Memory in animals

Retention and later use of information acquired through experience. Behavioural tests distinguish retention from immediate sensory responses and continuing access to external cues.

Sensory memory is a brief persistence of information from a sensory event after that event has ended. It can make recently presented information available for selection even when a person cannot report every item at once. Visual sensory persistence is one case; the experimental task and modality must be specified before assigning a duration or capacity to sensory memory.

George Sperling's1960 experiments separated information available after a brief visual display from information that could be reported serially. Observers saw arrays of letters. In partial-report conditions, a tone after the display selected which row to report, rather than announcing that row before presentation. Delaying the selection cue reduced the advantage of partial report. The critical logic was that a cue given only afterward could select information from more of the display than immediate whole-report performance appeared to show.^[ref]

For an illustrative calculation, suppose three rows contain four letters each. Whole report yields four correct letters, while a randomly chosen row can be reported with three correct letters. If each row is equally likely to be cued and comparably accessible, partial report estimates $3\times 3=9$ letters available near cue time. This is an inference under a sampling model, not a claim that nine letters were spoken on a single trial.

The example explains why report length and available information can differ. It also limits the interpretation: a post-display advantage is not unlimited long-term retention, and delays, masking or response demands can change what remains accessible. Sensory memory is studied through such contrasts, rather than identified solely with a participant saying that a display still feels visible.

A persistent physical change or ensemble associated with retaining information from experience. Reactivation, necessity for natural recall and sufficiency under artificial stimulation are distinct criteria; no one of them alone establishes a complete account of subjective recollection.

Memory for relationships among items, places, contexts or events. Flexible use of a learned relation in a new query is more informative about relational structure than repetition of a practiced stimulus–response choice.

An account assigning the hippocampus a role in organizing related experiences and permitting their flexible expression. The proposed computation can operate on spatial and nonspatial information. It is a theory whose parts require tests, rather than a synonym for every task affected by hippocampal damage.^[ref]

Information about an event's position in an evolving sequence or internal state. Changes in neural activity over time can supply a temporal context, but time covaries with behavior, expectation and other variables that require experimental controls.

Retention of the relative order of events. It need not involve knowing a calendar date or measuring elapsed duration. Tests should distinguish an ordered relation from judgments based only on the recency or strength of individual item traces.

Retention of behaviorally testable features of an event, often including integrated content, place and temporal information, without claiming to have established conscious reliving. The label separates evidence about a task from evidence about autonoetic consciousness.

Retention of which event occurred at which place and at which time or temporal position. Separate success on content, location and timing tests does not establish that their particular conjunction was retained; episodic binding is a stronger requirement.

The processes through which people retain and use information. Retention interval, material, awareness and response demands must be distinguished; they do not define one-to-one anatomical stores.

A human memory architecture in which new information must pass through a short-term store to reach a long-term store. A strong version is challenged if the specific short-term mechanism claimed to be obligatory is severely impaired while relevant long-term learning remains relatively preserved. Residual or alternative routes must be considered.

Retention beyond immediate temporary maintenance, including declarative and procedural forms. A person can have severe new declarative learning difficulty while retaining other enduring learning.

Retention over a brief interval. Immediate span is one assay, not a complete measure of all representations that can influence immediate behavior.

Chunking groups separate items into meaningful units for recognizing patterns or using short-term memory. Several digits can be treated as one familiar date, for example, so the same number of visible symbols can require different numbers of remembered units. The grouping draws on knowledge in long-term memory; inserting spaces alone does not make an unfamiliar pattern meaningful.

William G. Chase and Herbert A. Simon investigated these units by asking three chess players of differing skill to reconstruct positions and examining pauses and successive glances during reconstruction. The pattern of pieces placed together provided evidence about the structures a player perceived, rather than simply measuring the total number of pieces remembered.^[ref] Expertise can therefore improve performance by changing the representation of the material, without demonstrating an equal improvement in every unrelated short-term memory task.

Temporary retention of verbal material. Auditory presentation, phonological coding, rehearsal and output requirements can contribute differently to a measured span.

A task measuring the sequence length of digits that can be reproduced under specified conditions. Forward repetition and backward manipulation are different demands; modality and presentation rate also matter.

A change in processing an event because its representation has been activated before the event occurs. The direction and mechanism of the effect depend on the task and model.

In the SOP model, an associated cue or context activates elements of an expected event into secondary activation before that event occurs. It is conceptually distinct from residual activation caused by recent presentation of the event itself.

In the SOP model, a recent stimulus presentation leaves elements in secondary activation, making fewer elements available for primary activation on repetition. This supplies a transient response-decrement mechanism without requiring a new association.

A discrepancy between an associatively expected event and the event that occurs. Novel combinations of familiar stimuli can reveal it without introducing an entirely novel stimulus.

