Not in Your Genome

It turns out that if you begin an assertion with “it turns out” and sprinkle it with statistics and acronyms—especially if it’s expressed in the passive voice and followed by a footnote—up to 83 percent of the variation in whether people buy it is explained by their SCI (science credulity index) and 78 percent by their BDS (baloney detection score).¹

Here’s how it works. “It turns out,” writes Dalton Conley, the Henry Putnam University Professor in Sociology at Princeton, in his new book The Social Genome, “that almost every trait that has been studied is at least partially influenced” by genetic differences, including “about 40 percent” of the variation in how far people advance in school. “For income,” he continues, “it’s 70 percent. For cognitive ability, 75 percent.” Turning to the footnote, we find no evidence but instead a further assertion that PGIs (“polygenic indices”)—statistical treatments of genetic variants—are “incredibly useful for studying the social world.”

This is Conley’s central claim: that genetic analysis offers the key to understanding not just people’s biology but their social environment. To grasp what a PGI is, you need another acronym, GWAS, for “genome-wide association study.” This statistical practice emerged a couple of decades ago, after the Human Genome Project completed the first map of a composite human genome. A GWAS identifies tiny statistical correlations between millions of genetic variants and some measured trait in a sample of individuals. A PGI then takes a given person’s DNA variants, assigns a number to each based on the correlations of the GWAS, and creates a weighted sum intended to predict the person’s likelihood of having the trait.

Conley is a true believer in the power of the PGI. “We can take a saliva sample from a baby,” he writes, extract its DNA, calculate a PGI, and “predict that baby’s odds of completing college.” (He adds the caveat that the predictiveness of the PGI depends on the environment, and if the society changes, “all bets are off.” But this is trivially true and therefore not really a caveat, as we will see.) Elsewhere in The Social Genome he writes that PGIs “have become the FICO scores of human genetics” and declares that “the PGI, with its X-ray powers, has revealed the hidden logic of social life.”

According to Conley, a PGI is “a single number that summarize[s] a person’s genetic tendency for a particular disease or trait—for instance, blood pressure, height, or cognitive functioning.” But one of these things is not like the others. It’s easy to measure height and possible to measure blood pressure, though even in these cases there’s no causal pathway from DNA to trait, since height and blood pressure are highly polygenic, meaning that researchers have found minute correlations with thousands of regions of the DNA as well as epigenetic and environmental factors. “Cognitive functioning,” on the other hand, is difficult to define, let alone to measure. And yet throughout The Social Genome Conley assumes that nebulous personal qualities can be not only quantified as though they were straightforward physical characteristics but also correlated with genetic variants to predict the course of people’s lives.

The evidence that polygenic scores have this kind of predictive power is between extremely weak and nonexistent. (In fact Conley’s assertion that genes can help predict a baby’s odds of completing college is apparently contradicted by his own research: a 2022 article on which he’s listed twenty-seventh of forty authors states that “the PGI cannot be used to meaningfully predict” an individual’s educational attainment.) Statistical genetics hasn’t led to new understandings of diseases or medical therapies, as Conley acknowledges, and its validity remains unproven in the biomedical sciences.

In the social sciences, meanwhile, not only are the traits in question often difficult or impossible to define or measure; the correlations also have no demonstrated causal meaning. Conley twice repeats the familiar adage that “correlation is not causation.” You might find a statistically significant correlation between, say, a genetic variant that makes it difficult to metabolize alcohol, common in East Asian populations, and the use of chopsticks, but such a correlation would obviously be causally meaningless. (Conley even discusses an article titled “Beware the Chopsticks Gene.”) And yet his book posits lots of hypothetical “pathways” connecting PGIs with cognitive and behavioral traits. The implication is that these contain causal mechanisms, but neither Conley nor anyone else has ever identified any.

In 2022 we reviewed another book that ascribes cognitive and behavioral traits to genetic causes.² We think it’s important to confront these claims and the movement they represent, which Conley variously calls “sociogenomics,” “behavioral genetics (or BG as it is known),” “genoeconomics,” and “biosociology,” the last an inversion of the 1970s term “sociobiology.”³ This movement includes prominent faculty at elite universities and commands an increasingly wide audience. The Social Science Genetic Association Consortium, the main association for sociogenomics, founded in 2011, has grown into a multi-institution international collaboration that draws millions of dollars in grants and uses data from millions of people.

