Abstract
People with larger brains tend to score higher on tests of general intelligence (g). It is unclear, however, how much variance in intelligence other brain measurements would account for if included together with brain volume in a multivariable model. We examined a large sample of individuals in their seventies (n=672) who were administered a comprehensive cognitive test battery. Using structural equation modelling, we related six common magnetic resonance imaging-derived brain variables that represent normal and abnormal features—brain volume, cortical thickness, white matter structure, white matter hyperintensity load, iron deposits, and microbleeds—to g and to fluid intelligence. As expected, brain volume accounted for the largest portion of variance (~12%, depending on modelling choices). Adding the additional variables, especially cortical thickness (+~5%) and white matter hyperintensity load (+~2%), increased the predictive value of the model. Depending on modelling choices, all neuroimaging variables together accounted for 18–21% of the variance in intelligence. These results reveal which structural brain imaging measures relate to g over and above the largest contributor, total brain volume. They raise questions regarding which other neuroimaging measures might account for even more of the variance in intelligence.
Highlights
Half or more of the variance in human intelligence test performance is accounted for by the general factor of cognitiveS.J
In a large sample of older adults, we showed that a selection of brain imaging measurements account for about 20% of the variance in a latent trait of general cognitive ability
The highest percentages of variance accounted for were found in models using a ‘fluid’ factor of intelligence, but the percentages for a comprehensive, fifteen-test general intelligence factor were similar
Summary
Measures of the networks that support information transfer within the brain have shown predictive validity for cognitive ability; in the same sample analyzed in the present study, a general factor of brain white matter tract structure measured by diffusion tensor MRI accounted for about 10% of the variance in g (Penke, Muñoz Maniega et al, 2012). No studies to date have included all of these structural brain variables together in a single model to assess their incremental predictive validity for intelligence It is unclear whether they would each account for separate portions of variance, or whether the finer-grained variables would account for little after more global measures such as total brain volume are included. What is the best estimate of the percentage variance in g accounted for by the above brain measures when they are modelled together? Second, which brain imaging parameters have significant associations with g beyond total brain volume?
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