An organized body of prior knowledge that can shape encoding, interpretation and integration of new information. In flavour–place experiments, prior structured learning allowed rapid incorporation and consolidation of compatible new associations.^[ref]

A framework combining rapid learning with relatively separated episodic representations and slower learning with overlapping representations that extract shared structure. Hippocampal reinstatement supplies experience for interleaved cortical learning. The computational aim is to retain specific events while integrating new knowledge without destructive interference, not simply to duplicate short- and long-duration stores.^[ref]

Processes that stabilize or reorganize information after acquisition. Synaptic consolidation concerns cellular changes, whereas systems consolidation concerns changing interactions among brain regions. Better later performance need not mean a literal transfer of a complete memory between two stores.

Time-dependent reorganization of the neural circuits supporting retention and retrieval, often involving interactions between the hippocampus and cerebral cortex. Whether a particular detailed episode becomes independent of the hippocampus is a further empirical question.

An account in which reactivation of recently encoded representations during offline states modifies distributed neural circuits. Coordination of hippocampal sharp-wave ripples, sleep spindles and cortical sleep slow oscillations provides a candidate timing mechanism. Reactivation, strengthened behaviour and permanent cortical storage are distinct claims requiring different evidence.

Stabilization involving cellular and synaptic plasticity mechanisms after learning, rather than only continued sensory input or working-memory rehearsal.

Experience-dependent retention expressed in performance of a skill or procedure, which need not be accompanied by explicit recollection of training.

A motor-learning task in which a participant tries to keep a stylus on a moving target. Practice-related improvement and retention across sessions can assess skill learning separately from explicit recollection of the practice.

Memory accessible to flexible report or use concerning facts and events. It includes semantic memory and episodic memory; it is distinguished from preserved performance-based learning such as procedural memory.

An account in which the medial temporal lobe rapidly establishes and initially binds long-term declarative memory, with a changing dependence on the region during systems consolidation. The original strong restriction of the region to long-term memory must be distinguished from the broader claim that it is important for declarative learning.^[ref]

Memory for facts and conceptual knowledge, distinguished from recollection of a particular learning episode.

Knowledge of objects, events and their properties or relationships, assessed through verbal and nonverbal tasks. Failure on one naming task alone is insufficient to establish loss of this knowledge.

A representation of meaning or conceptual information. Degradation of such representations must be distinguished from difficulty regulating access or producing a response.

A representation of conceptual information not restricted to a single sensory input or verbal output format. Converging cross-modal deficits can support this hypothesis without proving that all features are stored in one homogeneous site.

The regulation of access to and use of conceptual knowledge according to a current goal. Sensitivity to competing meanings, cues and task requirements can help distinguish impaired control from degradation of representations.^[ref]

Identifying previously encountered material when it is presented again. Recognition provides external information unavailable in unaided recall, so a recognition–recall difference is not automatically evidence for separate memory stores.

A task assessing prior occurrence of an object. In spontaneous exploration versions, a preference for a novel over a familiar object is informative only with sufficient exploration and controls for sensory ability, motivation, spatial cues and pre-existing object preferences.

Retrieval of information about the context or associations of a prior event. A behavioural estimate depends on the assumptions of the task and model used.

A sense or signal of prior occurrence without retrieval of a specific associated context. Confidence in recognition is not uniquely equivalent to familiarity strength.

Access to retained information in response to internal or external cues. Cue reinstatement can alter test performance without changing the original episode, but controls are needed to distinguish retrieval from new learning.

Reactivation during retrieval of neural patterns or cells associated with a previous experience. It can be selective without reproducing every detail of the original sensory response. Correlated reinstatement and successful recall are not automatically a proof of causation.

Memory for the context in which information was acquired, such as the task performed when a word was encountered. Accurate item recognition can coexist with an incorrect source judgment. Source accuracy must be distinguished from confidence and from a stronger recognition signal.

An involuntarily occurring memory or memory-like impression. Frequency, sensory content, distress and sense of present occurrence should be assessed separately; an intrusion is not automatically a literal replay.

An experimental analogue using distressing film material to study later involuntary memories. It permits controlled manipulations and prospective diaries but is not equivalent to clinical trauma exposure or a diagnosis of PTSD.^[ref]

The proposal that sensory impressions preceding the worst moments of a trauma acquire predictive significance and are later reexperienced as warning signals. Descriptive clinical content supports investigation of the account, while causal conditioning and retrospective reconstruction require separate tests.^[ref]

Attributing a remembered or experienced representation to its source, such as an external event or an internally generated action. A source-monitoring account of voice hearing is a mechanistic hypothesis, not a definition or a uniquely established cause.

Restoration of a measurable response through cues related to an earlier experience. Reactivation does not by itself imply spontaneous retrieval, conscious recollection or unchanged retention of every original detail.^[ref]

Re-presenting a cue previously associated with learning, often during sleep, to bias subsequent reactivation. Unpaired-cue, waking-cue and sleep-stage controls help distinguish a memory-specific effect from sensory stimulation or arousal.

Re-expression of the temporal or sequential structure of a previously experienced activity pattern. Evidence should compare the observed sequence with appropriate shuffled or pre-experience baselines, rather than label any increased firing as replay.

