Neuropathology determines whether brain systems segregation benefits cognitive performance.

The human brain is a large-scale network, containing multiple segregated, functionally specialized systems. With increasing age, these systems become less segregated, but the reasons and consequences of this age-related reorganization are largely unknown. Thus, after characterizing age- and sex-specific differences in the segregation of global, sensorimotor, and association systems using resting-state functional MRI data, we analyzed how segregation relates to cognitive performance in both classical and eye movement tasks across age strata and whether this is influenced by the degree of neuropathology. Our analyses included 6,455 participants (30-95 years) of the community-based Rhineland Study. System segregation indices were based on functional connectivity within and between 12 brain systems. We assessed cognitive performance with tests for memory, processing speed, executive function, and crystallized intelligence and oculomotor tasks. Multivariable regression models confirmed that brain systems become less segregated with age (e.g., global segregation: standardized regression coefficient (ß) = -0.298; 95% confidence interval [-0.299, -0.297], p < 0.001) and that in older age this effect is stronger in women compared to men. Higher segregation benefited memory (especially in young individuals) and processing speed in individuals with mild neuropathology (not significant after multiple testing correction). segregation benefited crystallized intelligence in 46- to 55-year-olds. Associations between segregation indices and cognition were generally weak (ß ~ 0.01-0.06). This suggests that optimal brain organization may depend on the degree of brain pathology. Age-related brain reorganization could serve as a compensatory mechanism and partly explain improvements in crystallized intelligence and the decline in fluid cognitive domains from adolescence to (late) adulthood.

© 2025 The Authors. Published under a Creative Commons Attribution 4.0 International (CC BY 4.0) license.

PMID: 40937157