The central executive coordinates the subsystems and allocates attention, and has its own capacity limit

What it is

The phonological loop and the visuospatial sketchpad each store one kind of information, but deciding what gets priority when, when to switch tasks, and which distracting responses need to be suppressed is handled by working memory's third component, the central executive. It isn't itself a storage buffer — it's an attentional control mechanism. Like the two storage subsystems, the amount of coordination load it can handle at once is limited, and overloading it drags down performance.

Why it happens

The phonological loop and visuospatial sketchpad are relatively independent storage resources, each responsible for its own type of content. The central executive is a domain-general control resource that sits across both — whatever is currently being processed, verbal or spatial, any active decision-making, task-switching, or suppression of an irrelevant response has to go through this single unified control mechanism. Precisely because it's a generic resource that doesn't distinguish content type, two tasks landing in different storage channels can still compete with each other as long as both require the central executive to step in and actively coordinate. This is a different phenomenon from the independence between the storage subsystems, which each handle their own lane without interfering — the central executive's load is shared across channels.

Studying it

The classic manipulation for specifically loading the central executive is the random generation task: asking participants to call out a sequence of digits trying to make it as unpredictable as possible. This task places little demand on either the phonological or visuospatial store — what it mainly consumes is the control resource needed for active decision-making and suppressing habitual responses. Having participants perform this alongside another primary task disrupts that task noticeably, whether the primary task is verbal or spatial in nature — this indifference to channel, disrupting everything roughly equally, is exactly the evidence that the central executive is a domain-general control resource rather than one more content-specific storage subsystem.

Where it stops holding

Compared with the phonological loop and visuospatial sketchpad, the central executive's specific operation is characterized far less precisely — within the model it often ends up carrying "whatever function can't be pinned on the other two storage subsystems." That means performance drops shouldn't be blamed on "executive overload" as a catch-all unless there's actual evidence the task involves active decision-making, switching, or suppression rather than merely passive retention. A purely passive retention task — just holding onto a number that's already sitting there and doesn't need active processing — places comparatively little demand on the central executive.

Applying it

• When judging whether one action will interfere with another, don't just check whether they're "the same type of information" (both text, or both graphics) — also check whether both require the user to make an active judgment, switch attention, or suppress a habitual response. If both do, competition will occur at the central executive level even if one task is superficially verbal and the other spatial.
• Don't insert another interruption that also requires active judgment (an unrelated choice sprung on the user, say) while the user is focused on a complex task that itself requires active decision-making (multi-step spatial navigation, for instance), even if the interruption looks completely different in form — the conflict still happens at the control level.
• Content that genuinely only needs to be passively held, with no active processing involved, can be scheduled alongside other tasks with relatively little concern, since this kind of display places comparatively little demand on the central executive.
• Verification: compare how much an interrupting task affects primary-task performance when the interruption requires active decision-making/switching versus when it only requires passive retention. If the former causes a substantially larger disruption regardless of whether the channels match, the conflict is genuinely happening at the central-executive level rather than the storage-channel level.

Related

• Same group: A6.12.1 The phonological loop holds verbal and numeric information · A6.12.2 The visuospatial sketchpad holds visual and spatial information · A6.12.3 The two subsystems draw on relatively independent resources, so simultaneous information in different channels doesn't interfere · A6.12.5 Presenting the same information through both visual and verbal channels can raise working memory's effective capacity
• Nearby: A5.02 Divided attention and dual-tasking
• Search terms: central executive · random generation task · executive control · attentional switching