Ask a five year old to go upstairs, fetch their shoes, put the book on the shelf and close the window. Then watch. Something will come back, possibly a shoe, possibly nothing, and the window will still be open.
The standard adult interpretation is that the child was not listening. A better interpretation is that they were listening, and the instruction was larger than the space available to hold it.
What working memory is
Working memory is the mental workspace where information is held while something is done with it. It is what you use to hold a phone number long enough to dial, to keep track of where you are in a sentence, or to do arithmetic without writing anything down.
It has two properties that matter enormously.
It is small. Estimates of adult capacity vary with how you measure it, and every serious estimate is a small number of items. Not dozens. A handful.
It is brief. Without active maintenance, contents decay in seconds. Anything that interrupts the maintenance, a question, a noise, a thought, can empty it.
This is not a defect in some people. It is the architecture, and it is the reason writing was such a consequential invention.
Children have less of it, and the difference is large
Working memory capacity increases through childhood into adolescence. A young child's workspace is substantially smaller than an adult's, and there is wide variation between children of the same age.
Two consequences follow immediately. Instructions that are trivially manageable for an adult can exceed a child's capacity entirely. And children with smaller capacity in a given classroom will fail at tasks that have nothing to do with the skill being taught, because the task's incidental load consumed the space.
Teachers who understand this recognise a characteristic pattern: a child who can do the thing when it is presented one piece at a time and cannot do it when presented whole.
Load is a property of the task, not just the learner
The useful move is to stop thinking about capacity as a fixed property of the child and start thinking about load as a property of the task.
Every task imposes load from several sources. The intrinsic difficulty of the material. The way it is presented, including whether the instructions have to be held in mind while the work is done. And anything that has not yet become automatic, because unautomatic operations consume space that automatic ones do not.
That last point explains a great deal about early learning. A child who is still decoding words effortfully has little capacity left for comprehending the sentence. A child who has to work out each number fact has little left for the structure of the problem. This is why fluency in components matters: not because speed is valuable in itself, but because automaticity frees the workspace. We take this up in how reading is learned and number sense before arithmetic.
What exceeded capacity looks like from outside
Rarely like a memory problem. Usually like something else, which is why it is so frequently misread.
It can look like not listening, because the instruction was received and then lost. Like giving up, because the child cannot reconstruct what they were doing. Like carelessness, because a step was dropped. Like avoidance, because a task that reliably ends in failure is reasonably avoided.
None of those are attitudes. They are what a person does when the workspace is full, and adults do all of them too under sufficient load, which is worth remembering before attributing any of it to character.
The training industry and its disappointing record
Given how much working memory explains, the commercial logic is irresistible: train it and everything improves. A substantial market exists on that premise.
The evidence has not been kind. Trained tasks improve, which is unsurprising. Closely similar tasks improve somewhat. Transfer to genuinely different abilities, and to school attainment, has repeatedly failed to appear when trials are well controlled, adequately powered and use active control groups.
This is one of the clearer cases where a plausible mechanism and a large market produced claims well ahead of results, and it is worth understanding as a pattern, because the same shape recurs. See replication in plain words for the general phenomenon.
What does help is reducing the load
Not a technique for parents to administer, but a description of what works in settings that handle it well.
Instructions given one at a time rather than in sequences. Information left visible instead of held in the head, which is the entire principle behind writing things down. Sequences broken into steps with a natural pause at each. And repetition of the instruction without any implication that the child was not listening, because they were.
The general principle is that anything moved out of the head frees space in it. This is why checklists work for surgeons and pilots, populations not usually accused of poor concentration.
Why this concept earns its place
Because it replaces a moral explanation with a structural one. A child who forgets three of four instructions is not being difficult; the instruction was too big. A child who cannot follow a maths problem may understand the maths and be spending everything they have on the arithmetic.
That reframing does not solve anything by itself. But it changes what the adult concludes, and what the adult concludes tends to determine what happens next.