Working Memory Model

The Working Memory Model

In 1974, Alan Baddeley and Graham Hitch proposed the working memory model (WMM) as a replacement for the oversimplified STM component of Atkinson and Shiffrin's multi-store model. Rather than treating short-term memory as a single, passive store, Baddeley and Hitch argued that the memory we use whilst actively thinking, reading, or problem-solving is a complex, multi-componential system. The original model had three components; Baddeley added a fourth in 2000.

The Central Executive

The central executive is the most important and most complex component of the model. It is modality-free — not tied to any particular sensory system — and has a limited capacity. Its primary function is attentional: it controls and coordinates the slave systems, allocates cognitive resources, switches attention between tasks, and suppresses irrelevant information. Because of its supervisory role, the central executive is involved in any cognitively demanding task, from planning to problem-solving. It is also the least well understood component, partly because it is the most difficult to study in isolation from the other components.

The Phonological Loop

The phonological loop handles verbal and acoustic information. It has two sub-components. The phonological store (inner ear) holds sound-based memory traces for approximately 1-2 seconds before they fade. The articulatory rehearsal process (inner voice) refreshes these traces through subvocalisation — mentally repeating information to prevent decay.

Two key phenomena are explained by the phonological loop. The phonological similarity effect shows that words sounding similar to each other are harder to remember in sequence, because they produce overlapping and confusable traces in the phonological store. The word length effect shows that longer words are harder to retain than shorter ones, because the articulatory rehearsal process can only maintain approximately 2 seconds' worth of material — demonstrating that phonological loop capacity is time-based rather than item-based.

The Visuospatial Sketchpad

The visuospatial sketchpad (inner eye) processes and temporarily stores visual and spatial information. It is responsible for tasks such as mentally rotating shapes, planning a navigation route, or holding in mind the layout of a room. Evidence for its independence from the phonological loop comes from dual-task studies: performing a verbal task and a spatial task simultaneously causes far less mutual interference than performing two tasks in the same modality — suggesting they draw on entirely separate slave systems.

The Episodic Buffer

Added by Baddeley in 2000, the episodic buffer was introduced to address a key limitation of the original model: working memory can hold more integrated information than the separate slave systems alone could explain. The episodic buffer is a limited-capacity store that integrates information from the phonological loop, the visuospatial sketchpad, and long-term memory into unified, multi-dimensional representations called episodes. It acts as a temporary interface between working memory and LTM, controlled by the central executive.

Evidence for the Model

Dual-task studies provide the strongest experimental support. Baddeley and Hitch found that participants could perform two tasks using different slave systems simultaneously with relatively little disruption, but performance deteriorated sharply when both tasks competed for the same slave system. This demonstrates that working memory has separate, independently limited components rather than being a single unified store.

Neuropsychological evidence also supports the model. Patient KF, who suffered brain damage, showed severely impaired verbal STM alongside intact visuospatial STM — a dissociation consistent with separate phonological loop and visuospatial sketchpad components. Brain imaging studies have further linked phonological loop activity to left hemisphere language areas, and visuospatial sketchpad activity to right hemisphere visual areas.

Evaluation

The WMM is a significant advance over the multi-store model: it accounts for the active, processing nature of working memory and is supported by converging experimental and neuropsychological evidence. However, the central executive remains the most theoretically underspecified component — it is defined by its function without a precise mechanistic account, and risks being a theoretical catch-all. The model is also less informative about the broader relationship between working memory and long-term learning, though the episodic buffer partially addresses this gap.