Support NeoGAF

[Antiochus]

http://www.wired.com/wiredscience/2013/07/genetics-of-iq/

These kinds of research activities have already been on the radar for the vast majority of psychometricians and social scientists, but its only recently do you have mainstream media pieces starting to pick up the pieces.

Some people are smarter than others. It seems like a straightforward truth, and one that should lend itself to scientific investigation. But those who try to study intelligence, at least in the West, find themselves lost in a political minefield. To be sure, not all intelligence research is controversial: If you study cognitive development in toddlers, or the mental decline associated with Alzheimer’s disease, “that’s treated as just normal science,” says Douglas Detterman, founding editor of Intelligence, a leading journal in the field. The trouble starts whenever the heritability of intelligence is discussed, or when intelligence is compared between genders, socioeconomic classes, or—most explosively—racial groupings.

Since the 1990s, when a book called The Bell Curve (coauthored by a psychologist and a political scientist) waded into this last morass, attempts to quantify or even study intelligence have become deeply unfashionable. Dozens of popular books by nonexperts have filled the void, many claiming that IQ—which after more than a century remains the dominant metric for intelligence—predicts nothing important or that intelligence is simply too complex and subtle to be measured.

For the most part, an IQ test—the most common of which today is called the Wechsler Adult Intelligence Scale—is a series of brainteasers. You fit abstract shapes together, translate codes using a key, sort numbers or letters into ascending order in your mind. It’s a weirdly playful exercise, the sort of test you would expect to have no bearing on anything else. But studies make it clear that IQ is strongly correlated with the ability to solve all sorts of abstract problems, whether they involve language, math, or visual patterns. The frightening upshot is that IQ remains by far the most powerful predictor of the life outcomes that people care most about in the modern world. Tell me your IQ and I can make a decently accurate prediction of your occupational attainment, how many kids you’ll have, your chances of being arrested for a crime, even how long you’ll live.

Critics claim that these correlations are misleading, that those life outcomes have more to do with culture and environmental circumstances than with innate intellectual ability. And even IQ researchers are far from in agreement about whether scores can be validly compared between groups of people—men and women, blacks and whites—who experience very different environments even within the same country. Variations within groups are often greater than the variations between them, making it impossible to draw conclusions about someone based on their group.

But on an individual level, the evidence points toward a strong genetic component in IQ. Based on studies of twins, siblings, and adoption, contemporary estimates put the heritability of IQ at 50 to 80 percent, and recent studies that measure the genetic similarity of unrelated people seem to have pushed the estimate to the high end of that range.

This is an idea that makes us incredibly uncomfortable. “People don’t like to talk about IQ, because it undermines their notion of equality,” Detterman says. “We think every person is equal to every other, and we like to take credit for our own accomplishments. You are where you are because you worked hard.” The very idea of the American dream is undermined by the notion that some people might be born more likely to succeed. Even if we accept that intelligence is heritable, any effort to improve or even understand the inheritance process strikes us as distasteful, even ghoulish, suggesting the rise of designer superbabies. And given the fallout that sometimes results when academics talk about intelligence as a quantifiable concept—such as the case of Harvard president Lawrence Summers, who in 2006 resigned after suggesting that science is male-dominated due not to discrimination but to a shortage of high-IQ women—it’s no surprise that IQ research is not a popular subject these days at Western universities.

Zhao’s improbable rise at BGI began in the summer of 2009, when one of the firm’s founders, a geneticist named Wang Jian, noticed a skinny stranger lurking in the hall. “Hey, what are you doing here?” Wang asked the high school student with a spiky mess of hair. Zhao was 17, and he was there taking part in BGI’s science summer camp. “Why aren’t you in class?” Wang pressed.“It’s boring,” Zhao said.Wang took an immediate liking to him. On a hunch, he pushed Zhao into the hands of Li Yingrui, a recent college dropout who was already one of BGI’s leading scientists. “Do you know any Perl?” Li asked him. Perl is a programming language often used to analyze genomic data. Zhao admitted he did not; in fact, he had no programming skills at all. Li handed him a massive textbook, Programming Perl. There were only two weeks left in the camp, so this would get rid of the kid for good.
A few days later, Zhao returned. “I finished it,” he said. “The problems are kind of boring. Do you have anything harder?”
Perl is a famously complicated language that takes university students a full year to learn. So Li gave him a large DNA data set and a complicated statistical problem. That should do it. But Zhao returned later that day. “Finished.” Not only was it finished—and correct—but Zhao had even built a slick interface on top of the data.
The next morning Li marched into Wang’s office. “This guy is a genius,” he said. “You have to keep him.” So Zhao dropped out of high school, said good-bye to his mother and father—he is an only child, like most Chinese of his generation—and moved to Shenzhen to begin a new life.

Ask Zhao what draws him to IQ as a research subject and invariably he talks about the mysteries of the brain. He’s driven by a fascination with kids who are born smart; he wants to know what makes them—and by extension, himself—the way they are. But there’s also a basic pragmatism at work. By way of explanation, he points to the International Mathematical Olympiad, a tough competition that has helped define China’s approach to math. Two-thirds of students train for it, he says, and its judgment of the talent is so respected that for years high scorers were allowed to skip gaokao, the traditional college entrance exam. But only a tiny fraction of people have the mathematical gifts to be competitive, Zhao says, and this basically comes down to IQ. “You cannot ask a kid with low IQ to just work hard and then become a really talented mathematician,” he says. “It’s impossible.” And yet, Zhao says, that’s what is currently expected in China. He wants to stop the vast majority of Chinese students from wasting their time.[/B

On this part he is immensely correct. In fact that's what's largely happening on the Americana and European side of the Math Olympiads.

