In February 2026, Johns Hopkins Medicine announced a striking result from a federally funded trial that began in the late 1990s. Older adults who took a short computerized speed-training course, plus booster sessions over the next few years, had a 25% lower adjusted risk of a dementia diagnosis across two decades.
The number was real. It also described 264 people. When the researchers compared each complete training group, memory, reasoning and speed, with the group that got no training, none showed a statistically significant drop in dementia diagnoses.
Headlines and main results part company again and again in this field. “Cognitive stimulation” gets used for at least three different things: a group therapy for people who already have dementia, structured training for healthy older adults, and one-on-one rehabilitation aimed at daily tasks. Each has been tested in randomized trials. Each delivers something real. None has been shown to keep dementia away.
Three treatments share one label
The 2023 Cochrane review of cognitive stimulation draws the lines clearly. Cochrane, an international network, pools clinical trials using standardized methods. Its review separates the approaches this way:
- Cognitive stimulation is broad and social. People with dementia meet, usually in a small group, for activities that work the mind in general. Its aims include mood and quality of life, not only memory.
- Cognitive training is narrow and repetitive: guided practice on standard tasks that target one ability, such as memory or mental speed.
- Cognitive rehabilitation is individual and practical. A therapist works with one person on goals in everyday life.
Group stimulation and rehabilitation were built for people living with a dementia diagnosis. The best-known training trial enrolled people without one.
Group cognitive stimulation therapy for dementia
Picture a small circle of people in their late seventies with a trained leader: a chat about today’s news, then how things were done fifty years ago, a word game, a song. That, in outline, is cognitive stimulation therapy, or CST.
Cochrane’s plain-language summary lists the ingredients: talk about past and present events and topics of interest, word games, puzzles, music and hands-on creative activities. Sessions typically last about 45 minutes, twice a week, run by trained staff.
Researchers at University College London and Bangor University developed the program and tested it in a trial that ran from 1998 to 2001. Its standard version has 14 or more themed sessions; the original trial fit them into seven weeks.
That trial, published in the British Journal of Psychiatry in 2003, randomly assigned 201 people with dementia to CST groups or a control group. Afterward, the CST groups scored better on two cognitive tests and a quality-of-life scale. Put practically, for every six people who joined a group, one more person improved by four or more points on a standard dementia thinking scale than would have without it.
What 37 trials show
The 2023 Cochrane update, with Bob Woods of Bangor University as first author, pooled 37 randomized trials covering 2,766 people with mild or moderate dementia, at a median age of 79.7, in 17 countries. Most tested groups rather than one-on-one sessions.
- Thinking: a small benefit, rated moderate certainty. On the Mini-Mental State Examination, a 30-point screening test common in dementia care, CST participants finished about 2 points ahead of comparison groups. The review’s plain-language summary translates that into roughly a six-month delay in the decline usually expected at this stage.
- Communication and social interaction: clinically relevant gains, rated high certainty, though drawn from only five studies.
- Quality of life, mood and daily activities: slight improvements.
- Family caregivers: little or no change in their own mood or anxiety.
- Harms: none reported in any of the trials.
Three limits keep this from being a cure story. Most measurements came at the end of programs lasting a median of 10 weeks, and only eight of the 37 trials checked whether the benefit held up afterward. Results also varied a lot from study to study. And while gains were larger when groups met twice a week or more, and when dementia was mild rather than moderate, the reviewers labeled both patterns exploratory.
One more piece of context: three of the review’s six authors, Woods, Martin Orrell and Aimee Spector, were also authors of the original 2003 trial. Cochrane’s methods are built to limit bias, but readers deserve to know who is grading the work.
On this library’s evidence scale, a short-term benefit for thinking and communication in mild to moderate dementia counts as positive evidence. Whether that benefit lasts after a program ends is insufficient evidence, because so few trials looked.
Where guidelines stand
The UK’s National Institute for Health and Care Excellence, in its 2018 dementia guideline, tells clinicians to offer group CST to people living with mild to moderate dementia. It uses a softer verb, consider, for group reminiscence therapy and for cognitive rehabilitation or occupational therapy aimed at daily function. The same guideline says not to offer cognitive training as a treatment for mild to moderate Alzheimer’s disease.
The US picture is thinner. The American Academy of Neurology’s 2018 guideline on mild cognitive impairment, a stage of measurable decline short of dementia, said clinicians may recommend cognitive training, at Level C, the weakest of the recommendation levels it used. That guideline is now marked as retired, so it is not current advice.
