Working memory is the small mental workspace that holds a few pieces of information for a few seconds while you use them. It is what lets you do a sum in your head, follow a set of directions or keep track of an argument. Its capacity is smaller than most people think — about four chunks — and it is one of the abilities most closely tied to reasoning and IQ.
What working memory is
Psychologists use the term for memory that is being used rather than stored. Hearing a phone number and dialling it, multiplying 17 by 6 without paper, or reading a long sentence and still knowing who “she” refers to at the end: each needs a few items held active while something else happens. Without rehearsal, that information fades within seconds, or is pushed out by the next thing you attend to.
Working memory is also one of the three core executive functions — alongside inhibition and cognitive flexibility — and the one of the three most closely linked to intelligence.
Working memory vs short-term memory
The two terms are often used as if they meant the same thing. The distinction researchers draw is between holding and holding while working:
| Short-term memory | Working memory | |
|---|---|---|
| What it does | Keeps information briefly | Keeps information while you change or use it |
| Typical task | Repeat digits in the order you heard them | Repeat them backwards; read sentences while remembering each last word |
| Link to reasoning | Weaker | Stronger |
The second typical task is the “reading span” test, introduced by Meredyth Daneman and Patricia Carpenter in 1980: you read a series of sentences aloud and then recall the last word of each. It forces storage and processing to compete, which is exactly what working memory has to manage.
The Baddeley model
The most widely taught model comes from Alan Baddeley and Graham Hitch, who proposed in 1974 that working memory is not a single store but a system with separate parts. In its current form it has four:
- The central executive — directs attention and coordinates the other parts. It stores nothing itself.
- The phonological loop — holds sounds and words, kept alive by silent repetition (the voice in your head repeating a number).
- The visuospatial sketchpad — holds images, positions and routes.
- The episodic buffer — added by Baddeley in 2000 to explain how information from the other parts and from long-term memory is bound into single episodes.
A 2004 study of children aged 4 to 15 found that this three-part structure (loop, sketchpad, executive) is already in place from about age 6, and that each part grows substantially in capacity through the school years into adolescence.
How much can it hold? Seven, or four?
George Miller's 1956 paper on “the magical number seven, plus or minus two” is one of the most cited in psychology, and the source of the popular idea that we can hold about seven items. But people can hold more when they group items into chunks — FBI-CIA-USA is three chunks, not nine letters — and fewer when they cannot.
When Nelson Cowan reviewed the evidence in 2001, he described Miller's seven as more a rough estimate and a rhetorical device than a real limit. Under conditions that prevent rehearsal and chunking, capacity averages about four chunks, usually within a range of three to five. That smaller number is why a list of instructions falls apart after the third or fourth step. The same limit shapes memory techniques such as the memory palace, which work by packing more into each chunk.
Working memory and IQ
The relationship is strong enough that some researchers once argued the two were the same thing. Two 2005 papers in the same journal issue set out where the evidence stands:
- A meta-analysis of 86 samples put the correlation between working memory and general intelligence at about 0.48 — substantial, but well short of identity.
- A reanalysis of 14 latent-variable data sets with more than 3,100 young adults put the correlation between working memory capacity and fluid reasoning at a median of 0.72 — about half their variance shared.
Both sides agree the constructs are related and not identical; the size of the link depends on how carefully each is measured and whether the comparison is with reasoning specifically or with intelligence as a whole. The reasoning side of that comparison is explained on our page on fluid vs crystallised intelligence.
That is also why full-scale clinical tests measure working memory directly. The Wechsler adult scales report a Working Memory index alongside verbal comprehension, reasoning and processing speed — see how IQ tests work. A matrix-reasoning test does not measure working memory separately, but it draws on it: solving a hard matrix means holding several rules in mind at once.
How working memory is tested
- Backward span — hear a sequence of digits and repeat it in reverse order.
- Complex span — such as reading span: process something (read a sentence, check a sum) while remembering an item after each one.
- N-back — watch a stream of letters or positions and respond whenever the current one matches the one shown n steps earlier.
Online “working memory tests” usually use one of these formats. A single short task gives a rough indication of how you did on that task that day; it is not a clinical assessment.
Can you train working memory?
A 2008 study reported that training on a demanding n-back task raised scores on a fluid reasoning test, and that more training gave more gain. It launched a wave of brain-training claims. Later, larger evidence did not support the headline: a 2016 meta-analysis of 87 publications and 145 comparisons found that training reliably improves performance on working memory tasks in the short term, but found no convincing evidence of transfer to reasoning, verbal ability, reading or arithmetic when compared with an active control group. The broader question is covered on our page on whether you can increase your IQ.
What does help is working with the limit rather than against it: write steps down instead of holding them, give or ask for instructions one or two at a time, group information into meaningful chunks, and reduce distractions that compete for the same few slots. Our page on attention span covers the attention side of the same problem.
Working memory and ADHD
Working memory difficulties are one of the most studied features of ADHD. A 2012 meta-analysis found that children with ADHD, as a group, show large differences from typically developing peers on both verbal and visuospatial working memory tasks, with the size of the difference depending partly on how demanding the task was for the central executive. This describes group averages: it does not mean everyone with ADHD has weak working memory, and a low score on a working memory task is not a sign of ADHD on its own. Assessment is a job for a qualified professional.
Trust and scope notes
This page explains published research in general terms. It is not a diagnostic tool and does not assess ADHD or any other condition. Our IQ test measures reasoning; it does not report a separate working memory score.
IQ Revealed is not affiliated with, endorsed by or connected to the researchers, universities, journals or test publishers named on this page, including the publisher of the Wechsler scales. Their names are used only to identify the work being described.