Working Memory¶
Maintain a small, short-lived set of representations in an actively accessible workspace whose executive control can select, rehearse, transform, replace, and coordinate them for the task currently being performed.
Core Idea¶
Working memory is the cognitive architecture that keeps a small set of representations actively accessible while a person performs a task. Unlike passive short-term retention, it combines temporary maintenance with executive operations that select, rehearse, transform, reorder, replace, and coordinate the current contents. Its identity is the workspace, not the amount it can hold, the demand placed on it, or any one operation performed inside it.
Scope of Application¶
- Reasoning: retaining premises while deriving and checking an intermediate conclusion.
- Language: maintaining earlier words while integrating a sentence or discourse.
- Mental calculation: preserving partial results while executing the next operation.
- Planning: holding goals, constraints, and candidate actions in an active task model.
- Learning: coordinating new material with retrieved long-term knowledge.
- Executive control: replacing stale contents as task relevance changes.
Clarity¶
Working memory separates four neighboring questions that are often collapsed: What is the active workspace? How much can it hold? How much demand does a task impose? How are its contents updated? The architecture is Working Memory; the supply quantity is Working Memory Capacity, the demand is Cognitive Load, and real-time revision is Working-Memory Updating.
Manages Complexity¶
The abstraction replaces a loose catalogue of remembering, focusing, rehearsing, and manipulating with one bounded active-content system. Analysis can then track the current representations, the control process selecting them, the operations applied to them, and the capacity or duration constraints that force competition and eviction.
Abstract Reasoning¶
Working Memory supports state-tracking over an explicitly limited active set, supply-versus-demand comparison, and intervention reasoning about rehearsal, chunking, sequencing, and external offloading. It also licenses a boundary test: if a mechanism only selects an input, only stores it passively, or only measures a limit, it is related to the workspace but is not the workspace itself.
Knowledge Transfer¶
The mechanism transfers within cognitive psychology, neuroscience, education, interface design, and human-factors work wherever a human active workspace is the explanatory object. Queues, caches, scratchpads, and organizational dashboards are useful analogies, but their cross-domain structure belongs to more general primes; they do not become Working Memory without the cognitive architecture and executive-control commitments.
Example¶
While multiplying two multi-digit numbers mentally, a person must keep partial products accessible, direct attention to the next digit, update the retained total, and discard values that are no longer needed. Failure can reflect insufficient capacity, excessive task load, or faulty updating, but all three diagnoses presuppose the active workspace in which the values are maintained and manipulated.
Relationships to Other Abstractions¶
Current abstraction Working Memory Domain-specific
Parents (2) — more general patterns this builds on
-
Working Memory presupposes Attention Prime
Working Memory presupposes Attention because selective control determines which representations enter, remain active, and receive manipulation.Remove selective allocation and executive focus and the architecture cannot determine which incoming or retrieved representations enter the active set, remain accessible, or receive rehearsal and transformation. Attention supplies that control operation without itself supplying a short-lived storage-and-manipulation workspace.
-
Working Memory is part of Constraint Prime
Working Memory contains binding capacity and duration constraints that force rehearsal, replacement, and competition among active representations.The architecture contains a small simultaneous capacity and short persistence interval. Remove those restrictions and its overload regime, replacement pressure, rehearsal requirement, and contrast with long-term memory disappear. Constraint is therefore an internal constituent rather than a genus.
Children (3) — more specific cases that build on this
-
Cognitive Load Domain-specific presupposes Working Memory
Cognitive Load presupposes Working Memory because it is demand imposed on that active storage-and-manipulation architecture.Cognitive Load is the task-agent-format demand placed on an active workspace that temporarily maintains and manipulates representations. Remove Working Memory and there is no architecture whose finite supply the load can consume or exceed. The child is demand on that workspace, not a kind of Attention or a synonym for the workspace itself.
-
Working Memory Capacity Domain-specific presupposes Working Memory
Working Memory Capacity quantifies the simultaneous holding-and-manipulation supply of the Working Memory architecture.A capacity quantity is defined only for an architecture whose admissible active contents it measures. Remove Working Memory's temporary maintenance and manipulation workspace and no simultaneous supply remains to count.
-
Working-Memory Updating Domain-specific presupposes Working Memory
Working-Memory Updating operates on the Working Memory architecture's active content set, replacing and transforming items by current task relevance.The updating operation requires a bounded active workspace whose contents can be monitored, retained, transformed, replaced, and evicted. Remove that architecture and there is no working-memory content set for the executive function to update.
Hierarchy paths (2) — routes to 2 parentless roots
- Working Memory → Attention
- Working Memory → Constraint
Distinction from Neighbors¶
- Attention selects and prioritizes processing targets but does not by itself store and manipulate a short-lived content set.
- Working Memory Capacity measures the workspace's simultaneous supply; it is not the architecture.
- Cognitive Load is demand placed on the workspace by a task-agent-format combination.
- Working-Memory Updating is one executive operation over the workspace's contents.
- Short-term memory may denote passive retention without concurrent manipulation.
- Buffering manages a source-consumer rate mismatch and later release; that mechanism is not required here.
Notes¶
(New domain-specific intermediate; queued for Claude style harmonization and independent citation review.)