Coupling of sensorimotor and cognitive functions in middle- and late adulthood

The study investigates how aging relates to and potentially couples changes in sensorimotor and cognitive functions during middle to late adulthood. Motivated by evidence that sensory, motor, and cognitive declines share similar age trajectories and correlated age-related variance, the authors evaluated whether processing speed and executive functions (inhibition, task switching, working memory updating) are linked to sensorimotor abilities (listening in noise, postural control, functional mobility, tactile sensitivity). The central questions were: (1) Do sensorimotor measures (and/or processing speed) mediate age differences in executive functions? (2) Do individuals with relatively poor (or good) cognitive performance for their age also perform relatively poorly (or well) on sensorimotor tasks? (3) Does coupling strengthen with advancing age? The work is important for understanding mechanisms of cognitive–sensorimotor aging and for informing potential intervention and rehabilitation strategies during the critical transition from middle to late adulthood.

The paper reviews three major accounts linking sensory/sensorimotor and cognitive aging: (1) Cascade models propose that peripheral sensory degradation leads to downstream cognitive decline (and potentially reverse effects), including hypotheses like perceptual degradation and cognitive permeation where low-fidelity inputs draw more cognitive resources. Evidence for remediation effects (e.g., hearing aids, cataract removal) is mixed. (2) Common-cause hypotheses suggest shared senescent neural factors (e.g., white matter/myelin changes) drive parallel declines; processing speed has been posited as a central mediator, with prior studies (e.g., Berlin Aging Study) showing speed and sensory measures explain substantial age variance in cognition. (3) Compensation models argue older adults recruit additional cognitive resources to counteract sensorimotor decline; this recruitment may be permanent and more pronounced with age. Prior work indicates correlations between sensory and cognitive functions often strengthen in older age. The review also summarizes age effects and mechanisms in listening in noise (peripheral and central auditory changes, temporal processing, binaural and speech-on-speech masking), postural control (sensory reliability, central integration, SDA metrics), functional mobility (TUG as predictor of falls/disability), tactile sensitivity (peripheral and central changes), and executive processes (processing speed reductions, WM updating demands for speech-in-noise, inhibition’s role in noise susceptibility and balance, cognitive flexibility/task switching in auditory attention and mobility).

Design and participants: Cross-sectional study with four age cohorts: 46–55, 56–65, 66–75, 76–86 years. 112 adults recruited; cognitive screening via modified CODEX; only CODEX A/B included. Eight with C/D excluded; eight additional women from younger cohorts excluded to balance groups, yielding N=96 (24 per cohort). Small subset used hearing aids (6 participants). Demographics included education and estimated IQ using WAIS-III subtests (digit symbol substitution and digit span). Ethics approval obtained; informed consent provided; compensation €11. Procedure: Individual testing in lab or at home (~1.5 h). Listening-in-noise test always first; remaining tests randomized. Sensorimotor measures: - Listening in noise: Flemish Digits-in-Noise (DiN) test via tablet and calibrated headphones; speech at 65 dB A; adaptive SRT determination in broadband noise using triplet and digit scoring. - Postural control: Quiet stance on a Wii Balance Board for 30 s (4 trials after warm-up), eyes open fixed gaze; CoP sampled and processed (interpolation to 100 Hz, 13 Hz Butterworth low-pass). Stabilogram Diffusion Analysis (SDA) metrics: short-term diffusion coefficient (open-loop), long-term diffusion coefficient (closed-loop), and critical time interval. - Functional mobility: Timed Up-and-Go (TUG). Rise from chair, walk 3 m, turn, return, and sit; average time across 3 trials. - Tactile sensitivity: Von Frey monofilament light-touch thresholds on lateral forefoot using interlaced adaptive staircases (A/B) with 4-2-1 stepping; outcome: average intensity at step-size-1 turnaround points. Cognitive measures: - Processing speed: WAIS-III Digit Symbol Substitution (120 s score). - Working memory updating: 2-back task (letters; 20% targets; 300 ms ISI 1400 ms). Outcomes: hits RT and sensitivity d′; RT <100 ms or >1200 ms counted as misses. - Inhibitory control: Stroop (neutral Xs vs incongruent color words), responses via keypress; data trimming (delete first trial, RT <100 ms, and RT >2 SD within age/condition). Outcome: inverse efficiency score (IES = mean RT correct / accuracy) and interference (IES incongruent − IES neutral). - Task switching: Color–shape switch task (fixed color block, fixed shape block, mixed alternating block); trimming as above. Outcomes: general (mixing) costs = IES mixed − IES fixed; specific costs = IES switch − IES repeat within mixed. Statistical analyses: Variables transformed for normality (log or Box-Cox). Linear models with three a priori orthogonal age contrasts: (a) middle-aged vs older adults, (b) 46–55 vs 56–65, (c) 66–75 vs 76–86. Gender tested but excluded for no added fit. Causal mediation analysis (R ‘mediate’, non-parametric bootstrap) tested average causal mediation effect (ACME), average direct effect (ADE), total effect, and proportion mediated for processing speed and sensorimotor mediators on executive outcomes. Coupling analyses used within-age-group median splits on cognitive variables (processing speed, inhibition, switching, WM updating) as factors predicting sensorimotor outcomes (DiN SRT, TUG, postural short-term diffusion), with interactions with age contrasts; post hoc tests Bonferroni-corrected; Welch’s t when variances unequal.

