Lissa Harris

Intelligence may be written in our genes, but in a language we don’t yet know how to read.

This entry was posted in information. Bookmark the permalink.

1 Response to Lissa Harris

  1. shinichi says:

    The Ongoing Mystery of How DNA Affects Intelligence

    We know genes affect intelligence, but we’re years away from understanding how

    by Lissa Harris

    https://onezero.medium.com/the-ongoing-mystery-of-how-dna-affects-intelligence-722efb1a7918

    It’s the stuff of which sci-fi dystopias are made: Scientists have come up with a blood test for smartness.

    Until the past few years, the best science available would have been hard-pressed to reliably identify a snippet of DNA associated with something as complex as human intelligence. Today, thanks to recent advances in behavioral genetics, you can spit into a tube, send it off to a lab, and get a number known as a “polygenic score” that gives your rough odds of having high or low intelligence.

    The emerging science of genomics and intelligence has alarmed skeptics outside the small world of behavioral genetics, and for good reason. The history of research into how genes affect intelligence is littered with error, forays into eugenics, and, in some cases, outright atrocity. So it comes as little surprise that present-day science is currently being twisted and weaponized as fodder for racist demagoguery, whether geneticists like it or not.

    The newfound ability to look into a person’s genome for markers of intelligence promises — or maybe threatens — to bring about new futures. Some are darkly dystopian: “designer babies” bred for intelligence, children being closed off from educational opportunities in infancy, or the use of genetic markers to justify discrimination. Others are more optimistic: Some believe we could use genetics to develop better personalized education or intervene early to help children with learning disabilities.

    What do all these futuristic visions share in common: They’re way ahead of scientific understanding. Intelligence may be written in our genes, but in a language we don’t yet know how to read.

    Polygenic scores: Reply hazy, try again

    In 2018, geneticist Robert Plomin trained a public spotlight on polygenic scores for intelligence with his book Blueprint: How DNA Makes Us Who We Are. Plomin, one of the early pioneers of the twin and adoption studies that paved the way for current research linking genetics and human behavior, is an unabashed cheerleader for genetic intelligence testing.

    “Polygenic scores are the ultimate psychological test because, for the first time, they can tell our genetic fortunes,” Plomin writes. “Inherited DNA differences are the major systematic cause of who we are.”

    And indeed, Blueprint makes a compelling case for the heritability of intelligence. Hundreds of studies of twins and adoptive families have shown that intelligence is roughly 50 percent heritable, meaning that about half of the differences between people in cognitive ability are due to their genetic makeup. Scientists don’t know much about how genes influence intelligence, but recently, the accelerating speed and plummeting cost of genome sequencing technology has made it possible to map out where, zeroing in on parts of the human genome that seem to matter. A landmark study in 2017 led by Dutch geneticist Danielle Posthuma identified 52 genes associated with intelligence; a follow-up study in 2018 by the same team found 939 more.

    If heredity is half the story, environment is the rest. Half of what drives differences in intelligence can be chalked up to a variety of environmental influences: schooling, nutrition, or trauma in development, for instance. Most of those environmental factors remain just as mysterious as their genetic counterparts; apart from the most obvious influences, like lead poisoning in childhood, we simply don’t know much about how experiences shape intelligence.

    Despite talk of a “genetic test for intelligence,” polygenic scores aren’t themselves intelligence; they’re a statistical construct weakly associated with it.

    Nevertheless, once solid research linking specific genes to intelligence was published, it didn’t take long for personal genomics companies to develop consumer products around it. GenePlaza already has an “intelligence app” that gives users a polygenic score based on their variations of the 52 genes in Posthuma’s landmark study.

    That we can do this at all is nothing short of incredible. But at the individual level, the tests currently available are basically entertainment, the navel-gazing pleasures of astrology wrapped in a cloak of scientific respectability. Despite talk of a “genetic test for intelligence,” polygenic scores aren’t themselves intelligence; they’re a statistical construct weakly associated with it. You could have a high polygenic score for intelligence but a lower IQ, and vice versa.

    Because genetics is only half of what drives differences in intellect, polygenic scores will never be a crystal ball; for predicting individual intelligence or academic achievement, the scoring methods currently available are hardly better than a coin flip. Even if geneticists were to come up with a perfect polygenic scoring method that included every single genetic snippet associated with intelligence, it would still be missing all the other factors that influence cognitive ability. A curious person looking to measure their own intelligence would be a lot better off taking an IQ test, flawed as it is, than a genetic one.

