Introduction
The leaky bucket approach—formalised within the classical Modal Model of Broadbent, Waugh and Norman, and Atkinson and Shiffrin—conceptualises Short-Term Memory (STM) as a passive, unitary container of fixed volume. In contrast, the Working Memory (WM) framework pioneered by Baddeley and Hitch redefines STM as an active, multicomponent cognitive system capable of simultaneously storing and manipulating mental representations.
Structural and Functional Architecture
- Unitary Store vs. Modular System: The leaky bucket model conceptualises short-term memory as a single, modality-independent buffer constrained by a rigid structural capacity of 7 ± 2 chunks, as described by George Miller. Conversely, the working memory framework fractionates this storage into domain-specific slave systems—the phonological loop and visuospatial sketchpad—integrated via a multimodal episodic buffer and coordinated by an attentional central executive.
- Passive Receptacle vs. Dynamic Workspace: The leaky bucket metaphor treats short-term memory merely as a holding pen or an obligatory sequential gateway to long-term memory (LTM). In contrast, working memory functions as an operational mental workspace where maintenance directly supports concurrent executive operations such as reasoning, mental arithmetic, and language comprehension.
Mechanisms of Maintenance and Forgetting
- Loss Dynamics: In the leaky bucket view, forgetting occurs mechanically via spontaneous temporal trace decay across 18 to 30 seconds (as demonstrated by Peterson and Peterson) or displacement and overflow when incoming input exceeds capacity (Waugh and Norman). Working memory explains forgetting through representational interference, cognitive resource contention, and decay counteracted by active attentional refreshing or articulatory looping.
- Maintenance Strategies: The leaky bucket approach relies almost entirely on rote, maintenance rehearsal to circulate items and prevent leakage. Working memory employs active sub-vocal articulation, visual imagery manipulation, and dynamic attentional gating by the central executive.
Empirical and Neuropsychological Dissociations
Dual-task experiments conducted by Baddeley and Hitch demonstrated that saturating verbal digit span does not cause a catastrophic collapse of concurrent spatial reasoning or comprehension tasks, directly refuting the concept of a single-bucket bottleneck.
Furthermore, neuropsychological evidence such as Patient K.F. (studied by Shallice and Warrington)—who suffered impaired auditory short-term memory with a digit span of only 1 to 2 items but retained intact long-term learning and normal reasoning—falsified the leaky bucket premise that a preserved unitary short-term buffer is an indispensable prerequisite for higher cognitive functioning.
Conclusion
The shift from the leaky bucket model to the working memory model marks a fundamental transition from rigid structural bottlenecks to dynamic attentional allocation. Contemporary embedded-process frameworks, such as those proposed by Cowan and Engle, further extend this paradigm by viewing short-term memory not as an isolated container, but as the flexible deployment of top-down attention over activated long-term memory traces.