You are viewing a free preview of this lesson.
Subscribe to unlock all 10 lessons in this course and every other course on LearningBro.
The Working Memory Model (WMM) was proposed by Baddeley and Hitch (1974) as a direct critique of, and replacement for, the Multi-Store Model's concept of a single, unitary short-term store. Rather than a passive store that simply holds information before transfer to long-term memory, working memory is conceived as an active processing system that temporarily holds and manipulates information during complex cognitive tasks such as reasoning, comprehension, mental arithmetic and reading. The model is multi-component: a controlling attentional system directs two (later three) specialised subsystems, each handling a different kind of material.
Key Definition: Working memory is the limited-capacity system that temporarily stores and processes the information needed for ongoing cognitive tasks. It is not a single store but a set of interacting components.
This lesson covers the AQA Paper 1 Memory content on the working memory model: the central executive, the phonological loop (with its phonological store and articulatory control process), the visuo-spatial sketchpad (with its visual cache and inner scribe) and the episodic buffer, together with the features (coding, capacity and the controlling role of the central executive) of each component. It also requires the research that supports the model — dual-task studies, the word-length effect (Baddeley, Thomson and Buchanan, 1975) and the case study of KF (Shallice and Warrington, 1972) — which feed the evaluation. The WMM is best understood in relation to the MSM (the previous lesson): it accepts the existence of a short-term system but rejects the MSM's claim that it is unitary. You should be able to describe the components and their functions (AO1), apply the model to dual-tasking examples (AO2), and evaluate it through experimental, clinical and brain-imaging evidence (AO3).
The diagram below shows the full model: the central executive at the top directing the three subsystems; the phonological loop with its two parts (the phonological store and the articulatory control process); the visuo-spatial sketchpad with its two parts (the visual cache and the inner scribe); the episodic buffer; and the links from the episodic buffer and the loop and sketchpad to long-term memory.
flowchart TD
CE["Central Executive<br/>(attentional control;<br/>modality-free; very limited capacity)"]
CE --> PL["Phonological Loop<br/>(temporary verbal/acoustic store)"]
CE --> VSS["Visuo-Spatial Sketchpad<br/>(temporary visual/spatial store)"]
CE --> EB["Episodic Buffer<br/>(Baddeley, 2000)<br/>integrates the subsystems"]
PL --> PS["Phonological Store<br/>('inner ear')<br/>holds speech-based input"]
PL --> ACP["Articulatory Control Process<br/>('inner voice')<br/>sub-vocal rehearsal"]
ACP -.->|refreshes / re-codes| PS
VSS --> VC["Visual Cache<br/>('inner eye')<br/>stores form & colour"]
VSS --> IS["Inner Scribe<br/>records spatial arrangement<br/>& movement"]
PS --> LTM["Long-Term Memory"]
VC --> LTM
EB --> LTM
LTM --> EB
The central executive is the model's controlling component — the "boss" of working memory.
| Feature | Detail |
|---|---|
| Function | Directs attention, allocates processing resources to the subsystems, monitors incoming data and switches between tasks |
| Capacity | Very limited — it can attend to only a small amount of information at any one moment |
| Coding | Modality-free — it does not deal with one kind of material; it can process information in any form |
| Storage | It does not store information itself; it is purely an attentional/control system |
Baddeley likened the central executive to the supervisory attentional system in Norman and Shallice's model of action control: it intervenes when automatic, routine processing is insufficient and a task requires deliberate control. It is the most important component but the least specified — a point returned to in the evaluation.
The phonological loop deals with auditory and verbal information (speech-based material) and preserves the order in which items arrive. Baddeley subdivided it into two parts.
| Sub-component | Nickname | Function |
|---|---|---|
| Phonological store | "inner ear" | A passive store that holds speech-based information for roughly 1–2 seconds; the trace decays unless refreshed |
| Articulatory control process | "inner voice" | Active sub-vocal rehearsal that keeps items alive in the store, and which converts written material into a phonological code so it can enter the store |
Evidence — the word-length effect. Baddeley, Thomson and Buchanan (1975) found that participants recalled more short words (e.g., sum, wit, harm) than long words (e.g., opportunity, university, aluminium) in immediate serial recall. The interpretation is that the articulatory control process rehearses in real time, so longer words occupy more of the loop's limited time and fewer can be maintained before the phonological store's trace decays — implying the loop holds about as much as can be said in roughly two seconds. Crucially, the effect disappears when participants perform articulatory suppression (repeating an irrelevant sound such as "the, the, the"), because this ties up the articulatory control process and prevents rehearsal of either short or long words — strong evidence that the rehearsal mechanism, not simply the number of items, drives the effect.
The visuo-spatial sketchpad stores visual and spatial information — what objects look like (their form and colour) and where they are in relation to one another. Logie (1995) proposed that it, too, divides into two parts.
| Sub-component | Nickname | Function |
|---|---|---|
| Visual cache | "inner eye" | Stores information about visual form and colour (a passive store) |
| Inner scribe | — | Records the spatial arrangement of objects and handles movement and the planning of routes (an active component) |
The VSS has limited capacity (Baddeley suggested it can hold about three or four objects), which is why it is difficult to perform two demanding visual/spatial tasks at once — for example, mentally rotating an object while simultaneously navigating an imagined route.