A surge in the production of PGIs in recent years has led to their increasing prominence in popular writing and culture and their widening use outside research settings. For instance, in assisted reproduction, companies now offer to test embryos for PGIs that purportedly influence traits such as intelligence. Yet the claims of sociogenomics are dangerously misleading and essentially groundless. Its advocates use spurious scientism—an alphabet soup of acronyms and elaborate statistical practices that exploit vast, cheaply accessible genetic databases—to disguise social ideologies as scientific facts.

“Today, I am a self-proclaimed biosociologist,” Conley announces in chapter 3 of his book, but this wasn’t how he began. His dissertation in sociology, which he completed at Columbia in 1996, treated the “racial wealth gap” in relation to other social disparities. Later, however, he questioned his conclusions. “Perhaps,” he recalls thinking, “the parents’ assets, per se, didn’t drive a child’s success…but rather the underlying traits that led the parents to accumulate the assets.” He decided that he and other sociologists had been wrong when they “waved away the huge effects of genes.” From there he embarked on a quest to discover the genetic causes of “wealth accumulation,” which he aptly if puzzlingly likens to Ahab’s pursuit of Moby Dick.

To represent the relationship between genes and environment, Conley uses the metaphor of a Möbius strip. Imagine a twisted loop of paper that appears to have two surfaces, but if you trace either with your finger, you soon find yourself tracing the other, showing that there’s just one continuous surface. Genes and environment, Conley writes, are a Möbius-like continuum, because genes have different effects in different environments, while they also, he says, mold your social environment in various ways. They direct you toward genetically similar people. They send you to college or to a low-wage job in the service sector. They evoke certain treatment by your parents and teachers. They act within your family and friends, shaping their behavior toward you.

For many readers the claim that genetics affects social and economic success will carry a whiff of biological determinism in general and of scientific racism in particular. It has become standard in Conley’s field for authors to reassure readers on both counts. We needn’t worry that there’s a risk of scientific racism in sociogenomics, they argue, because their claims aren’t, for the moment, “portable.” Conley cautions that “PGIs have been developed and work within only a given ancestral population” and that “comparing PGIs across groups tells us nothing about genetic differences between those groups.” He also writes that it’s “fairly clear” that social causes such as Jim Crow and redlining were the “primary drivers of racial wealth gaps,” although elsewhere he says that in his dissertation he “overlooked” the importance of “pedigree” and that “genetics played a much bigger role in the lives of the children I studied than I had imagined.”

As for biological determinism more generally, Conley sometimes seems to avoid deterministic formulations, for instance by invoking “influences” rather than “causes.” He writes, for example, that society became more tolerant of homosexuality after “the discovery of the biological (read: genetic) influences on sexual orientation,” and he suggests the same might happen with addiction and depression once people understand that “genes influence important outcomes.” (The “gay gene” claims of the 1990s have been thoroughly debunked.)

In the same way, with his Möbius strip metaphor Conley at first glance appears to adopt a moderate position between nature-only and nurture-only extremists. “Blank-slaters,” he writes, won’t accept that “even the effects of the environment are partly driven by genes. Hereditarians, on the other hand, won’t appreciate that genes aren’t deterministic but part of a messy social process.” Conley warns against “genes-as-destiny” determinism and the risk of “genetic profiling.” He emphasizes that, counter to what “someone inclined toward eugenics” might think, environment makes a difference: “Altering the environment can affect how our genes work, thereby ameliorating genetic inequality.”

There’s nothing inherently antideterministic, however, in saying that genes work differently in different environments. Toss your cuckoo clock into a swimming pool and it will work differently, but this doesn’t reflect anything indeterminate in its mechanism. Meanwhile Conley betrays his crypto-determinism with his contention that there’s such a thing as “genetic inequality” in the first place. We’re “guided silently by our genes,” he says at one point; genetic analysis, he insists at another, will close “the loop between the cause of DNA and the effect of who we have become.”⁴ Although he acknowledges that scientists have given up the “gene for X” idea, he nevertheless invokes “genes for” all kinds of things, including verbal ability, beauty, extroversion, humor, voraciousness, a musical ear, manual dexterity, and slow-footedness. When referring to genes that cause a person “to seek new experiences,” he admits, “I am stipulating that there are such genes,” then claims there’s “growing evidence” for this. He doesn’t cite any.

