Page 3 of 12
What helped, and what did not
Memory games often improve performance on memory tests. The harder question is whether they also improve reading, everyday attention or wellbeing. These studies tested those wider benefits, with different results for different outcomes.
Memory practice mainly improves memory tasks
Working memory means holding information in mind while using it, such as remembering numbers during a calculation. Reviews of training found improvements on memory tests. They did not establish reliable improvements in reasoning, reading or arithmetic when the comparison group also did an activity. That comparison, called an active control, helps separate the effect of the exercises from the experience of taking part in a programme.
Read the working-memory studies and results
The numbers below compare improvement between groups on a common scale. Zero means no average difference; larger positive numbers favour training. They are not percentages or IQ points.
- Melby-Lervåg, Redick and Hulme (2016) reviewed 87 publications with 145 comparisons. Immediate gains appeared on verbal and visual working-memory tests. Comparisons with other activities did not provide convincing evidence of gains in general reasoning, word reading, reading comprehension or arithmetic.
- Sala and Gobet (2017) studied typically developing children. Gains on memory-related tests averaged 0.46. Gains on more distant skills averaged 0.12 across all studies, but only 0.03 in studies that both assigned groups randomly and used an active control. That last estimate did not establish an advantage over the comparison activity.
- Sala and colleagues (2019) combined earlier reviews. After accounting for publication bias and active controls, the estimated benefit on skills beyond working memory was about 0.01. Their wider analysis of the cognitive-training programmes examined also produced an estimate close to zero.
- Kassai and colleagues (2019) examined children's working memory, response inhibition and ability to switch tasks. Training helped on other tests of the same skill: an average effect of 0.44 across 43 comparisons. The estimated effect on a different one of those three skills was 0.11 across 17 comparisons, without a statistically clear advantage. This was a test of transfer between thinking skills, not a test of school grades or everyday behaviour.
Simons and colleagues (2016) reached a similar conclusion in their broad review: extensive evidence for getting better at trained tasks, less for closely related tasks, and little for distant skills or everyday performance. They also identified weaknesses in comparisons, expectations and outcome reporting.
Attention-training results are more limited than memory-test gains
In studies of people diagnosed with ADHD, computer training improved working-memory tests. The larger 2023 review also found a small reduction in inattention when ratings came from people judged likely not to know which training participants received. It found no clear immediate improvement in overall ADHD symptoms or hyperactivity and impulsivity.
Read the attention-training studies and results
- Cortese and colleagues (2015) reviewed 16 trials involving 759 children and adolescents. Ratings likely to be blinded showed a small overall symptom benefit, an effect of 0.20. That advantage was not clear when the analysis was limited to studies with active controls. Working-memory tests improved, but academic outcomes did not show a clear benefit.
- Westwood and colleagues (2023) reviewed 36 trials involving 2,234 participants, mostly children. On ratings likely to be blinded, effects were 0.12 for overall symptoms, 0.11 for hyperactivity and impulsivity, and 0.17 for inattention. Only the last was statistically clear. Verbal and visual working-memory effects were 0.38 and 0.49. A smaller set of follow-up studies suggested benefits in verbal working memory, reading comprehension and rated everyday thinking skills.
Behaviour improved in the comparison game too
Dovis and colleagues (2015) tested 25 sessions of a home computer game in 89 children aged 8 to 12 with ADHD. Children received full training, partial training, or a similar game without the intended training demands.
Full training improved visual-memory tests; both active versions improved inhibition tests. Behaviour ratings improved across all three groups, with no clear extra improvement from the active training. The study did not identify what caused that shared improvement. Its practical lesson is to compare groups, rather than credit the exercises for every change seen after starting them.
A school mindfulness course did not outperform usual lessons
The MYRIAD trial (2022) compared ten mindfulness lessons with schools' usual social-emotional teaching. It analysed 8,376 pupils from 84 UK schools, with an average age of 12.2. After a year, the mindfulness course had not produced better scores for depression risk, emotional and behavioural difficulties, relationships or wellbeing.
A separate analysis of the same trial found slightly worse depression-risk and wellbeing scores among pupils who already had more mental-health difficulties. The authors judged those differences small and below the level considered clinically important.
