioeconomic status. In 1923, the psychologist Edwin G. Boring, with no sense of irony, defined intelligence as “what intelligence tests test.” Intelligence testing became a big business after World War I, with universities and employers adopting intelligence tests as gatekeeping mechanisms for entering the middle class, making the correlation among IQ, educational attainment, and socioeconomic status a self-fulfilling prophecy. Today’s psychologists defend the validity of IQ tests by pointing to their ability to predict education and income, but none of these factors is independent of the others.
Early American eugenicists adopted a Mendelian model for the inheritance of intelligence, conceiving of it in binary terms: individuals were either normal or feebleminded, with “feeblemindedness” serving as a catchall term for those who appeared to have low intelligence or who exhibited behaviors that were thought to stem from low intelligence, such as poverty, crime, or promiscuity. Eugenicists recommended that feebleminded individuals be removed from the gene pool, through either sterilization or institutionalization.
There was an obvious problem with this proposal, however. Eugenicists understood feeblemindedness to be a recessive characteristic. This meant that an individual would be feebleminded only if they had inherited the feeblemindedness gene from both parents, while two normal parents could have feebleminded children if each carried a single copy of the gene. There was no way to tell which seemingly normal people harbored this deleterious allele because it doesn’t exist (nor does “feeblemindedness” as it was understood at the time). Certain genetic variants can reduce an individual’s cognitive capacity, but these could not be tested for in the 1910s, and the concept of feeblemindedness was much more capacious than cognitive impairment. Perhaps more importantly, though, public opinion in a democratic society like the United States likely would not have tolerated the sterilization of people who did not themselves appear to be feebleminded. Instead, eugenicists had to rely on individuals to reproduce appropriately. They therefore encouraged men to track their family histories and those of their prospective mates to minimize the probability of having feebleminded children. The Eugenics Record Office (ERO) attempted to facilitate this process by compiling genealogies of traits for anyone who would contribute family data, aiming to produce a nationwide genetic surveillance system.
By 1920, however, eugenicists had adopted a polygenic model of intelligence, one that underpins today’s polygenic embryo selection. Under this model, intelligence is akin to height: it can vary in nearly infinite gradations from very low to very high, and is influenced by numerous genes. Smarter individuals are thought to have more brightness-producing alleles and fewer dullness-producing alleles. This model turned IQ tests into direct tests of genetic quality. Eugenicists no longer had to rely on sterilization or on extensive genealogies such as those collected by the ERO. Under the polygenic model of intelligence, they could simply encourage more intelligent couples to have more children and less intelligent couples to have fewer. Beginning in the 1930s, the American Eugenics Society (AES) recommended that the government use tax policies to reduce the cost of childbearing for wealthier couples and thereby promote larger families among those it considered genetically superior. The majority of its efforts, however, focused on the private sector, where the AES encouraged teachers, clergy, and medical professionals to enact mechanisms of social control to ensure that couples had the number of children commensurate with their presumed genetic quality. The AES continued to promote this neoliberal and biomedical model of eugenics—which relied on couples making their own decisions, often in a medical setting—through the end of the 20th century, even after it changed its name to the Society for the Study of Social Biology and then to the Society for Biodemography and Social Biology.
During the Civil Rights Movement of the 1950s and ’60s, the race and class biases of intelligence testing came under scrutiny. Some localities—including New York, Philadelphia, Washington, DC, and the entire state of California—banned intelligence testing as a means of assigning students to remedial classes, as these classes typically served to warehouse students of color rather than remediating educational deficits. At the same time, scientists in a new subfield of psychology called behavior genetics, intimately entangled with the AES, turned to correlational (nongenetic) studies of twins and adoptees to demonstrate that intelligence (and every other human characteristic or social outcome) had a strong basis in biological heredity. Behavior geneticists like Arthur Jensen and Richard Herrnstein argued that studies showed the observed IQ gaps between Black and white Americans had genetic origins and were therefore immutable.