One way our DNA shapes us, Conley says, is “by eliciting a response” from others. He claims, for instance, that “kids who are attractive—as encoded by their genes—get more attention and praise from adults” and that “beautiful students’ grades and test scores are higher, on average, in part because of the positive investments they have received thanks to their pulchritude.” Conley doesn’t define “attractive” or “beautiful,” justify his assertion that people’s attractiveness is “encoded by their genes,” or confront the issue that “pulchritude” is famously in the eye of the percipient.

Nor does he have real evidence for his claim that different polygenic scores elicit different responses from others. Conley cites a study in which he says he and a collaborator showed that young children with higher PGIs for education receive more “cognitively stimulating” parenting. In the article itself, however, we see that the children’s PGIs for education didn’t correlate with statistically significant differences for two of the three parental behaviors studied—singing and reading to the children—but only for the third, playing with them.⁵ A genetic difference that elicits playing but not reading or singing? These aren’t distinct activities, as anyone who’s ever met a child, or been one, must know. Is ring-around-the-rosy singing or playing? Is Pat the Bunny playing or reading?

To make matters murkier, researchers in the other study Conley cites claim that adolescents with higher polygenic scores for education receive less parental attention than their siblings. That seems contradictory, but in a fine example of thrashing the data till it confesses, Conley surmises that everything changes after early childhood. “The same genes that evoke more reading or playing at age two might evince a completely different response” in adolescence, when report cards and test scores provoke parents to compensate for their less talented children’s “genetic disadvantages.”

Among Conley’s core assertions is that sociogenomic research has revealed an important genetic basis for income. This is an age-old theme in eugenics, as much as Conley is keen to distance himself from that tradition. From the late nineteenth through the mid-twentieth century, eugenicists fretted that the poor, who they assumed were biologically inferior, outreproduced the rich. Conley insists that positing “genetic differences for behavioral traits…doesn’t mean becoming a reactionary eugenicist,” but he shares with these theorists a belief that people are biologically unequal in ways that affect their ability to make money. He refers, for instance, to studies showing that “variation in genes in the US explain[s] 58 percent of the variation in income in the male population and 46 percent in the female population.” His source seems to be an article from 2012 reviewing earlier research based on a controversial and deeply flawed methodology known as twin studies, which, because Conley relies on it throughout The Social Genome, merits a brief historical excursion.

In 1875 Francis Galton—Charles Darwin’s cousin and the founder of eugenics—invoked a bunch of stories about twins to argue “that nature prevails enormously over nurture.” Conley writes that “Galton’s notions were infused with racial hierarchy” and that he mistakenly “ignor[ed] the environmental differences between households,” yet he observes that Galton “showed that mental traits varied in much the same way as physical traits (like height) and that they tended to cluster in families.” This is, to say the least, unscientific. Galton’s 150-year-old observations about “mental traits” reveal nothing beyond his Victorian prejudices about race and class. In the 1869 work Conley cites, Hereditary Genius, Galton takes “reputation” as an index of “ability” and expresses his conviction that eugenics would produce a race of men “as much superior mentally and morally to the modern European, as the modern European is to the lowest of the Negro races.”

In 1966 an English psychologist and geneticist named Cyril Burt published an article that purported to compare identical twins raised together and apart, arguing that intelligence depends mostly on “genetic constitution.” Three years later a University of California psychologist named Arthur Jensen, an admirer of Burt, published a paper invoking twin research, especially Burt’s, to argue that races differ in IQ and that no amount of social intervention could overcome genetic deficiencies.⁶ Then in 1974 a devastating book appeared by Leon Kamin, chair of the psychology department at Princeton, showing that Burt’s data were falsified. As for Jensen, in addition to using Burt’s fraudulent data, he counted results that suited him multiple times and omitted those that didn’t. Kamin also offered a powerful critique of twin studies, resting on the impossibility of isolating genetic factors from environmental ones.

