Study: 8 weeks of computerized cognitive training cut simulated driving infractions in young adults

What the study tested

Driving places heavy demands on attention, memory, executive function and visuospatial skill, and the authors note that human error may account for up to 90% of traffic accidents, with attention deficits linked to roughly 50% of injury-related incidents.^[1] Most prior work had focused on older drivers or on single skills such as hazard perception; the pilot asked whether an intensive, multidomain regimen could transfer to driving performance in young adult drivers.

The trial recruited from Universidad Nebrija in Madrid, Spain. Of 64 assessed for eligibility, 8 were excluded (6 did not meet inclusion criteria, 2 declined), leaving 56 randomized by a computer-generated sequence to an intervention arm or a passive waiting-list control. After dropout and exclusions for improperly recorded driving data, 50 were analyzed, 25 per group. Eligibility required a valid driver's license and ages 18–30; the final sample was 22 women, mean age 21.1 years (SD = 2.36), mean driving experience 2.84 years (SD = 1.61) and mean annual mileage 5,920 km (SD = 4,716). All had normal or corrected vision and hearing and no impairment on the baseline Cognitive Assessment Battery (CAB® PRO, CogniFit Inc., San Francisco). The groups did not differ at baseline on mileage (t(48) = 0.34, p = .595) or cognitive score (t(48) = 0.16, p = .872).^[1]

The intervention was the CogniFit dynamic computerized training system, drawing on a pool of 41 adaptive-difficulty games targeting executive function, attention, memory and visuospatial abilities. Each daily session ran three activities selected by the Individualized Training System, lasting about 15 minutes, completed 5 days a week for 8 weeks, a planned 40 sessions, on participants' own phones. The control group did no specific activity during the 8-week gap. The primary outcome was driving performance, scored as the summed count of traffic infractions across ten categories, from exceeding the speed limit to off-road driving and two collision types, recorded on a Simumak Simescar lite simulator over a 3.7 km urban route (moderate rain, slight fog, heavy traffic, four scripted hazards) lasting about 12 minutes, before and after the program. Analysis used a mixed-design ANOVA with factors Group (between) and Test Moment (within).

What it found

The omnibus ANOVA showed no main effect of Group (F(1,48) = 2.59, p = .114) but a significant main effect of Test Moment (F(1,48) = 13.99, p < .001, ∂η² = 0.226) and, critically, a significant Test Moment × Group interaction (F(1,48) = 4.92, p = .031, ∂η² = 0.093), the change over time depended on which group a participant was in.^[1]

As the table shows, the groups committed similar infractions at pre-test, but the trained arm fell from a mean of 22.60 (SD = 11.11) to 13.04 (SD = 4.15) while controls did not change. Mean driving speed did not change in either group, which the authors read as evidence that the safety gains came from greater accuracy, not from driving more slowly.^[1]

Breaking the total into categories (paired-samples t-tests, df = 24, trained group), five infraction types fell significantly: exceeding the speed limit (t = 3.769, p < .001, Cohen's d = 0.7537), off-road driving (t = 3.113, p = .005, d = 0.6226), aggressive driving (t = 2.179, p = .039, d = 0.4358), collisions with urban objects (t = 2.498, p = .020, d = 0.4995) and collisions with other vehicles (t = 3.161, p = .004, d = 0.6321). Collisions with other vehicles fell from 0.56 to 0.04, and off-road events from 4.04 to 1.96.^[1]

Adherence and dropout

The protocol prescribed one short session per day, 5 days a week, for a planned 40 sessions over the 8 weeks, with a 12-hour countdown after each to prevent overtraining. Of the 56 randomized participants, 6 did not complete the study, reported as a 3.36% drop-out, with three intervention-arm participants lost to non-adherence and one control participant lost to a missing post-evaluation. Two more were excluded for improperly recorded driving evaluations, leaving the final analyzed sample of 50 (25 per group).^[1]

Limitations, what this does not show

The authors were explicit that this was a pilot study whose findings should be treated as preliminary. The single largest limitation was the passive (waiting-list) control: with no active or sham task, the design cannot separate effects specific to the CogniFit training from the general benefits of any structured cognitive engagement, or from expectancy. Second, performance was measured only in a driving simulator that lacked vestibular and proprioceptive inputs and real crash consequences, so the authors cautioned that translation to on-road driving must still be verified. The trial included no follow-up assessment, so the durability of the gains is unknown, and the sample was a small, single-site group of university-recruited young adults, which limits generalisability. Because the study tested efficacy rather than mechanism, it does not identify which cognitive processes drove the improvement. None of these results establish that any product treats, prevents or cures a disease.

Where this fits in the broader evidence

The result extends a body of older-driver work to a younger group. Earlier trials reported that speed-of-processing training reduced at-fault crashes among older drivers,^[4] and a 2016 pilot found onboard cognitive training improved processing speed, working memory and driving safety in senior daily drivers.^[2] A 2021 systematic review and meta-analysis linked cognitive training, especially processing-speed training, to a substantial reduction in at-fault crashes among older drivers,^[3] while a separate randomized trial showed driver-inattention training reduced collisions in adolescents with ADHD.^[5] Against that backdrop, this trial is corpus-unique in showing infraction reductions in healthy young adult drivers. For related CogniFit research, see the research index, the companion study on cognition and gait in seniors, and the guide on brain training for older adults.