CST does reach some American settings. In a 2017 study, a Saint Louis University social work team ran six groups with 40 participants in Missouri and reported better scores afterward on a cognitive screening test and a depression scale, though not on quality of life. With no comparison group, it shows the program can be delivered here, not that it beat the alternative.
ACTIVE: a 20-year test of training in healthy older adults
ACTIVE, short for Advanced Cognitive Training for Independent and Vital Elderly, is a large, long-running trial of structured training in people without dementia. From March 1998 to October 1999 it enrolled 2,832 adults aged 65 to 94 in six US metropolitan areas. Thirty were randomized by mistake and excluded, leaving 2,802.
Three of the four groups got up to ten 60- to 75-minute small-group sessions over five to six weeks. One practiced memory strategies, one solved pattern-based reasoning problems, and one did speed-of-processing training: a visual search task, run on a computer, that grew harder as people improved. The fourth group had no contact with the trainers. Participants who finished at least eight of the ten sessions were eligible for a random draw for four-session boosters at 11 and 35 months.
What training changed, and what it didn’t
The first results, in JAMA in 2002, showed that training worked on what it trained. Right after the course, 87% of the speed group, 74% of the reasoning group and 26% of the memory group showed reliable gains on their own skill. At two years, though, the researchers detected no effect on everyday functioning.
By 10 years, reported in 2014, speed and reasoning gains were still measurable; memory gains had faded. All three trained groups reported less difficulty with instrumental activities of daily living, the more complex tasks of independent life. About 60% of trained participants, against 50% of controls, were at or above their starting level on that measure. Those were the participants’ own reports, not observed performance.
A 2017 analysis of the same decade used a study-built definition of dementia, combining interviews and test scores, and counted 260 cases. Speed training was linked to a hazard ratio of 0.71, roughly 29% lower risk, with a p value of .049, a whisker under the usual .05 cutoff for significance. Memory and reasoning training showed no significant effect.
The 20-year result
The 2026 analysis, led by Norma B. Coe of the University of Pennsylvania, linked participants to Medicare billing records through 2019. Diagnoses in claims come from participants’ ordinary medical care, independent of the study team. The analysis covered 2,021 people in traditional Medicare, 72% of the trial; 725 people enrolled in Medicare Advantage were left out because those records lack full claims. Over the two decades, 48.7% of the untrained group received a dementia diagnosis.
The main comparisons were all null. After adjustment, the hazard ratios were 0.85 for memory training, 0.87 for speed and 0.88 for reasoning, and none reached statistical significance. On this library’s scale, the claim that any of the three courses, given to everyone assigned to it, lowers dementia diagnoses was tested and not supported.
The positive finding came from a narrower slice: among speed-trained people who finished most of their sessions and were then drawn for boosters, 40% were diagnosed with dementia (105 of 264), against 49% of controls, an adjusted hazard ratio of 0.75. Speed-trained people who were eligible but not drawn for boosters had a hazard ratio of 1.01, no difference at all. Three features call for caution:
- The analysis plan. The study’s preregistration, a public record of planned analyses filed in advance, did not include the booster comparisons. The authors call that an inadvertent omission and note that boosters were part of the trial design from the start.
- Who qualified. Only people who completed most of the first sessions could be drawn for boosters, yet they were compared with the entire control group. People who stick with a program can differ in health and habits, and the authors acknowledge that selection bias after randomization remains possible.
- Financial interest. Karlene Ball, an author of the 2026 paper and first author of the original 2002 results, consults for and owns stock in the company that now markets the speed-training software, according to the paper’s disclosure statement.
The paper’s conclusion is measured: speed training “has the potential to delay the diagnosis…” of dementia, in the authors’ words. That makes the booster result preliminary evidence, a lead needing its own trial. An adaptive computer task repeated over three years may matter in a way a one-time course does not, but that has not been shown. Whether consumer brain-training apps reproduce any of it is a separate question, taken up in the library’s piece on those products.
Cognitive rehabilitation: practice on the tasks that matter
The third approach starts from the person’s own list. Cognitive rehabilitation addresses how impaired thinking affects daily function, and its goal-oriented version has been adapted for people with dementia. In it, someone with dementia and a family member sit down with a therapist and pick a few everyday goals. A goal might be using the phone without help, cooking a familiar meal again or keeping track of appointments.
The therapist works out where the task breaks down and builds a plan. Methods described by the GREAT trial researchers include memory aids and prompts, changes to the home, compensatory strategies, step-by-step practice of a skill, and relearning key information.
