Fluency drills that free the workspace — free
The practice portal's lower-level sets are perfect for making number facts automatic, so the counter is clear for the harder questions. Self-paced, instant feedback, no clock.
Open the practice portal →A child does each step correctly on its own, then falls apart the moment the steps have to be held together. It looks like they "don't get it". Far more often, they have simply run out of mental workspace. This is the story of working memory and cognitive load — the most practical idea in learning science, and one most of us were never told.
Give a struggling eleven-year-old a single instruction — "multiply these two numbers" — and they manage it. Ask them to "find the area, then subtract the smaller rectangle, then halve the result", and something visibly collapses. The pencil stops. The eyes glaze. "I don't get it," they say. But they just demonstrated that they do get each piece. What broke was not understanding. It was the attempt to hold four things in mind at once in a space that fits about three.
This is one of the most common and most misread moments in all of maths learning. A parent sees a child who "can't do problems". A teacher ticks a box marked "lacks problem-solving skills". The child concludes they are stupid. And the real culprit — a tiny, overloaded mental workspace — goes unnamed, even though naming it points straight at the fix.
Watch for this signature and you will see it everywhere. A student is fine with practice questions that drill one skill, then drowns in a word problem that combines three. They can recite times tables but lose the thread of a long-division question. They understand fractions on Monday and seem to "forget" them inside a harder topic on Friday. None of this is forgetting, and none of it is low ability. It is what running out of working memory feels like from the inside — and from the outside it is almost perfectly disguised as "not understanding".
Psychologists divide memory into two very different stores. There is long-term memory, which is effectively unlimited — a vast filing cabinet holding everything you know. And there is working memory, the small desk in front of the cabinet where you actually do your thinking. Anything you are consciously holding right now — a phone number, the next step of a sum, this sentence — is on that desk.
Think of a tiny kitchen counter. You can chop on it beautifully — but only a few ingredients fit at once. Try to lay out the whole recipe and things start sliding off the edges. Working memory is that counter. The filing cabinet behind you (long-term memory) is enormous, but the counter where the cooking actually happens stays small no matter how clever you are. The trick to cooking a complex meal isn't a bigger counter — it's prep: getting things off the counter and into reach.
Decades of research suggest the counter holds only around three or four "items" of new information at a time. That limit is roughly fixed and roughly universal — it does not grow much with age or talent. What changes between a novice and an expert is not the size of the counter but how much each item contains. To a beginner, "solve 3x + 6 = 18" is half a dozen separate items. To a fluent student it is a single, familiar chunk — one ingredient, already prepped — leaving the rest of the counter free for the genuinely hard part of a question.
When the counter overflows, the result is not a polite error message. It is a stall. The child stops, stares, maybe guesses, often gives up — and everyone in the room, including the child, reads this as a failure of understanding or character. This mislabelling is expensive. Treat overload as "doesn't understand" and you re-explain the concept, adding more to the overloaded counter. Treat it as "not trying" and you apply pressure, which (as our companion piece on maths anxiety explains) consumes even more of the very workspace that is already full. Both well-meant responses make the exact problem worse.
Reviewing the field, Raghubar, Barnes and Hecht (2010) found working memory bound up with mathematics at every level they examined — in moment-to-moment calculation, in the differences between children who find maths easy and hard, and as a predictor of how children's maths develops over years. Working memory is not a side issue in mathematics. It is, to a large degree, the engine room.
John Sweller's cognitive load theory (1988) reframed teaching itself around this limit. Sweller argued that because working memory is so easily swamped, the way material is presented can either overwhelm it or work with it. Poorly designed lessons spend the counter on the wrong things — confusing layouts, split attention, unnecessary searching — leaving nothing for actual learning. Well-designed lessons protect the counter for the ideas that matter. The implication is radical: a child's "failure" is often a design failure.
Ashcraft and Krause (2007) showed that maths anxiety works by occupying working memory — worry behaves like an extra, uninvited task on the counter. This is why anxiety and overload are so easily confused: both leave less room for the maths, and both produce the same stall. Often a struggling child is fighting both at once.
Maths is unusually demanding of working memory for reasons built into the subject. A multi-step problem requires you to hold the goal ("what am I trying to find?"), the intermediate results ("I've got 48 so far"), the method ("now I divide"), and the notation ("which bracket was I in?") — all at once, all in that small space, often while a clock ticks. Drop any one and the whole structure can fall.
Worse, maths hides its load. A page of algebra looks tidy and short, so adults underestimate how much is being juggled beneath it. And because the load is invisible, when a child buckles we reach for visible explanations — laziness, ability, attitude — and miss the quiet arithmetic of a counter that simply ran out of room.
One of cognitive load theory's most useful and counterintuitive findings concerns how we ask children to practise. The instinct — especially a well-intentioned, "let them figure it out" instinct — is to hand a learner problems and let them struggle toward the method. For a genuine beginner, Sweller's research suggests this is often the least efficient route, because unguided searching for a method floods the counter with dead ends, leaving little capacity to encode the method even when it's found.