Burt’s and Jensen’s reputations were fatally damaged, but the tradition of twin studies continued unchecked. In the later twentieth and twenty-first centuries, its proponents have often compared the resemblance of identical twins with that of fraternal twins, arguing that if identical twins have more similar life outcomes, the difference must be caused by genetics. The crucial assumption is that both kinds of twin share the same degree of environmental uniformity (“the equal environments assumption,” Conley writes, “or
EEA for short”). But this “quaint notion,” Kamin pointed out, is “clearly untenable”: identical twins have more similar environments because they’re treated more similarly.

Defenders of twin studies, including Conley, have proposed to vindicate the method by comparing “misclassified” twins—identical twins who grew up thinking they were fraternal or vice versa—with correctly classified twins. They assume that misclassified twins share the same level of environmental similarity as the kind of twin for which they’re mistaken, which if true would allow researchers to isolate the genetic from the environmental factors. Conley reports that the misclassified identical twins in his own study “were as alike as correctly classified identical twins.” Degrees of environmental similarity “didn’t really matter.”

In the study itself, though, the authors acknowledge that the extremely small size of their misclassified twins sample was a “limitation” of their analysis. It was too small to support any conclusion at all, according to one commentator, who noted that the results regarding high school grade point averages were based on just twelve misclassified fraternal twins and fifty-six misclassified identical twins. Moreover, the misclassified twins approach relies, as another article pointed out, on the assumption that twins’ experiences are determined by people’s explicit beliefs about them, a “hypothesis that twin researchers have not tested empirically” and that is clearly implausible. Identical twins, even if misclassified, resemble each other more closely than fraternal twins and must share more experiences on average. Genes are relevant in this context, but not in the way Conley suggests. For instance, identical twins might have a genetic variant that affects melanin production, leading to skin that burns easily, and so prefer indoor sports. Or they might have other genetic variants that govern the timing of melatonin release in the brain, making them early-rising “larks” or late-rising “owls” who spend hours together while everyone else sleeps.

A hardcore genetic determinist might say that’s just the genes acting through the environment. But that would be deeply misleading. DNA and other biological aspects of a person of course affect their experiences, but that doesn’t mean they have any meaningful or systematic relation to social or behavioral outcomes like educational attainment or income. “The fallacy can be illustrated by an analogy,” Kamin wrote. People were once enslaved on the basis of their genetically determined skin color, but we don’t conclude that “the black man carries slavery genes.”

Conley offers a similar example, linking the genetic causes of dark skin to high blood pressure via social discrimination: “Skin tone provokes discrimination from some interlocutors; discrimination causes stress; and stress causes high blood pressure.” He writes that “it’s irrational and immoral to treat someone differently on the basis of skin tone” and that “genetic effects can be just as irrational…as environmental influences.” Nevertheless, he uses the example to argue that genes also shape a person’s environment by evoking “rational” responses, such as putting a “precocious child into the honors track.” Again he treats indeterminate judgments about personal qualities as though they were well-defined physical characteristics with straightforward genetic causes. Skin tone is highly influenced by DNA through known mechanisms; “precocity” is not.

There are perhaps better examples of genetic variants that might evoke what Conley calls a “rational” response. “Lark” variants could induce your parents to enroll you in early-morning art classes, or “owl” variants in the after-school jazz band. But this does nothing to vindicate twin studies. On the contrary, it exemplifies the many contingent ways that genetically identical twins might share more experiences, even when misclassified, than fraternal ones do.

Moreover, despite their identical DNA and their many shared experiences, “identical” twins are far from identical. One of us is the father of identical twins and knows intimately how different they are. The other recalls that in her son’s elementary school there were identical twins whom we’ll call Nora and Jill. Asked how he told them apart, her son, age seven, patiently explained, “Nora looks Nora-ish and Jill looks Jill-ish.” Genetically identical twins are different from the very beginning, each in a continuous state of transformation through a contingent interaction of biology and environment that begins in utero and lasts throughout their lives. Conley writes that for any trait such as “education level,” you can derive its genetic “heritability” from twin studies, and “100 percent minus the heritability is the environmental influence.” But you can do this kind of arithmetic only with independent factors: you can’t add or subtract factors that interact, such as genetic and environmental influences. How much of a rainbow’s formation is due to moisture in the air and how much to sunlight?

