Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Comparative Spatial Memory and Prospective Cognition

Abstract. Comparative spatial memory and prospective cognition examine how animals retain information about locations and past events, anticipate future environmental or motivational conditions, and organize present actions accordingly. This fundamental research monograph reconstructs the historical development of cognitive-map research, hippocampal specialization in food-storing birds, episodic-like what–where–when memory, and prospective caching and tool-use experiments. It integrates original behavioral interventions, comparative neuroanatomy, neural recordings, field-linked cognitive research, and critical theoretical analyses across corvids, rodents, great apes, and cuttlefish. The evidence establishes durable spatial retention, context- and time-sensitive cache decisions, future-dependent resource allocation, and neural signatures linked to individual caching events. Distinct results support different levels of inference: successful anticipatory behavior does not uniquely identify conscious future simulation, and hippocampal correlates require further causal intervention. The synthesis assesses ecological and phylogenetic comparability, measurement validity, motivational controls, alternative associative mechanisms, and replicability. A final research framework specifies tests linking spatial codes, event representations, delayed resource valuation, and prospective choices, including operational comparisons with engineered systems in the Artificial Era. This work is a critical narrative synthesis of identified literature; it reports no new experimental data and is not a systematic review or meta-analysis.

Keywords: comparative spatial memory; prospective cognition; animal cognition; food caching; cache recovery; episodic-like memory; what–where–when memory; cognitive maps; hippocampus; place cells; future planning; scrub jays; Clark's nutcracker; chickadees; ravens; mental time travel; evolutionary psychology; comparative cognition; Artificial Era

The Research Object: Remembering Places and Preparing for Later Needs

Comparative spatial memory concerns how animals encode, retain, retrieve, and use information about the locations of objects, resources, routes, and events. Prospective cognition concerns the organization of present action by anticipated future circumstances. Their intersection is scientifically tractable when an animal places food in a location, subsequently remembers its location, and adjusts storage or retrieval according to what is expected to happen later. These operations form a sequence with multiple potentially separable components: locating a resource, selecting a storage site, binding a location to a particular event, retaining that information, recognizing changed conditions, choosing when to return, and adapting behavior when access or food value changes. The field asks which components are shared across taxa, which reflect ecological specialization, and which permit flexible adjustment to novel situations. The spatial-navigation tradition of Tolman (1948) and Morris (1981) supplies one foundation; experimental analyses of cache recovery and prospective caching in corvids Clayton & Dickinson (1998), Raby et al. (2007) supply another.

Prospective cognition is a family of functions, rather than a synonym for human episodic future simulation. A learned temporal contingency can allow an animal to arrive at a resource at the correct time; a representation of a future shortage can guide the allocation of food today; a retained intention can guide a delayed action after intervening behavior. These cases may have different representational requirements. Suddendorf & Corballis (2007) distinguishes levels of foresight according to flexibility and emphasizes the possibility that richer forms depend on multiple interacting capacities. Crystal (2013) develops criteria for investigating whether a rat retains a plan for a future event, including interruption and later reactivation. In the present synthesis, an observed future-benefiting behavior is evaluated by its predictive conditions, controls, and transfer properties before claims are made about a particular internal representation.

This monograph is a critical narrative synthesis of identifiable original experiments, neurobiological studies, and conceptual and review literature. It concentrates on food-caching birds, spatial-learning rodents, and selected comparative evidence from primates and cephalopods. A cross-taxon argument requires careful control of task affordances because a seed-storing bird, a laboratory rat, and an orangutan encounter distinct ecological and motivational structures. The principal research questions concern the geometry and content of spatial representations, the temporal organization of remembering, prospective allocation under changing motivation, hippocampal involvement, comparative validity, ecological fitness implications, and the relationship between neural mechanisms and observed behavior. The review neither reports new experimental observations nor claims the systematic coverage, protocol, or pooled effect estimates of a registered systematic review or meta-analysis.

Conceptual Boundaries: Navigation, Caching, Episodic-Like Memory, and Planning

Spatial navigation refers to movement guided by information about the environment, and spatial memory refers to persistence of such information beyond immediate perception. Navigation can depend on a learned sequence of turns, local cues, path integration, landmarks, or a more flexible representation of relative spatial relations. The term cognitive map has historically organized the last possibility, but its meaning depends on which transformations an animal can perform. Tolman (1948) proposed that learned spatial relations explain behavior beyond chained stimulus-response habits; Morris (1981) operationalized place learning through transfer from novel start positions in a water maze. A cache-recovery task adds an additional requirement: the target is generated by the animal's own earlier action, allowing later search to be evaluated against the actual sites and contents of its stored resources.

Episodic-like memory designates performance that satisfies behavioral criteria such as what happened, where, and when, without inferring subjective recollection from the response alone. Clayton & Dickinson (1998) applied these criteria to scrub jays retrieving foods with different perishability. Rodent paradigms developed related tests of unique events and their temporal or contextual features Babb & Crystal (2006), Kart-Teke et al. (2006). The label is a methodological achievement: it describes the content of information used in the task while preserving the difference between publicly observed discriminations and private experiential states. Event specificity, incidental encoding, integration of attributes, and flexible use are increasingly relevant extensions of the three-feature formulation, as surveyed by Boyle & Brown (2025).

