Across 1,396 healthy adults aged 36 to 108 (2,878 visits, mean 5.0 years of follow-up), five modifiable factors tracked with cognitive performance at a given age: lower education, diabetes, weak motor function, hearing loss and poor vision (all FDR q < 0.05). None predicted how fast cognition declined. Plasma Alzheimer's markers behaved differently: they linked to cognition only from age 65 onward and predicted faster decline, most consistently pTau217 positivity (0.058 SD per year on the PACC composite, p < 0.001). The risk-factor pattern held among pTau217-negative participants.
This two-component picture, a cognitive level shaped by modifiable exposures and a slope driven by emerging Alzheimer's pathology, challenges the intuitive assumption that fixing risk factors slows decline. It fits the cognitive reserve idea: education, sensory input and vascular-metabolic health may set a higher or lower starting point without changing the rate at which pathology erodes it. A person with hearing loss might score lower today yet decline no faster, which matters for how we interpret trials. Practically, managing diabetes, correcting hearing and vision, and staying physically capable may still buy years of better function, but this analysis does not show they delay decline. Limitations are real: observational data cannot prove causation, a cohort of healthy volunteers is likely more educated and healthier than the general population, and five years of follow-up is short for slow trajectories. This is a preprint that has not been peer reviewed, so findings may shift. The work is confirmatory-to-incremental, but its separation of level from slope is a useful framework.