The intelligence gene that turned out to be a thousand tiny ones

For years, studies reported that one gene or another was linked to intelligence. In 2012 a team led by Christopher Chabris tried to repeat several of these findings in large groups of people, and most did not hold up. They concluded that most reported links were probably false positives, the result of small studies and chance.

Researchers then moved to much larger studies that scan the whole genome at once. In 2018 a team led by Jeanne Savage and Danielle Posthuma combined data from about 270,000 people. They found 205 regions of DNA linked to scores on intelligence tests, involving around 1,000 genes. Each one had a tiny effect.

Twin and family studies suggest that genetic differences account for a substantial share of the differences in test scores between people in a given population. That is a statement about groups in a particular environment. It does not say how much of any one person's ability comes from genes, and it can change when environments change.

Intelligence, like height, depends on a great many genes and on environment, including nutrition, schooling, health and family life. No single gene makes someone clever, and genetic tests sold to predict a child's intelligence can tell you very little about an individual.

The AI angle: complex outcomes rarely have one cause. Be careful of any tool, genetic or AI, that claims to predict a person's ability or future from one score.

The idea behind this caseLesson: How machines learn →

1. 1Be sceptical of headlines about "the gene for" a complex trait.
2. 2Check whether a study was large and has been repeated by other teams.
3. 3Remember that heritability describes groups, not what an individual can achieve.
4. 4Be wary of genetic tests that claim to predict a child's intelligence or talent, and ask a doctor or genetic counsellor what a test can really tell you.