The study, conducted as part of the PASOCA project (Physical Activity and Sleep for Optimal Cognitive Ageing), followed 233 cognitively healthy, community-dwelling adults aged 55 and above in Flanders, Belgium, across three annual assessment points over two years. Participants, whose average age at baseline was 68.3 years and just over half of whom were women, wore a wrist-worn tri-axial accelerometer (an ActiGraph wGT3X-BT) for seven consecutive days at each time point. The raw accelerometer data were processed with the open-source GGIR software package, which autocalibrated the device signals, detected non-wear periods, identified sustained inactivity bouts to estimate sleep, and classified time into sedentary behaviour, light physical activity and moderate-to-vigorous physical activity using established Hildebrand cut-points based on the ENMO metric. This device-based approach avoided the recall biases that plague self-reported questionnaires and captured all three behaviours continuously across the full day.

Cognitive function, meanwhile, was assessed with the Cambridge Neuropsychological Test Automated Battery (CANTAB), a validated computerised testing platform administered on an iPad. Six tasks—including Delayed Match to Sample, Verbal Recognition Memory, Paired Associates Learning, Spatial Working Memory, One Touch Stockings of Cambridge and the Multitasking Test—yielded standardised composite scores for four distinct domains: short-term memory, long-term memory, executive function and processing speed. The researchers anchored their domain definitions in the Cattell-Horn-Carroll-Miyake cognitive framework, and all z-scores were computed relative to baseline performance, allowing consistent tracking of change over time.

What makes this study methodologically distinctive is its compositional data analysis (CoDA) framework. Because there are only 24 hours in a day, time spent in physical activity, sedentary behaviour and sleep is inherently co-dependent: more of one necessarily means less of another. Traditional regression approaches that treat these behaviours as independent variables suffer from multicollinearity and can produce misleading results. The CoDA framework circumvents this by normalising the behaviours to a 24-hour total and transforming them into isometric log-ratio (ILR) coordinates through a sequential binary partitioning strategy—first contrasting sleep with waking behaviours, then sedentary behaviour against physical activity, and finally light activity against moderate-to-vigorous activity. These ILR coordinates were then entered as time-varying predictors in linear mixed-effects models with random intercepts, capturing the nested structure of repeated measurements within individuals.

The modelling strategy was deliberately conservative. Missing data from participant attrition were addressed through multiple imputation by fully conditional specification, generating 20 imputed datasets whose estimates were pooled using Rubin’s rule. Covariates were grouped into predefined blocks—sociodemographic factors (age, sex, education, living arrangement), health-related variables (body mass index, hearing impairment, number of prescribed medications) and behavioural indicators (smoking, alcohol use)—and hierarchical models were compared with the D1 multivariate Wald test, allowing each cognitive outcome to retain only the covariates that genuinely improved model fit. This guarded against overadjustment bias while accounting for the well-documented influences of education, social isolation and health status on late-life cognition.

The headline finding is striking in its specificity: time-use composition was significantly associated with executive function (p = 0.005) but showed no significant relationship with short-term memory, long-term memory or processing speed. Crucially, the interaction between time-use composition and time was not statistically significant, meaning the association between daily behaviours and executive function remained stable across the entire two-year follow-up. This is the first study to establish such stability using repeated device-based measurements, addressing a major gap left by prior cross-sectional work that could only snapshot a single moment in the ageing process.

Exploratory post-hoc compositional isotemporal substitution analyses then probed which behaviour was driving the executive function association. These simulations hypothetically reallocated time in five-minute increments, up to 30 minutes, between pairs of behaviours while holding all else constant. The results pointed unambiguously to moderate-to-vigorous physical activity (MVPA) as the primary driver. Shifting 30 minutes from light activity to MVPA was associated with estimated executive function gains of 0.11 standard deviations at baseline and 0.17 at the first follow-up, with similar directional patterns at the second. Reallocating 30 minutes from sleep to MVPA produced smaller but still significant gains at baseline and follow-up 1, while shifts from sedentary behaviour to MVPA showed a similar but weaker pattern. Conversely, moving time away from MVPA—in any direction—was consistently associated with lower executive function, including a significant drop of 0.19 standard deviations when 30 minutes of MVPA was replaced with light activity at follow-up 1.