The wider evidence is mixed, not uniformly negative. Dunning and colleagues' 2022 review of 66 trials found small improvements in anxiety or stress and in measured mindfulness compared with other activities. Attention and executive-function benefits appeared against no-intervention comparisons, but not against active controls.
Read the mindfulness study details
MYRIAD randomised 85 schools; one withdrew before providing pupil data. The one-year group differences were 0.005 for depression risk, 0.02 for functioning and 0.02 for wellbeing, all close to zero. The research team also reported that more than 80% did not complete the required practice homework.
In the Dunning review, effects against active controls were 0.11 for anxiety or stress and 0.24 for mindfulness.
A separate small adult study by Pozuelos and colleagues (2019) found improved response inhibition and related brain responses after three weeks of breath-awareness practice. It measured performance on a button-press task, not the effect of a school curriculum on children's mental health.
Changing interpretations did not clearly reduce symptoms
Interpretation training asks people to practise less negative readings of ambiguous situations. A 2024 review by Sicouri and colleagues included 36 studies of children and adolescents. The training changed how participants interpreted situations, but did not produce clear improvements in anxiety symptoms, depression symptoms or negative feelings compared with the control conditions.
The estimated effects on anxiety and depression symptoms were 0.16 and -0.03.
Relational puzzles have not established a reliable IQ boost
SMART trains relationships such as “A is bigger than B; B is bigger than C.” Some studies found higher scores on IQ tests using pictures and patterns after training. The studies were small and had weaknesses in their methods, so they have not established a dependable intelligence benefit.
Read the SMART studies, including positive findings
- May and colleagues (2022) combined five small controlled studies. They found a moderate average benefit on nonverbal IQ tests, an effect of 0.57. All five studies had a high risk of bias, meaning their methods could distort the result. There was too little evidence to determine benefits in other cognitive or academic areas.
- McLoughlin and colleagues (2020) compared SMART with a coding activity in 12- to 15-year-olds. Among the participants analysed, nonverbal IQ scores rose by about 5.98 points with SMART and 1.85 points with coding. About 60% of participants dropped out, which makes the difference harder to interpret.
- Roche and colleagues (2023) studied 10- to 11-year-olds in school groups that were not assigned randomly. An apparent nonverbal IQ advantage was no longer statistically clear after accounting for starting IQ or attention.
- Meijer and colleagues (2026) compared eight weeks of SMART with working-memory training in older adults who reported cognitive concerns. The 105 completers improved more on relational reasoning with SMART, but not on nonverbal IQ, neurological status, self-reported executive functioning, depression or quality of life.
Time Gates’ relation puzzles are practice in working out relationships. We do not present them as a proven way to raise IQ.
Self-help apps have produced different results
In RETHINK (Funk and colleagues, 2025), 365 people aged 16 to 22 with high levels of repetitive negative thinking were offered six weeks of either a multi-strategy app, a concrete-thinking app, or a waitlist. Neither app clearly reduced depression symptoms, anxiety symptoms or repetitive negative thinking compared with waiting.
Use was low. In an additional analysis, researchers compared people who completed a minimum amount of the full app with the waitlist. That selected group reported a larger decrease in depression symptoms at follow-up. The analysis did not establish that low use caused the trial's main result.
A different trial, MyMoodCoach (Edge and colleagues, 2024), tested an unguided app in 236 people aged 16 to 24 with high worry or rumination. After six weeks, it improved worry, rumination, anxiety symptoms, depression symptoms and wellbeing compared with a waitlist.
Topper and colleagues (2017) also found benefits from a six-week programme delivered in a group or through guided online work. It combined several strategies. Its results do not isolate concrete thinking as the ingredient responsible, or show that adding guidance alone explains the difference between app trials.
What Time Gates takes from these findings
We choose exercises for the skills they directly practise. We will evaluate broader benefits by measuring those outcomes separately and comparing them with another activity. Better game scores alone are not evidence of better school performance or wellbeing.
We also want practice to fit ordinary life. In the #LIFEGOALS study, adolescents averaged only 26 minutes in the app over twelve weeks, and use fell sharply after the first week. That is a reason to test whether people actually use a product, not just whether they download it.
Our design choice is to keep sessions short and make returning after a missed day easy. We will test whether that approach supports regular practice, alongside the outcomes that matter to families.
Version 133 · Privacy · Terms of use · Appearance