Numvolt · Guides

How mental arithmetic works in your head, and why some methods are easier

Retrieval versus computation, where working memory runs out, why carrying is the expensive part, why anxiety costs accuracy, and what all of that means for how you practise.

By Roméo Gambino · Updated September 5, 2026

Two different ways to get an answer

When you answer an arithmetic question, one of two things happens. Either you retrieve the answer, the way you retrieve a familiar name, or you compute it through a series of steps. Seven times eight is usually retrieved. Twenty-three times seven is usually computed. These feel similar from the inside and they are not the same process at all, and almost everything that makes mental arithmetic easy or hard follows from which one you are doing.

Why retrieval feels instant

A retrieved fact costs almost nothing because the work was done years ago. You are not calculating, you are recognising. That is why a times table learned properly stays available under pressure while a method you half-remember collapses the moment you are distracted. The practical implication is unfashionable but real: memorised facts are not the opposite of understanding, they are what frees up room for it.

What happens when you compute instead

Computing means holding a partial result while you produce the next one, then combining them. Each step is easy on its own. What makes the whole thing difficult is that nothing is written down, so every intermediate number has to be kept alive while you work on something else. That is a memory problem wearing an arithmetic costume.

Where working memory runs out

DeStefano and LeFevre reviewed the role of working memory in mental arithmetic in 2004, and the picture that emerges is that the load is real and specific: holding intermediate results, keeping track of where you are in a procedure, and resisting interference from the numbers you have already used. When a calculation collapses, it usually collapses at one of those three points rather than at the arithmetic itself.

Why carrying is the expensive part

Carrying is the step that forces you to hold something while doing something else, which is precisely the load described above. This is why 48 plus 37 is harder than 40 plus 30 plus 8 plus 7, even though it is the same sum. Methods that avoid carrying are not tricks, they are ways of restructuring a problem so that it demands less of the part of you that is most likely to fail.

Why left-to-right suits the head

On paper you work from the right because carrying requires it. In your head that order is backwards, because it forces you to hold a growing tail of digits before any of them means anything. Working left to right gives you a usable approximation from the first step and keeps the number of things you are holding small. The method is not faster in theory, it is cheaper in working memory.

Why rounding then correcting is cheap

Turning 8 times 49 into 8 times 50 minus 8 replaces one hard step with two easy ones. It looks like more work and it is less, because the intermediate values are round numbers that are easy to hold. Almost every well-known mental arithmetic shortcut is doing the same thing: trading a step that is hard to hold for two steps that are not.

People use several methods and switch between them

Research on how people actually do arithmetic finds that they do not apply one procedure consistently. They retrieve when they can, compute when they cannot, and choose different routes for different numbers, often without noticing. Campbell and Xue examined cognitive arithmetic across groups with different educational backgrounds in 2001 and found meaningful differences in the mix of strategies used. The skill being trained is partly the choosing, not only the calculating.

Why the same problem is easy for one person and hard for another

Two people can meet 16 times 25 with entirely different experiences. One retrieves that 16 times 25 is 400 because quarters are familiar. Another halves and doubles twice. A third grinds through the standard algorithm and runs out of room. None of that is about intelligence. It is about which facts are automatic and which routes have been practised.

Anxiety takes up the same space

Ashcraft described in 2002 how mathematics anxiety carries cognitive as well as personal consequences, with intrusive worry consuming the same working memory that the calculation needs. That produces a self-confirming loop: worrying about being bad at arithmetic makes you worse at it in the moment, which confirms the worry. Knowing that the mechanism is competition for a limited resource, rather than a lack of ability, is genuinely useful.

What that means for pressure

If working memory is the bottleneck, then anything that consumes it degrades performance: being timed, being watched, being tired, being interrupted. A calculation you can do easily alone may fail in front of someone else, and that is a resource problem rather than a knowledge problem. Practising under mild time pressure helps, but only after the method is solid without it.

Writing things down is not cheating

If the constraint is holding intermediate results, then writing one down removes the constraint. There is nothing virtuous about carrying six digits in your head when a scrap of paper will do it for free. Save mental arithmetic for the cases where it is genuinely faster than reaching for something, which is most everyday cases, and stop treating every calculation as a test.

Estimate first, always

Producing a rough answer by rounding hard costs a second and catches the errors that matter, which are not small slips but answers wrong by a factor of ten. It also gives you something to hold onto if the exact calculation falls apart halfway. Most real arithmetic exists to support a decision, and a decision usually needs the right order of magnitude more than it needs the last digit.

How expertise actually develops

Not by learning more tricks. It develops by making more facts automatic, so fewer steps need computing, and by practising the choice between routes until it stops being a choice. That is why short frequent practice with immediate correction outperforms long sessions: you are building retrieval and selection, and both are built by repetition rather than by effort.

What this means for practice

Three things follow. Learn the facts that remove steps, which means tables, squares and the common fraction and percentage equivalents. Prefer methods that keep the number of held items small, which usually means left to right and rounding with correction. And correct every error at the moment it happens, naming which part broke: the method chosen, the arithmetic inside it, or a slip in holding a number.

Where Numvolt fits

Numvolt has a calm mode with no timer, where every wrong answer shows the method behind the correction rather than only marking it wrong, and timed sprints for when you want to measure speed and accuracy together. That order matters given everything above: build the method without pressure, then add pressure once holding the steps no longer takes all your attention.

Questions

Why is mental arithmetic so hard even when the steps are easy?

Because nothing is written down, so every intermediate result has to be held while you work on the next one. DeStefano and LeFevre reviewed the role of working memory in mental arithmetic in 2004, and the load falls on holding partial results, tracking where you are in a procedure and resisting interference from numbers already used.

Why is left-to-right easier in your head?

Because it gives you a usable approximation from the first step and keeps the number of things you are holding small. Working from the right forces you to hold a growing tail of digits before any of them means anything, which is what makes carrying the expensive part.

Is memorising times tables still worth it?

Yes, and for a specific reason. A retrieved fact costs almost nothing because the work was done years ago, so every fact you have automatic is a step you no longer have to compute. Memorised facts are not the opposite of understanding, they free up the room understanding needs.

Does math anxiety actually make you worse at calculation?

Ashcraft described in 2002 how intrusive worry consumes the same working memory the calculation needs, producing a loop where worrying about being bad at arithmetic makes you worse in the moment. The mechanism is competition for a limited resource rather than a lack of ability.

Is it cheating to write down intermediate results?

No. If the constraint is holding results in your head, writing one down removes the constraint. Save mental arithmetic for the cases where it is genuinely faster than reaching for something, which covers most everyday situations.