The alternative is the worked example: study a fully solved problem first, with each step visible, then try one yourself. Beginners who study worked examples typically learn more, faster, with less strain than those left to flounder. The twist is that this advantage fades — and can even reverse — as students gain expertise, at which point solving independently becomes the better workout. (Researchers call this the "expertise-reversal effect".) The practical lesson: match the support to the stage. Heavy scaffolding when it's new; remove it as fluency grows.
A novice learns a new method faster by studying two worked solutions than by battling two blank problems — because the battle spends working memory on searching, not on learning. Once the method is secure, flip it: now blank problems are the better practice. Support should fade as skill grows, not vanish all at once or linger forever.
You can read a lot of international maths success as, in part, good cognitive-load management. Here is how several systems approach it.
Singapore's Concrete–Pictorial–Abstract sequence is, among other things, a load-management device: physical objects and then pictures carry part of the thinking that abstract symbols would otherwise force onto the counter. The famous "bar model" externalises a word problem's structure onto paper, so the child reasons about a picture instead of holding the whole problem in their head.
Classrooms in high-performing Chinese cities lean heavily on carefully sequenced worked examples and "teaching with variation" — small, deliberate changes between problems that build a secure method before the load is increased. It is close to a textbook application of cognitive load theory, whether or not it is labelled as such.
Australia has become a hub for the explicit-instruction movement, with national bodies drawing directly on cognitive load theory to argue for fully-guided teaching of new material before independent practice. Sweller himself worked in Australia, and the debate there over "explicit vs inquiry" teaching is essentially a debate about how to respect working-memory limits.
England's "maths mastery" reforms borrow the Singapore logic: small steps, shared representations, and not moving on until a step is secure. The "I do, we do, you do" structure gradually transfers load from teacher to student — scaffolding that fades, exactly as the worked-example research recommends.
The United States has hosted the sharpest version of the argument — the long-running "math wars" between discovery-oriented and explicitly-taught approaches. Cognitive load theory weighs in clearly for novices: minimal guidance tends to overload beginners. The nuance is that guidance should fade as expertise grows, which both camps can, in principle, agree on.
You can lighten a child's cognitive load at the kitchen table without teaching a scrap of new maths.
Lead with worked examples for anything new. Show the method fully before asking for independent attempts, then fade the support as fluency grows. The pairing of a worked example with a near-identical problem to try ("example–problem pairs") is a reliable, low-effort upgrade to most practice sets.
Hunt down avoidable load. Split attention (a diagram here, its labels there), cluttered slides, and information the student must hold in mind while hunting for something else all tax the counter for no learning benefit. Integrate diagrams with their labels; keep the essential things in view; remove decorative noise.
Build fluency to free capacity. Every fact made automatic is a permanent reduction in load for every future topic that uses it. This is the deep reason fluency and understanding are partners, not rivals — a theme we develop in understanding versus memorising.
The practice portal's lower-level sets are perfect for making number facts automatic, so the counter is clear for the harder questions. Self-paced, instant feedback, no clock.
Open the practice portal →"Smart kids can just hold more in their heads."
The workspace is roughly the same small size for everyone. Experts seem to hold more only because their knowledge lets them treat big ideas as single chunks. The route to 'more capacity' is more secure knowledge, not a bigger brain.
"Brain-training apps will boost working memory and fix maths."
Working-memory games mostly make you better at the games. Transfer to maths is weak. Making facts automatic and knowledge secure is the reliable way to free up capacity.
"Letting children discover methods themselves is always deeper learning."
For novices, unguided discovery often overloads the workspace and teaches little. Worked examples first, independence later, tends to produce deeper learning with less strain.
It's your mind's small workspace — the place you hold and juggle information for a few seconds while you think. The "carry the one", the half-finished equation, the instruction you just heard: all of it sits in working memory. It's powerful but tiny, holding only a handful of items, which is why complex maths can overwhelm it.
Because a multi-step problem asks the workspace to hold the goal, the running total, the method and the notation all at once. When that exceeds capacity, the system stalls — the freeze you see. It almost always means the problem is overloading working memory, not that your child lacks ability. Writing each step down and making the basics automatic both help enormously.
Not much directly — its raw size is fairly fixed, and brain-training games show little transfer to schoolwork. But you can achieve the same effect indirectly: make number facts automatic so they stop taking up space, and build secure knowledge so big ideas can be recalled as single chunks. Functionally, that gives a child far more room to think.
For someone learning something new, usually yes — studying a fully worked solution shows the method without flooding the workspace with dead ends. As expertise grows, the advantage shifts toward solving independently. The skill is matching the level of support to the stage of learning.
They're cousins. Anxiety and overload both eat working memory, so they produce the same stall and often occur together. The difference is the lever: anxiety eases when you lower the stakes; overload eases when you lower the load. Many children need both, which is why we treat the two articles as a pair.
In the free assessment we watch how your child solves, not just whether they get the answer. That's usually enough to tell overload from a genuine gap — and to know exactly what to do next.
Book the free assessment