Planning for a future need requires selection of present behavior in relation to a later state of the world or organism. The strongest experiments compare conditions that differ in the future benefit while keeping current reward, training history, and immediate motivation as similar as possible. Current choice may nevertheless be shaped by cached stimulus values, associatively learned task rules, or general uncertainty-reduction procedures. Raby et al. (2007) tests anticipatory food allocation; Cheke & Clayton (2012) examines food-specific satiety and later preferences; Kabadayi & Osvath (2017) moves beyond cache-specific behavior to deferred tool use and bartering. These experiments support a spectrum of prospective performance, with different strengths of evidence about mental time travel. Such differentiation makes the scientific object more precise and allows theories to compete on discriminating predictions.

Historical Foundations of Spatial Representation

The experimental history of spatial cognition developed through investigations of mazes, place learning, and the conditions under which animals transfer learned information. Tolman (1948) articulated cognitive maps as an explanatory alternative for flexible maze navigation. The hypothesis acquired new experimental traction when O'Keefe & Dostrovsky (1971) reported hippocampal neurons in freely moving rats with location-related firing, providing a candidate neural substrate for spatial representation. This observation alone did not establish a complete map implementation; it made the relation between hippocampal activity and place a matter of direct measurement. Morris (1981) then showed rapid learning of a fixed hidden platform using distal cues and transfer from a new starting position, an influential behavioral test for relational rather than merely local place information.

Spatial-memory research in birds developed by exploiting naturally distributed caches. Field and laboratory observations showed that scatter-hoarding species must revisit numerous locations over time, often when local visual conditions have changed. Sherry et al. (1989) compared hippocampal anatomy across food-storing and non-storing passerines; Krebs et al. (1989) examined the association between food storing and hippocampal specialization. Sherry & Vaccarino (1989) went beyond correlation by testing the effect of hippocampal lesions on black-capped chickadee cache recovery. This combination of ecological observation, cross-species neuroanatomy, and causal intervention gave the program an unusually direct bridge between adaptive behavior and neurobiological mechanism.

Later work separated spatial duration, event content, and future deployment. Balda & Kamil (1992) tested Clark's nutcrackers after long delays and showed recovery above a random-search comparison even after many months. Clayton & Dickinson (1998) asked whether a bird uses information about the identity and age of different cached foods, while Emery & Clayton (2001) linked prospective cache protection to prior pilfering experience. The research focus expanded from where information is stored to how its retrieval is modulated by temporal, social, and motivational conditions. The historical sequence matters because modern debates about foresight inherit distinct evidential standards: geometric transfer, location-memory accuracy, event-specific recall, and the anticipatory selection of future-benefiting actions are not interchangeable endpoints.

Ecological Structure of Food Storage and Return

Scatter hoarding distributes resources across multiple cache sites. Its cognitive demands depend on cache density, spatial separation, seasonal resource availability, degradation rates, theft risk, and the opportunities to recover stores before competitors do. A cache is therefore both a location-memory problem and a delayed decision under uncertainty. Natural selection can favor accuracy, updating, or protective behavior without requiring the same solution in every species. Balda & Kamil (1992) provides direct experimental evidence that nutcrackers remember own-cache positions over lengthy intervals, while Sherry et al. (1989) places such performance in a broader comparative anatomical context. The ecological distribution of the tasks supplies hypotheses about the functions of memory but cannot by itself specify the neural algorithm or the content of a particular remembered episode.

Food caching also creates conflicts between present and future value. A bird that currently prefers one food may gain later from storing another; a site that is safe now may become less safe when another individual has observed the hiding action. Emery & Clayton (2001) found that experienced pilferers, after caching in view of another jay, changed cache locations when subsequently unobserved. Dally et al. (2006) demonstrated sensitivity to the identity of a former observer. These results connect memory for social context with future-oriented risk reduction. Their interpretation still requires separating memory of an observer's behavior from attribution of that observer's knowledge, and selecting among mechanistic alternatives remains an empirical rather than purely terminological question.

The comparative value of ecological tasks lies in their ability to make competence consequential and behavior repeatable. Yet ecological relevance must be balanced against control over exposure, search opportunity, and sensory guidance. Birds that hide their own food control stimulus placement, but they can also leave visual traces, reuse favored locations, or develop local search policies. Randomized sites, controlled delays, independent relocation of cues, and records of attempted but unsuccessful recoveries can separate those mechanisms. The ideal design measures both the animal's history of caching and its subsequent choices, then manipulates one factor at a time. This design logic prepares the more focused experiments discussed in later chapters.

Representational Frameworks and Their Testable Predictions

A landmark-association model proposes that the animal remembers features or directions associated with rewarded places. A metric or relational-map model predicts integration of more than one landmark into spatial relations that can guide novel approaches. A path-based model privileges accumulated movement and turns. An event-memory model binds the place to food identity, time, and the act of storage, whereas a prospective-choice model incorporates estimates of later need or availability. These models may operate together; the scientific question is when evidence necessitates each proposed representation. The transfer tests in Morris (1981) address place relations, the food-and-delay manipulation in Clayton & Dickinson (1998) addresses event content, and the future breakfast manipulations in Raby et al. (2007) address anticipation of changed availability.