Persistence of information across a delay. Measuring retention requires a specified retrieval test; a silent behavioural result is not sufficient to distinguish inaccessible information from erased information.

Processes through which an experience establishes information available to later retention and retrieval. A later failure to report an experience does not, by itself, identify whether encoding was unsuccessful.

Memory for a particular event and its context, including relations among what happened, where and when. A computational account must explain binding and context-sensitive retrieval, rather than only classification or familiarity.

Linking the content, context and temporal position belonging to a particular event. A binding representation distinguishes encountering an object in one context from encountering the same object in another context, even when all component features are familiar.

Consider two equally weighted sets of event triples: $\{(A,L,1),(B,R,2)\}$ and $\{(A,R,1),(B,L,2)\}$. Their separate distributions of item, place and time are identical, but $P(L\mid A)$ is respectively $1$ and $0$. Thus identical marginal knowledge does not determine the joint event representation. A behavioral test that can be solved from separate feature lists does not by itself establish episodic binding.

Temporary retention and use of information for ongoing cognition and action. Its explanation must account for capacity, precision, interference, prioritization and updating. Persistent neural activity and activity-silent working memory describe candidate mechanisms or states rather than interchangeable definitions of the behavioral capacity.

Temporary maintenance and manipulation of information in service of an ongoing task. Retaining intermediate arithmetic results or integrating a sentence with earlier text can tax working memory. A simple immediate span and a task requiring storage plus processing need not measure the same demand. Improvements in structurally similar memory tests are not sufficient evidence for improved academic achievement.^[ref]

Working memory training repeatedly practices tasks requiring temporary maintenance and manipulation of information, often adjusting difficulty as performance changes. Improving a practiced task, improving a different working memory task and improving classroom learning are separate outcomes.

A study using randomization of children with low working memory compared adaptive working memory training with comparison conditions and tested outcomes beyond the practiced exercises. Gains extended to some untrained working memory measures but did not establish improved academic progress or nonverbal reasoning.^[ref] This distinction motivates controls for practice and expectations, independent outcome tasks and follow-up measurements. A higher training score alone cannot show that a general cognitive capacity has increased; the claim depends on which additional tasks improve and for how long.

Replacing no-longer-relevant maintained information with currently relevant information. Working-memory updating must be distinguished operationally from simple maintenance and from manipulation without replacement.

A task requiring comparison of a current item with one presented a specified number of positions earlier. It combines maintenance, continuous replacement, comparison and response demands.

Retention through task-dependent network states without an uninterrupted, readily measurable content-specific activity pattern. Candidate mechanisms include temporary changes in synaptic efficacy. An absent decoding result is insufficient to establish silence, and perturbation-induced reactivation does not uniquely identify the underlying mechanism.

The hypothesis that sensory systems contribute representations used during retention, reusing feature-sensitive processing rather than requiring all memory content to reside in a separate abstract store. A decodable sensory pattern supports this idea without uniquely locating its maintenance mechanism.

Neural signals observed between encoding and a later memory-guided response. They may reflect retention, priority, anticipated action, time or control. Content-specific predictions and suitable task controls are needed before interpreting them as storage.

Temporary retention of sound-based or phonologically coded information. Articulatory rehearsal can support it, but demanding sentence processing also involves representations and computations beyond rehearsal.

Repeating a novel speech-like sequence. Performance jointly depends on auditory analysis, phonological retention and speech production, so it is not a pure test of any one of these components.

Refreshing information through overt or covert articulation. Concurrent speech can interfere with this process, providing a manipulation for testing its contribution to a task; changes in general task demands must also be considered.

The fidelity of a retained representation, assessed through the distribution of errors in continuous reports or a suitable behavioral model. Loss of precision is distinct from complete loss of an item; both can influence aggregate accuracy.

The amount of information that can be retained under specified task and measurement conditions. Behavioral estimates depend on stimulus complexity, interference and the assumed response model; a physiological plateau does not alone establish a universal fixed slot count.

Memory concerning events and information related to one’s own life. Mirror-guided localization of a body mark does not by itself demonstrate this capacity.

Dating a past event as more recent than its actual date. In prospective childhood-memory work, the same reported event can acquire progressively later age estimates. Shifts relative to the first interview establish dating instability, not the absolute accuracy of that first estimate.^[ref]

Children experience a distinctive apparatus that appears to produce smaller versions of toys, and later receive verbal recall, recognition or reenactment tests. Measuring vocabulary at the original event helps separate retained content from what the child can later express verbally.^[ref]

The relative scarcity of accessible autobiographical memories from the early years of life in later childhood and adulthood. It is not equivalent to inability to form memories during childhood. Encoding, retention, retrieval, verbal expression and dating can all affect the observed boundary. Prospectively following already documented early events can reveal loss of later access independently of adults’ retrospective dating.^[ref]

The particularly sparse later autobiographical access to events from infancy. Evidence for early conditioning or recognition challenges claims of global infant memory incapacity without necessarily removing this autobiographical phenomenon.