GREAT, run at eight centers in England and Wales and published in 2019, was the first large trial of the approach. The trial randomly assigned 475 people with mild to moderate Alzheimer’s, vascular or mixed dementia, each with a family member or other supporter willing to take part, to rehabilitation or to usual care. The rehabilitation group received ten weekly one-hour sessions at home over three months, then four maintenance sessions over six months. Nine of the ten therapists were occupational therapists; one was a nurse.
Large gains on the targets, none beyond them
On the goals people chose, the effect was large. Participants’ own ratings improved by an effect size of 0.97 at three months and 0.94 at nine months, and family members’ separate ratings agreed. Counting only the goals therapists worked on, the average gain was close to 3 points on a 0-to-10 scale on which a 2-point rise is considered clinically meaningful.
Everything else stayed flat: quality of life, mood, confidence in one’s own abilities, cognitive test scores and caregiver stress showed no significant group differences. A companion economic analysis found no cost-effectiveness on quality-adjusted life-years, a standard yardstick combining length and quality of life, though the program did look cost-effective when judged per unit of goal improvement.
The authors name their limits. Because usual care was the comparison, the design could not separate the specific methods from the effect of regular visits by an attentive professional. People without a family carer and people with rarer dementias were excluded.
Graded on its own target, everyday function in chosen tasks, rehabilitation has positive evidence. On cognitive test scores, the trial found no effect.
Why “puzzles keep dementia away” doesn’t follow
Most people have absorbed some version of the advice: do the crossword, learn a language, and your brain will stay young. The belief draws on observational studies, which track habits and note who develops dementia later, and which struggle to separate cause from consequence. Dementia’s silent early stages can make demanding hobbies less appealing years before diagnosis, so dropping the puzzles may be an early symptom rather than a cause. People who seek out mental challenge also differ in education, income and health.
Trials sidestep that by assigning the activity. When ACTIVE did so, with three different courses, none lowered dementia diagnoses over 20 years among the 2,021 participants whose Medicare records could be followed. A 2020 Cochrane review of computerized training lasting 12 weeks or more, in cognitively healthy people mostly 65 and older, rated the evidence low or very low quality; a slight early benefit showed no clear sign of lasting in the one trial that checked a year later.
The Institute of Medicine’s 2015 report on cognitive aging listed cognitive training among products to evaluate carefully before buying, next to supplements and medications. It also lists social and intellectual engagement among the steps it encourages, behind physical activity, cardiovascular risk control and medication review.
None of this makes mental activity pointless. CST improved communication, and rehabilitation helped people do things they cared about. What fails is the promise of insurance. What holds up is narrower and more useful: particular programs, delivered to particular people, improve particular things.
Questions to ask before a family member joins a program
If a memory clinic, senior center or care home offers a stimulation or rehabilitation program, these questions separate a well-built program from a loosely named activity hour.
- Who is it designed for? CST and goal-oriented rehabilitation were tested in people with mild to moderate dementia. A class for people without a diagnosis is a different intervention with different evidence.
- Is it based on a tested program? Ask whether staff follow a manual from a program that has been through trials, and what training they had. In the Cochrane review, 81% of studies reported using a treatment manual.
- How often does it meet? Twice-weekly groups showed larger cognitive gains than weekly ones in exploratory analysis.
- What should change, and how will anyone know? Good answers name outcomes the trials measured, such as conversation, mood or progress on specific goals. A promise to halt or reverse dementia is a red flag.
- What happens when it ends? Most trial benefits were measured on the last day. Ask about maintenance sessions or follow-on groups.
- What does it cost, and does any insurance pay? Ask before the first session.
- Does it sit alongside medical care? A good program adds to a care plan and never replaces any part of it.
When a clinician needs to weigh in
Whether a structured program suits someone already diagnosed, and at what stage, is a question for the clinician or memory service managing that diagnosis. Since the Cochrane analysis found bigger effects in mild dementia, the conversation is worth having early.
New or worsening memory trouble in someone who hasn’t been evaluated is a reason to talk with a clinician about assessment, and a training course is no substitute for one. What makes a change worrying, and how a work-up proceeds, are covered in the library’s guides to memory change and assessment. A sudden change is an emergency of a different kind. Sudden confusion, or sudden difficulty speaking or understanding speech, can signal a stroke. For those signs, the Centers for Disease Control and Prevention advises calling 911 at once instead of driving to the hospital.