Associative accounts deserve explicit experimental formulation. A discrimination acquired through prior encounters may predict where to cache or when to search without explicit simulation of tomorrow. Conversely, the existence of learning does not eliminate the possibility of sophisticated prospective representation; virtually all adaptive planning mechanisms develop from prior information. The informative comparison asks which history-dependent process can generate the observed conditional choice and whether that process predicts choices under novel combinations. Suddendorf & Corballis (2007) argues for more demanding criteria for mental time travel; Cheke & Clayton (2012) discusses how associative learning and future-oriented cognition may interact. A good experiment makes rival explanations predict different actions and includes trials designed to expose those differences.

A third explanatory level concerns implementation in neural systems. Correlation between larger hippocampal volume and accuracy at a particular spatial task can motivate a specialization hypothesis, but volumetric correlation alone cannot locate causal computations. The lesion evidence of Sherry & Vaccarino (1989) provides functional involvement, and the event-specific firing patterns reported by Chettih et al. (2024) identify a possible encoding mechanism. Neither necessarily establishes an internal phenomenology of remembering. The monograph maintains separate levels of description: ecological function, representational computations, behavior under experimental intervention, and neural realization. A strong synthesis connects them through converging tests rather than collapsing one into another.

Spatial Encoding: Landmarks, Geometry, and Relocation

Encoding a location requires a reference frame that survives the animal's departure. In a laboratory arena, this may comprise wall geometry, visual landmarks, odors, local texture, or self-motion cues. The distinction between proximal and distal cues is especially important in experiments that purport to test allocentric place memory. Morris (1981) varied the visibility and stability of the escape platform in a water maze, showing that rats could learn a fixed hidden place relative to room cues and approach it from a novel departure point. Such transfer gives stronger support to relational place learning than repeated approaches from an identical start. Nevertheless, an animal could combine several learned directions or landmarks without a single map-like representation, so additional transformations remain informative.

Caching provides an alternative to trained hidden-goal tasks. A bird chooses storage sites and later searches among them; the experimenter can record errors, first inspection, latency, site revisits, and the spatial distribution of recovery attempts. Balda & Kamil (1992) allowed nutcrackers to distribute seeds among many designated sites, then tested retention after different delays. Because the subject's own actions set the targets, the task has ecological validity and a tractable ground truth. Still, cache site preferences, exploratory search, scent traces, and displacement of substrate can influence performance unless controlled. A precisely coded first-choice measure and random-baseline model can reduce these ambiguities, while a range of spatial separations can reveal whether memory errors increase with representational similarity.

Landmark use varies with ecology, sensory capacities, and task design. A visual landmark rotated while local cache coordinates remain fixed can distinguish cue-following from room-centered search; replacing a landmark tests dependence on its identity; changing the release point tests movement-sequence accounts. An animal that succeeds in one manipulation may still fail in another because the required transformation differs. These controlled dissociations should be treated as a family of tests rather than a binary examination for possession of a cognitive map. Behavioral designs inspired by Tolman (1948) and Morris (1981) remain useful because they turn philosophical labels about maps into specified observable predictions.

Long-Term Retention and Interference in Cache Retrieval

Long-lasting spatial memory can be measured when food-storing animals are prevented from recovering their own caches until a scheduled test. In Balda & Kamil (1992), 25 Clark's nutcrackers experienced retention intervals of 11, 82, 183, or 285 days. Recovery remained better than random search at each interval, although some later-session errors increased in the longest condition. The study establishes that learned cache locations can guide behavior across a period much longer than typical short-delay laboratory memory tests. However, the randomization of birds to delay groups and the bounded laboratory setting do not by themselves demonstrate that every species member retains all caches at a comparable accuracy in a natural winter landscape.

The accuracy of recovery must be interpreted in relation to interference. A bird can store multiple foods at nearby sites, inspect already-emptied locations, or replace old caches with new ones, creating competition among memories with similar spatial addresses. Errors may reflect failure of encoding, retrieval interference, changed motivation, or a search policy that tolerates several unproductive probes. Separating these explanations calls for tests that vary the number of caches, site spacing, delay, food identities, and opportunities to revisit sites independently. Histories of attempted recovery are especially informative: a bird may remember a site but skip it because its expected value has changed. Clayton & Dickinson (1998) demonstrates that what and when can affect the decision to search in a location even when that location is remembered.

Long retention also creates a challenge for human assumptions about exact temporal representation. A response after months can be produced by durable association between place and food even if the animal has no detailed memory of the storage date. Conversely, a bird that avoids an aged perishable cache may be integrating temporal and content information without explicitly calculating elapsed clock time. Gould et al. (2012) reported accurate location performance by nutcrackers and conditional evidence for what-where-when use, with an unexplained dependence on the retention interval. Such outcomes demonstrate why one should measure spatial accuracy, temporal discrimination, and food-value expectations separately rather than combine them into a single score described as general memory.

