On this page
Some problems feel impossible not because they are deep but because there is too much to hold in your head at once: five numbers, two diagrams and a method all competing for space. Psychologists call the system you are straining working memory, and the most widely used model of it comes from Alan Baddeley and Graham Hitch. This post explains how they built the model, what each part does, and how to study with its limits in mind.
What question started it?
In 1974 Baddeley and Hitch published a chapter called simply "Working memory" in *The Psychology of Learning and Motivation*. Fifty years later, Hitch, Allen and Baddeley looked back on it in The multicomponent model of working memory fifty years on (2025).
They explain that their starting question was "What is STM for?" Several theories assumed that short-term memory (STM) was a general-purpose workspace for thinking. But some evidence did not fit. One neuropsychological patient had a severely reduced memory span yet showed no obvious general problem with thinking.
How did they test it?
Their method was to give healthy adults a memory load, such as a string of digits to keep in mind, while they did something else: verbal reasoning, understanding prose, or learning a list. If STM were the whole workspace, filling it should wreck the second task.
It did not. In the 2025 review they report that there was "moderate but far from catastrophic interference" when STM was filled to capacity, and little or no interference when the load was well below span. So STM had to be only one part of a larger system.
What are the parts of the model?
The authors describe their model as a "deliberately simplistic" interpretation, and Baddeley's own 2000 paper calls the 1974 version a three-component model:
Central executive. A limited-capacity system for attention and control, the part that decides what to do next.
Phonological loop. Originally called the articulatory loop: a store for speech-based information that fades quickly unless refreshed by silent rehearsal, the voice in your head repeating a number.
Visuospatial sketchpad. A store for images and spatial layout.
Evidence for the loop came from the word-length effect. In Baddeley, Thomson and Buchanan (1975), people recalled lists of short words better than lists of words that take longer to say, and, as the 2025 review explains, the effect disappeared when people were made to repeat an irrelevant word aloud, which blocks rehearsal. For the sketchpad, the team showed that keeping a stylus on a moving spot of light disrupted a visual imagery task but not a verbal one.
In that 2000 paper Baddeley proposed a fourth part, the episodic buffer: a limited-capacity store that binds information from the other parts and from long-term memory into single episodes. The 2025 review links it to the advantage people show for structured sentences over plain word lists.
Why does working memory matter for school?
Working memory is linked to how children do in maths and science. Gathercole and colleagues (2004) compared working memory test scores with national curriculum assessments in England. At 7, attainment in English and mathematics was associated with working memory scores. At 14, strong links remained with mathematics and science, but not with the higher-level English assessments. The authors suggest that the operations needed in mathematics and science "are constrained by the general capacity of working memory".
The phonological loop also matters for vocabulary. Baddeley, Gathercole and Papagno (1998) reviewed evidence that it plays a crucial role in learning the sound patterns of new words, useful to know if you are studying science in a second language.
How can you study with working memory in mind?
Offload onto paper. Write the given data, a labelled diagram and the target quantity before starting algebra. Every item on paper is one less in your head.
Use both stores. Pair a short verbal statement of the principle with a sketch. Words and pictures use different stores.
Avoid competing verbal input. Songs with lyrics compete with the phonological loop while you read or rehearse; see our post on music while studying.
Turn steps into chunks. Practice compresses procedures into single units, freeing space for the new part of a problem. Our post on George Miller and chunking shows how.
Make it meaningful. The episodic buffer works well with structured, meaningful material. Explain a derivation in sentences rather than memorising symbols in order.
What can a student take from Baddeley's way of working?
Baddeley and Hitch did not start with a grand theory. They asked a simple question, designed an experiment to test the popular answer, and built the smallest model that fitted the result, then kept revising it for fifty years. That is a good pattern for your own study too: test a method, see what fails, and adjust.
Some physics ideas, such as field lines or wave interference, take up a lot of working memory if you only read about them. The Fermi Simulation Lab has interactive simulations for ideas that are easier to see than to read, so the picture does some of the holding for you.
Next time a numerical feels too big, stop and write every given value and the target on paper before you try a single equation.
