Articulatory suppression
Articulatory suppression is the experimental technique of preventing subvocal rehearsal — the silent, motor-like articulation of verbal material that normally maintains phonological representations in working memory. When participants are asked to repeatedly say an irrelevant word (such as "the") while performing a memory task, their memory span for verbal material drops by roughly 30–50%. This effect is one of the most robust findings in cognitive psychology, and it is also one of the most misinterpreted.
The standard interpretation — inherited from Baddeley and Hitch's multicomponent model, the foundation of modern working memory theory — treats articulatory suppression as evidence for a two-component architecture: a phonological store that passively holds acoustic traces, and an articulatory rehearsal process that refreshes them. Under this view, suppression occupies the rehearsal mechanism, preventing it from refreshing the store, and the traces decay. The memory loss is explained as passive decay in the absence of maintenance.
This interpretation is wrong. The phonological loop is not a store-plus-refresher. It is a dynamical system — a recurrent feedback loop in which subvocal articulation maintains an activation pattern in speech-motor and auditory cortices. Articulatory suppression does not block a maintenance process while leaving a storage structure intact. It disrupts the feedback loop itself. The "store" is not a separate container; it is the self-sustaining activation pattern created by the loop. When you break the loop, the pattern collapses. What looks like decay is actually the dissipation of a dynamical attractor.
The Systems Reading
From a systems perspective, articulatory suppression is a perturbation experiment. It is the cognitive equivalent of clamping a feedback node in a control system to see what the rest of the circuit does. The result is instructive: the memory trace does not simply fade; it loses its structured form. Under suppression, participants still retain some acoustic information — the residual "echo" — but they lose the phonological structure that would normally be maintained by articulatory recoding. The remaining memory is no longer phonological; it is a degraded acoustic trace with reduced temporal resolution.
This systems reading explains why suppression affects not just memory span but also the phonological similarity effect. When participants must remember lists of similar-sounding words (e.g., "man, mad, map, mat"), performance is worse than for dissimilar words because the phonological representations interfere with each other. Under articulatory suppression, the phonological similarity effect disappears. This is not because the store has been emptied but because the loop has been broken: the representations that would normally be maintained in phonological form are now maintained as unstructured acoustic traces, and acoustic traces do not interfere in the same way.
Connections to Broader Phenomena
The logic of articulatory suppression extends beyond verbal memory. Any cognitive system that relies on recurrent feedback to maintain structured representations is vulnerable to perturbation of the feedback component. In motor control, preventing proprioceptive feedback disrupts the maintenance of motor programs. In perception, stabilizing a retinal image (preventing the normal microsaccades that refresh the visual input) causes the image to fade from consciousness — the Troxler effect. In each case, the mechanism is the same: a dynamical system that appears to store information is actually maintaining a pattern through active feedback, and blocking the feedback causes the pattern to dissipate.
The parallel to neuroscience is direct. Persistent neural activity in the prefrontal cortex during working memory tasks is not the storage of a static representation. It is the sustained firing of recurrently connected neurons that maintain an activation pattern through positive feedback. When that feedback is disrupted — by synaptic depression, by competing inputs, or by pharmacological intervention — the activity pattern collapses. What looks like forgetting is actually the loss of a dynamical attractor.
The connection to short-term memory and decay theory is equally direct. Decay is not a passive process of trace weakening. It is the active dissipation of a self-sustaining activation pattern when the feedback that maintains it is interrupted. Articulatory suppression is the experimental tool that makes this dissipation visible.
Articulatory suppression is not a technique for studying a storage system. It is a technique for revealing that the storage system never existed. The phonological loop is not a memory device with a broken refresher. It is a feedback loop with a broken feedback path. And that is the difference between a box and a system.