    Carl Zimmer, a New York Times science writer and author of the book She Has Her Mother’s Laugh: The Powers, Perversions, and Potential of Heredity, is highly cautious about the implications of polygenic scores. New insights linking genetics to intellect and achievement are powerful tools for studying how intelligence works in a more general way, Zimmer says. But at the personal level, the predictions made by polygenic scores are so weak as to be almost useless.

    “If you take 1,000 people and give them all scores, and then sort them into the top 20 percent scorers, the middle 20 percent scorers, and so on, you might start to make some interesting predictions,” Zimmer says. “[But] for any individual, it’s going to be terrible as a predictor. If you look at someone’s DNA, you cannot tell them how long they’re going to stay in school.”

    Separating the genetic from the environmental

    Now, when geneticists say that a trait is heritable, they don’t mean that genes control every aspect of it. Degrees of heritability fall on a scale — eye color, for instance, is 95 percent heritable, while height is 80 percent heritable.

    If intelligence is 50 percent heritable, as Plomin estimates, that doesn’t mean we turn out 50 percent similar to our parents. What that 50 percent figure means is that if you look at all the differences between people in cognitive ability, about half of those differences can be explained by genetic variation and the rest by environmental factors. Imagine two kayakers, each paddling down different rivers: If the two are moving downstream at different speeds, some of that difference in speed can be explained by the swiftness of the underlying currents in the two rivers and the rest to a host of other factors—how hard the kayakers are paddling, how much drag each boat has in the water, and so on.

    One of the trickiest things about the concept of heritability is that it can change, even if genes don’t. In populations subject to extreme conditions, where environmental influences loom large, the heritability of a trait might be lower. For instance, weight is highly heritable — about 70 percent, according to the research — but some researchers think it might be less heritable in places where food is scarcer. If malnutrition is widespread in a society, the harsh environment might blunt the influence of genes. Another field of research, epigenetics, is investigating how environmental influences on parents could actually change the way genes are expressed in their offspring.

    To someone battling with reading comprehension issues, the struggle is just as real whether the problem stems from genetic propensities or from having missed out on some critical piece of education in childhood.

    Apart from some relatively rare disorders, there aren’t many human conditions in which a single gene acts like an on/off switch for a trait. There’s nothing so simple as a “gene for intelligence” — or for obesity or depression, for that matter. Each of the genes that have been associated with intelligence accounts for just a vanishingly tiny fraction of the variance in the population. That’s the “poly” in a polygenic score.

    It is partly this complexity that makes the biology of intelligence so difficult to tease out. Scientists are still a long way from having a solid map of each of the thousands of genes (and so-called noncoding snippets of DNA) that are associated with intelligence, let alone teasing out how each one works in the context of a living human being.

    “There’s a huge amount of good, solid research on intelligence that’s been building up for decades. And yet, I think we’re a long way from really being able to say scientifically what intelligence is,” Zimmer says.

    Population problems

    Although fast, cheap gene sequencing is advancing the pace at which this science can be studied, some of the old troubles remain. One of the most pernicious problems in heredity science is that research done on one population doesn’t necessarily apply to another.

    The 80,000 people whose genomes were analyzed in Posthuma’s 2017 study, which led to the development of polygenic scores for intelligence, were all of European ancestry. So were the subjects in the follow-up study that found almost 1,000 more genes linked to intelligence.

    It’s possible that genetically different populations might have different mixes of genes associated with intelligence. If so, any genetic predictions based on these studies would be even less accurate for people of non-European ancestry.

    Scientists who study human subjects know their samples are biased. There’s even a common acronym for the types of populations that tend to be disproportionately represented in science papers: WEIRD—Western, Educated, Industrialized, Rich, and Democratic—a description that fits only about 15 percent of the world. The subjects of genetics research are also “overwhelmingly white”: About 80 percent of participants in large genome-wide studies are white, Vox reports. Correcting for this sort of bias is easier said than done, however: These are also the populations that tend to live near universities and volunteer for research experiments.

    Until science starts doing a radically better job of representing the full spectrum of human diversity, the results of any genetic research will come with a major caveat: “Does not necessarily apply to different populations.”

    Biased research sampling fuels a different problem: White supremacists often use intelligence research in bad faith as “evidence” of differences in intelligence between races. The Daily Stormer, a neo-Nazi white supremacist website, reacted to mainstream reviews of Plomin’s book with glee. “The future is looking very bright indeed,” one commenter wrote in response to an article about Plomin in Technology Review.

    Alt-right provocateurs have gotten a prominent assist from James Watson, the co-discoverer of the double helix. Watson has repeatedly asserted (without evidence) that white people are genetically predisposed to higher intelligence, to the dismay of many fellow geneticists. No science supports the claim that there are racial patterns to inherited human variation in intelligence, but many geneticists are fearful of engaging with the topic, despite widespread public misinterpretation of the science.