Sometimes Conley appeals to common experience to convince us that people’s genes govern their social environments. For instance, he writes, “Here’s an assignment: Go sit on a park bench and watch people.” He specifies that we should “note pairs or groups of people who seem to be strolling together,” confiding, “I like to do this.” The point is to notice that “like likes like.” Tall people stroll with other tall people, and the same for blond people and those with facial piercings.

Social sorting by sameness, Conley notes, is called “homophily.” He doesn’t explain the origin of the term, which was coined in 1954 by two sociologists, Paul Lazarsfeld and Robert Merton. In the same article they introduced “heterophily,” a tendency to form friendships between people who differ. Having both terms, the authors explained, helps avoid the fallacy of assuming that either one is universally applicable. It depends on the people, the circumstances, and the characteristic of resemblance or dissimilarity. (In our own experience in public parks, we haven’t noticed people grouped by height or hair color, and facial piercings seem to correspond mostly with age.) Lazarsfeld and Merton identified a pattern of value homophily in close friends, concluding that values “act both as cause and effect, modifying and in turn being modified” by friendship.

In Conley’s usage, the reasonable if not earthshaking idea that close friends often share values morphs into the extraordinary notion that people cleave to those who resemble themselves in every possible respect: “tattoos, weight, hair style, type of dog they are walking, or anything else you can think of with which to classify people.” All of this ostensible homophily, Conley says, is the result of the fact that “our genes select our friends; their genes select us.” He cites studies finding that friends tend to be the “genetic equivalent” of fourth cousins, or even second cousins, and the same for spouses.

These studies have been repeatedly criticized for overlooking the effects of “population stratification”: subpopulations of people differ genetically as a result of their different ancestry, and people often befriend and marry others in their subpopulation for geographical and social reasons. A 2014 refutation and a 2019 statistical reexamination both found, as the 2019 paper is titled, “No Evidence for Social Genetic Effects or Genetic Similarity Among Friends Beyond That Due to Population Stratification.” Conley recognizes the confounding presence of population stratification as a general matter but then disregards how it undermines the studies he cites to justify his claims about “homophily.”

His effective dismissal of population stratification is also apparent in his assertion that not just socioeconomic inequality but biological inequality has been “widening” in recent decades. With the zeal of the convert, Conley writes that he had previously overlooked the importance of biology in inequality. It was only after “wandering in the desert” of social science for a decade that he “dove headfirst into genetics” and discovered the social importance of worsening “genetic inequality.” As evidence of this phenomenon, he cites a study of the geographic clustering in Great Britain of polygenic scores associated with traits including “openness,” “conscientiousness,” and “educational attainment,” from which he concludes that people in depressed coal-mining towns in the UK are “genetically unequal” to Londoners. Setting aside the problems inherent in the idea of scientifically measuring a person’s “openness,” which is like proposing to use a screwdriver to analyze a poem, the study’s authors present their findings as indicative of “genetic consequences of social stratification”—people in different population groups have different ancestry—rather than genetic causes. Moreover, several of them published a subsequent paper, which Conley doesn’t cite, showing that controlling for population stratification “significantly decreased the heritability for socioeconomic status (SES)-related traits, most strongly for educational attainment and income.”

Emigration and upward mobility, like friendship and marriage, are social phenomena to which Conley says “genetic sorting” holds the key. The “best and the brightest,” driven by genetic predispositions for “cognitive ability,” “grit, savviness, ambition,” and “skill, guts, and energy,” leave home in search of opportunities, creating genetic polarization by geographical region. As evidence, Conley mentions a study claiming to reveal “The Genetics of Success”: how a child’s PGI for educational attainment predicts their life trajectory. But the study’s authors subvert their own conclusion by writing that the children with higher PGIs “were born into better-off homes” and had “grown up in families with more economic resources.” There’s no reason to think PGIs, rather than socioeconomic status in childhood, made the difference to the outcomes they studied, such as socioeconomic status in adulthood.

