Working Memory: How It Limits Learning and What Helps

Working memory is the small mental workspace where you hold information for a few seconds and work with it: keeping a carried digit while you finish a sum, holding the start of a sentence while you read the end, or juggling three steps of a proof. It can only hold about four chunks of information at a time, which makes it the main bottleneck when you're learning something new.

You can't make that workspace much bigger. Years of research on brain-training programs show they don't produce gains that carry over to schoolwork. What you can do is need less of it: offload, simplify, and build knowledge in long-term memory so each new idea takes up less room.

This guide explains Baddeley's working memory model, how much working memory can hold, what the training research really shows, and study strategies that work within its limits.

What is working memory?

Working memory is the system that keeps a small amount of information active and available while you use it for something else: reasoning, reading, solving a problem, following a lecture.

Working memory vs short-term memory

The two terms overlap, and researchers use them in slightly different ways. Nelson Cowan's 2008 review offers a useful distinction:

• Short-term memory is holding information briefly, like repeating a phone number until you dial it.
• Working memory is holding information and doing something with it, which takes attention. Doing mental arithmetic, or keeping a question in mind while you search a text for the answer, are working memory tasks.

The difference matters. Cowan notes that simple, routine holding tasks don't predict much about academic ability, while the attention-demanding tasks usually labelled "working memory" correlate well with aptitude measures.

For studying, this means the hard part isn't keeping information for a few seconds. It's keeping it while you think. Anything that does the holding for you, such as paper, a labelled diagram, or a formula you know cold, frees attention for the thinking.

Baddeley's model of working memory

The most widely taught model comes from Alan Baddeley and Graham Hitch, who proposed it in 1974. They described working memory as a set of parts rather than one store. In 2000, Baddeley added a fourth component.

Two practical points follow from the model:

1. The verbal and visual parts are partly separate. A diagram with a spoken or short written explanation spreads the load across both. That's one reason dual coding can help.
2. The central executive is the scarce part. Switching between tasks, resisting distractions, and coordinating information all draw on it. Each time a notification pulls your attention away, it has to drop and rebuild what you were holding.

How much can working memory hold?

Less than most people think. George Miller's famous 1956 estimate was "seven, plus or minus two" items. In 2001, Cowan reviewed the evidence and concluded that when people can't rehearse or group items, the real limit is about four chunks, with a range of three to five.

A chunk is a unit you already know: a letter, a word, a formula you recognize at a glance. A chemistry student holds "H₂SO₄" as one chunk; a beginner may need four or five. That's why the same lecture can feel easy to one student and overwhelming to another. Our chunking guide shows how to build bigger chunks on purpose.

Working memory is also brief. Without active rehearsal, information slips away within seconds, especially when something else grabs your attention. Researchers still debate how much of that loss is pure time-based decay and how much is interference from new information (the same debate shapes theories of why we forget), but the practical lesson is the same: if it matters, write it down.

Why working memory matters for learning

Everything you learn passes through working memory first. When a new topic throws too many unfamiliar pieces at you at once, some of them never make it into long-term memory. This is the core idea behind cognitive load theory.

Two findings show how working memory and knowledge interact:

• It predicts some school skills. A 2018 meta-analysis of 197 studies by Peng Peng and colleagues found a moderate correlation between working memory and reading (r = .29).
• Knowledge can stand in for it. In the same analysis, the link between working memory and reading comprehension disappeared once decoding skill and vocabulary were taken into account. When the words and concepts are already in long-term memory, there's less to juggle.

That second point is the hopeful one. You can't expand the workspace much, but every fact and pattern you master in long-term memory frees up space the next time you meet it.

Can you improve working memory?

You can get better at working memory tasks, but the gains don't spread to other skills. In 2013, Monica Melby-Lervåg and Charles Hulme analyzed 23 studies of working memory training programs. Training produced short-term improvements on working memory tests, but verbal gains faded by follow-up, and there was no convincing evidence of improvement in verbal or nonverbal ability, attention, word decoding, or arithmetic.

A larger 2016 meta-analysis by Melby-Lervåg, Thomas Redick, and Hulme pooled 87 publications with 145 comparisons. Against active control groups, here is how much each type of outcome improved right after training:

The authors also found that how much people improved on working memory tests had no relation to how much they improved on far-transfer measures. Their conclusion was that these programs produce short-term, specific training effects that don't generalize. If an app promises better grades through memory games, this is the evidence to weigh it against. The minutes it asks for are better spent on retrieval practice in the subject you're actually tested on. Our neuroplasticity guide covers brain-training claims in more detail.

Signs you're overloading working memory

• You reread the same sentence three times and still can't say what it meant.
• You lose your place in a multi-step problem and have to start over.
• You follow each step of a worked solution but can't explain the whole method.
• Your notes are a transcript of the slides because you had no capacity to decide what mattered.
• You feel confused, not challenged: there are too many new terms to connect anything.

If two or more of these sound familiar, the fix is less juggling, not more effort. Each strategy below takes something off your mental plate.

8 study strategies that work with working memory limits

1. Write intermediate steps down. Paper is free working memory. In math and science, write every step instead of holding it in your head.
2. Learn the basics to automaticity. Times tables, unit conversions, key formulas, core vocabulary. When they come back instantly, they stop using working memory. See how to memorize formulas.
3. Study a worked example before solving. For brand-new problem types, read a fully worked example and explain each step to yourself, then try a similar problem on your own.
4. Put labels on diagrams. Text that sits next to the part of the diagram it describes saves you from holding one while searching for the other.
5. Take one new idea at a time. Pause lecture videos after each new concept and summarize it before moving on.
6. Turn off notifications while learning new material. New material needs the most attention, so protect those sessions most.
7. Chunk and name. Group related items into a few named units so they take one slot each.
8. Build long-term memory with retrieval practice. Every concept you can recall instantly is one less thing to juggle later. Retrieval practice is the most reliable way to get there.

Worked example: a stoichiometry problem without overload

How many grams of water form when 8.0 g of hydrogen gas burns completely in oxygen?

A student who tries this in their head has to hold the balanced equation, two molar masses, a mole ratio, and two intermediate answers at once, which is well over four chunks. Here's the low-load version:

1. Write the balanced equation: 2H₂ + O₂ → 2H₂O. Now it's on paper, not in your head.
2. Write the route as one chunk: grams H₂ → moles H₂ → moles H₂O → grams H₂O. Once you've practiced this route a few times, it becomes a single familiar pattern.
3. Do one step per line: 8.0 g ÷ 2.0 g/mol = 4.0 mol H₂. Then 4.0 mol H₂ × (2 mol H₂O / 2 mol H₂) = 4.0 mol H₂O. Then 4.0 mol × 18.0 g/mol = 72 g H₂O.
4. Check against a known pattern: mass is conserved, so 8.0 g of hydrogen plus the 64 g of oxygen it uses should give 72 g of water. It does.

After a week of practice with this route, step 2 is a chunk in long-term memory, molar masses of common substances are automatic, and the same problem needs a fraction of the working memory it took on day one. That shift, not a bigger workspace, is what learning looks like. For more on practicing problems, see how to study math effectively.