Comparative Neuroanatomy and the Hippocampal Specialization Hypothesis

Spatially demanding ecologies motivate the prediction that natural selection will be associated with neural specializations supporting memory for places. Krebs et al. (1989) examined 52 birds representing 35 passerine species or subspecies and found a larger hippocampal complex relative to body and brain size in food-storing groups than in non-storing groups. Sherry et al. (1989) also compared the relative hippocampal complex across food-storing and non-storing passerine families. These anatomical relationships were foundational because they connected an identifiable natural behavior with a brain structure already implicated in mammalian spatial learning. The cross-species design supports an association, but related taxa share evolutionary history, habitat, diet, and other features that complicate attribution to food storing alone.

Developmental comparisons add temporal information to anatomical patterns. Healy & Krebs (1993) compared a food-storing and a non-storing corvid to investigate the emergence of hippocampal specialization. Biegler et al. (2001) linked hippocampal size to the duration of spatial memory in comparative behavioral testing. These approaches suggest that ecological specialization is related to memory-demanding behavior, while leaving open the relative contributions of inherited development, environmental experience, and task-specific strategy. A larger hippocampus is an anatomical measure, not an interchangeable index of intelligence, and allometric adjustments are central when comparing species with different brain architectures.

Causal lesions provide a stronger, though still circumscribed, test of function. In black-capped chickadees, Sherry & Vaccarino (1989) reported that bilateral hippocampal damage reduced cache recovery to chance and impaired memory for places more than an associative cue task. The intervention supports a necessary contribution of the avian hippocampal complex to this form of spatial behavior under the tested conditions. Lesions can also alter working memory, search monitoring, or broader network activity, so they do not reveal an exclusive single-purpose module. Combining selective interventions with tasks that distinguish item, location, delay, and navigation would specify more precisely which computation depends on hippocampal circuits.

Neural Codes: Place Representation and Event-Specific Signatures

The proposal that hippocampal activity represents spatial position gained early support when O'Keefe & Dostrovsky (1971) described location-correlated firing in the freely moving rat. In later comparative research, it became important to ask whether avian hippocampi represent position similarly and whether a memory for a particular cached event can be distinguished from a representation of that site's geometry. Place codes support information about where an animal is, but repeatedly caching different items in the same site introduces an identity problem. The neural system must, in some manner, distinguish episodes or update expected contents while preserving navigational information.

High-density recordings in caching chickadees described by Chettih et al. (2024) directly address this problem. During individual caching events, sparse patterns of hippocampal activity formed distinct event-associated signatures described as barcodes. Patterns reappeared briefly during later interactions with the corresponding caches and were distinguishable from conventional spatial place activity. This supplies evidence that an event marker and a general spatial code can coexist in the same circuit. An event-specific neural pattern is an experimentally observed correlate; identifying it as the mechanism of episodic recollection remains a hypothesis requiring perturbation, prediction of future retrieval, and examination across contexts. The results do not by themselves demonstrate the subjective experience of remembering.

In 2025, Payne & Aronov (2025) recorded hippocampal signals while chickadees alternated between distant visual search and locomotion. Place-related cells were activated both during presence at a location and, in some conditions, when gaze was directed toward that location from afar. The study described rapid relations between head saccades, internal prediction, and visual observation. These findings sharpen a distinction between remembering a place and being physically there: spatial codes can support remote attention to possible destinations. They also suggest an important bridge between spatial memory and prospective navigation, because a later route can be evaluated while the bird remains elsewhere. Whether the same neural operations support food-specific future needs and nonspatial planning is an additional experimental question.

What, Where, and When: The Experimental Logic of Event Memory

In Clayton & Dickinson (1998), scrub jays cached preferred perishable wax worms and less perishable peanuts at different sites. The experiment manipulated the delay before recovery, so that wax worms were desirable soon after caching but no longer worthwhile after a longer interval for birds that experienced spoilage. Those birds preferentially searched for worms at the short delay and shifted their recovery preference at the longer delay. This design showed a functional integration of food identity, cache location, and temporal information. It operationalized what-where-when memory with behavior rather than verbal reports. The result is stronger than simple place memory, because knowing a location alone cannot explain the observed food- and interval-specific searching.

What-where-when criteria have since been tested in mammals and invertebrates. Babb & Crystal (2006) used a rat paradigm that dissociated aspects of a specific earlier event, while Kart-Teke et al. (2006) tested integrated object, place, and temporal information in Wistar rats. Jozet-Alves et al. (2013) reported that cuttlefish adapted foraging to food identity, location, and time since feeding, based on replenishment contingencies. These different paradigms show that temporal discrimination linked to context is found beyond corvids. Yet differences in training, retention intervals, reward ecology, and item recognition prevent immediate equivalence between their memory representations. Evidence of integrated temporal information in one task establishes a capacity under those conditions, not a species-wide claim about autobiographical memory.