Polygenic indices, according to Conley, reveal “the causal arrows of all kinds of influences” and provide “a genetic thread we can follow to trace out the path of causes and effects.” But these causal threads never gain material specificity. By what mechanism might certain genetic variants affect a person’s income, or whether they continue their studies, or their propensity to leave their depressed area and migrate to a land of opportunity? Just how might nucleotides in particular spots in a person’s genome shepherd them toward others with similarly placed nucleotides? How would a child’s nucleotide placements influence their parents to play with them?

Of course, DNA has a causal role in producing plenty of phenotypic traits: we’ve mentioned known genetic causes for being a morning lark or a night owl or being prone to sunburns. But if you want to say that DNA sends people to Harvard, as Conley implies, not because of legacy admissions but because Harvard students, their parents, and their grandparents all share a superior genetic profile for “cognitive ability” and other qualities, you need a plausible causal mechanism that stretches from DNA through the synthesis and disposition of proteins to the Harvard admissions office. Conley doesn’t specify one, nor has anyone else, and it’s not for lack of trying. Only after decisively failing, about twenty years ago, to find links between individual sequences of DNA and traits including cognitive abilities did social scientists turn to the alphabet soup of GWASs and PGIs. In the process they abandoned any pretense of seeking causal mechanisms to justify the connections they assert between DNA and social and behavioral outcomes. In a kind of magical thinking, Conley and the others speak and write as though their PGIs themselves had causal powers.

Conley assures readers that we can rectify “genetic inequality” by making environmental adjustments to compensate for it, and he mentions some great ideas, such as “free early childhood education for all.” But by representing human inequality as fundamentally biological, The Social Genome reifies the social hierarchies Conley says he wants to rectify. He cites an article on which he’s listed as an editor suggesting that the early-twentieth-century eugenicist R.A. Fisher was right about the biological basis of social class—that in England between 1600 and 2022 “social status was mainly determined by genetic inheritance.” Social welfare programs, it says, “did not change in any measurable way” the “persistence of social status across generations.” This paper is what Conley cites in the footnote most proximate to his claim that while we inherit wealth socially, we “inherit biologically…the ability to make wealth.” He doesn’t mention that the research has been repeatedly discredited, most definitively by a reanalysis showing that its “conclusions are based on the conflation of genetic and non-genetic transmission (for example, of wealth) within families.” Undeterred, at various points Conley chides sociologists for treating the transmission of wealth as a social phenomenon, since “BG” (behavioral genetics) shows it to be “usually” biological; claims that we find ourselves born into a certain position in society thanks to our “ancestors’ genes”; and suggests that our forebears passed their advantages to us “culturally or even (gasp) biologically.”

The “(gasp)” cutely hints that the author is saying something shocking and bold, but in fact he is reprising a timeworn theme: social hierarchies derive from innate differences. Citing no evidence, Conley asserts toward the end of The Social Genome that some people are genetic “orchids,” keenly sensitive to the environment, while others are “dandelions” that can survive anywhere. “Fortunate human ‘orchids,’” he observes, “end up more beautiful (i.e., healthier and more successful) than the plain old dandelions, which have a ‘ceiling’ of achievement.”

This pseudoscientific distinction among humans—some intrinsically beautiful and successful, others inescapably plain and limited—recalls the 1997 dystopian movie Gattaca, whose plot turns on the evils of eugenics and the triumph of the human spirit over misguided genetic determinism. Near the start of The Social Genome Conley recounts having tried to persuade a fertility doctor to help him and his wife conceive “the world’s first PGI-optimized baby” by quoting from Gattaca. The doctor, perhaps having better understood the film, said it wasn’t feasible. But that was then. Conley reports that he and his family just barely missed out: the same year his son was born, another family got to PGI-optimize their baby. Now, as we’ve mentioned, companies in the assisted reproduction market offer screening of embryos for PGIs that are supposedly causally associated with cognitive and behavioral traits.

In another curious passage, Conley mentions Donald Trump’s frequent invocations of “good genes,” acknowledging that “only white supremacists compare humans to thoroughbreds in what’s called ‘racehorse theory.’” Depending on how you feel about white supremacists, you might assume Conley is headed toward repudiating racehorse theory. But you would be mistaken. As compared with racehorses, he says, humans make an even “more interesting game to calculate the odds on.” There is perhaps comfort to be found in the thought that this is a losing proposition: the field Conley calls BG amounts to a similar acronym we’re too polite to mention.