Does the world need another IQ test? There are many well-validated tests, and also a series of short tests suitable for large-scale surveys, many of which take less than 10 minutes, and several useful ones that take less than 5 minutes. However, if you’re looking for a good measure of the variety of human achievement, it might be worth stretching the web even further to capture every gem of intellectual prowess. In that case, who would sit down to conduct such an evaluation, given that the gold standard Wechsler tests take more than an hour?
One approach, in which Roberto Colom’s Madrid team has stood out, is to capture the assessment in the form of a computer game. This makes it accessible to a much wider audience and increases the number of more far-reaching intelligence tests that can be used as part of genetic studies of intelligence. In fact, the “gamification” is quite slight, just a superficial sheen, but it seems to have been enough.
Now Robert Plomin’s team has added more to their own published game-based intelligence test, and has exciting new results to report. The names of the authors are a good list of the new wave of intelligence investigators.
They note that genome-wide association studies have not yet accounted for as much variation as can be gleaned from twin studies. Critics have called this “missing heritability,” which is not the point. We know from twin studies that intelligence is inherited, and genome-wide association studies are trying to identify the genes responsible for this outcome. (We know that genetics is powerful in real life, now we have to prove it in theory.) Part of the problem is that larger studies have brought together disparate test results, so the team has designed a 40-item intelligence game that produces a reliable measure (internal consistency = .78, two-week retest reliability = .88 ) of g that they have given to 4,751 young adults from their study of twins.
This new measure of g, which also produces reliable verbal and nonverbal scores, was substantially correlated with standard measures of g collected at earlier ages (r from .42 at 7 years to .57 at 16 years). Pathfinder showed substantial twin heritability (.57, 95% CI = .43, .68) and SNP heritability (.37, 95% CI = .04, .70). A polygenic score calculated from GWA studies of five cognitive and educational traits accounted for 12% of the variation in g, the strongest DNA-based g prediction to date. The widespread use of this attractive new measure will promote research not only in genomics, but also in the biological, medical and behavioral sciences.
Therefore, they have both verbal and non-verbal scores and can generate a gene-based prediction that accounts for 12% of the variance of intelligence, a good result by current standards. The test used item response theory to select the most powerful items and obtain the maximum predictive power from the fewest number of items.
The software is free, and if more genome-wide association studies adopt it, the explained variance may increase from 12% to 30%, assuming that test heterogeneity is a complicating factor that would exceed universal use of this test. . The great advantage of a polygenic risk score for intelligence is that you can get a prediction from birth onwards, which overcomes the problem that the first intelligence tests do not become reliable until about 11 years old and gain in precision at from then on. Early predictions could be a more accurate way to assess whether early teaching has an effect on later intelligence.
In my opinion, it is not worth including the proof of the missing letter. Sure, it’s a basic process measure, but a bit risky in factor terms.
This large sample of 4,751 25-year-olds shows significant sex differences in favor of men. The authors do not comment on this, but it fits the emerging pattern of a male intellectual advantage in adulthood.
Heritability was 57%, shared environmental influence was 8%, and multivariate polygenic scores predicted up to 12% of the variance. The latest finding, that 12% of the variance of Pathfinder g can be predicted by DNA, makes it the strongest polygenic score predictor of g reported to date. Although 12% is only one-fifth of the heritability estimate from the twin study, we expect that the addition of Pathfinder g in large biobanks will improve the performance of GWAS 21 meta-analytical analyzes by increasing the sample size and decreasing the heterogeneity of cognitive measures. It should be possible to use the brute force method to increase the sample size, especially with less heterogeneity of measures, to close the gap from missing heritability of 12% to SNP heritability of approximately 30%.
It will be difficult to exceed 30% of the variance, to reach the 60% heritability revealed by twin studies. That will require whole genome sequencing.
The bottom line is that the team has created a good new 15-minute IQ test that correlates well with the many longer assessments used over the years on their large sample of twins. It also has good predictive power. If it is more widely adopted and the few bits of the explanatory English language translated into other languages, it could be a very useful contribution to GWAS ‘great research on the genetic basis of intelligence.
You can get Pathfinder here: