Longitudinal associations between loneliness and cognitive performance in older adults: evidence from intensive longitudinal assessment

Higher loneliness has been linked to cognitive decline in older adulthood, yet its directionality remains unclear. Most prior studies rely on infrequent assessments, obscuring short-term variability and longer-term change. We examined longitudinal associations between loneliness and cognition in both directions using ecological momentary assessment (EMA) across annual measurement bursts.

Data were drawn from 318 older adults (ages 70-90) in the Einstein Aging Study. Participants completed up to six annual 14-day EMA bursts, including repeated measures of momentary loneliness and concurrent smartphone-based tasks assessing spatial working memory, processing speed, and associative memory. Multilevel growth models tested whether baseline loneliness predicted cognitive changes and whether baseline cognitive performance predicted changes in loneliness across bursts.

Limited evidence was observed for longitudinal associations in either direction. Baseline loneliness did not predict change in most cognitive domains, and baseline cognitive performance did not predict change in loneliness. The only exception was associative memory, where higher baseline loneliness predicted a lower likelihood of error-free performance over time. Loneliness increased modestly across bursts, and depressive symptoms predicted higher baseline loneliness and steeper increases over time.

Baseline cognitive performance did not predict subsequent changes in loneliness. When examined in the other direction, associations between loneliness and cognition appeared limited and domain-specific, with only associative memory showing a longitudinal association with baseline loneliness.