Baddeley added the episodic buffer in 2000 because the original 1974 model had no general store: it could not explain how information from the (verbal) phonological loop and the (visual) sketchpad was combined, nor how working memory drew on and fed back to long-term memory.
| Feature | Detail |
|---|---|
| Function | A temporary store that integrates information from the phonological loop, the visuo-spatial sketchpad and long-term memory into a single, coherent multi-modal "episode" |
| Capacity | Limited — approximately four chunks of information |
| Coding | Multi-modal (it can hold information in any combination of codes) |
| Control & links | Controlled by the central executive; provides the crucial bridge to long-term memory, allowing information to flow both ways |
Key Definition: The episodic buffer is a limited-capacity, multi-modal temporary store that binds information from the other working-memory components and from long-term memory into integrated episodes.
A useful way to consolidate the model is to compare the components on the dimensions the specification emphasises — what kind of information each handles (coding) and how much it can hold (capacity) — and to note that, crucially, the whole point of the model is that these capacities are separate.
| Component | Coding / type of material | Capacity | Active or passive? |
|---|---|---|---|
| Central executive | Modality-free (any type) | Very limited | Active (controls) — but stores nothing |
| Phonological store | Acoustic / speech-based | About 2 seconds' worth of speech | Passive (store) |
| Articulatory control process | Acoustic / speech-based | Limited by rehearsal time | Active (rehearses) |
| Visual cache | Visual (form, colour) | A few objects | Passive (store) |
| Inner scribe | Spatial / movement | Limited | Active (manipulates) |
| Episodic buffer | Multi-modal | About four chunks | Active (integrates) |
The separation of capacities is what generates the model's signature prediction. Because the phonological loop and the visuo-spatial sketchpad have independent limited capacities, a verbal task and a visual task can be performed concurrently with little mutual interference — they draw on different pools of resources — whereas two verbal tasks (or two visual tasks) must share one pool and so compete. This is precisely what the dual-task evidence below shows, and it is the single most important reason the WMM is preferred to the MSM's single, shared short-term store.
Consider what happens when you read a sentence and try to hold its meaning while reading on. The articulatory control process re-codes the printed words into a phonological form so they can enter the phonological store; the central executive directs attention to the relevant words and suppresses distractions; if the sentence describes a scene (e.g. "the red car turned left at the church"), the visuo-spatial sketchpad may build a rough mental image, with the visual cache holding the car's appearance and the inner scribe tracking its movement and position; and the episodic buffer binds the verbal and visual elements together and links them to relevant knowledge in long-term memory (what a church looks like, what "turned left" means). The model thus depicts comprehension not as passive storage but as the coordinated, simultaneous activity of several specialised subsystems — exactly the kind of complex cognition the MSM's unitary STM cannot capture.
A central strength of the WMM is the support from dual-task studies. When participants perform two tasks that draw on the same component (e.g., tracking a moving light and simultaneously imagining the angles of a block capital letter — both visuo-spatial), performance on both deteriorates; but when the two tasks use different components (e.g., the tracking task plus a verbal task such as counting aloud), performance is barely affected and both can be done well. This matters because the pattern is exactly what the WMM predicts — interference only when two tasks compete for one limited subsystem — and is precisely what the MSM's unitary STM cannot explain, since a single store should produce interference for any two concurrent tasks. The implication is that the experimental evidence does not merely fit the WMM but actively discriminates it from its predecessor, which is the strongest kind of support a model can have.
The model is further supported by the case study of KF (Shallice and Warrington, 1972), a patient with brain damage following a motorcycle accident. KF's phonological loop was severely impaired — his immediate recall of verbal/auditory material, such as digit span, was reduced to only one or two items, and he was especially poor at recalling material presented by ear — yet his visuo-spatial short-term memory and his ability to process visual information were largely intact. This dissociation matters because it is very difficult to explain if STM is a single store: damage to one store should impair all short-term memory, not selectively the verbal kind. The implication is that verbal and visual short-term memory rely on separable systems, exactly as the WMM claims, and that the deficit is in a specific subsystem rather than in STM as a whole. The same logic underlies findings that some patients show the opposite pattern (impaired visual but intact verbal short-term memory), which together approach a double dissociation — the gold standard for arguing two systems are functionally independent.
A third strength is biological. PET and fMRI studies show that different working-memory tasks reliably activate different brain regions: verbal/phonological tasks engage left-hemisphere language areas (around Broca's area and the supramarginal gyrus), visuo-spatial tasks engage right-hemisphere parietal and occipital regions, and tasks demanding attentional control engage the prefrontal cortex. This matters because it provides a biological line of evidence that converges with the behavioural (dual-task) and clinical (KF) evidence — three independent methodologies pointing to the same conclusion of separable components. The implication is that the WMM's components are not merely convenient theoretical fictions but appear to correspond to physically distinct neural systems, which considerably strengthens the model's validity.
The most serious limitation is that the central executive is vague and probably oversimplified. It is defined as an attentional control system, but its precise mechanisms are not specified, and it is sometimes criticised as a "homunculus" — a little person in the head that conveniently does whatever the data require without itself being explained. Evidence suggests the executive is not unitary: Eslinger and Damasio's patient EVR performed well on reasoning and planning tests (apparently intact executive function) yet showed grossly impaired decision-making in everyday life, implying the executive comprises separable sub-functions rather than a single resource. This matters because the central executive is the model's most important component — it controls everything else — so vagueness at its core weakens the whole account. The implication is that, although the WMM is far more detailed than the MSM about storage, it remains underdeveloped about control, and a complete model would need to fractionate the executive into specified subprocesses.
Subscribe to continue reading
Get full access to this lesson and all 10 lessons in this course.