The strongest unresolved issue is the distinction between event memory and familiar rules applied to event-related cues. Delays may be represented as elapsed duration, temporal order, or change in internal or external context. An animal may select the correct resource by learned interval-dependent value without reactivating a representation of the individual caching act. Gould et al. (2012) documents a useful ambiguity in nutcrackers: accurate spatial recovery and food-specific choices occurred, but performance differed in a way that was difficult to explain solely by what-where-when competence. More demanding tests can manipulate whether time information was encoded incidentally, introduce unexpected recovery questions, and vary the remembered content after only one relevant episode. The aim is to isolate the content and flexibility of retrieval.

Prospection as Anticipatory Allocation Rather Than Immediate Reward

An anticipatory act is a present behavior whose usefulness depends on later conditions. Evidence for prospective cognition therefore depends on whether the organism's choice changes with future circumstances while immediate payoffs are held constant. This design distinguishes prospective allocation from routine caching or hoarding. It also limits the claim: a bird can optimize later food availability without constructing a vivid imagined scene of tomorrow. Suddendorf & Corballis (2007) explains why future-directed action admits mechanisms with different degrees of flexibility. Studying that continuum avoids assigning a single theoretical meaning to all delayed-benefit behavior. The decisive question is the specificity of information about a later need, opportunity, or contingency that controls the action.

The relation of prospective cognition to spatial memory can be represented as a sequence of conditioned decisions: encode where a resource is, evaluate when it is likely to be available or edible, predict what the organism will need, and select a site or action in the present. Some steps may be computed by cached values learned over many trials; others may recruit retrieval of unique past events or flexible recombination of several features. Crystal (2013) argues that rigorous prospective tests should distinguish representation of a particular future time from generalized expectancy, and should consider whether a plan survives intervening behavior. A future reward that follows an action is thus an insufficient test unless task conditions discriminate between mechanisms.

The breadth of the construct has implications for comparative inference. Cache protection by corvids tests social uncertainty, food-specific caching tests future consumption, and tool retention by apes tests deferred instrumental use. A taxonomic comparison requires explaining what each behavior demands: temporal prediction, motivation management, inhibitory control, memory for location, choice of useful means, or the capacity to act when a relevant situation has not yet arrived. Raby et al. (2007), Osvath & Osvath (2008), and Kabadayi & Osvath (2017) demonstrate distinct experimental entry points. Their convergence supports the study of prospective functions across lineages, while variations in control conditions and learning history keep the scope of particular cognitive claims open to examination.

Western Scrub-Jays and the Breakfast-Planning Paradigm

The experiments reported by Raby et al. (2007) created a setting in which western scrub jays experienced different future breakfast conditions in separate compartments. In the first paradigm, one compartment was associated with food availability in the morning and another with the absence of breakfast. When birds were subsequently allowed to cache, they preferentially deposited food in the area associated with future hunger. The manipulation was important because storage occurred before the future deprivation and the choice was directed toward a later location-specific need. Rather than infer foresight solely from the fact of caching, the experiment compared spatial allocations under experimentally learned differences in the following morning's conditions.

In the second paradigm Raby et al. (2007) assessed whether birds stored particular food types according to which food would be missing in each compartment on the following morning. The reported differential caching linked food identity, future location, and anticipated supply. The result is a conditional prediction about the value of a particular resource at a later time, not simply a tendency to cache more in general. Its significance is also methodological: place is not just a retrieval destination but an index for differing future outcomes. The evidence challenges a strong assertion that animals can respond only to current motivational needs. It is consistent with future-oriented behavior in a controlled ecological task.

The design cannot uniquely identify an imagined future experience. Birds had learned histories in the compartments, and their subsequent caching could be guided by learned associations between place, morning deprivation, and expected food type. A trial-unique manipulation, generalization to unfamiliar compartment combinations, and controls for learned policies would sharpen the representational claim. The interpretation is therefore appropriately layered: the behavioral result shows condition-sensitive prospective allocation; a model of future motivational representation offers one explanation; the existence of conscious mental time travel is a separate proposition. Suddendorf & Corballis (2007) provides the more stringent comparative standard, while Boyle & Brown (2025) emphasizes the value of distinguishing ecological function from the particular process that fulfills it.

Specific Satiety, Future Desires, and Motivational Control

A key experimental problem is whether planning can diverge from what an animal wants at the moment of choice. If present hunger determines every food decision, current preference might explain caching without anticipation of a changed future preference. Specific satiety offers a controlled manipulation: after consuming one type of food to satiety, a bird may temporarily value another. In Cheke & Clayton (2012), Eurasian jays faced circumstances in which present and later food preferences differed, and their caching behavior was evaluated against future opportunities. The study reported that birds could adjust storage to two distinct later needs despite a conflicting current motivation. This broadens the range of prospective behaviors beyond the initial breakfast paradigm.

The mechanism remains an empirical question because satiety itself is learned and may trigger associative rules about which food becomes valuable after a delay. The fact that a task permits motivational dissociation makes it more discriminating than one in which present and future preferences align, but it does not eliminate all history-dependent accounts. Cheke & Clayton (2012) explicitly discusses how associative learning and prospective cognition might operate together. A stronger test would manipulate the temporal order of prefeeding, change a prospective food's usefulness unpredictably, and examine whether animals generalize without extensive repeated reinforcement. Such designs would help distinguish a fixed food-switching rule from flexible consideration of an expected later state.

