Among 2,510 Indians aged 60 or older assessed twice (2017-2019 and 2022-2024) in the HCAP/LASI cohort, the link between years of schooling and cognitive change differed significantly across 18 states and union territories (change-score F = 2.97; ANCOVA F = 4.11; both P < .001). State estimates ranged from -0.056 SD per schooling year in Odisha to +0.021 SD in Kerala. Pooled, the association reversed from -0.022 SD to +0.031 SD once baseline cognition was adjusted.
The sign reversal is the real story. More-educated people start higher, so they have more to lose and regress toward the mean; the unadjusted "steeper decline" likely reflects this statistical artifact rather than education speeding up aging. Once baseline is held constant, schooling looks mildly protective, in line with the cognitive-reserve literature from Western and Chinese cohorts. The geographic heterogeneity suggests that schooling quality, literacy, health systems, and nutrition differ enough across states that one national coefficient misleads. Kerala's pattern fits its long-standing investment in education and public health. Limitations are material: schooling years are a crude proxy for educational quality, 18 state estimates rest on modest subsamples, attrition between waves may be selective, and the analysis is observational and cannot establish causation. This is a preprint that has not been peer reviewed, so figures may shift. For readers, the practical message is modest: lifelong learning and education access probably support cognitive resilience, but the effect depends on context. The work is incremental, a useful methodological caution rather than a paradigm shift.