Expanding understanding of link between genetics and cognitive ability
Researchers at the Wellcome Sanger Institute, Amsterdam University Medical Center, and their collaborators, built on previous studies by statistically guessing problem-solving test scores for UK Biobank participants who never completed the test. This reduced the bias that comes from analysing only the people who chose to take the test and increased the number of people from around 270,000 to more than 455,000.
This study, published today (5 October) in Nature Genetics, found that adding the estimated scores increased the number of common DNA changes associated with differences in test scores from 390 to 550, a 40 per cent increase.
The gain was even larger for rare genetic variants. The number of genes where rare genetic changes are linked to differences in cognitive ability rose from 5 to 26. This includes 14 that were already known to play a role in neurodevelopmental conditions.
It was well-established before this study that genes influence cognitive ability, alongside environmental factors. This study helps identify some of the genes involved to help us better understand these factors and their impact.
UK Biobank is a large research resource of biological, health and lifestyle information, containing data from around half a million people in the UK. Amongst other things, the participants were asked to complete a problem-solving test called the verbal numerical reason test. However, around 40 per cent of its participants never completed this test, and those that did were more likely to have higher education qualifications. This bias in the sample population, known as ascertainment bias, can lead to problems in downstream genetic studies.
To increase the size and representativeness of the dataset for studying cognition, the researchers used other information about these participants, including health, behaviour and socioeconomic measures, to statistically guess the likely test scores.
They then analysed the larger dataset to understand more about how genetic changes may influence cognition. In total, the team identified 550 common genetic variants, each with a very small effect on test scores. They also found rarer genetic differences that can have much larger effects, pinpointing 26 genes, all associated with lower test scores. Fourteen of these genes are already known to cause neurodevelopmental conditions, and four had previously been linked to measured cognitive test scores or education in UK Biobank.
The remaining eight had no strong previous evidence linking them to cognition or neurodevelopmental conditions. However, the team showed that in patients with neurodevelopmental conditions these genes had a higher frequency of genetic changes, implying that they probably do contribute to risk.
Most people carrying these rare variants in UK Biobank do not have a diagnosis of a neurodevelopmental disorder. This suggests these rare genetic effects can influence cognition even without causing people to reach a clinical threshold for diagnosis.
This research helps add to the wider understanding of how genetic changes influence cognition and could give insights into neurodevelopmental conditions. The researchers stress that the results are strictly to improve the scientific understanding of cognition at a group level, not to make individual predictions about cognitive ability. The estimated scores should not be treated as equivalent to directly measured intelligence, and the researchers emphasise that better measurement of cognition in large studies remains the most robust way forward.
“The challenge was making sure the estimated scores still captured cognitive ability, rather than simply reflecting the other information we used to calculate them. We tested that extensively. Through this approach of statistically guessing unmeasured scores, we were able to expand the study to nearly half a million people, making it the largest study of its kind to date. These results will help us learn more about how cognition relates to social and health inequalities.”
Dr Abdel Abdellaoui, co-senior author at Amsterdam University Medical Center
“The findings from genetic studies can be biased if they are based on a non-random subset of people. In this case, we were able to include information from over 170,000 additional participants, to give a more representative picture of how genetics influences cognition scores and to uncover rare DNA differences linked to cognition that we previously did not have enough statistical power to find.”
Dr Daniel Malawsky, co-first author at the Wellcome Sanger Institute
“By mapping genetic variants across nearly half a million people, we’ve shown that damaging DNA differences in the same genes that cause severe neurodevelopmental conditions can also affect cognitive traits across the wider population. Learning more about these impacted genes could help us better understand how the brain develops and why DNA changes in these genes lead to neurodevelopmental conditions in some people but not others.”
Dr Hilary Martin, co-senior author at the Wellcome Sanger Institute
More information
A full Q&A written by the research team explaining more about the study, along with the background and impact of their work, can be found here: https://github.com/dmvandenberg/fluid-intelligence-imputation
Publication:
D. M. van den Berg, D. S. Malawsky, W. Huang, et al. (2026) ‘Imputation of fluid intelligence scores reduces ascertainment bias and increases power for analyses of common and rare variants’. Nature Genetics. DOI: 10.1038/s41588-026-02787-5
Funding:
This research was funded in-part by Wellcome. A full acknowledgement list can be found in the publication.