In 1970, the Nobel Prize–winning physicist William B. Shockley proposed a eugenic sterilization program that would have paid individuals who agreed to be sterilized $1,000 per IQ point below 100. Although this program was, on its surface, race-neutral, Shockley pointed out that one of its supposed benefits would be that approximately 85 percent of African Americans would qualify for a sterilization payment. Shockley and Jensen were part of a cohort of scientists sponsored by the overtly racist Pioneer Fund with the aim of mobilizing science to resegregate American public education in the wake of the 1954 Supreme Court opinion Brown v. Board of Education. When Herrnstein, together with the political scientist Charles Murray, published The Bell Curve: Intelligence and Class Structure in American Life in 1994, they disproportionately cited racist studies funded by the Pioneer Fund to support their calls for eviscerating the welfare state. Socioeconomic inequality, they said, reflected genetic diversity and therefore could not be overcome through social policy.
By that time, the eugenicist idea that intelligence had a strong genetic basis was gaining public support in the burgeoning fertility industry. Obstetricians and gynecologists had long sought to use their positions for eugenic ends, often pressuring poor patients and patients of color to undergo sterilization. Those who worked with infertile patients typically recruited sperm donors with presumed high intelligence. Doctors with particularly high opinions of themselves sometimes used their own sperm, usually unbeknownst to their patients. In 1979, the businessman Robert Klark Graham, inspired by the Nobel Prize–winning geneticist (and eugenicist) Hermann Muller, opened a sperm bank in Escondido, California—colloquially known as the Genius Sperm Bank—whose aim was to provide the sperm of Nobel laureates to highly intelligent white women in heterosexual marriages. Up to that point, donor insemination had typically involved fresh sperm, and patients had little choice in the matter of whose sperm they used. At Graham’s Repository for Germinal Choice, couples could take eugenics into their own hands, choosing from a color-coded catalog of sperm. Donors underwent no genetic testing; rather, the Nobel Prize itself served as supposed evidence of genetic quality, never mind that the prize is often awarded late in life and donors’ sperm might thus be of poorer quality. Ultimately, Graham was able to secure the sperm of very few Nobel laureates—Shockley was the only one who publicly admitted to having donated (as a septuagenarian)—so Graham broadened his search criteria to include less-decorated scientists and self-made businessmen.
The Repository for Germinal Choice closed its doors in 1999, having produced about 200 babies, though none from Nobel Prize donors. By that point, however, the fertility industry had embraced Graham’s consumer-choice model. One of his assistants opened the California Cryobank, which quickly became the country’s leading sperm bank, locating branches in college towns for easy access to donors with elite educations. Other sperm banks quickly followed suit. IVF was also available by then, making egg donation possible, and eggs from well-educated donors typically fetched a premium, reflecting broad acceptance of the idea that intelligence is encoded in our DNA. This belief generated support for the $2.7 billion Human Genome Project, which promised to unlock the secrets not just of the human body but also of the human mind.
As late as 2012, however, scientists had found no specific genes with any bearing on intelligence in the normal range, and no variants that increase intelligence. This was not for lack of trying. Genomic Prediction co-founders Stephen Hsu and Laurent Tellier had both worked on a project at BGI (formerly the Beijing Genomics Institute) aimed at identifying the genetic correlates of high intelligence, which was ultimately unsuccessful. Couples using donor sperm or eggs could choose gametes based on their donors’ intelligence or education, but those doing IVF with their own gametes could not yet select embryos according to the expected intelligence of the resulting children.
This changed in 2013, with the publication of the first genome-wide association study (GWAS) of educational attainment. A GWAS assesses millions of single-nucleotide polymorphisms (SNPs, pronounced “snips”) across the genome, testing each for a statistical correlation with the outcome in question—in this case, years of schooling completed. The process produces a formula to calculate individual polygenic scores (PGS), which represent an individual’s genomic propensity for experiencing the given outcome. Because GWAS assesses so many correlations, and because each one is expected to be minuscule, GWAS requires enormous sample sizes. The Social Science Genetic Association Consortium (SSGAC)—the organization behind the educational GWAS—settled on educational attainment as its outcome of interest because there simply weren’t enough people who had been both genotyped and intelligence-tested. Nearly everyone who had provided a sample for genomic research, however, had also reported their highest level of education.
