Cognitive reserve, resting-state network organization, and executive functioning in healthy aging

As the global population ages, understanding cognitive decline in healthy older adults becomes essential. Cognitive decline varies between individuals due to both biological and environmental factors, highlighting the need to identify mechanisms that support preserved cognitive functioning in aging. Cognitive reserve (CR) has been proposed as a key factor supporting preserved cognition in aging; however, its relationship with resting-state network efficiency, inter-network communication, and executive functioning (EF) remains unclear. The present study examined whether CR is associated with specific resting-state cross-modal network properties and whether it influences, in healthy older adults, the relationship between network efficiency or communication and executive performance. Resting-state fMRI data from 289 cognitively unimpaired participants aged 50 years and older were obtained from the Human Connectome Project-Aging (HCP-A) dataset. Functional connectivity matrices were constructed using the Schaefer 400-parcel atlas, with analyses focused on the default mode network (DMN), frontoparietal network (FPN), and dorsal attention network (DAN). Network topology was characterized using within-network global efficiency and between-networks participation coefficients. Primary analyses used an 80% network density, with sensitivity analyses across multiple thresholds. CR was operationalized as a composite measure including education, crystallized abilities, and physical activity. Executive functions were assessed using NIH Toolbox measures of inhibitory control/selective attention (Flanker inhibition task), cognitive flexibility (Dimensional Change Card Sorting Task (DCCS)), task switching (Trail Making Test), alongside a measure of working memory (List Sorting Task). At the 80% network density threshold, models predicting CR from network measures did not reach statistical significance, indicating that CR levels were not consistently reflected by intrinsic network organization in this sample. However, CR emerged as the most robust positive predictor across all cognitive domains. In contrast, resting-state network efficiency and participation coefficients demonstrated domain-specific effects; notably, TMT performance was significantly predicted by age, CR, DMN efficiency, FPN efficiency, and DMN-DAN participation. A significant interaction between CR and DAN efficiency was observed. Across other executive domains, interactions were generally absent, suggesting largely additive rather than stable moderating effects. Taken together, these findings suggest that while CR is a robust predictor of executive functioning, its relationship with resting-state organization is threshold-dependent and domain-specific. The lack of consistent interactions across all domains suggests that CR may support cognitive performance through parallel additive pathways rather than a universal moderating effect on intrinsic network organization. These results reinforce the idea that CR-related neural adaptations are subtle and may not be fully captured by static resting-state metrics, highlighting the need to consider both additive and interactive mechanisms in the aging brain.