What older adults do on screens may matter more than total screen time

By Tarun Sai LomteReviewed by Susha Cheriyedath, M.Sc.Sep 29 2026

Researchers combined four aging cohorts with genetic analyses to examine whether the type and duration of screen use could help explain differences in cognitive aging.

Study: Screen time modalities and neurocognition in older adults: multi-cohort and Mendelian randomization evidence. Image Credit: fizkes / Shutterstock

In a recent study published in the journal npj Digital Medicine, researchers examined associations between television viewing, computer use, and cognition among older people and used Mendelian randomization to investigate potential causal relationships.

Global dementia prevalence is projected to reach 152 million by 2050. The accelerating normalization of living alone and the aging population have transformed the relationship between digital technology and older adults. Older people are increasingly becoming active users of screen media. Whether this behavioral shift contributes to cognitive aging, or whether certain forms of digital engagement are associated with greater cognitive resilience, remains to be investigated.

About the study

In the present study, researchers investigated the associations of television viewing and computer use with cognition in older individuals. The study used a combined approach integrating observational epidemiology and genetic causal inference. First, the team assessed associations between computer use and television (TV) viewing time and cognitive domains across four aging cohorts.

Cohorts included the Survey of Health, Aging, and Retirement in Europe (SHARE), the English Longitudinal Study of Aging (ELSA)–Harmonized Cognitive Assessment Protocol (HCAP), the Health and Retirement Study (HRS)- HCAP, and the National Health and Nutrition Examination Survey (NHANES). Cognitive indicators included processing speed/orientation, delayed recall, immediate recall, and animal fluency; a global cognition score was computed from these indicators.

Multivariable-adjusted linear regression models examined associations between screen time and cognitive outcomes. Mendelian randomization (MR) analyses were performed to investigate potential causal associations of genetically predicted computer use time and TV viewing time with neuroimaging outcomes, cognitive outcomes, and dementia subtypes. MR analyses used summary statistics from large-scale genome-wide association studies.

Neuroimaging outcomes included regional brain volume, cortical surface area and thickness, and resting-state functional connectivity, while cognitive outcomes covered overall and domain-specific measures. Four dementia subtypes were considered: Alzheimer’s, frontotemporal, Lewy body, and vascular dementias. An exploratory two-step MR mediation analysis was used to examine whether neuroimaging phenotypes might mediate the associations between screen time and cognitive function.

Findings

The observational analyses included 8,996 participants, with mean ages ranging from 69.2 to 76.2 years. Around 40%–49% of participants across cohorts engaged in prolonged TV viewing (≥ four hours per day), while 30%–40% reported no computer use. Computer non-use and prolonged TV viewing were associated with older age, physical inactivity, lower cognitive scores, greater burden of chronic diseases, and lower socioeconomic status.

TV viewing time was associated with poorer global cognition. By comparison, computer use was associated with higher global cognition. Effect estimates were highest for two to three hours of computer use/day in the ELSA-HCAP, HRS-HCAP, and NHANES cohorts, and for up to one hour per day in the SHARE cohort.

At four or more hours/day, effect estimates for computer use attenuated. This meant that the positive association with cognition may not increase with longer computer use. Domain-specific results broadly matched the overall pattern, with TV associations most consistent for immediate recall and computer-use associations more consistent for processing speed/orientation and delayed recall, though results varied across cohorts. MR estimates of TV viewing time showed nominal inverse associations with total cortical and insular surface area and rostral anterior cingulate thickness, but none of these cortical findings survived multiple-testing correction.

MR estimates for computer use time showed a nominal negative association with the pars triangularis surface area, which also did not survive multiple-testing correction. MR estimates for TV viewing time were associated with lower left insula volume after stringent multiple-testing correction, while evidence for the right insula was weaker. No regional brain volume association for genetically predicted computer use time survived multiple-testing correction, with nominal positive associations observed in the left posterior cingulate and left supramarginal gyrus and nominal inverse associations in the left cuneus and left lateral orbitofrontal region.

MR estimates for TV viewing time and computer use time exhibited nominal inverse and positive associations with functional connectivity in frontal and parietal-frontal regions, respectively, but none of the resting-state functional connectivity findings survived multiple-testing correction. MR analyses linked TV viewing time to lower cognitive function and performance, whereas computer use time was associated with better cognitive performance; however, its association with the separate overall cognitive function outcome was not statistically significant.

There was no convincing evidence that genetically predicted computer use or TV viewing time was associated with the risk of dementia subtypes. Finally, two-step MR suggested that lower left insula volume accounted for about 15% of the association between genetically predicted TV viewing and poorer cognitive performance. The cohort questionnaires recorded computer-use duration but not its purpose, context, or cognitive demands, and did not provide harmonized measures of smartphone use. The genetic datasets were largely of European ancestry, and the MR models could not capture the non-linear patterns seen in the observational data. The authors cautioned that computer use should be treated as a broad marker of digital engagement rather than a direct measure of cognitively active screen use.

Conclusions

Taken together, computer use and TV viewing had contrasting associations with cognition. TV viewing was inversely associated with global cognition, while computer use was positively associated. Two-step MR suggested that lower left insula volume accounted for part of the association between genetically predicted TV viewing time and poorer cognitive performance. These results support distinguishing among different types and contexts of screen use in cognitive risk evaluation, rather than relying on total screen time.