Read this topic

Behavioural Science & Sociocultural Psychiatry · Paper A

Working memory

The idea

Working memory is not a box for storage — it is a small mental workbench with four separate tool trays, and each tray can break independently of the others. A person can hold a phone number perfectly (one tray intact) while being completely unable to picture a room's layout (a different tray broken) — and neither finding tells you anything about whether they can form a new long-term memory at all.

How it works

Working memory, in Baddeley and Hitch's model, is an active system of four components. The CENTRAL EXECUTIVE allocates attention, switches strategy, and links to long-term memory. The PHONOLOGICAL LOOP holds verbal/acoustic information via a store plus rehearsal. The VISUOSPATIAL SKETCHPAD holds visual/spatial information, and the EPISODIC BUFFER binds information from the other components and long-term memory into one ordered representation.
This replaced the older, unitary 'short-term memory' concept (Atkinson-Shiffrin), which treated brief storage as one passive box rather than an active system with specialised, separable parts.
The phonological loop and visuospatial sketchpad dissociate on dual-task testing. Articulatory suppression (repeating an irrelevant sound aloud) disrupts verbal recall but not spatial recall. A competing spatial-tracking task disrupts spatial recall but not verbal recall instead — this is how the two-component split was originally demonstrated experimentally.
The central executive is closely related to, but not the same as, executive function proper. The central executive allocates resources within working memory itself, while executive function is the broader system of planning, switching and inhibition that draws on working memory as one of its resources.
Working memory is tested clinically by digit span (forward and backward) and by dual-task paradigms; a normal digit span reflects an intact phonological loop but says nothing about the visuospatial sketchpad, the episodic buffer, or any long-term memory system.
In practice, this is why a patient can score normally on digit span testing yet still have severe, disabling amnesia — working memory and long-term episodic memory are entirely separate systems, and clinicians must test both rather than inferring one from the other.

How to spot it

A vignette describing verbal recall disrupted by repeating an irrelevant sound aloud (articulatory suppression), with spatial recall unaffected, is describing the phonological loop specifically.
A vignette describing normal digit span but an inability to hold a spatial layout or mentally rotate a shape is describing a selective visuospatial sketchpad deficit.
A vignette describing difficulty switching attention between two working-memory subtasks, with broader real-world planning and social judgement intact, is describing the central executive, not executive function proper.
A vignette describing binding of verbal, visual and background-knowledge content into one ordered mental representation held briefly is describing the episodic buffer, not episodic (long-term) memory.
A vignette describing a normal digit span alongside a severe inability to form new long-term memories is demonstrating the working-memory/episodic-memory dissociation directly.
A vignette describing preserved skills (riding a bicycle) despite dense amnesia is describing procedural memory, a long-term system unrelated to working-memory capacity.

Tell it apart

Working memory vs short-term memory — working memory is an active, four-component system that manipulates information; short-term memory (the older term) describes a single passive store without these separable parts.
Central executive vs executive function — the central executive allocates attention within working memory itself; executive function is the broader control system (planning, switching, inhibition) that uses working memory as one resource among several.
Phonological loop vs visuospatial sketchpad — verbal/acoustic content disrupted by articulatory suppression (phonological loop) vs visual/spatial content disrupted by a competing visual-tracking task (visuospatial sketchpad); they dissociate cleanly in both testing and lesion studies.
Episodic buffer vs episodic memory — the episodic buffer is a SHORT-TERM working-memory component that binds information briefly; episodic memory is a LONG-TERM, hippocampally-dependent system for recollecting past events — the shared word 'episodic' does not mean shared mechanism.
Working memory vs episodic memory — a normal digit span (working memory intact) does not rule out severe amnesia (episodic memory impaired); the two dissociate completely, most dramatically in classic amnesic syndromes.
Working memory vs semantic memory — semantic memory (long-term facts and concepts) and procedural memory (long-term skills) are further long-term systems, both independent of working memory capacity.
Working memory vs sustained attention — poor backward digit span can reflect a genuine central-executive/working-memory deficit OR a primary attention deficit; the two must be teased apart rather than assumed.

Watch for

⚠️ Using 'working memory' and 'short-term memory' interchangeably as if they name the same model — they describe different theoretical frameworks (active multi-component system vs passive single store).
Assuming a normal digit span means 'memory is fine' overall — it only speaks to the phonological loop, not to episodic, semantic or procedural long-term memory.
Conflating the central executive with executive function proper because they share the word 'executive' — one is a working-memory subsystem, the other is the broader control system.
Confusing the episodic buffer (a working-memory component) with episodic memory (a long-term system) purely because of the shared word 'episodic'.
Attributing poor performance on a working-memory task automatically to a working-memory deficit without checking whether a primary attention problem is the real cause.
Assuming a single working-memory deficit is global — a patient can have an isolated phonological-loop or visuospatial-sketchpad deficit with the other components, and all long-term systems, completely intact.

