sciencebriefs
10:00in productionCh. 1 · A model built to replace/ 10:00 · ceiling 15 min
Neuroscience · Life sciences

Baddeley's model of working memory

Alan Baddeley and Graham Hitch · 1974

Working memory isn’t one thing — it’s three things arguing over your attention.

Baddeley and Hitch’s 1974 model reframed working memory as a set of interacting, domain-specific components — not a single bottleneck. Dual-task interference, the word length effect, and amnesic patient profiles provide direct empirical support. Later additions (episodic buffer in 2000, renamed awareness buffer in 2026) extend but do not overturn the core claim of fractionation. It remains indispensable — not because it is complete, but because it is falsifiable, componential, and repeatedly predictive.

NeuroscienceLife sciencesMemoryPsychologyNeuroscienceMemory across the network →
Chapters & takeaways5
  1. 1:25
    A model built to replace

    It replaced Atkinson and Shiffrin’s unitary short-term store with a multi-component architecture.

  2. 2:34
    Evidence from interference

    Dual-task experiments showed verbal and visuospatial processing run in parallel — not in series.

  3. 4:12
    How words fit in memory

    The phonological loop’s capacity depends on speech rate — not number of items.

  4. 5:22
    Memory systems are separable

    Amnesic patients proved short-term and long-term memory can break apart.

  5. 6:26
    An update, not an overhaul

    The episodic buffer was added in 2000 — then renamed the awareness buffer in 2026.

Worth your time?

Yes. Study the whole thing.

4.5/ 5
What works
  • dual-task dissociation
  • word length effect
  • amnesic patient dissociation
What does not
  • proves anatomical localisation
  • quantifies central executive bandwidth
  • explains developmental trajectory
  • accounts for individual variation in dual-task cost
Study it if
  • cognitive psychologists
  • clinical neuropsychologists
  • education researchers
Skip it if
  • computational neuroscientists seeking biophysical mechanisms
  • AI architects building memory modules without behavioural constraints
The written brief1 min read

What the work claims

Working memory is not a unitary short-term store. It comprises at least three functionally distinct, interacting components: a phonological loop for verbal information, a visuospatial sketch pad for visual and spatial information, and a central executive that controls attention and coordinates them.

How it was done

Baddeley and Hitch used dual-task paradigms in 1974 to test interference across perceptual domains. They measured performance when participants performed two tasks simultaneously — one verbal, one visuospatial — and compared it to single-task baselines. They also conducted a 1975 word length effect study using recall tests with short and long words, and tested articulatory suppression by having participants repeat an irrelevant sound.

What holds up

Dual-task findings hold: cross-domain task pairs (e.g., verbal + visuospatial) show near-additive performance; within-domain pairs impair performance. The word length effect holds: short words are recalled better than long ones, and articulatory suppression abolishes this effect. Amnesic patient data hold: intact short-term memory coexists with impaired long-term memory.

What does not

The model does not prove the anatomical reality of its components. It does not quantify capacity limits beyond the word length effect’s dependence on pronunciation time. It does not resolve how the central executive operates mechanistically — only that it coordinates subsystems.

Why it matters beyond the lab

It shifted applied fields — from education (designing dual-modality instruction) to clinical neuropsychology (interpreting amnesia, ADHD, stroke deficits) — by providing a testable, component-based framework for memory failure, not just a label for ‘poor recall’.

Is it worth your time

Yes. It redefined how psychologists model memory: not as a passive store but as an active, multi-component system. Its empirical scaffolding — dual-task dissociation, word length effect, amnesic patient profiles — remains foundational for experimental design in cognitive psychology.

Same field · Neuroscience4 of 45
13:00
Henry MolaisonIn 1953 surgeon William Scoville removed most of both hippocampi from 27-year-old Henry Molaison to control severe epilepsy, leaving him unable to form new long-term memories while his memory of his own distant past remained largely intact. Working with him from 1957 onward, Brenda Milner showed through tasks like mirror-tracing that Molaison could still learn new skills and improve with practice despite having no conscious memory of ever practising, establishing a lasting distinction between declarative and procedural memory. A 2014 digital reconstruction of his brain after death found roughly half his hippocampal tissue had survived and revealed an undetected prefrontal lesion, complicating but not overturning decades of findings built on his case.
13:00
Mirror neuronRecording single neurons in the premotor cortex of macaque monkeys, Giacomo Rizzolatti's group at the University of Parma, working with Giuseppe Di Pellegrino, Luciano Fadiga, Leonardo Fogassi and Vittorio Gallese, found cells that activated both when a monkey performed a grasping action and when it watched another individual perform the same action. Rather than staying a specific finding about monkey motor cortex, mirror neurons became the proposed basis for human imitation, empathy, language and even autism. Subsequent scrutiny, including failed replications of key predictions and pointed methodological criticism, has left much of that broader theory in doubt, even as the original cellular finding in monkeys is not seriously disputed.
13:00
REM reboundWilliam Dement's experiments on selectively waking sleepers the moment REM began showed that the brain compensates specifically for lost REM sleep, not for lost sleep in general — a narrow but well-supported finding about how sleep is regulated.
13:00
SynesthesiaSynesthesia, in which stimulating one sense involuntarily triggers an experience in another, was first medically documented by Georg Tobias Ludwig Sachs in 1812 and empirically surveyed as early as 1876 by Gustav Fechner, but prevalence estimates stayed wildly inconsistent, from one in four people to one in a hundred thousand, because most studies recruited self-identified synesthetes rather than a general population. A carefully designed study of 500 university students in Edinburgh and Glasgow found a prevalence of 4.4 per cent across nine documented types, and separate work found the specific male-to-female ratio long cited for the condition was inflated by the same self-selection problem. Synesthetes can be distinguished from people simply describing a metaphor by their consistency on the same associations tested years apart, but a definitive neurological explanation for why the crossing happens remains unsettled.
Up next in Science

X-ray crystallography

William Henry Bragg and William Lawrence Bragg · 11:22

They didn’t see atoms — they measured how atoms *must* be spaced, using nothing but angles, film spots, and arithmetic.

Play nextBack to Neuroscience
11:22