- Age effects across domains: - Listening in noise: Older groups had worse SRTs than middle-aged; significant differences between adjacent cohorts; SRTs worsened by ~0.15 dB SNR per year on average (accelerating with age: ~0.16, 0.19, 0.32 dB SNR between successive cohorts). - Postural control: Path length increased with age; short-term diffusion coefficient (open-loop sway) higher in older vs middle-aged; no age effects for long-term diffusion or critical time interval. - Functional mobility: TUG times increased significantly with age; significant differences both between middle-aged groups and between older groups. - Tactile sensitivity: Minimal age sensitivity; only the 46–55 vs 56–65 groups differed. - Cognition: Processing speed decreased with age; Stroop interference increased; general (mixing) switch costs increased with age (specific switch costs unchanged); WM updating (d′) decreased in older vs middle-aged. - Mediation of age-related variance (ARV) in executive functions: - Processing speed significantly mediated age effects: - Task switching: ACME=0.12 (p<0.01), ADE=0.11 (p<0.05), total=0.24 (p<0.01), proportion mediated=0.52 (p<0.005). - Inhibitory control: ACME=0.01 (p<0.01), ADE=0.01 (p<0.05), total=0.02 (p<0.05), proportion mediated=0.41 (p<0.01). - Working memory updating: Mediation not reliable. - Functional mobility (TUG) significantly mediated age effects: - Task switching: ACME=0.09 (p<0.05), ADE=0.14 (p<0.05), total=0.23 (p<0.01), proportion mediated=0.42 (p<0.05). - Inhibitory control: ACME=0.01 (p<0.01), ADE=0.01 (p<0.05), total=0.02 (p<0.005), proportion mediated=0.41 (p<0.005). - Working memory updating: Mediation not reliable. - Listening in noise and postural control did not reliably mediate age effects on executive outcomes. - Coupling analyses (median splits within age groups): - Listening in noise: No overall main effects for cognitive median splits; a significant interaction for processing speed showed coupling only in the oldest group (76–86): slower processing speed associated with higher (worse) SRTs (ΔM=0.07 dB SNR; t(22)=3.45; p=0.002). - Functional mobility (TUG): Significant main effects for all cognitive splits—processing speed (t=2.30; p=0.032), task switching (t=3.06; p=0.004), inhibitory control (t=3.26; p≈0.002), WM updating (t=2.55; p=0.026). Interactions indicated stronger coupling with age for processing speed and WM updating. Post hoc: processing speed coupling significant in 56–65 (ΔM=0.008; p=0.012) and 76–86 (ΔM=0.01; p=0.018); WM updating coupling significant in older adults (ΔM=0.01; p=0.017) but not middle-aged. - Postural control (short-term diffusion): Only processing speed showed coupling with age interactions; older adults with faster processing speed had lower short-term diffusion (better control): overall older ΔM=0.23 (p=0.008); pronounced within older groups 66–75 vs 76–86 ΔM=0.40 (p=0.005). Overall, processing speed showed the most consistent coupling with sensorimotor performance, and functional mobility showed the strongest reciprocal links with executive functions; coupling generally intensified with age for speed and WM updating.

Findings support robust age-related declines across most sensorimotor and executive functions starting in middle age, with acceleration in later decades. Mediation analyses align with common-cause perspectives by showing that processing speed—and to a similar extent functional mobility—accounts for a substantial portion of age-related variance in task switching and inhibitory control. However, listening in noise and postural control did not mediate cognitive aging, and coupling between cognitive control and these sensorimotor domains was weak, suggesting that beyond general factors (e.g., speed), domain-specific mechanisms substantially shape performance in listening-in-noise and balance tasks. Coupling was strongest between cognition and functional mobility and between processing speed and postural control, particularly in older adults, consistent with the idea that shared resources and/or compensatory recruitment become more consequential with age. The limited coupling for listening in noise may reflect task-specific demands (speech-weighted noise less taxing than informational maskers), peripheral hearing status, and central auditory temporal processing contributions not fully captured by the chosen measure. The results highlight the heterogeneity of sensorimotor–cognitive interdependencies in aging and underscore the need to consider both general and domain-specific factors when designing interventions.

The study demonstrates robust age effects on cognitive and sensorimotor functions emerging in middle adulthood and accelerating in late adulthood. Processing speed and functional mobility show reliable sensorimotor–cognition coupling and mediate age effects on executive control (task switching, inhibition), with coupling increasing with age, particularly for processing speed. In contrast, coupling between cognitive control and listening-in-noise or postural control was limited, implying significant domain-specific contributions. These findings suggest that interventions targeting processing speed and mobility may yield broader benefits, while improvements in listening-in-noise and balance may require domain-specific approaches. Future work should employ longitudinal designs, richer auditory and postural task variants (e.g., informational maskers, higher postural complexity), and broader sensory-neural measures to disentangle mechanisms and inform targeted rehabilitation strategies for aging individuals, including those with hearing impairment.

- Cross-sectional, correlational design precludes causal inference; longitudinal data are needed to adjudicate cascade, common-cause, and compensation accounts. - Sample and task choices may limit generalizability: listening-in-noise assessed with speech-weighted noise (less cognitively demanding than informational maskers); tactile sensitivity measured at the foot may be less sensitive to age-related change; postural task had relatively low complexity, possibly underestimating age effects. - Hearing aid use was relatively low in the sample; hearing status variability and cutoff criteria for DiN may influence coupling estimates. - Mediation results may depend on specific operationalizations (e.g., processing speed via digit symbol; mobility via TUG) and transformation choices. - Educational differences across cohorts may contribute residual confounding despite modeling.