Cognitive training, mobility and driving safety in older adults: the evidence

Why mobility and driving are cognitive questions in aging

Walking and driving feel automatic, but both lean heavily on attention, processing speed and executive function, the same capacities that decline with age. Slow gait predicts morbidity, falls and loss of independence, and roughly 50% of people who begin a physical-exercise program drop out within the first 3–6 months, which has driven researchers to ask whether training the mind, rather than the body, could move physical function.^[1] On the road, the demand is similar: one pilot trial reported that human error may account for up to 90% of traffic accidents, with attention deficits linked to roughly 50% of injury-related incidents.^[2]

The 2024 report of the Lancet standing Commission on dementia identified 14 modifiable risk factors that, in its modelling, were associated with around 45% of dementia cases worldwide, and placed cognitive and physical activity within a broader prevention frame.^[7] Against that backdrop, the narrow question this page examines is concrete: when older adults trained cognition, did measurable real-world function, how they walked and how they drove, change in controlled studies?

Can cognitive training change how older adults walk?

The clearest positive signal came from a randomized single-blind pilot trial led by Joe Verghese and colleagues at the Albert Einstein College of Medicine, published in The Journals of Gerontology Series A in 2010.^[1] From a referral pool of 45 screened individuals, 24 sedentary adults aged 70 and over with a baseline gait velocity below 1.0 m/s were randomized 12 per arm to 8 weeks of thrice-weekly computerized cognitive remediation or a wait-list control. Gait was measured on an instrumented walkway under usual normal-pace walking and a demanding “walking while talking” dual task.

Among the 10 completers in the training arm, walking-while-talking gait velocity rose from 36.7 to 56.7 cm/s, a gain of 19.9 cm/s over controls (p = .05), while the 10 wait-list controls did not change in either walking condition (normal-pace p = .57; dual-task p = .70).^[1] On the dual task, all 10 trained participants improved by at least 4 cm/s versus 3 of 10 controls (odds ratio 3.5, 95% CI 1.5–8.0), and speed of processing improved by 1,662 ms (95% CI 155–3,168, p = .03). The full methods and effect sizes are summarized on the study page for cognitive remediation and gait in sedentary seniors.

That result did not generalize cleanly to a more impaired population. In a single-blind randomized trial of 47 community-dwelling older adults with cognitive impairment (mean age 81.16 years), Embon-Magal and colleagues reported that 8 weeks of thrice-weekly computerized training improved global cognition (F(1,44) = 17.43, p < .001), working memory, divided attention and processing speed, but gait speed and stride-time variability did not change under single- or dual-task walking (all p ≥ .142).^[3] The contrast is detailed on the study page for cognition versus gait in seniors with cognitive impairment. Read together, the two trials suggest that any transfer to walking depended on the sample's starting point: the gait gain appeared in higher-functioning sedentary seniors but not in a frailer, lower-cognition group.

Does cognitive training transfer to driving safety?

Driving is the most demanding everyday task many older adults perform, and it has the longest evidence trail. In the ACTIVE-derived analysis by Karlene Ball and colleagues, published in the Journal of the American Geriatrics Society in 2010, speed-of-processing training was associated with fewer at-fault motor-vehicle collisions among older drivers across subsequent years of follow-up.^[4] A 2021 systematic review and meta-analysis by Bridget Fausto and Jerri Edwards reinforced the pattern, linking cognitive training, especially processing-speed training, to a substantial reduction in at-fault crashes among older drivers.^[5]

The newest and corpus-unique evidence came from a 2024 randomized controlled pilot trial by Jose Tapia, David Sánchez-Borda and Jon Andoni Duñabeitia in Cognitive Processing.^[2] Although it recruited young adult drivers (N = 50, ages 18–30, mean 21.1 years) rather than seniors, it is the first to measure infraction counts after a multidomain regimen. Participants trained about 15 minutes a day, 5 days a week for 8 weeks, then drove a high-fidelity simulator over a 3.7 km urban route. Total traffic infractions in the trained arm fell from a mean of 22.60 to 13.04 (about 42%; within-group t(48) = 4.21, p < .001), while the passive control group did not change (p = .706), and a significant Test-Moment × Group interaction (F(1,48) = 4.92, p = .031) confirmed the effect was specific to the trained arm.^[2] Mean driving speed was unchanged, which the authors read as accuracy gains rather than slower, more cautious driving. The methods, category-level breakdown and limitations are on the study page for cognitive training and simulated driving performance.

How do gait and falls connect to cognitive decline?

The interest in walking is not only about mobility for its own sake. Verghese and colleagues reported in 2007 that quantitative gait dysfunction, slow speed and high variability, especially under dual-task conditions, was associated with a higher risk of later cognitive decline and dementia.^[6] That makes the walking-while-talking measure used in the 2010 gait trial more than a fitness metric: it is an ecologically valid index that reflects how attention is allocated while moving, and slowing on it predicts both falls and cognitive trajectory.^[1][6]

This is why the mobility and driving literatures sit together. Both ask whether strengthening attention and processing speed shows up in a downstream, safety-relevant behavior, a steadier dual-task gait, or a cleaner drive. The Lancet Commission framed physical inactivity and the conditions that accompany frailty among the modifiable contributors to dementia risk, situating these behavioral outcomes within a wider prevention picture rather than treating them as isolated endpoints.^[7]

What the evidence does NOT show

The honest reading of these studies is narrower than the headlines. The mobility and driving trials were small or preliminary: the 2010 gait trial analyzed 20 participants and, by its authors' own power calculation, would need about 400 seniors for a confirmatory test, its between-group difference in normal-pace gait was not statistically significant.^[1] The 2024 driving trial used a passive control with no active sham, measured gains only in a simulator rather than on real roads, included no follow-up, and recruited young adults, so it cannot be read as evidence about senior on-road driving.^[2]

Crucially, transfer to walking was not universal: in the trial of older adults with cognitive impairment, cognition improved while gait did not move at all.^[3] None of these studies established that cognitive training treats, prevents or cures any disease, slows dementia, or prevents falls or crashes in the general population; they reported changes on specific performance measures in defined, mostly small samples. The gait–dementia link reported by Verghese is an association in observational data, not evidence that improving gait alters cognitive outcomes.^[6] Larger, active-controlled, on-road and longer-follow-up trials would be needed before any of these signals could be called a reliable, generalizable effect.

What this means in practice

For older adults, families and clinicians weighing the evidence, the picture is cautiously encouraging but bounded. Randomized trials reported that computerized cognitive training was followed by measurable gains in dual-task gait velocity in some sedentary seniors,^[1] and a broader older-driver literature, including a meta-analysis, linked processing-speed training to fewer at-fault crashes.^[5] But these are research findings in specific populations, not guarantees of individual benefit, and they sit alongside a trial in which gait did not change.^[3] Decisions about driving fitness or fall risk should be made with a qualified clinician, not on the basis of any training program.

To see how this fits the wider field, the cognitive-decline evidence guide sets the spectrum from healthy aging to impairment. Related leaves cover brain training in healthy older adults, the evidence after stroke and post-operative recovery, cognition in multiple sclerosis, the sleep and insomnia literature, persistent long-COVID brain fog, and the effect of stress on memory. For named-author summaries with DOIs, see the research index.