Over the years there have been many attempts to study the relationship that gender, class, race and much more besides have with our intelligence quotient, or IQ. The most notorious example is perhaps the 1994 book The Bell Curve, which discussed race (whatever that is) and intelligence, and sparked fears that the "cognitively challenged" would outbreed the "cognitive elite", thus "dumbing down" America.
Today, however, I can report hard evidence of what most of us have suspected all along: previously reported links between general intelligence and "race" are bunkum. In the journal Neuron, an article by me, along with Adrian Owen and Adam Hampshire of the University of Western Ontario, shows that the unitary nature of "IQ" is too simplistic to capture the spectrum of human intelligences.
The origins of IQ lie in 1904, when the English psychologist Charles Spearman observed that cognitive performance was linked across different tasks. In other words, if you have a good memory, you tend to be good at reasoning too. He proposed that there was a single dominant general intelligence factor called "g", a simple idea with radical implications. Once you can measure g you can generate an intelligence quotient (IQ) and probe its links to race, gender and so on.
But did Spearman's idea of a single "generalised intelligence" say more about the limits of his tests than about the abilities of the human mind? My attempt to find an answer began five years ago, after a conversation with Owen, who is best known for his work on reading the minds of patients in a vegetative state. With Hampshire, he had devised a reliable way to carry out cognitive tests online to monitor rehabilitation after brain injury, the effect of smart-drug trials and so on.
We used a dozen tests to probe the broad range of cognitive skills believed to depend on general intelligence: planning, reasoning, attention and working-memory performance. With some fanfare, we launched them with the help of New Scientist, Discovery and The Daily Telegraph.
Of the 110,000 participants, around 60,000 completed the half-hour test and a questionnaire. After removing the young (under 12) and the very old (over 70), we were left with 44,600 sets of data, including more than a million data points.
Hampshire then came up with a clever way to find out whether g really exists by "fractionating" people's performances. Using principal-components analysis (a trick pioneered by Spearman) to find underlying structure within a large body of data, he could ask whether there is a single factor, such as g. The answer, as revealed in Neuron, is an emphatic no. You can explain the observed variations in intelligence with at least three factors: short-term memory; reasoning; and, finally, the verbal factor. But no fewer.
We wanted to be sure. Did the three factors activate three separate brain circuits? Hampshire used MRI to study the brains of 16 participants. Each of the three different factors identified by the principal-components analysis did indeed correspond to a different brain network.
We can now say, with certainty, that the idea that populations can be compared using a single measure of intelligence is dead.
More usefully, and controversially, it could help disentangle the effects of genetics, lifestyle and education on these three factors and, in turn, the effects of these three factors on other aspects of our lives. Intelligence is correlated with many aspects of wellbeing, including success at work and lifespan, and these measures could provide deeper insights.
Roger Highfield works for the Science Museum Group in London. Help him investigate by measuring your own intelligences using his new test.
This article was originally published by WIRED UK
