Cognitive training for healthy older adults: what the 2024–2026 evidence shows

How big is the cognitive-aging question?

A 2024 scoping review in The Journal of Prevention of Alzheimer's Disease projected that 14,298,671 people in the European Union alone could be affected by dementia by 2050, framing cognitive training as a candidate within a "window of opportunity for delaying the development of dementia, especially considering the absence of a known cure"^[3]. Against a worldwide brain-fitness market of $295 million in 2009, up nearly 35% on the prior year, the 2017 systematic review set out to test whether the literature behind those products supported their claims^[1].

The Lancet Commission on dementia estimated that around 40% of dementia cases were associated with modifiable risk factors, raised to roughly 45% across 14 factors in 2024, cognitive engagement being one population-level factor among many^[6].

Does cognitive training improve cognition in healthy older adults?

For specific trained abilities, several controlled trials reported gains, though size and durability vary. The benchmark is the ACTIVE trial: 2,832 adults aged 65 or older randomized to reasoning, memory or speed-of-processing training, where the reasoning and speed groups showed benefits that, with boosters, persisted over 2, 5 and possibly 10 years^[5]. Across the 13 studies in its top tier, the 2017 review found 7 with improved processing speed, 5 with improved memory or working memory, and 3 with improved attention^[1].

Recent work points the same way. A 2026 controlled study from the Universidad de Castilla-La Mancha in Spain followed 45 adults over 60 (23 trained, 22 control) through 12 weeks of app-based training; the trained group gained 132 points (SD 98.55) on a 0–800 global cognitive score versus 18 points (SD 71.53) in controls (p < .001), with parallel gains on the Montreal Cognitive Assessment^[4]. The 2024 scoping review of 34 studies across 27 apps found the field targets a few domains: 79.4% trained memory and 76.5% executive functions^[3].

How good is the evidence, and which programs hold up?

The 2017 systematic review is the most rigorous attempt to grade this field. From 32 commercial programs and 7,985 records screened, the authors narrowed to 7 programs with eligible clinical trials and 26 supporting studies; 11 were removed for having no empirical evidence at all. Across the 18 trials without follow-ups, methodological quality was high, a mean PEDro score of 7.05, with 12 rated high, 6 moderate, and none poor^[1].

The review sorted the 7 programs onto a three-level evidence ladder. Only two reached Level I, at least two well-designed randomized trials, one high and one at least moderate quality on the PEDro scale^[1]. One Level I program was evaluated across 3 studies (1 high-quality, 2 moderate), including a randomized double-blind trial in 155 healthy older adults where the training was more effective than an active control for visuospatial working memory, learning and attention^[1]. Company-funded studies scored higher on quality (PEDro 7.4) than independently funded ones (6.3), and the authors warned that "caution must be taken regarding any potential bias due to selective reporting"^[1].

Dose, by contrast, is wildly inconsistent: the 2024 scoping review reported weekly training time of 20 to 600 minutes and durations from a single week to 12 months^[3], and cautioned that this spread "may lead to inconsistent findings and limit the ability to compare and generalize the results"^[3].

What did the strongest null result find?

The most demanding study in this set returned a null primary outcome. A 2016 doctoral trial through the City University of New York randomized 69 cognitively healthy adults aged 80 and older (mean age 85.81; mean MMSE 29.06) to an adaptive computerized training program (n = 39) or an active games control (n = 30), each in 20-minute sessions every other day for 24 sessions^[2].

The two were statistically indistinguishable: no significant program-by-time interaction appeared on the global cognitive composite (F(1, 55.99) = 0.198, p = .658) or any specific domain^[2]. The one significant moderator was education, not the program: participants without a college degree improved on overall cognition (t(13) = 2.59, p = .023) regardless of program^[2].

This null fits broader warnings that home-based, unsupervised training is unlikely to benefit unimpaired older adults, particularly a high-functioning sample over 80 with little room to improve^[2].

What the evidence does NOT show

It does not show that training prevents dementia or reverses decline. The 2017 review concluded that cognitive training "may provide benefits in preventing/slowing progression of cognitive decline" but "may not necessarily reverse the disease trajectory," and did not establish a disease-modifying effect^[1]. The Lancet Commission's modifiable-risk figure is a population-level association, not an outcome of any one product^[6].

Effects are domain-specific and often do not transfer. The 2026 Spanish study found no significant improvement in everyday memory failures, depression, sleep or quality of life, even as global scores rose^[4]; the 2016 trial detected no benefit over a simpler active control at all^[2].

Samples are small and the field is conflicted. Across the 34 studies in the 2024 scoping review, average enrollment was about 39 participants and only 4 reached 100 or more^[3]. Of the trials graded in 2017, 12 were company-funded against 6 independent, with the company-funded ones scoring higher on quality^[1]. Larger, longer, independent trials with follow-up remain the missing piece.

What this means in practice

Read together, the evidence supports a measured reading: structured computerized training has produced reliable gains on the abilities it trains, processing speed, memory and attention, across several randomized trials, with the ACTIVE study suggesting some gains persist for years^[5][1]. But the benefit is bounded to trained domains, at least one rigorous trial in the oldest-old found no advantage^[2], and none of this evidence establishes prevention of dementia.

For how this fits the cognitive-decline spectrum, see the evidence-based guide to cognitive decline. Related reviews cover long-COVID brain fog, rehabilitation after stroke, cognition, mobility and driving, multiple sclerosis, chronic insomnia, and stress and memory. The primary studies behind this page: the 2017 review, the 2016 null trial, the 2024 scoping review, and the 2026 study over 60.