Motivational control is also not equivalent to simple self-denial. An animal might cache one food because its future marginal utility is higher, or retain a tool because carrying it has an expected payoff, even when the present preferred reward remains available. Prospective choice is best quantified as the conditional difference in action under experimentally varied future need, with immediate physiological state and task history measured independently. The same analytic logic applies to food scarcity, competitor presence, and delayed tool access. Across Raby et al. (2007), Cheke & Clayton (2012), and Kabadayi & Osvath (2017), the most informative contrast is what current behavior reveals about sensitivity to a future state that has not yet occurred.

Cache Protection, Social Memory, and Expected Pilferage

A cache that can be stolen illustrates how past social experience shapes future resource security. Emery & Clayton (2001) studied scrub jays that had previously stolen others' caches and compared their behavior with birds lacking that experience. Experienced pilferers re-cached food after being watched while storing, but did so when allowed to recover privately. This pattern links a remembered observation context to later protective action and implies that the original storage site was evaluated in terms of possible future loss. The same observed sequence can be described functionally without asserting that the cacher attributes a fully human-like propositional belief to the observer.

Subsequent studies complicated the memory demand. Dally et al. (2006) reported that scrub jays could keep track of which particular bird had watched a caching event, adjusting later re-caching accordingly. This identity-sensitive behavior indicates more than a general reaction to the mere physical presence of any conspecific during recovery. Yet an individual-specific association between observer and theft risk is a plausible alternative to explicit representation of another bird's mental state. Experiments that distinguish an observer's line of sight, prior access, and genuinely mistaken information can test more precise models. The spatial-memory mechanism must also track which caches were visible to which observer at which time.

Cache protection belongs within prospective cognition because a bird can alter present storage or later retrieval to reduce expected future loss. It also interacts with memory updating: moving a cache requires retaining a new location and revising expectations about an old one. The process is therefore cognitively richer than retrieving a static list of stored points. Clayton et al. (2005) documents sensitivity of caching behavior to recovery conditions, while Emery & Clayton (2001) and Dally et al. (2006) show how social history modifies cache management. The functional outcome is adaptive management of stored resources; which social representation underlies it must be investigated using controlled informative and uninformative observer conditions.

Beyond Caches: Apes, Ravens, and Delayed Instrumental Action

Prospective evidence outside food caching provides an important test of generality. Osvath & Osvath (2008) examined chimpanzees and orangutans on future tool-use tasks requiring selection and retention of an appropriate implement. Success requires acting before access to the later use situation and managing immediate opportunities in light of deferred value. These experiments are not identical to corvid caching tests: tool availability, transfer to a future site, and reward expectation create different possible learning routes. Their comparative significance lies in testing whether deferred useful-object selection is available outside one species' food-storage repertoire.

In Kabadayi & Osvath (2017), ravens selected objects for later tool use and bartering, with reported delays of up to seventeen hours. The tasks tested future-directed behavior in activities other than caching, addressing the hypothesis that corvid foresight is strictly a specialized response to food hoarding. The research reported flexible performance and self-control in these contexts. Redshaw et al. (2017), however, emphasized that simpler mechanisms could still explain aspects of selection, making the study an example of how productive criticism develops stronger comparative tests. Object choice may depend on discriminative training histories, learned exchange values, or prospective representation; experiments must isolate these accounts by manipulating future contingencies.

The theoretical gain of comparing apes, corvids, and rodents is a map of component capacities. One species may excel at cache location, another at selecting a future tool, and another at temporal discrimination in an arena. Successful performance in one domain does not imply an identical cognitive architecture or shared evolutionary origin. Boyle & Brown (2025) proposes approaching episodic memory through functions that may vary across species and contexts. For a broader comparative synthesis, methods should include untrained transfer, different reward types, controlled present motivation, and measures of interference between plan formation and execution. These tests can reveal whether deferred-action competence is domain-specific, more general, or built from partially shared mechanisms.

Comparative Methods: Naturalistic Observation and Controlled Intervention

Comparative inquiry benefits from integrating naturalistic and laboratory methods. Field studies establish ecological conditions: where caching occurs, how long food remains hidden, how competitors behave, and whether cache recovery influences survival. Laboratory tasks can manipulate delay, food type, obstacle, visual access, and future availability while preserving accurate records of each bird's prior behavior. Their combined value is inferential complementarity. An association between food storing and hippocampal anatomy Krebs et al. (1989) motivates causal questions, a lesion experiment Sherry & Vaccarino (1989) tests involvement of a brain region, and a prospective allocation experiment Raby et al. (2007) manipulates expected future benefit. No one of these methods establishes all the others' conclusions.

An experimental comparison must control what is visible and valuable at each phase. At encoding, the researcher must document exposure, deliberate versus incidental attention, and the opportunity to use scent or substrate traces. During retention, access to the site, rehearsal, new learning, and changes in motivation matter. At retrieval, choice sets, response order, reward contingencies, and the chance of indiscriminate search must be accounted for. Trials in which the animal chooses not to retrieve can be ambiguous between forgetting and rational non-recovery. The perishability manipulation in Clayton & Dickinson (1998) illustrates why knowing an item's location and choosing to inspect it are analytically separate.

