You can drive and hold a conversation. You cannot read a book and hold the same conversation. Both use short-term memory, so why does one combination work and the other collapse? The multi-store model has no answer. This model does.
📘 What you need to know
The working memory model (WMM) was devised by Baddeley & Hitch (1974) as a more dynamic and flexible model.
It covers short-term memory only — it explains the “here and now”, not the types or functions of LTM.
STM is treated as a complex information processor, not a static unitary store.
Four units: the central executive (CE), the phonological loop (PL), the visuospatial sketchpad (VSS) and the episodic buffer (EB), added in 2000.
Cognitive load theory (CLT) says working memory has limited capability, so learning suffers when it is overloaded.
Dual-task studies test this by having participants do two things at once.
Baddeley et al. (1973) is the key study.
The four components
The episodic buffer was a later addition. It exists because the model needed a way for the other components to talk to long-term memory, which the original 1974 version had no route for.
The central executive
The driving force of the model. It decides which tasks go to which slave system, rather like the managing director of a company. It is not really a memory store but a command centre, deciding what to pay attention to and what to ignore, then directing the slave systems accordingly. It is probably the most important element of the WMM — and by far the hardest to test.
The phonological loop
Handles acoustic information, spoken or written. Written words are encoded by vocalising them in your mind, so visual information becomes acoustic. It splits into two parts:
the phonological store, the “inner ear”, which holds speech-based information for a couple of seconds
the articulatory control system, which translates written words into speech so they can be kept looping in a holding pattern — repeating a phone number until you can type it in
The visuospatial sketchpad
Handles visual and spatial information: the colour of a flower, the arrangement of windows in a house, recognising a friend, giving directions to a stranger. It also splits in two:
the visual cache, the “inner eye”, storing form and colour — a purple triangle, a tall white vase
the inner scribe, holding spatial and movement information such as how to cross a crowded room. It rehearses information which is then stored in the visual cache.
The episodic buffer
Added in 2000 to acknowledge that the CE and slave systems must communicate with LTM to work properly. Working memory has to respond to the current situation — forgetting which side of the road to drive on would cause serious problems. The EB arranges information into “packets” and, when needed, moves it to the slave systems in a set sequence, so events feel continuous rather than out of joint. In a conversation you see the other person’s lips move at the same moment you hear the words.
Notice the pattern. Every slave system splits into a storage part and a rehearsal part. Learn that structure once and both the PL and the VSS become easy to reproduce.
Cognitive load and dual tasks
Cognitive load theory proposes that working memory has limited capability for processing new information. Learning and task success are impaired when it is overloaded, because it cannot handle too much competing information at once. The load is simply too heavy.
A dual-task procedure has participants perform two tasks simultaneously to see how they compete for cognitive resources — drawing while listening to music, or drawing while mentally visualising shapes. It assesses the effect of divided attention on performance.
This is the whole argument for separate slave systems in one picture. If STM were a single store, both panels would fail equally — and they do not.
Key study: Baddeley et al. (1973)
🧩 The study in four steps
Aim — to investigate whether the VSS and the PL exist as separate components, using a dual-task activity.
Method — a dual-task lab experiment. Participants followed a spot of light with a pointer around a circular path (the tracking task) while imagining block capital letters such as H, T, F and E in their heads.
The two conditions — in one, participants gave verbal yes/no responses about features of the imagined letter, for example whether it had a top, middle or bottom line, while tracking. In the other, they had to silently visualise those letter features while tracking.
Findings and conclusion — participants in the silent visualisation condition had great difficulty doing both at once. Tracking and letter imagery were competing for the limited resources of the VSS, whereas tracking plus a verbal response used separate components — the VSS and the PL. There may therefore be separate slave systems, and overloading one with two tasks stops it functioning properly.
Why the letters matter. Imagining a block capital is a visual job, so it lands on the VSS. Saying “yes” is a verbal job, so it lands on the PL. Change nothing but the response type and performance changes — that is the experiment in one sentence.
Evaluating the WMM and cognitive load
Strengths
Replicable. Most dual-task studies are lab-based with standardised procedures, so they can be repeated to check for consistency.
Captures the dynamic side of memory. The research shows memory being used in the moment to complete a complex task, rather than as a static store.
Better fit with everyday experience. It explains why some task combinations are easy and others are impossible.
Limitations
Validity problem. Since participants had to visualise letters in their heads, they may have become distracted and thought about something else entirely, which nobody could check.
Extraneous variables. Some people naturally have better memories or longer concentration spans, some are more visual or more auditory learners, and some are dyslexic or dyspraxic. The findings may therefore not be due to cognitive load, and the research does not account for individual differences.
The CE is barely tested. The evidence supports the slave systems far better than the component the model calls most important.
Link to concepts
Perspective — the WMM provides a much more detailed and dynamic model of STM than the MSM, explaining how responses to acoustic and visual information occur at the same time rather than in a linear, static fashion. Most researchers now say “working memory” rather than “short-term memory”, because the term captures the here-and-now nature of a system working hard within restricted duration and capacity.
Causality — there is very little evidence about how the CE actually directs attention towards the slave systems. It is almost impossible to test, so it remains hypothetical, while the VSS and PL are far easier to measure. By not properly explaining the role of LTM, the model also ignores key factors in how LTM and the EB combine.
Exam practice
WORKED EXAMPLE
Explain one model of memory. [3 marks]
1. Structure
The WMM has a central executive directing three subsystems: the phonological loop, the visuospatial sketchpad and the episodic buffer.
2. FunctionThe CE allocates attention; the PL handles sound, the VSS handles visual and spatial information.
3. Evidence
Baddeley et al. found two visual tasks clashed, while a visual and a verbal task did not.
Structure + function + evidenceSay “slave systems” once. It is the model’s own term and examiners look for it.
WORKED EXAMPLE
Discuss cognitive load theory. [ERQ paragraph]
Claim
CLT holds that working memory has limited capacity, so overload impairs performance.
Evidence
Baddeley’s silent visualisation condition overloaded the VSS and tracking accuracy collapsed.
LimitationHowever extraneous variables such as concentration span or dyspraxia could produce the same result.
Judgement
The theory has good supporting evidence, but individual differences make it hard to attribute any single result to load alone.
Supported, but not isolatedNaming a specific extraneous variable beats the phrase “individual differences” on its own.
💡 Exam tip
Learn the two sub-parts of the PL and the VSS. Four small terms that lift a description into the top band.
Say the WMM explains STM only. Students lose marks by treating it as a full model of memory.
The EB was added in 2000, long after 1974. That date shows you know the model developed.
Use the driving-and-talking example. It makes cognitive load obvious in one line.
“The CE remains hypothetical” is the sharpest criticism available. Use it.
Baddeley is a lab experiment, unlike Glanzer and Cunitz’s list task — know which is which.
⚠️ Common mix-up
The CE is not a store. It directs attention; it does not hold information.
The PL is not “hearing”. Written words also enter it, once vocalised internally.
The visual cache and inner scribe are not the same. Cache stores form and colour; scribe handles space and movement.
The WMM does not replace the MSM. It replaces one box within it.
Two tasks are not automatically hard. Only tasks needing the same system clash.
Cognitive load is about capacity, not motivation. Participants were trying just as hard in both conditions.
Up next: Multi-Store or Working Memory: Which Fits Better? — you now know both models. The next page is about how to compare them without simply describing them twice.
Want this explained one-to-one?
Book a free session with an experienced IB Psychology tutor and get your trickiest topics made simple.
