Cues

Abstract

A cue, which MeSH classifies under learning, is any stimulus that guides behavior or gains access to stored information; in the study of memory a retrieval cue is the prompt that recovers a target from storage. Its power is relational, not intrinsic: the encoding specificity principle holds that a cue aids retrieval only to the extent its features were encoded with the target, so a strong associate can fail where a weak one that accompanied study succeeds. The same construct recurs under other names, from the reinstated environment of context-dependent memory to the predictive signal that orients attention and the conditioned stimulus that triggers relapse, each resting on the logic that a partial pattern reactivates a fuller one. Three interactive demonstrations let the reader vary cue-to-trace overlap, reinstate an encoding context, and orient attention with a predictive spatial cue.

Keywords: cue, retrieval, encoding specificity, context-dependent memory, engram

Cognitive psychology uses the word cue for a family of stimuli that share one function: a partial input recovers or directs a larger response. The retrieval cue that recovers a name, the smell that returns a childhood scene, the arrow that speeds a visual judgment, and the sight of a bar that provokes a craving are all cues in this technical sense. What unites them is that none carries the full response within itself; each instead makes contact with a representation already stored or prepared, and the quality of that contact—not the salience of the cue alone—determines the outcome. The encoding specificity principle formalizes this for memory, holding that a retrieval cue is effective only insofar as the target was encoded in its presence (Tulving & Thomson, 1973). Because cues are the point of access to stored knowledge, understanding them clarifies why information that is demonstrably available can nonetheless be momentarily inaccessible (Tulving & Pearlstone, 1966).

What a Cue Is

In the broadest sense a cue is a signal that occasions a response. The technical usage in cognitive psychology narrows this to stimuli that operate by contact with a stored or prepared representation rather than by carrying the response themselves. A retrieval cue does not contain the target; it addresses a memory that already exists. A spatial cue does not perform the perceptual judgment; it biases where attention is allocated before the target appears (Posner, 1980). A conditioned cue does not produce a response through any intrinsic property; it does so through a learned relationship to an outcome (Bouton, 2004).

This shared structure explains why a single English word covers so much ground. The defining property is relational: a cue is effective in proportion to what it has come to stand for, not in proportion to its physical intensity. A faint odor can be a powerful retrieval cue while a loud tone is a poor one, if the odor was bound into an experience and the tone was not. The consequence, developed in the sections that follow, is that the memory system's limits are frequently limits of access rather than of storage—a distinction that the study of cues brought into focus (Tulving & Pearlstone, 1966).

Retrieval Cues and Encoding Specificity

The central principle governing retrieval cues is encoding specificity: the features present when a target is encoded determine which later cues can recover it (Tulving & Thomson, 1973). A cue that was processed alongside the target at study is effective at test; a cue that is semantically related to the target but was not part of the encoding episode may fail, even when it seems obviously helpful. Tulving and Thomson demonstrated the counterintuitive form of this: a strong associate presented at test could be a worse cue than a weak associate that had actually accompanied the word during study.

The principle reframes the classic distinction between availability and accessibility. Tulving and Pearlstone (1966) had shown that participants who could not freely recall list words often produced them immediately when given category labels as cues—the words were available in storage all along but not accessible without the right prompt. Encoding specificity supplies the rule for which prompt will work: the one whose features overlap with the encoded trace.

A corollary is the cue-overload principle. A cue's power is diluted as the number of targets it subsumes grows; a category label that points to two studied words recovers them more reliably than the same label pointing to twenty. The most striking evidence is recognition failure of recallable words: under encoding specificity, a target can be recalled to a copy-cue that matches its encoding yet fail a recognition test that presents the target itself in a mismatched context, because recognition, too, depends on cue–trace overlap rather than on the target's mere presence (Watkins & Tulving, 1975).

Context-Dependent Memory

If encoding binds incidental features of the environment to the target, then the environment itself becomes a retrieval cue, and reinstating it should aid recall. The best-known demonstration is the underwater experiment of Godden and Baddeley (1975): divers who learned word lists on land or submerged recalled more when tested in the same environment than in the other, a crossover interaction that isolates context as the operative cue rather than any main effect of place.

