The Working Memory Gap Was Real, but the Cause Stayed Open

What the research found

Working memory is the mind’s short-term workspace. It is what lets a child hold a phone number in mind long enough to dial it, or keep the start of a sentence in view while reading to its end. Researchers have long suspected that this workspace behaves differently in children with dyslexia. A 2022 study in Scientific Reports set out to measure that difference carefully, using both behaviour and brain activity at the same time.

The team, led by Jie Wang and Urs Maurer, studied 74 native Cantonese-speaking children in Hong Kong, aged seven to eleven and in their second or third year of school. Forty-six had dyslexia; twenty-eight were typically developing readers who served as a comparison group. Each child completed a set of n-back tasks. In these, an item appears on screen, and the child decides whether it matches the item shown one or two steps earlier. Some tasks used Chinese characters (a verbal load); others used checkerboard patterns (a visual load). The one-back version is easier; the two-back version demands more holding and updating.

The behavioural gap was clear. Control children responded faster and more accurately than children with dyslexia across the tasks. On the verbal one-back task, controls answered in about 779 milliseconds against 932 for the dyslexic group, and their accuracy, measured as sensitivity, was higher too: 3.49 against 2.66. The pattern held on the harder visual two-back task, where controls scored 1.31 against 0.94. These are group averages, not verdicts on any individual child, and the ranges within each group overlapped.

At the same time, the researchers recorded electrical activity from the scalp using EEG. They did not look at the usual event-related spikes but at rhythmic brain activity in particular frequency bands. During the phase when children were holding information in mind, the dyslexic group showed higher frontal-midline theta activity, while control children showed higher upper-alpha power at the back of the head. In plain terms, the two groups were not just performing differently; their brains appeared to be doing the retention work in measurably different ways.

The headline result combined both streams. Feeding the behavioural scores and the brain measures into a statistical model, the researchers could sort children into the correct group 82.4 per cent of the time overall. The model correctly identified 87.0 per cent of the children with dyslexia and 75.0 per cent of the controls. That is a genuinely high hit rate for a research classifier, and it suggests working memory signals carry real information about dyslexia.

What it means for you and your child

It is tempting to read a result like this as a diagnostic test in waiting. It is not one, and the authors do not present it as such. An 82 per cent classification rate sounds impressive, but it also means the model was wrong for roughly one child in five. A quarter of the typically developing children were misclassified as dyslexic. No responsible clinician would decide a child’s future on those odds. This was a study of how groups differ on average, run on a sample of 74 children in one city, learning one writing system.

The deepest limit is one the authors state plainly. They cannot say whether the working memory differences cause reading difficulty or simply accompany it. It may be that a weaker mental workspace makes learning to read harder. It may equally be that years of effortful, frustrating reading shape how the workspace develops. A single snapshot taken at one moment, comparing two groups, cannot separate these possibilities. Cause and consequence look identical in a photograph. You would need to follow children over years to tell them apart.

Two further cautions matter for how much weight to give this. The authors note that the two groups were not matched on non-verbal intelligence, so some of the measured gap could reflect general reasoning rather than reading-specific memory. They also did not measure the children’s oral language ability, which sits close to both memory and reading. These are honest admissions, and they narrow what the study can claim.

So what can a parent reasonably take from it? That working memory and reading difficulty travel together is now well supported, and this study adds careful evidence to that. If your child struggles with reading and also loses track of multi-step instructions, or forgets the beginning of a sentence by its end, those two things are probably related, and neither is a matter of effort or attention alone. That understanding can change the tone at home. It invites patience and structure rather than pressure.

What the study does not license is a leap to any particular fix. It tested no intervention. It does not show that working memory training improves reading, because it did not train anything or measure change over time. Programmes that promise to lift reading by drilling memory games go well beyond what evidence like this can support. The sound response remains the ordinary one: a proper assessment by a qualified specialist, targeted reading support with a track record, and instructions broken into smaller, memorable steps. This paper deepens the picture of what dyslexia involves. It does not hand us a shortcut around it.