Lesson 4 of 11 · 18 min
The working memory model
Explain Baddeley and Hitch's working memory model, its central executive, phonological loop, visuospatial sketchpad and episodic buffer, and how it has developed from 1974 to the present.
What you will be able to do
- Describe the four components of the working memory model and what each does.
- Use the model to predict whether two tasks done at the same time will interfere.
- Describe the evidence for the model, including dual-task studies and the word-length effect.
- Explain how the model developed from Baddeley and Hitch (1974) to the present, and why it has greater explanatory power than the multi-store model's short-term store.
You can drive along a familiar road while holding a conversation, but you stop talking when you need to merge into fast traffic. You can listen to music while you sketch, but not while you read. If short-term memory were one small store, any two tasks should compete for it equally. They don't, and the pattern of which tasks clash is the evidence for the working memory model.
Baddeley and Hitch (1974) argued that short-term memory is not a passive store where items wait to be rehearsed, but an active workspace we use to hold and manipulate information while we reason, read, calculate and plan. They called it working memory, and divided it into components.
The working memory model as Baddeley revised it in 2000. The 1974 model had only the central executive and its two slave systems.
| Component | What it does | Example |
|---|---|---|
| Central executive | Directs attention to the task, switches between tasks and coordinates the other components. Limited capacity; stores nothing itself. | Deciding what to do next while cooking a meal from a recipe |
| Phonological loop | Holds speech and sound. The phonological store (the "inner ear") holds sounds for about 2 seconds; the articulatory control process (the "inner voice") rehearses them by silent repetition. | Repeating a phone number to yourself |
| Visuospatial sketchpad | Holds and manipulates visual information (what things look like) and spatial information (where they are). | Picturing the route from your house to the station |
| Episodic buffer | Added by Baddeley (2000). A limited-capacity store that combines information from the loop, the sketchpad and long-term memory into single, multi-sensory episodes. | Remembering a scene from a film: what was said, what it looked like, and what it meant |
Key idea
The prediction that matters
Two tasks that need the same component interfere: performance on one or both falls. Two tasks that need different components can be done together with little loss. The central executive is the exception: almost every demanding task draws on it, so two tasks that both need a lot of attention clash whatever else they use.
The performance figures are illustrative: they show the direction of the effect, not measured values.
A student tries to remember a list of spoken words while doing each of these at the same time. Which task will reduce recall of the words the most?
The evidence.
- Dual-task studies. Baddeley and Hitch (1974) had participants hold a string of digits while doing a verbal reasoning task. If STM were a single store, a full load of digits should have stopped the reasoning. It did not: reasoning slowed as the digit load increased, but participants kept reasoning with few errors. STM must have more than one part.
- The word-length effect. People recall more short words (sum, wit, hate) than long words (university, opportunity, aluminium) from a list, even though both are single words. The phonological loop holds about as much as you can say in about 2 seconds, and long words take longer to say. When participants repeat an irrelevant word aloud (articulatory suppression), the effect disappears, because they can no longer rehearse at all.
- Case studies. KF, whose STM for heard words was very poor while his visual STM was better, fits a damaged phonological loop and an intact sketchpad. The multi-store model, with one STM, cannot explain this.
The word-length effect disappears under articulatory suppression. Which component of working memory does this finding provide evidence for?
Development of the model, from 1974 to the present.
- 1974. Baddeley and Hitch proposed a central executive with two "slave systems", the articulatory (later phonological) loop and the visuospatial sketchpad.
- 1980s. The loop was split into the phonological store and the articulatory control process, which explained the word-length effect and why written words are recoded into sound to be rehearsed.
- 2000. The model could not explain how people combine information from different sources with long-term memory, why people with dense amnesia can still recall a paragraph of prose immediately (far more than the loop could hold), or how chunks that depend on LTM form. Baddeley added the episodic buffer to do that job.
- Since then. Research has refined the buffer (Baddeley and colleagues now see it as a store the central executive and other components feed, rather than one it controls), and brain imaging has linked the components to different areas: speech-based storage and rehearsal to areas of the left hemisphere, visuospatial storage more to the right, and the central executive to the frontal lobes (Baddeley, 2003).
Switch between the 1974 and 2000 versions. Which task has nowhere to go in the 1974 model?
Worked example
Explain why working memory explains more
In a dual-task experiment, participants learnt a list of words while doing one of three secondary tasks: repeating "the, the, the" aloud, tracing a pattern with a finger, or nothing. Recall was much lower with repeating "the" than with tracing, and tracing lowered recall only slightly compared with doing nothing.
Explain how the working memory model accounts for these results, and why the multi-store model cannot.
Step 1: Identify the component each task needs
Learning words loads the phonological loop. Saying "the, the, the" also loads the loop's articulatory control process. Tracing a pattern loads the visuospatial sketchpad.
Step 2: Explain the large drop
Repeating "the" uses the same component as rehearsing the words, so the words cannot be rehearsed and recall falls a lot.
Step 3: Explain the small drop
Tracing uses a different component, so the words can still be rehearsed. The small drop comes from the central executive, which coordinates both tasks and has limited capacity.
Step 4: Why the multi-store model cannot explain it
The multi-store model has a single short-term store. Any second task should take up the same limited capacity, so it predicts similar interference from both secondary tasks. It cannot explain why one task interferes much more than the other.
Try it yourself first, then check the working one step at a time.
Answer
Repeating "the" and learning words both need the phonological loop, so they interfere strongly; tracing needs the visuospatial sketchpad, so it interferes little (only through the shared central executive). A single STM predicts equal interference from both, so the multi-store model cannot explain the difference.
Method
Assign each task to a component.
Same component: strong interference. Different components: little (central executive only).
Show what a single store would predict, and that it does not fit.
Watch out for
Saying tracing has no effect at all: both tasks still draw on the central executive.
Note
Limitations
- The central executive is the least understood part. It is described by what it does (directing attention), and critics say it explains little until its own components are identified.
- The capacity of each component is hard to measure separately.
- The episodic buffer was added to explain findings after they appeared, which makes the model flexible but harder to falsify.
- The model is about short-term processing; it says little about how information is stored in long-term memory.
Exam tip
Explanatory power
When a question asks why the working memory model has greater explanatory power than the multi-store model, name a finding and show that one model explains it and the other cannot: "Dual-task studies show two tasks can be done together if they use different components. The multi-store model's single STM predicts they should always interfere; working memory explains it with separate components."
Summary
Working memory (Baddeley & Hitch, 1974) is an active workspace with separate components, not a single passive store.
Central executive: directs attention and coordinates; stores nothing. Phonological loop: speech and sound (store plus articulatory rehearsal). Visuospatial sketchpad: images and locations.
The episodic buffer (Baddeley, 2000) combines information from the other components and LTM into episodes.
Tasks that use the same component interfere; tasks that use different components can be done together.
Evidence: dual-task studies, the word-length effect and articulatory suppression, and KF.
It explains more than the multi-store model's single STM, but its central executive is vague.
Finished the lesson?
Try the 2 quick checks above and the lesson completes itself, or mark it complete now.