Prospective tests add a second temporal distinction: the contingency known now and the need expected later. An animal might experience a future condition through repeated training, or may encounter a truly novel combination that requires transfer. These conditions support different inferences. Raby et al. (2007) manipulated the distribution of later breakfast opportunities; Cheke & Clayton (2012) manipulated present and anticipated food values; Osvath & Osvath (2008) examined deferred tool use. Strong replication should preserve the conceptual contrast while changing incidental cues, enclosures, and training procedures. Ecologically grounded tasks should be paired with discriminating controls rather than relying on realism alone as evidence of a cognitive mechanism.

Measurement: Search Accuracy, Temporal Binding, and Choice Models

A recovery outcome can be represented through initial site choice, proportion of correct probes, number of unrewarded inspections, latency to first correct probe, route efficiency, and recovery distribution across trials. Each measure emphasizes a different operation. First-choice accuracy isolates a decision before feedback; cumulative recovery may increase with repeated searching; latency combines memory with motivation and movement speed. In Balda & Kamil (1992), correct probes were compared against chance under a known spatial arrangement, making long-delay performance interpretable. Statistical analysis should also model clustering by individual and cache site, repeated exposure, and differences in possible number of attempts.

What-where-when experiments require separate diagnostic checks for item identity, location, and temporal information. A bird might remember a location perfectly while failing to distinguish how old the stored food is. Conversely, a food-value discrimination may arise from a general duration rule without retrieving a unique past episode. Clayton & Dickinson (1998) offers a distinctive joint manipulation of food type and delay; Gould et al. (2012) shows why surprising interactions with retention interval limit strong generalization. Useful analyses include conditional prediction matrices in which each combination of what, where, and when generates a distinct expected response under competing models. A study should not collapse these data into an unexamined composite score.

For prospective behavior, the target measure is the dependence of action on a manipulated future condition. A difference in caching between compartments can be evaluated while controlling current food availability, baseline preference, and previous compartment histories. A deferred tool choice can be scored before later reward access, and a transfer phase can test whether the chosen object remains appropriate under changed conditions. Kabadayi & Osvath (2017) and Redshaw et al. (2017) exemplify how a positive result and its mechanistic interpretation can be discussed separately. Confidence intervals, individual trajectories, pre-registered exclusion rules, and transparent handling of nonparticipants are necessary to understand the robustness of comparative claims; the present review does not infer numeric effect magnitudes from abstracts that do not report them.

Phylogenetic Inference and Species-Appropriate Comparability

A comparison between species is informative only when the chosen measures reflect comparable psychological operations. Food-caching birds have extensive experience hiding resources, whereas non-caching birds may have little reason to visit hidden, unrewarded sites. Identical apparatus can impose unequal motivational demands and therefore produce differences unrelated to memory. An ecologically standardized design aims to match task function instead of merely matching apparatus. The comparative anatomic results in Sherry et al. (1989) and Krebs et al. (1989) thus support a specialization hypothesis, but a modern reanalysis should account for phylogenetic covariance and shared environmental factors.

Phylogenetic nonindependence matters because members of related species inherit many characteristics from common ancestors. Statistical comparisons that treat every species as an independent observation can overstate precision. Environmental variables such as seasonality, spatial range, food-storing intensity, and habitat complexity may covary with lineage. Evidence for a common mechanism across birds and mammals requires more than similar performance scores. O'Keefe & Dostrovsky (1971) and Chettih et al. (2024) identify distinct neural research programs that can be compared at the level of representational functions, while homologous anatomical regions require independently supported evolutionary relationships.

The scope of the evidence is also constrained by taxonomic sampling. Corvids and chickadees are disproportionately represented in cache studies because their behavior is experimentally accessible. Jozet-Alves et al. (2013) shows that a what-where-when foraging paradigm can be adapted to cuttlefish, but the ecological task differs from bird cache recovery. Generalizing from these cases to an entire class of animals requires attention to sensory ecology, habitat, feeding strategy, age, developmental history, and training. A defensible synthesis reports which behaviors were demonstrated in which species and experimental conditions, then formulates broad evolutionary claims as hypotheses requiring wider independent sampling.

Causal Identification, Alternative Mechanisms, and Replication

The greatest interpretive hazard in prospection research is an inference from successful future-benefiting action to one privileged internal mechanism. Several candidate explanations can fit the same behavioral pattern: learned expected value, stimulus-response rules, event retrieval, flexible causal modeling, and prospective simulation. The distinguishing experiment forces these accounts to disagree. For example, changing tomorrow's food availability while preserving the local cues associated with previous rewards tests whether the animal can update a prediction. A transfer trial with unfamiliar objects or compartments can test generalization. Suddendorf & Corballis (2007) made such flexibility central to stronger mental-time-travel claims; Redshaw et al. (2017) applied mechanistic caution to ravens' future object selection.