Context-dependent memory is reliable but bounded. The meta-analysis of Smith and Vela (2001) established that environmental context effects are genuine yet moderate, and that they shrink or vanish when the material is meaningful enough to generate its own internal cues, or when participants mentally reinstate the study context at test even in a new place. This mental reinstatement result is practically important: it implies that the benefit is carried by the cue relationship, not by physical presence, so imagining the encoding context can substitute for returning to it. The bounded, moderated character of the effect is itself evidence for the cue account: context helps precisely when it was the most diagnostic prompt available and other cues were weak.

Cues in Attention

The attentional sense of cue concerns not what is retrieved but where processing is directed. In the spatial cueing paradigm of Posner (1980), a cue that predicts the location of an upcoming target speeds responses when it is valid and slows them when it is invalid, relative to a neutral baseline. The difference—the validity effect—indexes the covert orienting of attention in advance of the target, without any movement of the eyes.

The paradigm decomposes the cue's influence into a benefit (faster responding at the cued location) and a cost (slower responding when attention must be redirected from an invalidly cued location). This cost–benefit structure shows that a predictive cue commits limited processing resources ahead of the stimulus, and that the commitment has consequences when the prediction is wrong. The attentional cue thus differs from the retrieval cue in its object—a location or feature to be prioritized rather than a trace to be recovered—while preserving the defining relational logic: the cue matters because of what it predicts, not because of its own salience.

Cues in Learning and Motivation

In associative learning a cue is a stimulus that has acquired significance through its relationship to an outcome. A conditioned stimulus signals a biologically important event and comes to elicit anticipatory responses; a discriminative stimulus signals when a response will be reinforced. Bouton (2004) showed that these cue–outcome relationships are not erased by extinction but inhibited, so that the original response renews when the contextual cues change—direct evidence that context operates as a cue that gates whether a learned association is expressed.

The clinical form of this principle is cue-reactivity. Stimuli repeatedly paired with drug use—paraphernalia, places, moods—come to elicit craving and physiological arousal, and the meta-analysis of Carter and Tiffany (1999) established cue-reactivity as a robust, measurable phenomenon across substances. Because these effects are cue-driven, they are context-dependent in exactly Bouton's sense: extinguishing a craving response in the clinic does not remove it in the environments where the cues were learned, which is one reason cue-exposure treatments transfer poorly and relapse follows a return to old settings. The motivational cue and the retrieval cue are the same construct viewed through behavior: a partial stimulus reactivating a fuller learned pattern.

Neural Basis of Cue-Driven Retrieval

At the neural level a retrieval cue works by pattern completion: a fragment of the original input reactivates the distributed neuronal ensemble—the engram—that was potentiated during encoding, reconstructing the fuller pattern (Josselyn & Tonegawa, 2020). Optogenetic studies that label and later reactivate the neurons active during learning show that artificially driving the ensemble can substitute for the natural cue, expressing the memory in its absence, while silencing the ensemble blocks cue-driven recall. The engram is the physical referent of encoding specificity: the cue must contact the ensemble that encoding created.

Retrieval is not a passive readout. Frankland, Josselyn, and Köhler (2019) describe retrieval as an active, hippocampally and prefrontally coordinated reconstruction in which the cue initiates a competitive process among candidate ensembles, and the same act of retrieval can modify the trace it recovers. Ryan and Frankland (2022) extend this to forgetting, arguing that much apparent forgetting is a reversible failure of the cue to access an engram that remains physically intact—an adaptive change in retrievability rather than erasure. This returns the neuroscience to Tulving and Pearlstone's distinction: availability is a property of the engram, accessibility a property of the cue's contact with it.

Table 1

Worked Example

Consider a simple feature-overlap model of encoding specificity that the first demonstration computes exactly. A target is encoded together with a set of n contextual features. At test, a retrieval cue reinstates some of those features (matches) and may add others that were never encoded (intrusions, which do not help). Let the cue effectiveness be the proportion of encoded features the cue reinstates, E = m / n, where m is the number of matching features. Model the recall probability as P = P_max × E, with a ceiling P_max = 0.90 reflecting the fact that a perfectly matched cue still cannot exceed the availability of the trace.

Take a target encoded with n = 4 features (say, cold, blue, quiet, evening). A copy cue that reinstates all four gives E = 4/4 = 1.00 and P = 0.90 × 1.00 = 0.90. A partial cue reinstating two features (cold, blue) gives E = 2/4 = 0.50 and P = 0.90 × 0.50 = 0.45. A semantically related but non-encoded cue (warm, red) reinstates none of the encoded features, so E = 0/4 = 0 and P = 0, even though the cue feels related—the counterintuitive result Tulving and Thomson reported.

