Do brain-training games work? What the evidence can tell us

Brain-training games can produce better performance on the tasks people practise. That does not automatically mean better everyday memory, broader intelligence or improved performance on unrelated tasks. Research findings depend on the activity, participants, comparison group and outcome being measured.

The useful question is therefore: work for what? Finding a game enjoyable, learning a particular word and demonstrating a lasting change outside the game are three different outcomes. This guide separates them and looks at both a widely cited review and more recent evidence. Sources were checked on 16 October 2026.

Four outcomes that should not be bundled together

• Practised-game performance: doing better at the game you repeatedly play.
• Learning particular material: remembering a word meaning or a specific rule you encountered.
• Transfer: performing better on a different task. Near transfer concerns closely related tasks; far transfer concerns more distant ones.
• Everyday outcomes: changes in activities outside the test setting, such as managing ordinary tasks. These need their own measurements.

The 2016 review by Simons and colleagues found substantially stronger support for gains on trained tasks than for distant transfer or everyday cognitive performance. Its scope was published intervention studies cited by brain-training companies, rather than every possible form of learning. It also identified shortcomings in study design and analysis. That makes it a useful framework for judging a claim, not a reason to assume every training approach has identical results.

Learning a word is a specific achievement

Suppose you encounter frugal, check its meaning and later recall it when asked. The thing learned is identifiable. It is a different claim from saying the activity improved your general memory capacity.

A 2008 experiment by Karpicke and Roediger illustrates that distinction. Students learned foreign-language vocabulary pairs. Continuing to retrieve already-correct items helped their recall of those pairs a week later compared with dropping them from further testing. The outcome concerned the studied material under a particular practice procedure.

That study does not establish that an arbitrary word game improves everyday remembering or overall language ability. It also does not show that Word Sense implements the same learning procedure. A game containing words is not automatically a structured vocabulary course.

A controlled trial: game gains without broader advantages

In a 2017 trial by Kable and colleagues, the analysed sample comprised 128 healthy young adults who completed ten weeks of either Lumosity training or ordinary online video games. The comparison involved another activity, not simply doing nothing.

The training group improved on its practised games. However, the researchers found no advantage over the comparison group on their decision-making outcomes or untrained cognitive measures. A separate retest group also improved on cognitive tests, illustrating why before-and-after gains alone do not establish a training effect.

There are limits: 166 people were originally randomised and 38 were lost to follow-up. The participants were young adults, and the study tested one training programme and schedule. It does not settle what every activity does for every population. It does show why a rising in-app score cannot stand in for an independent outcome.

Newer evidence is not uniformly negative

A systematic review by Zhu and colleagues, published online in 2024 and in a 2025 journal volume, examined cognitive training in healthy middle-aged adults. It included 19 articles in its narrative synthesis and eight in meta-analyses, with searches through July 2023.

The pooled analyses found positive results for executive function, verbal memory and working memory, but not attention/processing speed or language. The authors also discussed some transfer findings. These results deserve acknowledgement rather than a blanket claim that training never affects another task.

The review combined different interventions, including strategy-based and computerised training. Most domains had low-to-moderate certainty of evidence; some estimates varied substantially between studies. Limited data prevented several subgroup and maintenance analyses. It would be an overreach to turn these results into a promise about any commercial game, a few minutes of daily play, or lasting everyday benefits.

Why studies and summaries can disagree

A review that combines strategy instruction, specific working-memory exercises and game collections is asking a different question from a trial of one app. Even two tests called “memory” may ask people to do different things. Their findings should be compared at the level of the actual activity and outcome, not just the marketing category.

The studies above also show why the comparison matters. An active alternative helps separate the effect of the proposed training from repeating a test or taking part in an engaging activity. Follow-up matters for claims about persistence. A statistically detectable change on a laboratory task still needs interpretation before anyone calls it a meaningful everyday benefit.

A practical way to read the next claim

When you see “improves memory”, look for what participants practised, what was measured separately, who took part and what the comparison group did. Check whether the result lasted and whether the product being advertised was actually studied. A study about a different programme is not product validation.

You can choose a game because you enjoy interpreting clues, remembering an arrangement or working through a deduction. If you set a five-minute break, that is a personal boundary, not a research-backed dose. Enjoyment does not require a claim of general cognitive improvement.

Puzzle Lantern offers those activities: Word Sense for clues, Recall for patterns and Sudoku for deductions. None of the studies discussed here tests Puzzle Lantern. Play it for the puzzles and enjoyment it offers, without treating it as a proven cognitive intervention.