Use it again

First ask whether the vignette is describing a SHORT-TERM, actively-manipulated holding task (working memory) or a LONG-TERM storage/recollection task (episodic, semantic or procedural memory) — this single distinction resolves most confusion.
If it is working memory, identify which of the four components is implicated by the content: verbal/acoustic (phonological loop), visual/spatial (visuospatial sketchpad), attention-switching/resource-allocation (central executive), or binding-into-one-representation (episodic buffer).
If digit span is reported as normal, do not extend that finding to any other memory system — check explicitly what is being asked about episodic, semantic or procedural memory before assuming they are also intact.
If poor performance on a working-memory task is described, check whether a separate attention or motivation problem, rather than a genuine central-executive deficit, better explains it.
If the word 'executive' appears, check whether the vignette is describing a working-memory-internal control function (central executive) or a broader real-world planning/social-judgement failure (executive function proper).

In detail

Hook

Ask someone to remember seven digits while they repeat 'the-the-the' out loud, and their recall falls apart. Ask them to remember the same seven digits while tracking a moving dot with their eyes instead, and recall is almost untouched. Same task, same digits — completely different outcome, because two separate systems were doing the holding.

Plain definition

Working memory is an active, limited- system for holding and manipulating information over seconds, made of four separable components (central executive, phonological loop, visuospatial sketchpad, episodic buffer) — it is distinct from the older, simpler 'short-term memory' idea of one passive storage box, and distinct from long-term memory systems (episodic, semantic, procedural), which store information for minutes to a lifetime rather than seconds.

Critical caveat

A normal digit span does NOT mean a patient's memory is generally fine. It only confirms an intact phonological loop. It says nothing about the visuospatial sketchpad, the episodic buffer, or — critically — any long-term memory system, all of which must be tested separately.

First principles

Central executive: allocates attention and strategy within working memory; related to, but not identical with, executive function proper.
Phonological loop: holds verbal/acoustic content; disrupted by articulatory suppression.
Visuospatial sketchpad: holds visual/spatial content; disrupted by a competing visual task.
Episodic buffer: binds content from the other components and long-term memory into one ordered representation — a short-term function despite its name.

Structured walk-through

Imagine you test a patient's digit span and it comes back entirely normal — it is tempting for you to conclude their memory is generally fine, but when you then ask what they had for breakfast, they cannot say, because that question tests a completely different, long-term system that one normal screening result never touched, and confirming it needs its own dedicated diagnostic assessment.

First decide whether the vignette describes a short-term, actively-manipulated holding task (working memory) or a long-term storage/recollection task (episodic, semantic or procedural memory). If it is working memory, identify the content type: verbal (phonological loop), spatial (visuospatial sketchpad), attention-allocation (central executive), or multi-modal binding (episodic buffer). If digit span is reported as normal, explicitly check what the question is asking about long-term memory before assuming it is also intact — the two systems dissociate completely and do not predict one another, and no amount of reassurance about one system should be read across to the other without its own dedicated test.

Look-alikes table

Presentation	Component/system	Key separator
Verbal recall worsens with articulatory suppression, spatial recall spared	Phonological loop	Disrupted by verbal, not visual, interference
Spatial recall worsens with visual distraction, verbal recall spared	Visuospatial sketchpad	Disrupted by visual, not verbal, interference
Difficulty switching between two working-memory subtasks, real-world planning intact	Central executive	Narrower than executive function proper
Normal digit span, dense amnesia for recent events	Episodic memory (impaired)	Working memory and episodic memory dissociate
Preserved skills despite amnesia	Procedural memory	Independent of working-memory

Diagnosis

Assessment uses digit span (forward and backward) for the phonological loop and central executive, spatial span tasks (e.g. Corsi block-tapping) for the visuospatial sketchpad, and separate formal testing (recall/recognition tasks with delay) for episodic memory — no single test covers all four working-memory components or distinguishes working memory from long-term memory on its own.

Management

Where a working-memory deficit is identified as part of a broader cognitive assessment (e.g. in , traumatic brain injury, or early neurodegenerative disease), management follows the underlying cause per relevant NICE guidance (for example, NG87 for attention-deficit hyperactivity disorder or NG97 for dementia), alongside compensatory strategies (external memory aids, reduced task-switching demands, chunking of verbal information) rather than a working-memory-specific pharmacological treatment.

Exam pearl

When a vignette reports a normal digit span, do not let that finding answer a question about a different memory system — check specifically what episodic, semantic or procedural memory testing shows before concluding anything beyond the phonological loop.

In one line

Working memory is four separable components, not one box, and a normal result on one component (usually digit span) never guarantees that any other component, or any long-term memory system, is also intact.