Brain Training Games: What Makes One Work

What separates a cognitive training game from a game

Four things, and a game needs all four to be called training, whether the family calls it a brain training game or a cognitive one. It targets a named mental task rather than a general idea of thinking. Its difficulty moves — up when the player succeeds, down when they struggle, without the player having to ask. It gives feedback within a second or two, so the player can tell whether the last action was right. And it produces a record that can be compared across sessions, which is what makes the word training mean anything.

Mahjong has three of the four. It demands working memory, planning and social attention, and people play it for hours. What it does not have is a difficulty ladder or a comparable record, so a care home can say that residents enjoy it and play it well, and cannot say whether anything changed over eight weeks. That is not an argument against mahjong; it is an argument for being precise about which claim the home is making.

The distinction matters commercially as much as clinically. A tablet app with bright colours and a cheerful voice is easy to sell to a worried daughter and hard to justify to an occupational therapist. Choosing a games app for older adults takes the buyer's side of that question, and this note takes the designer's.

Which cognitive domains a game can actually target

The domains that survive a tabletop or a touchscreen are the ones with a visible, scorable output. The table maps each to a task a care worker can set up this afternoon, because the best test of whether a game works is whether a person will play it twice.

Two cautions belong with the table. A game that mixes all five domains in one round is measuring the mix, not any one of them, and a change in the score cannot be attributed. And a domain is not the same as an everyday task: getting better at matching pairs on a tablet is getting better at matching pairs on a tablet, and the honest claim stops there unless the person also reports that something in the day got easier.

What the evidence supports, and what it does not

The World Health Organization's 2019 guidelines on risk reduction of cognitive decline and dementia recommend that cognitive training may be offered to older adults with normal cognition or mild cognitive impairment, and rate the certainty of that evidence as low. That is a permission, not a promise, and it is the correct frame for a care home: offering structured cognitive activity is reasonable, and claiming that it prevents dementia is not supported.

Read the wording closely, because it does two jobs. It says the intervention is low-risk and may be offered, which is enough to justify a session in an activity programme. It also says the evidence is weak, which means a home should not promise a family that a game will hold back a diagnosis, and a vendor should not print a claim that it will.

What the evidence is stronger on is the reason the session happens at all: structured activity, done with other people, is a defence against the afternoons when nobody comes. The case for activity as a defence against loneliness in care homes is more robust than the case for any particular game improving any particular score.

Designing a session that survives a care home afternoon

Thirty minutes, one domain, four to six people, and a staff member who is playing rather than supervising. That shape works because it fits the gap between lunch and tea, because one domain keeps the scoring meaningful, and because a group of six is small enough that nobody waits more than a minute for a turn.

Start below the player's level. The first two rounds should be ones the group will certainly pass, because the point of the opening is to establish that this is a thing they can do. Difficulty rises after success, not on a timetable. Stop while it is still going well — a session that ends on a win is a session that gets repeated, and a session that ends on three failures is a session that gets refused next week.

Have a stop rule for each player written into the plan: three consecutive errors, a change in mood, a request to stop, or a physical sign such as rubbing the eyes or shifting in the chair. The rule is not a judgement about the person. It is what lets a care worker offer the activity to someone with moderate impairment without spending the session afraid of getting it wrong.

Recording a score without turning the session into a test

Record three things per session and no more: which task, the highest level reached, and how long the player stayed engaged. Those three are enough to see a trend over a month, and they are few enough that a care worker will still be filling them in during the fourth week.

Do not turn the record into a cognitive assessment. A game score is not a MoCA, it is not a diagnosis, and a fall in the score is not evidence of decline — it can be a poor night, a new medication, a hearing aid left in the bedroom, or a room that was too noisy. What a falling score justifies is a conversation with the nurse and, if the pattern holds, a referral, not a conclusion written on a chart.

The same rule protects the family. A daughter who is shown a graph of her mother's declining scores will read it as a prognosis. A daughter who is shown which game her mother chose to play three times this week will read it as an afternoon that went well, which is what the record actually shows.

What to avoid, and where our own product stands

Avoid time pressure as a default. A countdown turns a task into a race and a race into a failure, and the residents most likely to be offered a cognitive game are the ones least able to win one. Where speed is the domain being trained, measure it quietly rather than displaying it. Avoid competition between residents, which produces one winner and five people who now know they were slowest. Avoid anything that looks like a children's toy, because an adult asked to play one is being told something about how they are seen.

And avoid the claim that the game is doing something medical. It is a structured activity with a score. If it is well designed, it is also the best thirty minutes of some residents' afternoons, and that is worth doing without inflating it.

Sanospark is the cognitive games line SilverGuard is building, and its games are in demo development rather than on sale. We describe it that way on the Sanospark site because the design rules above are the ones we would have to meet, and a demo is the honest way to find out whether we have.