HPC › Learning › Glossary
Working Memory is a limited-capacity cognitive system that temporarily holds and manipulates information during active thought. Four interacting components govern it: a central executive that directs attention, a phonological loop for verbal material, a visuospatial sketchpad for spatial content, and an episodic buffer that integrates information across modalities. Its genuine capacity is approximately four independent chunks.
The term supersedes the older concept of short-term memory: working memory is active and multicomponent, not a single passive store.
How it works
Baddeley and Hitch proposed the multicomponent model in 1974, replacing the earlier notion that short-term memory was a single passive store 1. Four interacting components organise working memory. The central executive, anchored in prefrontal cortex circuits, functions as an attentional controller rather than a storage buffer, directing which information is maintained or suppressed during complex cognitive tasks 3. The phonological loop preserves verbal information through subvocal rehearsal; without active rehearsal, memory traces decay within approximately two seconds 1. The visuospatial sketchpad holds visual and spatial content independently. An episodic buffer integrates information across modalities into coherent representations that can pass into long-term memory 3.
The genuine capacity limit of working memory is approximately four independent chunks, correcting Miller's influential but contested 1956 figure of seven, plus or minus two. That larger number arose from chunking and rehearsal strategies that inflated apparent capacity in the original experiments 2. A chunk is any unit compressed through prior learning: for an expert chess player, a recognisable board configuration is one chunk; for a novice, each piece is a separate item. Under controlled conditions that exclude rehearsal and chunking effects, capacity consistently falls near four discrete items 2.
Example
A medical professional reviewing a patient chart must hold the current diagnosis, three recent test results, and the proposed treatment simultaneously while reading incoming lab data. Each new piece of information competes for one of the roughly four available slots. When the number of active items exceeds capacity, earlier items are displaced, which increases the risk of missed connections between findings.
This is why experienced practitioners offload information into written notes or structured protocols: reducing active memory demand lets the central executive focus on the decision at hand.
Why it matters
Working memory capacity is among the strongest predictors of academic achievement and fluid intelligence 3. Learners with lower capacity are disproportionately harmed when instructional designs impose unnecessary cognitive load. The four-chunk ceiling means that complex skills with many simultaneous demands will routinely overwhelm novices, explaining why tasks that experts perform automatically remain effortful for beginners. Experts have compiled long-term memory schemas that allow them to treat multi-step procedures as single chunks, freeing capacity for higher-order processing 2.
When cognitive load exceeds working memory limits during instruction, new information cannot be encoded into long-term memory, regardless of the learner's attention and motivation 4. Instructional methods that reduce extraneous load, such as worked examples, spatially integrated layouts, and segmented presentation, directly improve retention by preserving available capacity for schema formation 4. Understanding this constraint reframes what effective teaching requires: the central task is managing the information throughput entering a system with a firm upper bound.
What is the difference between working memory and short-term memory?+
Short-term memory refers to the passive retention of information over brief intervals. Working memory is the broader, active system that not only holds information but also manipulates it during ongoing cognitive tasks. Baddeley and Hitch's multicomponent model replaced the single-store concept with four interacting components, including a central executive that actively governs attention.
How many things can working memory hold at once?+
Controlled experiments indicate a genuine limit of roughly four independent chunks, correcting Miller's influential 1956 figure of seven, plus or minus two. That larger number arose from chunking and rehearsal strategies that inflated apparent capacity. The true limit falls closer to four when those strategies are controlled in experimental conditions.
How does working memory affect learning and academic performance?+
Working memory capacity is one of the strongest predictors of academic achievement. When instruction imposes more cognitive load than available capacity allows, new information cannot be consolidated into long-term memory. Well-designed instruction reduces this load through worked examples, eliminating split-attention layouts, and breaking complex content into smaller segments.
Can working memory capacity be trained or improved?+
Working memory training programmes show improvements on trained tasks but limited transfer to broader cognitive abilities. Gains tend to be task-specific rather than reflecting an increase in underlying capacity. More robust performance improvements come from reducing extraneous cognitive load in the learning environment, such as through better instructional design, than from attempting to expand the limit itself.