The bottom line
“Cognitive stimulation” is three things, each with its own verdict. For people with mild to moderate dementia, group CST brings a small, short-term lift in thinking and a clearer gain in communication, with no reported harms. Goal-based rehabilitation helps people get better at the daily tasks they choose. And in healthy older adults, the ACTIVE trial improved trained skills for a decade but, in its main analysis, did not lower dementia diagnoses over 20 years.
Judge any program by the question it can answer: what will this help this particular person do, and how will we know when it has?
Sources
- Cochrane Database of Systematic Reviews. “Woods B et al. "Cognitive stimulation to improve cognitive functioning in people with dementia." Cochrane Database Syst Rev, 2023. PubMed. DOI: 10.1002/14651858.CD005562.pub3.” pubmed.ncbi.nlm.nih.gov
- British Journal of Psychiatry. “Spector A et al. "Efficacy of an evidence-based cognitive stimulation therapy programme for people with dementia: randomised controlled trial." Br J Psychiatry, 2003. PubMed. DOI: 10.1192/bjp.183.3.248.” pubmed.ncbi.nlm.nih.gov
- University College London (REF2014 impact case study). “Cognitive Stimulation Therapy – a new therapy for dementia.” impact.ref.ac.uk
- National Institute for Health and Care Excellence (UK). “Dementia guideline NG97: Recommendations.” nice.org.uk
- Neurology (American Academy of Neurology). “Petersen RC et al. "Practice guideline update summary: Mild cognitive impairment [RETIRED]: Report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology." Neurology, 2018. PubMed. DOI: 10.1212/WNL.0000000000004826.” pubmed.ncbi.nlm.nih.gov
- Journal of Gerontological Social Work. “Stewart DB et al. "Making a Difference: A Study of Cognitive Stimulation Therapy for Persons with Dementia." J Gerontol Soc Work, 2017. PubMed. DOI: 10.1080/01634372.2017.1318196.” pubmed.ncbi.nlm.nih.gov
- JAMA. “Ball K et al. "Effects of cognitive training interventions with older adults: a randomized controlled trial." JAMA, 2002. PubMed. DOI: 10.1001/jama.288.18.2271.” pubmed.ncbi.nlm.nih.gov
- Journal of the American Geriatrics Society. “Rebok GW et al. "Ten-year effects of the advanced cognitive training for independent and vital elderly cognitive training trial on cognition and everyday functioning in older adults." J Am Geriatr Soc, 2014. PubMed. DOI: 10.1111/jgs.12607.” pubmed.ncbi.nlm.nih.gov
- Alzheimer's & Dementia: Translational Research & Clinical Interventions. “Edwards JD et al. "Speed of processing training results in lower risk of dementia." Alzheimers Dement (N Y), 2017. PubMed. DOI: 10.1016/j.trci.2017.09.002.” pubmed.ncbi.nlm.nih.gov
- Alzheimer's & Dementia: Translational Research & Clinical Interventions. “Coe NB et al. "Impact of cognitive training on claims-based diagnosed dementia over 20 years: evidence from the ACTIVE study." Alzheimers Dement (N Y), 2026. PubMed. DOI: 10.1002/trc2.70197.” pubmed.ncbi.nlm.nih.gov
- Johns Hopkins Medicine. “Cognitive Speed Training Linked to Lower Dementia Incidence Up To 20 Years Later (news release, February 9, 2026).” hopkinsmedicine.org
- International Journal of Geriatric Psychiatry. “Clare L et al. "Individual goal-oriented cognitive rehabilitation to improve everyday functioning for people with early-stage dementia: A multicentre randomised controlled trial (the GREAT trial)." Int J Geriatr Psychiatry, 2019. PubMed. DOI: 10.1002/gps.5076.” pubmed.ncbi.nlm.nih.gov
- Health Technology Assessment (NIHR). “Clare L et al. "Goal-oriented cognitive rehabilitation for early-stage Alzheimer's and related dementias: the GREAT RCT." Health Technol Assess, 2019. PubMed. DOI: 10.3310/hta23100.” pubmed.ncbi.nlm.nih.gov
- Cochrane Database of Systematic Reviews. “Gates NJ et al. "Computerised cognitive training for 12 or more weeks for maintaining cognitive function in cognitively healthy people in late life." Cochrane Database Syst Rev, 2020. PubMed. DOI: 10.1002/14651858.CD012277.pub3.” pubmed.ncbi.nlm.nih.gov
- National Academies of Sciences, Engineering, and Medicine (Institute of Medicine). “Cognitive Aging: Progress in Understanding and Opportunities for Action (2015), Summary.” nationalacademies.org
- Centers for Disease Control and Prevention. “Signs and Symptoms of Stroke.” cdc.gov