The cue-overload principle appears when one cue subsumes several targets. If a single category cue was encoded with k targets and can effectively address only a fixed retrieval capacity c = 2 items, then the per-target recovery probability scales as min(1, c/k). For k = 2 targets the cue recovers both (2/2 = 1.00); for k = 8 targets the same cue recovers on average only 2/8 = 0.25 of them. Doubling the load from 4 to 8 halves per-item recall from 0.50 to 0.25—the quantitative signature of an overloaded cue, and the reason a distinctive cue that points at one target beats a generic cue that points at many.

Discussion

The value of treating cue as a single construct is that it exposes a common mechanism beneath phenomena that are usually taught in separate chapters. Encoding specificity, context-dependent memory, attentional orienting, conditioned responding, and cue-reactivity are not five unrelated facts; they are five settings of one relationship, in which a partial pattern makes contact with a stored or prepared fuller one and the quality of that contact governs the result. This is why the same experimental logic—manipulate the match between a prompt and what was encoded, hold everything else constant—recurs across the domains, and why the crossover interaction (better performance when prompt and encoding correspond, regardless of the absolute quality of either) is the shared fingerprint.

The construct also reorganizes how memory failure is understood. The intuitive model treats forgetting as loss from storage; the cue account treats a large share of it as a failure of access to a trace that remains available (Tulving & Pearlstone, 1966; Ryan & Frankland, 2022). The practical payoff is direct: performance can often be improved by engineering better cues—reinstating context, generating distinctive encoding features, or matching study and test conditions—rather than by adding storage. It also warns against a symmetric error in the clinic, where cue-reactivity means that a response extinguished in one context is not gone but merely inhibited, ready to renew when its cues return (Bouton, 2004).

Current Directions

The most active current work on cues is at the neural level, where the engram framework has made the abstract notion of cue–trace contact physically testable. Optogenetic labeling of the ensembles active during encoding, and their selective reactivation or silencing, now allows researchers to substitute an artificial cue for a natural one and to ask directly whether a memory that cannot be recalled is absent or merely inaccessible (Josselyn & Tonegawa, 2020). A convergent line reframes forgetting itself as regulated retrievability: Ryan and Frankland (2022) argue that engram cells can become transiently inaccessible to natural cues while remaining intact, so that forgetting is often an adaptive, potentially reversible change in cue access rather than the destruction of a trace. This reconceptualization, still contested, would unify the behavioral cue literature with cellular neuroscience: the encoding specificity principle and the availability–accessibility distinction become statements about which cues can reach a given ensemble. Open questions concern how the brain arbitrates among competing ensembles that a single cue partially matches, and how retrieval-induced modification of the trace (Frankland et al., 2019) interacts with the reversibility of forgetting.

Common Misconceptions

A good retrieval cue is one that is strongly related to the target.

Relatedness in the abstract is not what matters; overlap with the encoding episode is. A strong pre-existing associate can be a worse cue than a weak one that happened to be processed with the target at study (Tulving & Thomson, 1973). The belief persists because most helpful cues are also related, so the confound is rarely exposed outside the laboratory.

Forgetting means the information is gone from memory.

Much forgetting is a failure of access, not of storage: items that cannot be freely recalled are often produced immediately when the right cue is supplied (Tulving & Pearlstone, 1966), and current neuroscience treats a large share of forgetting as reversible loss of cue access to an intact engram (Ryan & Frankland, 2022).

If a craving is extinguished in treatment, the cue no longer triggers it.

Extinction inhibits a cue–outcome association within its training context but does not erase it, so the response renews when the original cues and settings return (Bouton, 2004). This is why cue-reactivity predicts relapse and why context matters to the durability of cue-exposure therapy (Carter & Tiffany, 1999).

Glossary

Accessibility.

The retrievability of a stored item at a given moment, determined by the cues currently available; distinct from whether the item exists in storage.

Availability.

The presence of an item in memory storage, whether or not it can currently be retrieved; a property of the trace rather than of the cue.

Conditioned stimulus.

A cue that, through learned pairing with a biologically significant event, comes to elicit an anticipatory response.

Context-dependent memory.

Superior retrieval when the environment or internal state at test matches that at encoding, because context features serve as retrieval cues.

Cue-overload principle.

The dilution of a cue's retrieval power as the number of targets associated with it increases; a distinctive cue subsuming few items outperforms a generic one subsuming many.

Cue-reactivity.