A major problem with GWAS is that the process is informed by and perpetuates racist assumptions about the structure of human genetic diversity. To minimize spurious findings, GWAS is typically limited to samples that are thought to be ancestrally homogeneous. GWAS of educational attainment has only been done on individuals with exclusively European genetic ancestry. Yet the very idea of European (or African or Asian) genetic ancestry depends on the belief that continental boundaries divide human genetic diversity into natural categories, which is not the case. The human genome varies gradually across space, without sharp discontinuities. No method of segmenting human populations is more natural than any other. And yet genomic reference samples were collected to match the continental structure of the race concept in the United States, which was constructed through hundreds of years of social, political, and economic discrimination, including genocide and enslavement. This sampling strategy perpetuates the dangerously incorrect idea that humans come in a distinct set of natural varieties that correspond to continental race categories. Because GWAS of educational attainment includes only individuals who are genetically proximate to a “European” reference sample, it is a bit like a drunk looking for his keys under the lamppost, not because that is where he dropped them but because that is where the light is. GWAS will only find signals that are present in the reference sample, which means it will produce better predictions for individuals who are more genetically similar to the reference sample than less so.
The results of the first GWAS of educational attainment were not impressive. The study identified three variants with a statistically significant relationship to educational attainment, each of which increased schooling by approximately one month. The PGS constructed from this GWAS accounted for about two percent of variance in educational attainment among white Americans, leaving 98 percent unexplained. Nevertheless, scientists wrote breathless op-eds and books for popular audiences presenting the SSGAC’s GWAS of educational attainment as evidence that genes play a decisive role in determining a person’s intelligence and therefore their position in the social order. These works naturalized socioeconomic inequality, contending that the poor have not been exploited by the capitalist system but rather received an unlucky draw in the “genetic lottery.”
Genomic Prediction’s Stephen Hsu and Princeton University biosociologist Dalton Conley both published articles in Nautilus envisioning a future in which the average IQ increased dramatically through the widespread use of PES aimed at maximizing the PGS for educational attainment. Nick Bostrom, founder of the Future of Humanity Institute—which was recently dubbed “eugenics on steroids”—contended that PES could increase global human capital. Hsu has even suggested that parents may soon use gene-editing tools like CRISPR to alter embryos to increase their predicted intelligence. The United States prohibits the use of federal funding for human germline gene editing research, but there are no restrictions on research that uses private funds. Currently, neither Genomic Prediction nor any other company reports on embryos’ PGS for educational attainment, but they do allow prospective parents to download raw data for each embryo, which can then be uploaded to an online service that will calculate the PGS for educational attainment.
If this was the “positive” side of molecular eugenics, it also had its “negative” side. Because the SSGAC’s GWAS of educational attainment included only individuals with exclusively European ancestry, the resulting PGS were lower in magnitude and less predictive for people of color. Modern-day race scientists have pointed to these artifacts as evidence that the racist claims of behavior geneticists like Arthur Jensen and Richard Herrnstein were right all along: observed differences in IQ between racially defined groups, they contend, have a genetic basis. The results were predictably devastating. On May 14, 2022, a white nationalist opened fire in a supermarket in an African American neighborhood of Buffalo, killing 10 people and wounding three others. The screed he posted online in an attempt to justify his actions cited one of the SSGAC’s GWAS of educational attainment as evidence that African Americans were genetically inferior to white Americans.
The SSGAC has, so far, published a total of four educational GWAS. As sample sizes have increased, so too has the predictive power of the resulting PGS. The most recent GWAS, using a discovery sample of over three million people, produced a PGS that accounts for approximately 16 percent of the variance in educational attainment among white Americans. Through this research, however, scientists have learned that both education and genetics are more complicated than they initially thought. As sociologists have long known, educational attainment is not a straightforward reflection of intelligence but rather the result of a long and complex series of social interactions.
