Do brain games for seniors work? What the biggest trial found
The honest summary of brain games for seniors is narrower than either the marketing or the backlash suggests. Training a specific mental skill reliably improves that skill, the improvement can last a decade, and it does not reliably spread to anything you have not trained. That is the finding of the largest randomised trial in this area, and it has held up for more than twenty years.
The trial is ACTIVE, short for Advanced Cognitive Training for Independent and Vital Elderly. Karlene Ball and colleagues published the two-year results in JAMA in 2002, and George Rebok and colleagues published the ten-year follow-up in the Journal of the American Geriatrics Society in 2014.
It recruited 2,832 volunteers aged 65 to 94 from senior housing, community centres and clinics across six US metropolitan areas, with recruitment running from March 1998 to October 1999 and the first follow-up completing in December 2001. That scale and that duration are why it remains the reference study for this question rather than one of the hundreds of smaller ones that followed it.
How the trial was built
Participants were randomly assigned to one of four groups: ten group sessions of memory training, which targeted verbal episodic memory, with 711 people; ten sessions of reasoning training, targeting problems that follow a serial pattern, with 705; ten sessions of speed-of-processing training, targeting visual search and identification, with 712; or a no-contact control group of 704.
A randomly chosen 60 per cent of the trained participants were offered four booster sessions eleven months later, which lets the trial test whether topping up the training matters. Thirty participants were incorrectly randomised and excluded from the analysis.
This structure is the reason the trial can answer the question properly. Because each arm trained one ability and was measured on several, the design separates improving at the thing you practised from getting generally sharper. Most smaller studies in this field cannot make that separation, which is why their results are so often reported as though training one skill improved cognition in general.
“No training effects on everyday functioning were detected at 2 years.”
What improved, and by how much
Every intervention improved the ability it targeted, and the effect was durable to two years, with the authors reporting significance at better than one in a thousand for all three.
The split between abilities is the most useful number in the whole study. Immediately after training, 87 per cent of the speed group and 74 per cent of the reasoning group showed reliable cognitive improvement, against 26 per cent of the memory group. Training speed of processing worked for most people; training memory worked for about a quarter.
Booster sessions helped, and only in two arms. In speed, 92 per cent of those who received a booster improved against 68 per cent of those who did not. In reasoning it was 72 per cent against 49 per cent, and both of those gaps were still there at the two-year follow-up.
The finding that matters most, and gets quoted least
No training effects on everyday functioning were detected at two years. That is the trial's own wording, and it is the sentence any honest answer to this question has to include.
The authors gave a reason for it rather than leaving it hanging. Functional decline across all four groups, including the controls, was minimal over two years, so there was very little decline for the training to prevent. They concluded that a longer follow-up would be needed to see whether the effect appeared.
That is a meaningful caveat and it is not a get-out. Taken on its own, the two-year result says that getting substantially faster at visual search did not make people measurably better at managing their own lives inside that window. It is also the result that the first wave of brain-training marketing in the 2000s was built on top of, which is why regulators later took an interest in how those products were advertised.
What happened after ten years
The follow-up changed the picture in both directions. Reasoning and speed-of-processing training still showed effects on their targeted abilities a decade on, with effect sizes of 0.23 and 0.66. Memory training effects on memory performance were no longer maintained.
On daily life, participants in all three training groups reported less difficulty with instrumental activities of daily living than controls, with effect sizes of 0.48 for memory, 0.38 for reasoning and 0.36 for speed. At a mean age of 82, about 60 per cent of trained participants were at or above their own baseline self-reported function, against 50 per cent of controls.
Read that carefully, because it is where overstatement creeps in. The ten-year daily-function result is self-reported, and the two-year result on measured everyday function was null. A difference of ten percentage points in whether people say they are managing as well as before is a real and modest finding.
So which brain games are worth an older adult's time?
By this evidence, the ones that train speed of visual processing and pattern reasoning, because those are the two arms whose effects persisted for ten years. The memory arm is the one that faded, which is awkward because memory is what most commercial brain-training products advertise.
The practical translation is unglamorous. Visual search tasks, where you find a target among distractors under time pressure, and serial pattern problems, where you work out what comes next in a sequence, are the trained skills in question. Plenty of ordinary puzzle games contain both.
What the trial does not license is the claim that a particular commercial product delivers these results. ACTIVE tested structured training delivered in group sessions by trainers, not an app, and we have looked separately at [whether brain-training products live up to their marketing](/brain/do-brain-training-games-work-what-the-research-says).
Does any of this prevent dementia?
ACTIVE was not designed to answer that, and it should not be cited as if it were. Its outcomes were cognitive performance and everyday functioning in people living independently, not the incidence of a disease.
The broader evidence on whether cognitive training reduces dementia risk remains unsettled, and the honest position is that improving performance on a trained task is a different claim from altering the course of a neurodegenerative condition.
Anyone selling a game on the basis that it prevents dementia is making a claim the strongest trial in this field did not test. That is the single most important thing a reader can take from this piece before spending money, and it applies equally to free games that carry the same implication in their store descriptions.