Craving and physiological arousal elicited by stimuli previously paired with drug use or reward; a robust, measurable determinant of relapse.

Cue.

Any stimulus, internal or external, that guides behavior or recovers stored information by making contact with a prepared or stored representation.

Discriminative stimulus.

A cue signaling that a particular response will be reinforced, thereby setting the occasion for that response in operant learning.

Encoding specificity principle.

The rule that a retrieval cue is effective only to the degree its features were encoded together with the target.

Engram.

The physical memory trace—a distributed ensemble of neurons potentiated during encoding—that a retrieval cue must contact to recover a memory.

Pattern completion.

The neural process by which a partial input reactivates the full stored ensemble, the mechanistic basis of cue-driven retrieval.

Renewal.

The return of an extinguished conditioned response when the contextual cues change, showing that extinction inhibits rather than erases the original association.

Retrieval cue.

A prompt that recovers a target from storage; its effectiveness is set by overlap with the encoded trace.

Validity effect.

In spatial cueing, the response-time advantage for validly cued locations over invalidly cued ones, indexing the covert orienting of attention.

Key Researchers

Alan D. Baddeley (b. 1934). Emeritus Professor of Psychology at the University of York; with Godden (1975) demonstrated context-dependent memory in the underwater diving experiment, establishing the reinstated environment as a retrieval cue. ORCID - Wikipedia - Google Scholar

Mark E. Bouton. Robert B. Lawson Green and Gold Professor of Psychology Emeritus at the University of Vermont; showed that extinction is context-dependent, so contextual cues gate whether a learned association is expressed and drive renewal and relapse. Faculty Page - Google Scholar

Sheena A. Josselyn. Senior Scientist at the Hospital for Sick Children and Professor at the University of Toronto; established how a retrieval cue reactivates the specific neuronal ensemble—the engram—that encoded an experience. ORCID - Wikipedia - Google Scholar

Michael I. Posner (b. 1936). Emeritus Professor of Psychology at the University of Oregon; devised the spatial cueing paradigm showing that a predictive cue orients covert attention in advance of a target. Wikipedia - Wikidata

Steven M. Smith. Professor of Psychological and Brain Sciences at Texas A&M University; reviewed and meta-analyzed environmental context-dependent memory, defining when reinstating physical context cues aids recall and when mental reinstatement suffices. Faculty Page - Google Scholar

Endel Tulving (1927-2023). University Professor Emeritus at the University of Toronto; formulated the encoding specificity principle and the availability-versus-accessibility distinction that anchor the modern theory of retrieval cues. Wikipedia - Wikidata - Google Scholar

Frequently Asked Questions

What is a cue in psychology?
A cue is any stimulus, internal or external, that guides behavior or recovers stored information by making contact with a prepared or stored representation rather than carrying the response itself (Tulving & Thomson, 1973).

What is the encoding specificity principle?
It is the rule that a retrieval cue is effective only to the extent that its features were encoded together with the target, so a prompt present at study outperforms one that is merely related to the item (Tulving & Thomson, 1973).

What is the difference between availability and accessibility?
Availability is whether an item exists in storage; accessibility is whether it can be retrieved right now given the available cues, and items that are available can be temporarily inaccessible until the right cue is supplied (Tulving & Pearlstone, 1966).

What is context-dependent memory?
It is the finding that recall improves when the environment or internal state at test matches that at encoding, because contextual features act as retrieval cues, as in the land-versus-underwater diving experiment (Godden & Baddeley, 1975).

Does context always improve memory?
No; environmental context effects are genuine but moderate, and they shrink when material generates strong internal cues or when a person mentally reinstates the study context at test even in a new place (Smith & Vela, 2001).

How do cues direct attention?
A predictive spatial cue orients attention covertly before a target appears, speeding responses at validly cued locations and slowing them at invalidly cued ones, a pattern called the validity effect (Posner, 1980).

Why do drug-related cues trigger relapse?
Stimuli paired with drug use acquire the power to elicit craving and arousal, and because these cue-outcome associations are inhibited rather than erased by treatment, they renew when the original cues and contexts return (Carter & Tiffany, 1999; Bouton, 2004).

How does a cue recover a memory in the brain?
A retrieval cue works by pattern completion: a fragment of the original input reactivates the distributed neuronal ensemble, or engram, that was potentiated during encoding, reconstructing the fuller memory (Josselyn & Tonegawa, 2020).

References