Our DNA, it turns out, has been shaped by a much longer series of social interactions. Governments and social institutions, most notably the family, play an enormous role in access to education and marriage partners, forging correlations between educational attainment and genomes that show up in GWAS. Within-family analysis indicates that only about 25 percent of the predictive effect of the most recent educational PGS works through biochemical mechanisms, while the remaining 75 percent reflects correlations between an individual’s educational PGS and the environment in which they are raised, including their parents’ income and education, and the quality of schools they attend. For this reason, individuals with low PGS for educational attainment typically experience more educational success if they are adopted by well-educated parents or attend wealthier schools. Even the proportion of the effect that works through biochemical mechanisms might not actually increase cognitive function. It might instead make a person taller (which has a positive effect on education) or more attractive (which has been shown to elicit a positive response from teachers). Untangling these effects is beyond the capacity of today’s scientists. Regardless, a child’s socioeconomic status plays a larger role in their educational trajectory than does their DNA. Children with high PGS for educational attainment but low parental income graduate from high school and college at lower rates than children with low PGS for educational attainment but high parental income.
For all of these reasons, polygenic embryo selection is likely to have little appreciable effect on a child’s chances of getting into a top-100 college or university. Imagine that a couple has 10 embryos from which to choose. Regardless of which one is selected, the child will grow up in the same house, with the same parents and siblings. All of the embryos also have the same two biological parents, which matters because our DNA is not a random selection of all possible variants; it is a random selection of the variants in our parents’ DNA. There will, therefore, be less variation in the educational PGS within a batch of embryos from the same biological parents than in all embryos created by a given IVF clinic or in the population at large. On average, parents who choose the embryo with the highest educational PGS will see an increase of about half a year of schooling as compared with choosing one of their embryos at random. The margin of error, however, is so wide that, statistically speaking, the gain is indistinguishable from zero.
¤
Whether Titan Invictus Collins or Industry Americus Collins attend a more selective university than their unselected older brothers will likely have little to do with their DNA and more to do with other factors that enter into the college admissions process, and with everything that leads up to it. Yet this fact has not penetrated the public discussions about polygenic embryo selection. Both advocates and critics are focused on the ethics of using genetics to increase babies’ intelligence, under the assumption that polygenic embryo selection works the way that companies like Genomic Prediction and poster families like the Collinses claim. Advocates sometimes concede that educational PGS are not terribly powerful at this point. But they are also quick to point out that such instruments will become more predictive as GWAS gets even larger. And they point to the supposedly imminent invention of in vitro gametogenesis—the conversion of somatic cells (e.g., skin cells) into sperm or egg cells—which would dramatically increase the number of embryos available to a couple and thus increase the chances of an outlier at the top of the PGS distribution. Yet these claims neglect the fundamental finding of research in social and behavioral genomics, which is that DNA has little direct bearing on social outcomes like educational attainment and college admission.
While critics warn that choosing embryos for their predicted educational attainment will further entrench existing inequalities, advocates argue that doing so is a duty. The self-described “former academic, current surf instructor” Jonathan Anomaly, for example, says that parents have a moral imperative to give their children the best brains money can buy, not just for the sake of the children themselves but also for the sake of society. However, by assuming that PES will one day work as claimed—even if it does not do so yet—both sides are buying into and thereby furthering the larger eugenicist project of attributing socioeconomic inequality to genetic variation.
It is unlikely that larger GWAS, more precise PES, in vitro gametogenesis, or even germline genome editing will make our society more like the one portrayed in the 1997 film Gattaca. What could bring us there, however, is widespread adoption of the belief that intelligence is primarily genetic and that it can be altered through PES and CRISPR. Eugenics has never worked primarily by producing more intelligent babies. Instead, it has worked by creating the illusion that intelligence is primarily genetic, absolving governments of responsibility for ameliorating social inequality. The illusion leads to policies that enhance the life chances of those who are already privileged while further diminishing the life chances of those who are not.
¤
Featured image: Winslow Homer. School Time, ca. 1874. National Gallery of Art, Collection of Mr. and Mrs. Paul Mellon (2014.18.19). CC0, nga.gov. Accessed August 20, 2024.
