The Effect Did Not Survive Blinding

What the research found

In 2022, a team led by Shuxian Chen published a meta-analysis in the journal Frontiers in Psychology. A meta-analysis pools the results of many separate studies to see what pattern holds across them. This one gathered 17 randomised controlled trials of cognitive training in children with attention deficit hyperactivity disorder, covering 1,075 children aged between three and eighteen. Every trial had randomly assigned children either to a training programme or to a comparison group, which is the design that gives us the best chance of seeing whether the training itself caused any change.

The researchers asked a sharper question than most. Rather than simply asking whether cognitive training works, they asked which factor makes it work. They separated two kinds of programme. The first trains a single mental skill, usually working memory, drilled on its own. The second, which they call multiple cognitive training, works across several domains at once, such as working memory, inhibition, attention, flexibility and planning. They then compared the two.

The headline result was that the type of training mattered a great deal. Multi-domain training was associated with a meaningful reduction in inattention, with a standardised effect of −0.51 (95% confidence interval −0.72 to −0.29). It also moved hyperactivity and impulsivity, at −0.31, and everyday executive function behaviours, at −0.50. Working memory training on its own did essentially nothing. Its effect on inattention was 0.04, on hyperactivity 0.13, and on executive function 0.09, and none of these were statistically distinguishable from no effect at all. The single most popular form of brain training, drilled working memory, did not register.

The authors also looked at how much training children received. Neither the length of each session nor the total number of sessions predicted benefit. The one dosing factor that did was frequency, how often the training happened, which was statistically significant for inattention and for hyperactivity. Their careful conclusion was that multi-domain training at a moderate frequency may carry wider benefit than single-skill drilling.

Then comes the finding that should temper all of the above. The apparent gains were reported by parents and clinicians who knew which children had been trained. When the researchers restricted their analysis to the small number of trials that used blinded raters, people who did not know which group a child was in, the effects on both inattention and hyperactivity were no longer statistically significant. Only 6 of the 17 trials reported any blinded measurement. On formal assessment of study quality, just 3 of the 17 were judged to be at low risk of bias, while 11 were at high risk. The authors name the likely culprits plainly: parental expectancy bias and the Hawthorne effect, the tendency for people to change simply because they know they are being observed and helped.

What it means for you and your child

The useful message here is a comparative one, and it is worth holding onto. If a family is going to spend time and money on cognitive training, a programme that works across several thinking skills has more evidence behind it than one that drills working memory alone. Standalone working memory apps, which are widely marketed to parents of children with ADHD, did not separate from doing nothing in this analysis. That is a reasonable thing to know before you buy.

But the larger and more honest message is about certainty, and it cuts the other way. The benefits that looked encouraging were visible mainly to observers who were hoping to see them. When the paper stripped that hope out, by looking only at blinded raters, the benefits thinned to nothing that could be confidently distinguished from chance. We cannot yet tell how much of the reported improvement was a real change in the child and how much was a change in how a hopeful parent perceived that child. Both are understandable. Only one of them is what you are paying for.

This does not mean cognitive training is useless, and it is important not to overread it in the opposite direction. A meta-analysis of mostly higher-risk trials cannot prove a treatment ineffective any more than it can prove it effective. What it can do is tell you where the evidence is currently thin and where the reasonable expectation should sit. Right now, the evidence for cognitive training producing durable, observer-independent change in ADHD is weak. The authors themselves note that medication remains the first-line treatment, and they do not claim that training gains transfer to broad real-world behaviour.

For a parent, that suggests a calm and specific stance. Treat multi-domain cognitive training as a plausible supplement worth trialling, not as a foundation and not as a substitute for the interventions with stronger backing. Set a clear, observable target before you begin, ideally something a person outside the household could notice, such as a teacher’s report or a homework routine that runs without you standing over it. Give it a defined window. If the only evidence of improvement is your own sense that things are better, treat that gently but sceptically, because this study shows exactly how easily that sense can outrun the underlying change.

Above all, resist the marketing that presents brain training as a settled solution. This is one meta-analysis, built on trials the authors judged mostly imperfect, and it points to a discipline that has not yet earned confident claims. The most valuable thing it offers is not a product recommendation but a habit of mind: ask who measured the result, and whether they knew what they were hoping to find.