Researchers wanted to know if brain scans could predict how well someone with bipolar disorder would score on an IQ test, not just spot general differences between bipolar patients and healthy people as a group, but actually estimate an individual’s specific scores. To do this, they used resting-state fMRI to map patterns of connectivity between different brain regions. Feeding this connectivity data into a prediction model, they tried to forecast each patient’s verbal IQ, performance IQ, and overall full-scale IQ.
The model worked: it could meaningfully predict all three IQ scores from brain connectivity data alone. The connections that mattered most weren’t confined to one brain region. They spanned networks involved in vision, touch/movement, and two networks responsible for directing attention. That makes intuitive sense, since taking an IQ test requires paying attention to and processing sensory information. Interestingly, many of these same connections have shown up in prior research as abnormal in bipolar patients, linking them to known attention and cognition problems in the disorder. The researchers also found that a lot of the same brain connections predicted both verbal and visuospatial IQ, and that these shared connections tended to cluster more on the left side of the brain (a pattern that differs from typical healthy brain organization), hinting that bipolar disorder may shift how the brain divides labor between its two hemispheres.
This is an early but promising step toward using brain scans as an objective, quantifiable tool to assess cognitive functioning in bipolar patients. It is potentially useful because traditional IQ testing can be hard to administer reliably when someone’s mood symptoms are active. The authors are careful to note real limitations, though: this was a single dataset from one research site, didn’t account for medication effects or the wide variability in how bipolar disorder presents across patients, and used a relatively simple statistical model. So while the results are encouraging, they’ll need to be replicated in other patient groups before this could move toward real clinical use.
Link to full article: https://www.sciencedirect.com/science/article/pii/S0361923025000504