Two things flash at once — an animal in the center, a sunflower at the edge. Catch them both. The better you get, the faster it goes. This is the same dual-attention, speed-of-processing style of training that made headlines when a 20-year study linked it to a 25% lower dementia risk.
In February 2026, researchers published a remarkable long-term follow-up in the journal Alzheimer's & Dementia: Translational Research & Clinical Interventions. It came from the NIH-funded ACTIVE trial (Advanced Cognitive Training for Independent and Vital Elderly) — the largest randomized controlled trial of cognitive training ever run in healthy older adults.
Beginning in the late 1990s, 2,802 participants aged 65 and older (average age 74) at six U.S. sites were randomly assigned to one of four groups: memory training, reasoning training, speed-of-processing training, or a no-training control group. Training ran for ten sessions of 60–75 minutes, twice a week, over five to six weeks. About half of each training group was then invited back for shorter "booster" sessions roughly one and three years later.
Two decades on, researchers checked Medicare claims for 2,021 of the original participants to see who had been diagnosed with dementia — including Alzheimer's disease, vascular dementia, and other forms. The result surprised even the study team:
Study co-author Marilyn Albert, PhD, of Johns Hopkins, and colleagues call the results promising while stressing this is one study — diagnoses came from insurance claims — and more research is needed.
The training in the study wasn't Sudoku, crosswords, or memorization. It was a dual-attention visual task built on a research paradigm called the Useful Field of View (UFOV) — a measure of how much visual information your brain can take in from a single glance.
In the study's exercise, an object appeared briefly in the center of the screen while a second target appeared somewhere in the periphery, surrounded by distracting images. Players identified the center object and pointed to where the peripheral target had been. Get both right, and the next flash gets faster. Miss, and it slows down. Over time, players extract more information from ever-shorter glimpses — a process called implicit learning: learning by doing, not memorizing.
Double Take is built on the same training principles: a center decision (bird or butterfly?) plus a peripheral search (where was the sunflower?), with adaptive flash speeds and distractor mushrooms that appear as you improve.
It's our own original game inspired by the paradigm — the program used in the actual research is a commercial product. What both share is the core skill being exercised: fast, accurate, divided visual attention. Your "speed" score is the shortest flash you can still catch completely — lower milliseconds, sharper eye.
It seems counterintuitive: the training with the long-term dementia signal wasn't memory practice at all. The leading interpretation is that processing speed is the bottleneck for the whole cognitive system. Before your brain can store anything, it has to take it in. As we age, processing slows and information gets captured incompletely — like a photo from a shaky camera.
A memory can't be sharper than the perception it was made from. Training the brain to extract information faster and more completely may give the memory system cleaner raw material to encode — so more of daily life actually "sticks."
Double Take-style training is really an attention workout: you split attention between two locations, ignore distractors, and commit both answers to short-term memory in a fraction of a second.
Selective and divided attention decide what enters memory in the first place. Strengthening the gatekeeper may improve everything downstream — names, where you parked, what someone just said. Researchers also note the training's real-world echoes in earlier ACTIVE results: better-maintained everyday functioning and safer driving among speed-trained participants.
Adaptive speed tasks push the visual and attention networks to respond at their limits — exactly the kind of demand thought to drive neuroplasticity, the brain strengthening and re-tuning its own connections.
One hypothesis is that this builds cognitive reserve: extra processing efficiency that lets the brain compensate longer if age-related changes or early disease begin, delaying the point where symptoms interfere with daily life. Because the training works through repetition and reaction rather than conscious study, this "learning by doing" pathway may generalize more broadly than practicing one memorized skill.
Perhaps the most practical takeaway: the benefit belonged to those who came back. Initial training alone wasn't enough — the booster sessions a year and three years later made the difference.
For your own routine, that argues for short, regular sessions over one heroic effort: a few minutes of Double Take a couple of times per week, kept up over time. Your best speed is saved on this device, so you can watch it improve session by session — most players see rapid gains in the first weeks as implicit learning kicks in.
Don't strain to stare harder — the skill develops through relaxed repetition, exactly the implicit learning the research describes.
More free brain exercise: Memory Match, Daily Word Game, or Sudoku.