She Has Her Mother's Laugh: The Powers, Perversions, and Potential of Heredity (41 page)

Burt's scandal stained all of twin research, leading many to dismiss it as bad science. Yet just because a field attracts a fraudster doesn't make all the discipline's findings wrong. Hundreds of well-designed
twin studies have come to the same conclusion: Identical twins have closer intelligence test scores than fraternal twins. Even when identical twins are raised apart, their intelligence test scores stay more similar than siblings raised together. These studies have led scientists to estimate the heritability of intelligence test scores as roughly 50 percent. That's substantially lower than Burt's claim of 80 percent, but it still indicates that heredity has an important role to play in intelligence that should not be dismissed.

As these more rigorous studies piled up,
they drew criticisms of their own. Some researchers complained that they relied on the assumption that the only difference between fraternal and identical twins is their genes. Some research suggested that this might not be the case. Because identical twins look alike, they may be treated as interchangeable. Fraternal twins may look different enough from each other that they have an experience more akin to their ordinary siblings.
In a 2015 study, a group of researchers investigated twins who experienced bullying, sexual abuse, and other kinds of trauma. They found that identical twins had more similar experiences than fraternal twins. If one identical twin was abused, it was more likely that the other one was, too.

But researchers who have looked closely at the experiences of twins have concluded
these effects are weak or nonexistent. One such study was even
carried out by a skeptic of twin studies, a Princeton sociologist named
Dalton Conley. Conley realized he could investigate the experiences of twins by studying the surprising number of twins who get misclassified.

Some identical twins are recorded as fraternal at birth, and fraternal ones as identical. A genetic test can easily reveal the true nature of newborn twins, but doctors apparently don't bother with it much. In a 2004 study in Japan, researchers found that hospitals misclassify as many as 30 percent of twins. In the Netherlands, researchers tested the DNA of 327 pairs of twins and then asked their parents what kind of twins they were. Nineteen percent of the parents gave the wrong answer.

If genetic differences weren't important, then fraternal twins misclassified as identical should end up more similar to each other. You'd also expect that identical twins would be robbed of this powerful experience if their parents and teachers and everyone else around them treated them like fraternal twins. But Conley and his colleagues discovered no such thing. The cases of mistaken identity had no effect on how the twins turned out. Identical twins ended up more like each other in a range of traits, even when they didn't know they were identical. They reached more similar heights. Their risk of depression was closer. Their grades in high school were more alike. The only explanation for these similarities was heredity.

Every behavior scientists have studied turns out to be partly heritable,
from smoking to divorce rates to watching television. At this point, it would be astonishing if intelligence
weren't
heritable. But twin studies on intelligence could not say what exactly was being inherited—which genetic variants, in other words, influence people's scores on intelligence tests.

To hunt for those variants, scientists followed the trail blazed by the researchers who studied height. Scientists first linked genes to height by studying people with growth disorders like Laron syndrome. The first genes tied to intelligence also turned up in
studies of intellectual disorders such as PKU. Those early discoveries brought tremendous benefits to children.
They made it possible to test for a growing list of disorders and search for ways to treat them—a special diet for some, a special education program at school for others.

But when it came to the heritability of intelligence, those genes were practically meaningless. The severe mutations that lead to intellectual developmental disorders are very rare. PKU affects only one in ten thousand people, for example. For the population as a whole, those variants say nothing about why some people score higher on intelligence tests than others.

At the start of the twenty-first century, behavioral geneticists were full of hope that DNA sequencing technology and a map of the human genome would let them quickly find more genes that influence intelligence. “In a few years,
many areas of psychology will be awash in specific genes responsible for the widespread influence of genetics on behavior,” Robert Plomin and John Crabbe predicted in 2000.

At first, it looked as if the deluge was on its way. Researchers identified genes that seemed like good candidates for influencing intelligence and studied them in ordinary people. One of those genes, called COMT, encodes an enzyme in the brain. The enzyme keeps the neurotransmitter dopamine in check. It does so by finding the dopamine molecules and chopping them to pieces. One variant of COMT produces an enzyme that chops slowly, allowing the brain's level of dopamine to rise. The slow-chopping variant is fairly common. (Checking my genome, I discovered that I carry one copy of it.) Many scientists suspected that the different versions of COMT might have some effect on intelligence test scores, because dopamine is crucial for memory, decision-making, and other mental tasks. The slow-chopping variant was letting more dopamine build up in people's brains, improving their performance.

In 2001,
Michael Egan, a researcher at the National Institute of Mental Health, led a study to test this idea. He and his colleagues gave 449 people an exam known as the Wisconsin Card Sorting Test. It is really just a simple game. The scientists showed the volunteers cards with circles, squares, crosses, and stars. The shapes came in different numbers and colors. The
object of the test was to find sets of cards that matched each other according to a rule—a rule that Egan's team didn't tell the volunteers. Through trial and error, the volunteers eventually figured out the rule, but Egan would then switch it, leaving the volunteers to figure out the new rule. The scientists measured how quickly they made this discovery.

Egan and his colleagues found that people with the slow-chopping variant of COMT performed slightly better at the game. Their success prompted other scientists to decide to see about COMT for themselves. A number of them also found a link to intelligence.

It was an exciting discovery, but before long, the excitement curdled into disappointment. Later studies on more people
failed to find any effect from the slow-chopping version of COMT. Other researchers tested out
other candidate genes for effects on intelligence, only to watch their promising associations collapse.

In hindsight, searching for candidate genes was a strategy pretty much guaranteed to fail.
Our brains use 84 percent of our twenty-thousand-odd protein-coding genes. Each type of neuron uses a distinctive combination of those genes, and it turns out the brain is made up of hundreds of cell types—so many that scientists will not finish its catalog for a long time. To think we could just reach into this jumble and pluck out a single gene that had a clear-cut role in intelligence was to pretend we know more about the brain than we really do.

