“Go upstairs, put your reading book in your bag, and bring me your water bottle.” Ten seconds later the child is upstairs doing something else entirely, and the reasonable adult conclusion is that they were not listening. Before settling on that, it is worth counting what the sentence actually asked for — because it is not one request, and hearing was only the first of several things that had to go right.

1. Hear It — Which Is Not the Same as Detect It

The sound has to be separated from everything else arriving at the same moment: a television, a sibling, traffic, the noise of whatever the child was doing.

That separation is active work, not a passive property of having functioning ears. A child can pass a hearing test and still find speech hard to pull out of a noisy background, and the two are genuinely different questions.

2. Hold It, in Order, While the Rest Arrives

By the time the sentence ends, the beginning has to still be available. Three actions means three items held simultaneously, in sequence, through the several seconds it took to say them.

This is the part most often described as the whole problem — a working memory limitation — and it is real. But the research below suggests it is not the entire account.

3. Stop What You Were Already Doing

An instruction arrives while something else is in progress, and that something else usually has more immediate pull than the request.

Suppressing an ongoing action to substitute a different one is effortful in its own right, separate from remembering what the new action is. A child can retain the instruction perfectly and still not execute it, because the stopping did not happen.

4. Convert Words Into Movement

This is the step that turns out to matter more than expected, and it is where the research gets interesting.

In one study, 32 children encoded simple spoken instructions while performing distractor tasks designed to interfere in different ways — articulatory suppression, backwards counting, and a motor suppression task. They then recalled the instructions either by saying them or by physically performing them.

The distractors impaired performance, as expected. But the enactment advantage — better memory when children performed the instruction rather than repeating it aloud — remained intact in every condition. The authors conclude the effect does not rely solely on working memory resources (British Journal of Developmental Psychology, 2023).

That last clause is the one worth carrying. If performing beats repeating even when working memory is under load, then instruction-following is not purely a storage problem, and treating it as one misses part of what is happening.

5. Which Encoding Strategy Helps

A 2026 study followed the question into the classroom, on the grounds that encoding and recalling spoken instructions plays a key role in learning there.

In the first experiment, children aged 8 and 9 heard instructional sequences of two, three and four actions while using one of four strategies — motor imagery, action observation, self-enactment or plain verbal rehearsal — then recalled either orally or by enacting. The second experiment ran a similar design with adolescents aged 12 to 14 on four-action sequences (Behavioral Sciences, 2026).

The point of comparing several action-based strategies rather than testing one is that they differ in what they cost. Imagining an action, watching someone else do it, and doing it yourself are not equally practical in a kitchen or a classroom, and knowing whether they work equally well is what makes the finding usable.

What This Does and Does Not Mean

It does not identify why any particular child did not follow any particular instruction. These are small experimental studies — 32 children in one — using deliberately simple instruction sequences under controlled conditions, and they were designed to isolate a memory effect rather than to explain household mornings.

It does not distinguish between the possible failure points either. Nothing here tells a parent whether hearing, holding, stopping or starting was the step that did not happen, and that is precisely the diagnostic question.

It does not license removing the instruction from the child. The enactment advantage is about how something is encoded, not an argument for asking less.

What it does support is narrower and useful: that following an instruction is a multi-step operation, that memory for instructions is measurably better when they are performed rather than recited, and that this advantage is not explained by working memory alone. Where a child appears to hear everything and act on none of it, the sensory-load version of that question is a different one about filtering, and the tiredness version is the September one.