What the trial cannot tell you
Three limits, all of them real.
The participants were volunteers who were living independently and well enough to attend ten group sessions, which is not the same population as people already in decline. The training was delivered by people in rooms, so nothing here transfers automatically to an app used alone at home. And the ten-year daily-function benefit is based on what participants said about their own difficulty rather than on a measured task.
There is also the question nobody can settle from this design: whether the benefit came from the cognitive content or from ten structured sessions in a group with other people and a purpose. Both arms got that, and the controls got no contact at all.
The practical answer
If an older adult enjoys puzzles, the evidence supports playing them and does not support expecting much beyond getting better at them. The abilities most likely to show durable improvement are processing speed and reasoning, and the effect on daily life, where it appears at all, is modest and self-reported.
If the goal is maintaining independence rather than scoring higher on a game, the evidence base for physical activity, hearing correction, social contact and sleep is stronger than the evidence base for brain games, and none of those requires a subscription.
And if a product advertises memory improvement specifically, note that memory was the arm where only 26 per cent improved at all and where the effect had gone by ten years. That is the arm the marketing leans on hardest, and on this evidence it is the weakest of the three.
Frequently asked
Do brain games for seniors actually work?
They reliably improve the specific skill you train and do not reliably spread beyond it. In the ACTIVE trial of 2,832 adults aged 65 to 94, every intervention improved its targeted ability and the effect lasted two years, but no training effects on everyday functioning were detected at two years. At ten years, reasoning and speed-of-processing gains persisted while memory gains did not, and self-reported daily function was modestly better in all three training groups than in controls.
Which brain games are best for elderly people?
On this evidence, tasks that train speed of visual processing and pattern reasoning rather than memory. Those were the two ACTIVE arms whose effects on their targeted abilities were still present at ten years, with effect sizes of 0.66 and 0.23. Visual search under time pressure and serial pattern problems are the actual trained skills, and many ordinary puzzle games contain both. Memory training was the arm that faded.
What was the ACTIVE trial?
The largest randomised trial of cognitive training in older adults. It recruited 2,832 volunteers aged 65 to 94 from senior housing, community centres and clinics in six US metropolitan areas, and randomly assigned them to ten group sessions of memory, reasoning or speed-of-processing training, or to a no-contact control group. Two-year results were published in JAMA in 2002 and ten-year results in the Journal of the American Geriatrics Society in 2014.
Do brain games prevent dementia?
ACTIVE was not designed to test that and should not be cited as if it were. Its outcomes were cognitive performance and everyday functioning in people living independently, not the incidence of a disease. Improving performance on a trained task is a different claim from altering the course of a neurodegenerative condition, and anyone selling a game on a dementia-prevention basis is making a claim the strongest trial in this field did not test.
How long do brain training gains last in older adults?
Longer than most people expect, for two of the three abilities tested. In ACTIVE's ten-year follow-up, reasoning training still showed an effect on reasoning performance with an effect size of 0.23, and speed-of-processing training still showed an effect on speed with an effect size of 0.66. Memory training effects on memory performance were no longer maintained at ten years, despite having been durable at two.
Did brain training help older adults with daily life?
Modestly, and only in the longer follow-up. No training effects on everyday functioning were detected at two years, which the authors attributed partly to minimal functional decline across all groups in that window. At ten years, all three training groups reported less difficulty with instrumental activities of daily living than controls, and at a mean age of 82 about 60 per cent of trained participants were at or above baseline self-reported function against 50 per cent of controls.
Do booster sessions make brain training more effective?
In two of the three ACTIVE arms, yes. Four booster sessions eleven months after the original ten were offered to a randomly chosen 60 per cent of trained participants. In the speed arm, 92 per cent of those boosted showed reliable improvement against 68 per cent of those who were not. In reasoning it was 72 per cent against 49 per cent. Both gaps were still present at the two-year follow-up, and the memory arm did not show the same benefit.
Why did memory training work so much less well?
The trial reports the result rather than explaining it, and the gap is stark: 26 per cent of memory participants showed reliable improvement immediately after training, against 74 per cent in reasoning and 87 per cent in speed. The memory effect was also the only one of the three that had disappeared by the ten-year follow-up. That matters commercially, because memory improvement is what most brain-training products advertise.
Can an app replicate the ACTIVE trial results?
Nothing in the trial establishes that. ACTIVE delivered structured training in ten group sessions led by trainers, with a defined curriculum, in person. An app used alone at home differs in the content, the structure, the social setting and the accountability, any of which could matter. Citing ACTIVE to support a specific commercial product is citing a study of something else, which is the most common misuse of this research.
What helps older adults more than brain games?
For maintaining independence rather than improving a game score, the evidence base for physical activity, correcting hearing loss, social contact and sleep is stronger than the evidence base for cognitive training, and none of those requires a subscription. ACTIVE's own design raises a related question it cannot answer: whether part of the benefit came from ten structured sessions in a group with a purpose rather than from the cognitive content itself.