As candidate genes failed to reveal genes linked to intelligence, scientists turned to genome-wide association studies. By searching through genetic markers scattered across the genome, scientists would let the genes speak for themselves.

Ian Deary led the first genome-wide association study on intelligence. As part of his research on the Scottish Mental Survey, he and his colleagues sequenced DNA from some of the test takers. Adding DNA from people who had volunteered for other studies, they analyzed 3,511 people all told. The researchers scanned half a million genetic markers to see if any of them were correlated with high or low levels of intelligence. Nothing on that scale had ever been attempted before. And yet, as Deary and his colleagues
reported in 2011,
they failed to find even one gene with any clear effect on people's intelligence test scores.

The experiences Joel Hirschhorn and his colleagues had with height had prepared Deary for this kind of disappointment. Complex traits can be influenced by hundreds or even thousands of genes. Their common variants may be so weak that small studies will fail to reveal them. Making matters worse, intelligence is not an obvious trait that can be accurately measured with a simple tape measure. Psychologists may use different intelligence tests depending on who they're studying, what aspect of intelligence they want to study, or how much time they have to examine each person. To try to amass a lot of subjects for their research, scientists often merge smaller studies that used different intelligence tests. The mismatch among the tests can spread a blanket of fog over the influence of genes.

Despite these challenges, genes continued to send signals from their hiding places. Peter Visscher's test for genetic similarity, which he had previously used on height, also confirmed that intelligence test scores were heritable. In fact, he could account for much of the missing heritability of intelligence. The precise amount was different depending on the age of the people the scientists studied. When Visscher and his colleagues examined twelve-year-old children, they could account for
a staggering 94 percent of the heritability of intelligence.

The first genes associated with intelligence came to light in a roundabout way. Medical surveys often ask how long people stayed in school, and it turns out that
educational attainment is a modestly heritable trait. The correlation between identical twins is stronger than for fraternal twins; full siblings raised together are more similar in their schooling than half siblings. By some estimates, about 20 percent of the variation in how long people stay in school is explained by the variation in people's genes.

In 2013, a team of scientists headquartered at Erasmus University Rotterdam brought together data from dozens of medical studies. They looked for variants in the DNA from more than 100,000 people that
correlated with their educational attainment. They found dozens of variants that were more common in people who finished more school than in those who left early.

How long people stay in school depends on a lot of factors, such as motivation and attention. But intelligence plays a part, too: A small amount of the variation in the years people spend in school is explained by their intelligence test scores. The Rotterdam team suspected that a few of the genetic variants that influence educational attainment also influence intelligence. They picked out sixty-nine variants from their educational attainment study and investigated whether any of them had a link by examining 25,000 people who had provided DNA and had taken intelligence tests. In 2015,
they reported three variants. Each of the three variants could lift a person's IQ score only three-tenths of a single point. They did not explode like fireworks. They merely popped like champagne bubbles.

These successes spurred other researchers to merge their studies, hoping to find more variants in larger groups of people. In a 2017 study, an international team analyzed
nearly 80,000 people. They found fifty-two genes, which they were then able to confirm by turning to other groups of people and finding them once more. Altogether, the genes still account for only a small percent of the variation in people's test scores. And when the scientists looked at each gene's function, no compelling story of biology emerged. A few of the genes control the development of cells throughout the body. A few others are responsible in particular for various tasks inside neurons. Some may work in hidden pathways that scientists have yet to uncover.

If intelligence, like height, turns out to be omnigenic, the fifty-two genes will be only the start of a long list that will keep growing for years. Perhaps there will be a core of genes that shapes the brain in ways that influence intelligence test scores. But the search will take scientists out to rings after rings of more distant networks of genes. And even if scientists gain all this knowledge, they will still be a long way from a complete understanding of intelligence itself.

To Galton, Pearson, and their fellow hard-core hereditarians, intelligence seemed a case in which nature trounced nurture. Henry Goddard even convinced himself that all feeblemindedness could be
explained by a single Mendelian mutation. In the extreme form of this view, intelligence was like blood types. Your blood type has nothing to do with whether your parents told you to turn off the television, or whether you ate three decent meals a day, or if you got chicken pox in grade school. Your type was fixed as soon as your parents' genes came together to form a new genome.

Intelligence is far from blood types. While test scores are unquestionably heritable, their heritability is not 100 percent. It sits instead somewhere near the middle of the range of possibilities. While identical twins often end up with similar test scores, sometimes they don't. If you get average scores on intelligence tests, it's entirely possible your children may turn out to be geniuses. And if you're a genius, you should be smart enough to recognize your children may not follow suit. Intelligence is not a thing to will to your descendants like a crown.

As hard as it has been for scientists to tease out the genes involved in intelligence, mapping the influence of the environment is even harder. It requires
venturing into a daunting wilderness that lies outside the mathematical tranquility of genome-wide association studies. Psychologists who want to figure out the environment's contribution to intelligence have to take into account kindness and trauma, the biochemistry of the womb and the impact of stress on the brain. The influences of the environment cannot be snapped apart into distinct chunks the way genetic variants can.
They ramify each other, forming the mycelium of experience.

One reason for this complexity is that intelligence, like height, develops. In an embryo, it does not yet exist. Children need a few years of growth and experience before they can get a meaningful, predictive score on an intelligence test. All along that path, experiences can influence how intelligence develops, and different experiences can lead to different intelligence test scores. While the environment can probably influence intelligence in many subtle ways, scientists understand a few strong effects best.