One unexpected nuance deserves attention: hypothetically shifting time from light activity to sedentary behaviour was also associated with modestly higher executive function scores at baseline and follow-up 1. The authors caution against overinterpreting this, but suggest that the cognitive context of sedentary activities—reading, social interaction, computer-based tasks—may partially explain it, echoing prior evidence that cognitively engaging sedentary behaviours differ from passive screen time. It reinforces a growing argument in the field that the quality and context of behaviours matter, not just their raw durations. At the same time, the researchers stress that nothing here diminishes sleep’s established importance for brain health; because MVPA occupied only about 4–5% of the day in this cohort, even small absolute reallocations represent a large proportional increase in vigorous activity, and trimming light activity or sitting time rather than sleep may be the more realistic route for most older adults.

Why would vigorous movement, in particular, benefit the brain’s executive machinery? The study’s discussion draws on established neurobiological mechanisms from the physical activity literature. Higher-intensity exercise stimulates the release of brain-derived neurotrophic factor (BDNF), insulin-like growth factor 1 (IGF-1) and vascular endothelial growth factor (VEGF), molecules that support angiogenesis and neurogenesis. It also increases lactate production, which can serve as an auxiliary energy source for the brain, while reducing pro-inflammatory cytokines such as tumour necrosis factor-alpha and oxidative stress. Structural brain changes have been documented in response to aerobic training as well—most famously, a 12-month moderate-intensity aerobic exercise programme produced significant hippocampal volume increases in adults aged 55 to 80. The present study did not measure these biomarkers directly, so the mechanistic chain remains inferential, but the behavioural-to-cognitive link it documents is consistent with this biological framework.

The authors are candid about their study’s limitations. The two-year follow-up may be too short to capture meaningful cognitive decline in a sample of healthy, highly educated volunteers—a classic healthy-volunteer bias that limits generalisability. Cognitive scores improved over time in short-term memory and executive function, a pattern most plausibly attributable to practice effects from repeated testing, which ranged from 0.01 to 0.20 standard deviations and fell within the ‘optimal’ range documented in longitudinal cognitive research. The GGIR algorithm classifies inactivity but cannot distinguish standing from sitting, and the study did not capture behavioural context or sleep quality. Still, the longitudinal design, device-based measurement across all 24 hours, compositional statistical framework and multi-domain cognitive battery represent a substantial methodological advance. The message for ageing populations is actionable: protecting or increasing moderate-to-vigorous physical activity—ideally by trading in sedentary or light-activity time rather than sleep—may help safeguard the executive functions that underpin independence, planning and quality of life in later years. Future interventional studies must now test whether deliberately restructuring the 24-hour day can produce those cognitive benefits in practice.

Subject of Research: Longitudinal associations between 24-hour movement behaviours and cognitive function in cognitively healthy adults aged 55 and above

Article Title: Longitudinal associations between 24-hour movement behaviours and cognitive function in adults aged 55 and above

Article References: Marent, P.-J., Cardon, G., Albouy, G., & van Uffelen, J. (2026). Longitudinal associations between 24-hour movement behaviours and cognitive function in adults aged 55 and above. Journal of Activity, Sedentary and Sleep Behaviors, 5(1), Article 9. https://doi.org/10.1186/s44167-026-00100-7

Image Credits: AI Generated

DOI: 10.1186/s44167-026-00100-7

Keywords: physical activity, sedentary behaviour, sleep, cognitive ageing, executive function, moderate-to-vigorous physical activity, compositional data analysis, accelerometry, CANTAB, memory, processing speed, healthy ageing