Cognitive network plasticity across divergent aging trajectories: an exploratory graph-theoretic study

Interindividual variability in response to cognitive training remains a key barrier to developing precision approaches for maintaining neurocognitive health across the lifespan. Characterizing changes in system-level cognitive reorganization may provide insight into mechanisms of cognitive reserve, resilience, and plasticity that shape divergent aging trajectories and vulnerability to age-related cognitive decline. This exploratory study applies a cognitive network framework to examine how relationships among cognitive functions change and reorganize following training. We conducted a secondary analysis of open-access data from a randomized controlled trial comparing computerized cognitive training (Lumosity) with an active control (crossword puzzles), including 4471 participants (mean age 38.7 ± 15.1 years; range 18–80) from the retained analytic sample. Participants were stratified into low- and high-performing groups based on baseline cognitive performance. Cognitive networks were estimated using partial Spearman correlations among seven cognitive tasks, controlling for age, education, estimated training exposure, and self-reported psychological variables indexing affect and cognitive failures. Permutation-based thresholding ensured robust network estimation, and graph-theoretical metrics quantified training-related changes in network integration and segregation. Both interventions were associated with improvements in overall cognitive performance, with larger gains observed among low-performing individuals undergoing computerized training. Network-level changes were modest but systematic: low-performing participants in the computerized training group showed increases in global integration, whereas high-performing participants exhibited reductions in clustering and local efficiency. These findings demonstrate that cognitive network metrics capture subtle, performance-dependent patterns of plasticity expressed as changes in system-level organization, beyond mean cognitive gains and demographic age. Such system-level markers may help elucidate heterogeneity in neurocognitive aging and inform strategies to promote cognitive health.