Storage (memory)¶
The retention of encoded information across time for later recognition or retrieval in a memory system.
Core Idea¶
Memory storage is the persistence of encoded information across time so that it can shape a later response. It is analytically distinct from encoding, which forms the representation, and retrieval, which accesses it, although real memory processes interact. A storage claim requires some retained state and a later criterion for detecting its influence.
Psychology uses several models of that persistence. Short-term and long-term stores are useful in some accounts; others model explicit traces or changes in distributed connection strengths. The abstraction does not require one disputed neural location, an exactly fixed item capacity, or a claim that forgetting always means complete erasure.
Structural Signature¶
Sig role-phrases:
- Encoded content — Supplies the representation to be retained, not the act of retention itself. It is constitutive. Counterfactual: Without content there is nothing for a memory system to preserve.
- Retaining substrate or state — Carries content or associations across a delay, whether modeled as a store, trace, or changed network weights. It is constitutive. Counterfactual: Without persisting state a later response cannot depend on prior encoding.
- Retention interval — Separates an immediately present stimulus from information maintained for later access. It is constitutive. Counterfactual: Collapse the interval to zero and storage becomes indistinguishable from current presentation.
- Maintenance and interference — Modify how accessible the retained state remains under rehearsal, decay, or competing input. It is central. Counterfactual: Ignoring these conditions misreads a failed recall as proof that no information was ever stored.
- Retrieval or recognition test — Provides evidence that earlier content still affects a later response. It is central. Counterfactual: Without a later test, retention is postulated but not observed in the behavioral case.
What It Is Not¶
- Not encoding. Transforming incoming information into a representation is a prerequisite, not the same question as its persistence.
- Not retrieval. Failure to recall can reflect a poor cue even if a retained state remains.
- Not literal filing in one brain location. Store, trace, and weight are model-dependent descriptions.
- Not permanent by definition. Retention may be brief and may degrade or become inaccessible.
- Closest near-miss. Encoding changes how incoming information is represented; storage concerns its persistence, even though the processes interact.
Scope of Application¶
- Immediate and working-memory tasks. Examine how briefly maintained information supports a response after presentation ends.
- Longer-delay memory. Study persistence and later access across intervals not covered by immediate maintenance alone.
- Formal memory models. Represent persistence as explicit traces, store states, or altered associative weights.
- Experimental diagnosis. Use rehearsal, interference, and changed retrieval cues to distinguish explanations of later performance.
Clarity¶
Separating encoded content, persistence, and later access prevents the statement 'it was forgotten' from concealing multiple possibilities. A distractor-sensitive recency effect concerns brief accessibility under a particular task; it does not settle the existence or anatomy of every long-term store. The analyst should name the model, retention interval, response test, and competing explanation.
Manages Complexity¶
Storage reduces a history of past experience to a state that can still affect the present. A model then chooses how to represent that state—as items, traces, or distributed weights—so it can predict later recall or recognition. This compression makes experiments tractable but omits many details of neural implementation and can fail when an assumed store architecture is too rigid for the evidence.
Abstract Reasoning¶
- Specify what information was presented and how it was encoded in the task.
- Identify the interval over which information must persist after the stimulus is absent.
- State the model of the retained state and its maintenance or interference conditions.
- Choose a later recall, recognition, or other response that could reveal prior information.
- Compare performance under changed delays, distractors, and cues before attributing a failure to loss of storage.
- Keep behavioral evidence distinct from claims about an exact physical storage site or universal capacity.
Knowledge Transfer¶
The storage concept transfers literally across memory experiments and models that ask how previously encoded information persists for later access. Computer-storage language can help explain it, but a brain is not thereby shown to use a literal file system. The more general prime-like idea is persistence of information, while the entry's evidence and models belong to cognitive memory research.
Examples¶
Canonical¶
In a serial-recall task described by the article's dual-store discussion, the last items in a just-presented list are often recalled well because they remain readily accessible at test. Inserting a distracting task before recall can reduce this recency advantage. The contrast is evidence about short-term retention and interference under that model, not proof that every memory is stored in two literal boxes.
Mapped back: Encoded content → items from the presented list; Retaining substrate or state → short-term-store state in the model; Retention interval → delay between presentation and recall; Maintenance and interference → distracting task disrupts accessibility; Retrieval or recognition test → serial recall of list items.
Applied / In Practice¶
In the article's multi-trace model, an encoded item is represented as an attribute vector and added to a trace matrix. Later, a probe is compared with stored traces, yielding a basis for recognition or cued recall. The model demonstrates one formal way to implement storage; it should not be presented as a settled anatomical account of how the human brain stores every memory.
Mapped back: Encoded content → attribute vector for an item; Retaining substrate or state → trace matrix; Retention interval → period between insertion and probe; Maintenance and interference → overlap with other traces may affect discrimination; Retrieval or recognition test → similarity comparison against the probe.
