Working Memory Exercises: What Works and What Doesn't

Working memory is the small mental workspace you use to hold a few things in mind while you do something with them: a phone number while you dial it, the first half of a sentence while you read the second, three rules while you solve a puzzle. It is limited for everyone, and when it overflows you drop things.

So it is natural to want exercises that make it bigger. This guide covers what working memory is, what decades of training studies actually show, and the exercises and habits that are worth your time because they help you get more from the capacity you have.

What working memory is

The most widely used model comes from Alan Baddeley and Graham Hitch, who in 1974 proposed that working memory is not one store but a small system. Baddeley later added a fourth part.

Working memory differs from short-term memory mainly in the "working". Short-term memory is holding a list. Working memory is holding the list and doing something with it, such as reordering it, comparing it or calculating with it. Researchers such as Randall Engle argue that much of what we call working memory capacity is really the ability to control attention: keeping the right things active while blocking the rest.

How much it holds

You may have heard "seven, plus or minus two", from George Miller's famous 1956 paper. Later work revised that down. Nelson Cowan's 2001 review concluded that when people cannot rehearse or group items, adults hold about four chunks, typically somewhere between three and five.

The key word is chunk. A chunk is any unit your long-term memory already knows. "C, A, T" is three letters, but "CAT" is one chunk. Chess masters in the classic studies by Chase and Simon could recall a real game position after a glance because they saw a few familiar patterns, not 25 separate pieces. When the pieces were placed at random, their advantage largely disappeared.

Why it matters for reasoning

Working memory capacity is closely related to fluid intelligence, the ability to solve new problems. Matrix puzzles are a good example: you hold one rule about shape, another about rotation and a third about count, then check each candidate answer against all three. That is also why working memory sits near the center of general ability, the g factor. People with a large working memory span show up on our list of signs of high IQ, though a bigger span is a tendency, not a guarantee.

Can you train working memory? What the evidence says

This is where most "working memory exercises" articles overpromise, so it is worth being precise.

In 2008, a study by Susanne Jaeggi and colleagues reported that training on a demanding "dual n-back" task raised scores on a fluid reasoning test. It caused a boom in brain-training apps. Then came the replications and meta-analyses.

• Melby-Lervåg and Hulme (2013) pooled 23 controlled studies. Training produced reliable short-term gains on working memory tasks, but there was no convincing evidence that it carried over to other skills such as verbal or nonverbal ability, attention control, word decoding or arithmetic.
• Melby-Lervåg, Redick and Hulme (2016) updated this with 87 publications and 145 comparisons. Gains on similar working memory tasks were reliable. Gains on intelligence, reading and arithmetic were not, once training groups were compared with an active control group doing some other engaging task.
• Simons and colleagues (2016) reviewed the evidence cited by brain-training companies and concluded that training improves the trained tasks, sometimes closely related tasks, and rarely anything further.
• Sala and Gobet (2019) pooled meta-analyses across working memory training, video games, chess and music. The effect on general cognitive ability was minimal, and the variation between studies was largely explained by study quality.

Psychologists call this the difference between near transfer (getting better at tasks like the one you practiced) and far transfer (getting better at different things, like reasoning or school results). Working memory training gives you near transfer. Far transfer is the part that has not held up.

Working memory exercises compared

Here is how common exercises stack up, including the strategies that work by reducing load rather than enlarging capacity.

Exercises worth doing

The useful exercises are not about making the box bigger. They make each item smaller, keep the box clear, or build knowledge so that more fits in a single chunk.

1. Practice chunking on purpose

When you meet a string of details, group them. Read a card number as four blocks, not 16 digits. Turn a shopping list into categories (dairy, vegetables, cleaning). Before a meeting, reduce your points to three headings, each with details underneath. Within a week this becomes automatic, and it is the single strategy with the clearest link to how capacity actually works.