The project has already weathered at least one serious false start. Zhao’s original plan made sense on paper: Get IQ scores and DNA samples from a large number of smart people, compare their genomes to normal people, and after some statistical magic, voilà! The genetic markers for IQ would pop right out. So Zhao returned, naturally, to Renmin, the high school he had just abandoned. He worked with Yang Rui, a brilliant Renmin graduate who had worked on the human genome project at BGI as a teenager. Zhao knew nothing about psychometrics, the branch of psychology behind the IQ test. Yang’s psychometric research at Yale and Brown universities made her the perfect collaborator.
But they ran up against an unexpected problem: There was no Chinese IQ test, and Renmin students’ English skills weren’t consistent enough to conduct the test in English. Moreover, the IQ test is supposed to be given as a one-on-one interview, so when Zhao and Yang tried to administer it to whole classes of students at once, the result was chaos and useless data. Worst of all, getting blood samples from students proved nearly impossible, as wary parents began to raise alarms. Principal Liu had to call a meeting with parents to calm their nerves. The collaboration with Renmin was quietly put on hold.
The fortunes of the research effort changed in October 2010, when Zhao met Steve Hsu, a professor who was spending a year in Taiwan while on sabbatical from the University of Oregon. (He is now vice president for research and graduate studies at Michigan State University.) A theoretical physicist who likes to apply his mathematical muscles to a wide range of complex problems, Hsu was intrigued by a news article about BGI that briefly mentioned Zhao’s project. A few emails later, he flew over to give a talk at BGI and soon became a permanent collaborator and adviser to the team.

Since most of the variation in IQ is heritable, scientists have long searched for genetic differences that might account for it. The reason we haven’t found them, Hsu theorizes, is because there aren’t any single genes or even a handful of genes with a big effect on IQ. Instead, the thinking goes, there are as many as 10,000 different locations in the genome where a mutation can affect IQ. According to Hsu’s rough model, all humans carry a few hundred of those 10,000 possible mutations, and each mutation has a tiny negative cost to IQ, on the order of half an IQ point.

If this is right, then the difference between a brilliant 150-IQ person and an average 100-IQ person comes down to DNA typos at perhaps 100 of those 10,000 places. Other traits—like height, for example—seem to work the same way, and an ongoing study into the genetics of height has begun to find relevant mutations. Most geneticists who have studied intelligence agree with this theory in broad strokes. At the very least, says Kevin Mitchell, a geneticist at Trinity College Dublin in Ireland who studies brain development, Hsu’s basic theory of many deleterious mutations “is far more plausible than the alternative”—that is, more plausible than the idea that the mutations are building up IQ, not knocking it down.
The only way to unravel the genetics of a trait scattered among 10,000 possible DNA variants is to use something called GWAS (genome-wide association study). Rather than identifying the variants that cause a trait, as can be done with so-called Mendelian traits like finger length or earwax type, you find the variants associated with the trait.
But—and this is crucial—the implications of this math are that it will take far more than a few thousand genomes to solve the puzzle of intelligence. Given the small sample size they have so far, Hsu hopes they’ll start by finding one or two genes associated with intelligence. A recent Dutch study required more than 125,000 genomes to isolate three variants associated with educational attainment; to create a genomic predictor of IQ, Hsu says, it could take 1 million or more.

The good news for Zhao is that cheap DNA sequencing, together with more creative ways of obtaining DNA, means that a million genomes could be in reach within five years. The genomes don’t all need to be geniuses, because the IQ-affecting genetic markers they’re looking for—DNA typos that drag down intelligence—are more often carried by the 100-IQ people. That is, it’s a relative dearth of these mutations that gives people higher IQ, according to the theory.

Assuming Zhao and his team succeed, there are implications that will trouble many people. Hsu is confident that through embryo screening during IVF, any genetic markers for intelligence that their team discovered would inevitably be used to select for more intelligent babies. Children tend to fall within a spread of 13 IQ points above and below the average IQ of their parents. But sometimes the apple can fall twice as far from the tree—that is, two parents with 100 IQs producing a child with an IQ of 126. Hsu puts the chance of such a positive outlier at around 2 or 3 percent, and it depends mostly on which sperm meets which egg.
If parents use IVF to conceive, then a genetic test—an extension of the screening tests for genetic diseases that are already routinely done on embryos—could let them pick the smartest genome from a batch of, say, 20 embryos. “It’s almost like there are 20 parallel universes,” Hsu says. “These are all really your kids.” You’re just choosing the ones with the greatest genetic potential for intelligence. But effectively, you could be giving an unborn child a boost in IQ above their parents. As Hsu sees it, this is no Faustian bargain. “Aren’t we doing them a great service?” Over the long term, he proclaims, this would “improve the average IQ of the species by quite a bit.” He hopes governments will even provide it for free; Singapore, he predicts, would be the first to sign up.

If anyone wants a good source of this, they are welcomed to read Steve Hsu's personal blog, which devotes more than a few posts to his project:

http://infoproc.blogspot.com/

Here are two of talks that give an excellent overview of genetics and intelligence:

http://www.youtube.com/watch?v=62jZENi1ed8

http://www.youtube.com/watch?v=FgCSkGeBUNg