Structural Tensions¶
T1 — Brief Access versus Durable Retention. A short-term store explains recent-item accessibility and distractor sensitivity, while longer retention requires different assumptions about persistence and organization. Treating them as identical erases the evidence that motivated separate models; treating them as wholly disconnected overstates a disputed architecture.
Diagnostic: What observation requires one store versus a transition between stores?
T2 — Explicit Traces versus Distributed Association. A trace model preserves item-like records that a probe can compare, while a weight model distributes associations over a fixed network. The first gives a direct record metaphor at a scaling cost; the second has compact shared state at the cost of less transparent item boundaries.
Diagnostic: Does the task require a separate item record or can altered association weights explain retrieval?
T3 — Retention Inference versus Retrieval Failure. A missed response can result from information being lost, inaccessible, or inadequately cued. Equating nonrecall with absent storage simplifies measurement but can misclassify the underlying process.
Diagnostic: Could a changed cue reveal retained information after the failed test?
Structural–Framed Character¶
Memory storage is mixed-structural: retention across an interval is a testable relation, but particular “store” models interpret how it occurs. Evaluative weight: retained information is not inherently good memory; accuracy and usefulness depend on the task, whereas the entry asks whether prior encoding still affects later recognition or retrieval. Human-practice-bound: organismal retention can occur without a psychologist, but encoding protocols, delays, and recognition tests are human choices used to infer it. Institutional origin: cognitive theories distinguish short- and long-term stores in different ways; no institution creates the underlying phenomenon or licenses one model as its only mechanism. Vocabulary travels: persistence and later access compare with computing, but a neural trace is not thereby a disk file. Import versus recognize: a new experiment measuring retained cognitive influence recognizes the same phenomenon; importing “memory storage” for database writes changes the carrier and evidence test.
The portable skeleton is prior information leaving a state with a later causal effect. It is a future-prime candidate here because no reviewed strict prime parent for that retention relation is asserted in the current DAG. Trace organization, interference, recall, and recognition remain questions of cognitive memory rather than properties of every persistent state. Its character: an empirically inferred psychological relation with multiple competing internal models.
Structural Core vs. Domain Accent¶
Skeletal core. Prior information leaves a state that affects later behavior. Domain-bound accent. In cognitive memory the content, maintenance conditions, delay, and recall or recognition tests define the phenomenon. Replace those with bytes written to a disk and one has a different storage process. Why not a prime. The general persistence relation may travel, but claims about short-term stores, traces, interference, and retrieval belong to the memory domain.
Instantiates / Related Primes¶
-
Approved unparented root. The frozen graph asserts no parent that necessarily entails cognitive retention across a delay.
-
Related, not asserted as parents. Encoding and retrieval flank storage; neither alone entails persistent memory state.
Neighborhood in Abstraction Space¶
Storage (memory) sits in a crowded region of the domain-specific corpus (35th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.
Family — Memory Storage, Retrieval & Encoding (9 abstractions)
Nearest neighbors
- Exosomatic Memory — 0.91
- Cue-dependent forgetting — 0.90
- Atkinson–Shiffrin memory model — 0.89
- De Novo Protein Synthesis Theory of Memory Formation — 0.88
- Recall (Memory) — 0.88
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
- Encoding. Tell: Forms or transforms a representation; ask whether persistence after input ends is at issue.
- Retrieval. Tell: Accesses previously retained information; ask whether the question is the maintained state or its later access.
- Working memory. Tell: Is one family of active maintenance models; ask whether the claim concerns all memory retention or a particular working-memory task.
- Physical computer storage. Tell: May share the persistence metaphor; ask whether the substrate and evidence concern cognitive memory.
References¶
- Frozen Wikipedia discovery revision: https://en.wikipedia.org/wiki/Storage_(memory) (revision 1364360433).
- Preserved source candidate: https://semanticscholar.org/paper/145272a828eb27ef0e9e5095a5696058df5320c0
- Preserved source candidate: http://www.garfield.library.upenn.edu/classics1990/A1990DB80900001.pdf
- Preserved source candidate: https://www.zora.uzh.ch/id/eprint/151291/1/Thalmann.et.al.Chunking.final.pdf
- Preserved source candidate: http://www.nature.com/scientificamerican/journal/v215/n1/pdf/scientificamerican0766-90.pdf
- Preserved source candidate: https://serendipstudio.org/biology/b103/f97/projects97/Warren.html
- Preserved source candidate: https://escholarship.org/uc/item/741429sd
- Preserved source candidate: http://www.brightfocus.org/alzheimers/about/understanding/anatomy-of-the-brain.html
- Preserved source candidate: https://web.archive.org/web/20171117025519/http://neuroscience.uth.tmc.edu/s4/chapter07.html
The frozen Wikipedia revision is discovery provenance. The retained source set was reviewed for identity, formal or operational relation, and scope. The encyclopedia's structural synthesis is bounded to those claims; a thin authority surface is recorded as a nonblocking source-strengthening repair rather than concealed.