    Designer babies are a bust — for now

    Advances in behavioral genetics tend to set off public fears about “designer babies” and the creation of a genetically endowed aristocracy. With the advent of CRISPR, a recently developed gene-editing technology that Chinese geneticist He Jiankui claims to have already used on human embryos, it’s becoming an ever more plausible worry. But with current science and fertility technology, we’re just not there yet.

    Let’s say you want to use polygenic scores to design a really smart baby. Technically, grading embryos based on polygenic scores is possible, but it’s unlikely to get results. Fertility clinics sometimes offer genetic screening for simple traits, like biological sex or a disease that arises from a single gene. But with dozens or hundreds of genes involved in scoring, and the same two parents contributing genes, it’s unlikely that the polygenic scores of a handful of embryos created in a round of IVF will be vastly different from one another.

    Even aside from the inherent limitations of IVF, and the fact that polygenic scores don’t predict much about how intelligent an individual will actually grow up to be, there’s a deeper uncertainty: We don’t know what these genes do. Already, there is evidence that some of the recently identified genetic sequences associated with intelligence are also associated with autism and depression, a link that may make clinics (and parents) think twice about attempting to blindly use polygenic scores to grade embryos.

    A few unscrupulous clinics might opt to traffic in parents’ hopes and fears, and a few wealthy and desperate souls might take them up on it. But for now, the idea that research on the heredity of intelligence might result in a rush on designer babies lies squarely in the realm of science fiction.

    The allure of infophilia

    In American society, we seem willing to tolerate vast levels of environmentally imposed inequality, but the idea of genetic inequality feels deeply disturbing, especially to those inclined toward ideals of social justice.

    Some of our discomfort stems from people’s tendency to look at genes as “real” in a way that other influences aren’t. But to someone battling with reading comprehension issues, for instance, the struggle is just as real whether the problem stems from genetic propensities or from having missed out on some critical piece of education in childhood. What’s more, we already have access to better predictors of individual achievement. It’s just that we apparently care more about genes.

    Personal genomics is just one lens of many through which we can look at human intelligence — a thing that, after all, is a property of people, not of nucleotides and genes.

    “For now, knowing a child’s polygenic score predicts his ultimate educational attainment about as well as knowing his family’s income, and slightly worse than knowing how far his mother went in school,” psychologist Paige Harden wrote in a 2018 op-ed for LeapsMag called “Genetic Test Scores Predicting Intelligence Are Not the New Eugenics.” “These pieces of information are also readily available about children before they are born, but no one is writing breathless think-pieces about the dystopian outcomes that will result from knowing whether a pregnant woman graduated from college.”

    Susan Oyama, a philosopher of science who has long been critical of the division of influences into “nature” and “nurture,” calls this attitude “infophilia”: the idea that genes are in some way more real, more essential, than the living beings in which they are found. Everything that influences our development is real, she says.

    Seen in this light, personal genomics is just one lens of many through which we can look at human intelligence — a thing that, after all, is a property of people, not of nucleotides and genes.

    Putting the new science to work

    Still, just because polygenic scores aren’t very useful predictors of individual intelligence doesn’t mean they can’t be useful to educators. One of the most promising uses for polygenic scores is to study environmental factors — for instance, using genetic information to amp up the statistical power of studies that measure the impact of educational strategies.

    In any experiment, if you can control for a variable — in this case, genetic diversity — you can reduce the number of subjects you need to test your hypothesis. By boosting the statistical power of smaller studies, polygenic scores for intelligence and educational achievement might help make studies on the effectiveness of different educational strategies both more powerful and less expensive — and therefore more likely to yield useful results.

    That’s a prospect that appeals to Harden, who wrote an op-ed for the New York Times last July called “Why Progressives Should Embrace the Genetics of Education.”

    “Talking about including genetics as a variable in statistical models doesn’t have the same dark allure as eugenic proposals to screen embryos or assign children to schools based on their genotypes,” she writes. “But the widespread use of polygenic scoring in research aiming to understand how environments shape children’s lives will yield big payoffs for knowing how to maximize a child’s potential.”

    Turns out the big payoff for the blockbuster new science of intelligence might just be, well, more science. If that sounds insufficiently futuristic, that’s just how science moves: one tiny, shuffling footstep at a time.

    “I think it’s important for people to understand just how modest the claims are that scientists are making,” Zimmer says. “Sometimes it seems that they’re making some sweeping statement about all of human nature, when in fact they’re just looking at a cloud of data points and seeing a very gentle trend in it and don’t even know what’s causing that trend. That’s the nature of this kind of big psychological research these days.”

Leave a Reply

Your email address will not be published.