What Is Working Memory? Definition, Capacity, and Its Link to IQ

What is working memory?

Working memory is the brain system that provides temporary storage and manipulation of the information needed for complex cognitive tasks such as reasoning, learning, and language comprehension. It is a workspace, not a filing cabinet: information is held there only while you are actively using it, typically for a matter of seconds.

Two properties define it. It is remarkably flexible — you can hold almost anything in it, including something you have encountered only once. And it is severely limited: only a handful of items can be maintained at any moment before earlier ones start to fade or get displaced.

What is the difference between working memory and short-term memory?

The two terms are often used interchangeably, but the distinction is precise and it matters. Short-term memory is storage. Working memory is storage plus manipulation.

That reversal in the example is not a trivial difference. Saying digits back in order taps storage. Saying them back in reverse requires you to hold the sequence and operate on it — and it is that manipulation component that predicts reasoning ability.

How many things can working memory hold?

Around four — not seven. The figure most people have heard comes from George Miller's 1956 paper “The Magical Number Seven, Plus or Minus Two.” It is one of the most cited papers in psychology, and one of the most misread. Miller himself presented the number as a rough estimate and a rhetorical device, not a measured capacity limit.

Nelson Cowan revisited the question in 2001 in a paper deliberately titled “The Magical Number 4 in Short-Term Memory.” When you control for the tricks people use to inflate their apparent capacity — silent rehearsal, and chunking separate items into larger units — the limit of the focus of attention settles at roughly four chunks, plus or minus one. Neuroscientific estimates of how many distinct objects can be maintained at once converge on a similar three-to-four range.

A chunk, importantly, is not an item — it is a meaningful unit. An expert chess player can hold an entire board position in a few chunks because the patterns are familiar; a novice sees twenty unrelated pieces. This is why capacity looks larger in domains you know well, without the underlying limit having changed at all.

The Baddeley–Hitch model of working memory

In 1974, Alan Baddeley and Graham Hitch proposed that working memory is not one thing but a small system of cooperating components. Their model remains the standard framework in cognitive psychology.

The model is not without critics. The central executive has been called a homunculus — a component defined largely by what it is supposed to accomplish rather than by how it works — and the episodic buffer's precise function remains debated. But the basic insight, that verbal and visual material are held by partly separate systems under a common attentional controller, has held up well.

Does working memory affect intelligence?

Yes, and the relationship is one of the strongest in all of differential psychology. Working memory capacity and fluid intelligence — the ability to reason about novel problems, independent of learned knowledge — correlate at a median of about r = .72. That means the two constructs share roughly half their variance.

That figure comes from Kane, Hambrick and Conway's (2005) reanalysis of fourteen datasets drawn from ten published studies, covering more than 3,100 young adults. Individual estimates in the literature run as high as r = .85.

Why should holding things in mind predict reasoning at all? Shipstead, Harrison and Engle (2016) argue that two complementary abilities are doing the work: maintenance — keeping the information you need available despite distraction — and disengagement — discarding information and hypotheses that have stopped being useful. Reasoning tasks demand both. A person who cannot let go of a failed approach is as impaired as one who cannot hold the problem in mind.

Why this matters for your IQ score

This is not an abstract connection. If you have ever taken a professionally administered IQ test, your working memory was measured directly, and it changed your score.

Which IQ tests measure working memory?

The Wechsler Adult Intelligence Scale — the most widely used individually administered IQ test — measures working memory as a full index that feeds directly into the Full Scale IQ. On the WAIS-IV (2008), the Working Memory Index was built from Digit Span and Arithmetic, with Letter–Number Sequencing as a supplemental subtest. On the current WAIS-5 (2024), the core working memory subtests are Digit Span Sequencing and Running Digits. The children's version, the WISC, is structured similarly. In research settings, working memory is more often measured with complex span tasks or the n-back.

Can low working memory lower your IQ score?

Yes. Because the Working Memory Index feeds the Full Scale IQ, a low working memory score pulls the composite down even when abstract reasoning and vocabulary are strong. This is a common and clinically important pattern: it appears frequently in ADHD and in specific learning difficulties, where reasoning ability may be entirely intact while the mental workspace is not. The effect is well enough recognised that, when assessing for ADHD on the WAIS-IV, substituting Letter–Number Sequencing for the Arithmetic subtest has been advised — Arithmetic depends on mathematical computation, which co-occurring maths difficulties can depress for reasons that have nothing to do with working memory.

Why do psychologists measure it separately?

Because averaging hides the thing you most need to know. Consider someone with a Verbal Comprehension Index of 120 and a Processing Speed Index of 75. Their Full Scale IQ lands somewhere in the 90s — a number that describes nobody. It understates their reasoning and overstates their fluency, and it would be actively misleading as a summary of that person.

The Wechsler manuals address this directly. When the Working Memory or Processing Speed index is depressed by a secondary condition, clinicians are advised to report the General Ability Index — a composite built from the verbal and reasoning indices only — in place of the Full Scale IQ. Reporting working memory separately is what makes that judgement possible, and it is a large part of why a clinical assessment tells you more than a single number ever can.

Can you improve your working memory with brain training?

You can reliably get better at the training task. There is no convincing evidence that this makes you more intelligent.

The distinction psychologists use is between near transfer — improvement on tasks resembling the one you practised — and far transfer — improvement on genuinely different abilities such as reasoning, reading comprehension, or arithmetic. Near transfer is real and repeatable. Far transfer is where the claims collapse.

An early study by Jaeggi and colleagues (2008) reported that n-back training improved fluid intelligence, and it launched an industry. Redick and colleagues (2013) failed to replicate it. The decisive evidence arrived with Melby-Lervåg, Redick and Hulme's (2016) meta-analysis of 87 publications and 145 experimental comparisons: reliable gains on working memory tasks themselves, and no convincing evidence of improvement on measures of intelligence or any other far-transfer outcome once training groups were compared against actively treated controls.

What does help is less marketable: sleep, managing stress and anxiety (both of which consume the very attentional resources working memory depends on), and reducing unnecessary load — writing things down, breaking problems into steps, and building genuine expertise so that information chunks more efficiently. None of these raise your capacity. They stop you wasting the capacity you have.

What are the signs of poor working memory?

Everyday working memory difficulties tend to look like inattention rather than forgetfulness. Commonly described experiences include:

• Losing the thread of a sentence, instruction, or question halfway through.
• Walking into a room and forgetting the purpose that sent you there.
• Struggling with multi-step instructions unless they are written down.
• Losing your place in mental arithmetic, or in a task interrupted midway.
• Rereading the same paragraph because the earlier part has already faded.

Working memory: myths and facts