The P300 event-related potential as a physiological signature of cognitive reserve: links to working memory in aging

Cognitive reserve (CR) may help explain why older adults differ in cognitive functioning despite comparable age- or pathology-related changes. However, the electrophysiological expression of CR across stages of cognitive decline remains unclear. This study examined whether CR predicts P300 event-related potential, and whether working memory (WM) capacity modulated the CR-P300 link across cognitive status groups. Seventy-one older adults [healthy controls (HC), n = 31, Mage = 67.3, SDage = 7.19; mild cognitive impairment (MCI), n = 28, Mage = 70.8, SDage = 7.06; moderate cognitive impairment (MoCI), n = 12, Mage = 70.6, SDage = 6.50] completed a visuospatial WM task (Corsi block-tapping test). Using 32-channel EEG, P300 ERP responses were recorded during an auditory oddball task. CR was quantified using the Cognitive Reserve Index questionnaire, including education, working activity, and leisure activity domains, and global cognition was assessed using the Montreal Cognitive Assessment. Group-stratified hierarchical regression models tested (1) CR prediction of P300 amplitude and latency after adjusting for age and global cognition, and (2) CR × WM interaction. A whole-sample hierarchical regression model was used to test whether the CR-P300 amplitude association differed across cognitive status groups. In HC, CR positively predicted larger P300 amplitude. In MCI, the CR coefficient at mean WM was negative, whereas CR × WM interaction was positive, suggesting that CR-related amplitude patterns may depend on preserved WM capacity. The whole-sample model provided evidence that the CR-P300 amplitude association differed between HC and MCI. In MoCI, amplitude estimates were unstable and showed wide confidence intervals, supporting only exploratory interpretation. CR and CR × WM showed no reliable associations with P300 latency in HC or MCI. However, in MoCI, domain-level latency analyses, occupational and educational reserve showed comparatively larger negative coefficients, suggesting possible associations with shorter P300 latency. However, confidence intervals crossed zero. Thus, latency findings were treated as exploratory and less stable than amplitude effects. These findings suggest that the electrophysiological expression of CR may differ across cognitive status, with P300 amplitude showing preliminary evidence of CR-related variation in healthy aging and WM-dependent modulation in MCI. P300 latency appeared less sensitive to CR-related variation in the present sample.