Back

Patterns of brain-wide associations reflect socioeconomics

Marek, S.; Donohue, M. R.; Karcher, N.; Hoyniak, C.; Chauvin, R.; Meyer, A.; Miller, J.; Van, A. N.; Wang, A.; Baden, N.; Suljic, V.; Scheidter, K.; Monk, J.; Whiting, F.; Ramirez-Perez, N.; Krimmel, S.; Metoki, A.; Paul, S. E.; Gorelik, A.; Hendrickson, T. J.; Malone, S. M.; Schwarzlose, R.; Cardenas-Iniguez, C.; Herting, M.; Petersen, S.; Luby, J.; Randolph, A.; Shanahan, M.; Turkheimer, E.; Kay, B. P.; Gordon, E. M.; Laumann, T. O.; Barch, D.; Fair, D. A.; Tervo-Clemmens, B.; Dosenbach, N. U.

2025-12-13 neuroscience
10.64898/2025.12.10.693206 bioRxiv
Show abstract

Previous brain-wide association studies (BWAS) cross-sectionally linked a specific behavioral trait, most commonly IQ or psychopathology, to variation in brain function or structure. Here, we expanded the focus of BWAS from effect sizes to interpretability and generalizability by mapping 649 variables to brain function and structure. We compared the resultant BWAS maps to other types of brain data to annotate the BWAS patterns. Socioeconomic status (SES) -- not IQ or psychopathology -- showed the strongest associations with both resting-state functional connectivity (RSFC) and cortical thickness in the Adolescent Brain Cognitive Development (ABCD) Study. A principal exposome brain pattern, anchored to sensory and motor cortex, captured 34% of the variance across all BWAS maps. This exposome pattern was strongly correlated with the SES and IQ BWAS maps and non-BWAS maps of sleep (EEG), norepinephrine (PET), and stimulants (drug trial), but not cognitive activation maps (task fMRI). Adjusting for SES, reduced brain-IQ associations by 40%. Brain with IQ associations did not generalize, as they could no longer be detected in subsamples drawn from only higher SES backgrounds, while brain with SES associations remained strong in higher-IQ-only subsamples. These findings reveal SES as the principal axis of population-level brain variation, possibly stemming from the sleep deprivation and heightened stress associated with lower SES, since socioeconomics can only indirectly affect the brain.

Published in Science (predicted rank #2) · training set

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.