2. Rehearse out loud, then test yourself

For something you need to hold briefly, like directions, say it once out loud in your own words. That uses the phonological loop. For something you need to keep, close the page and recall it a few minutes later. Recall moves it out of working memory and into long-term memory, which is where knowledge stops competing for space. Our guide on how to improve learning speed covers this in detail.

3. Learn the domain, and capacity seems to grow

Experts appear to have a bigger working memory in their field. They do not. They have richer long-term knowledge, so a whole pattern counts as one chunk. If you want to hold more spreadsheet logic, more chess, or more code in your head, the most reliable exercise is deliberate practice in that domain.

4. If you like n-back, use it honestly

Brain-training games are not harmful, and some people enjoy them. Treat them as a game or a warm-up, not as an IQ intervention. If you want to work on reasoning, practicing the actual kind of problem (for example matrix puzzles) at least trains what you will be tested on.

Strategies that protect the capacity you have

Because the limit is real, the biggest gains usually come from not wasting it.

Reduce cognitive load

John Sweller's cognitive load theory, from 1988, shows that learners solving problems with heavy search demands can learn less, because the search uses up working memory needed for building understanding. Practical versions: study a worked example before trying a problem alone, put diagrams next to the text that explains them, and break multi-step instructions into numbered steps.

Offload to paper and screens

Writing a list, sketching a diagram or setting a reminder is called cognitive offloading. Risko and Gilbert's 2016 review describes how people naturally use the world as external memory, and how doing so frees internal resources for the task in hand. Before a demanding task, write down everything you might need. During it, park stray thoughts on a notepad instead of trying to keep them in mind.

Single-task where it counts

Switching between tasks has a measurable cost: people are slower and make more errors right after a switch, even when the switch is predictable, as Stephen Monsell's review describes. For thinking-heavy work, close messages, silence notifications and finish one chunk before starting another. We look at this in an office setting in working memory at work.

Manage anxiety and sleep

Worry competes for the same space. Ashcraft and Kirk found that people with high math anxiety showed reduced working memory capacity during math tasks, which helps explain why anxious test-takers underperform. Sleep matters too: a meta-analysis by Lim and Dinges found that short-term sleep deprivation impairs attention and working memory. Feeling sharp after a short night is not a reliable guide.

Common mistakes

• Treating it as unlimited. Trying to juggle eight things in your head and then blaming yourself for errors. Accept the limit and design around it.
• Expecting brain games to transfer. The trained task improves. Everything else mostly does not.
• Rereading instead of recalling. Rereading feels productive but builds familiarity, not retrievable knowledge.
• Multitasking during deep work. Each switch reloads context and costs accuracy.
• Confusing anxiety with ability. A crowded mind under stress is not a small mind. Lower the stress and performance often recovers.

To check your own span, try the digit span test. If you want to see working memory in action on reasoning problems, the LifeScore IQ test is 25 timed visual puzzles where several rules have to be held at once.

Questions people ask

What is the best exercise to improve working memory?

There is no exercise shown to raise general working memory capacity in a way that carries over to everyday life. The most useful habits are chunking, writing things down, single-tasking and practicing the specific skill you need, because they let you do more with the capacity you have.

Does dual n-back increase IQ?

The 2008 study that suggested it did has not held up well. Later meta-analyses comparing n-back training with active control groups found improvements on n-back and similar tasks, but no reliable gains on intelligence tests.

How many things can working memory hold?

Around four chunks for most adults when rehearsal and grouping are not possible, according to Cowan's 2001 review. The older figure of seven, plus or minus two, overestimates it because people were chunking.

Is working memory the same as short-term memory?

Not quite. Short-term memory is holding information briefly. Working memory is holding it while you use it, which adds attention control and makes it more closely linked to reasoning.

Does working memory decline with age?

Fluid abilities, including working memory, tend to peak in early adulthood and decline gradually, while knowledge keeps growing. Older adults often compensate with better chunking and expertise, which is another reason domain practice pays off. Our guide to the age at which IQ is highest shows when each ability peaks.