Causal conclusions also depend on what has been randomized. In Raby et al. (2007), the experimental manipulation establishes that differential later breakfast histories influence subsequent caching behavior within the task. It does not randomize the entire animal's lifetime of learning or directly reveal phenomenal foresight. Cheke & Clayton (2012) manipulates present satiety and estimates its relation to later caching decisions, yet internal physiological state may change several appetitive and attentional processes together. Sherry & Vaccarino (1989) causally implicates the hippocampal complex through lesions, but broad neural perturbations make precise computation difficult to isolate. Each of these studies is informative precisely because its causal target can be specified.

Reproducibility requires attention to individual animals, colony-level dependence, species, environmental enrichment, trial structure, exclusions, and complete reporting of attempted responses. Many landmark studies in comparative cognition used specialized colonies and tasks, which can be difficult for other laboratories to reproduce identically. Independent conceptual replications should retain the key predicted contrast while varying secondary details. Null findings also need adequate power, suitable motivation, and successful task engagement before counting as evidence against a capacity. The narrative synthesis presented here does not claim to have independently reanalyzed raw datasets or completed a systematic replication census; instead it distinguishes results verified from accessible primary reports from general mechanisms still in need of discriminating tests.

Temporal Scales, Development, and Individual Differences

The relevant time scale varies from seconds of route selection to hours of future food availability and months of cache retention. Mechanisms that support accurate short-delay place choices may differ from those necessary to retain thousands of sites or to anticipate seasonal shortages. Balda & Kamil (1992) documents nutcracker recovery after intervals extending to 285 days, and Raby et al. (2007) examines anticipation of the following morning. An animal may be competent at one interval while using simpler mechanisms at another. Research should estimate retention functions and future-condition sensitivity across multiple delays rather than selecting a single arbitrary test interval.

Developmental experience can change the practical problem posed by a spatial task. A young food-storing bird must acquire caching routines, learn properties of the food and terrain, and experience the risks of theft. Healy & Krebs (1993) investigated development of the hippocampal complex in a food-storing and a non-storing corvid. Emery & Clayton (2001) found a behaviorally consequential distinction between birds with and without prior pilfering experience. These findings caution against treating cognitive specialization as a fixed adult species property, independent of the individual's history of opportunities and learned information. Developmental designs should include rearing environment, age, caching experience, and opportunities for social observation as candidate moderators.

Individual variation also creates opportunities to study ecological consequences. Welklin et al. (2024) linked measured spatial cognitive ability with longevity in wild food-caching chickadees, offering a field-based association between performance and a life-history outcome. This kind of evidence extends the research beyond the laboratory and raises questions about fitness-relevant mechanisms. Yet association with longer life does not on its own prove that a measured spatial skill caused survival differences. Health, territory quality, foraging opportunity, and other correlated traits may contribute. Strong future programs would integrate repeated individual cognitive measurements, longitudinal field tracking, environmental covariates, and ethically appropriate natural or experimental interventions.

Fitness, Selection, and the Ecological Function of Memory

An evolutionary explanation asks how a capacity affects resource acquisition, survival, or reproductive outcomes, and why its organization appears in particular lineages. Cache recovery can improve access to stored resources during periods when external food is scarce, while accurate updating may reduce wasted searches. The findings of Balda & Kamil (1992) show durable location knowledge in a species for which delayed recovery is ecologically important; Krebs et al. (1989) suggests a relationship between storage ecology and relative hippocampal size. These data provide converging grounds for specialization hypotheses, yet the existence of an adaptive use does not identify the historical selection pathway or prove a one-to-one adaptation for every aspect of the hippocampus.

Recent field-linked evidence extends the functional inquiry. Welklin et al. (2024) reported an association between spatial cognitive ability and longevity in wild food-caching chickadees, connecting behavioral measures with a consequential life-history outcome. The result is a powerful reason to examine cognitive variation under natural selection, and also a reminder that ecological associations must be interpreted using appropriate causal models. Birds with higher task performance may differ in territory, health, energetic reserves, and behavioral traits that also affect survival. Longitudinal designs can estimate whether a cognitive measure predicts outcomes after measured confounders, while experiments may clarify which memory operations actually alter foraging success.

As Boyle & Brown (2025) argues, episodic memory can have multiple functions across environments and species, including resource management, learning, and possibly more general planning. A capacity that originated in one ecological context may subsequently become useful in another. For the comparative spatial-memory literature, this means that food caching should be treated as a particularly informative natural assay, rather than an exhaustive account of why event memory exists. Evolutionary arguments become more rigorous when they specify the trait, its causal contribution, the selective circumstances, plausible alternatives, and the evidence needed to differentiate historical function from present-day usefulness.

Integrating Spatial Maps, Event Codes, and Prospective Policies

The central synthesis is a layered research architecture. Spatial maps or location-dependent codes supply information about places and paths. Event-specific representations differentiate a particular act of caching from another act at the same or a nearby location. Value and temporal information estimate whether a resource is useful at a later time. Control mechanisms organize choice under competing immediate and deferred goals. The classical hippocampal observations of O'Keefe & Dostrovsky (1971), what-where-when results of Clayton & Dickinson (1998), and event-specific chickadee recordings of Chettih et al. (2024) each illuminate a different layer. None alone establishes a complete model of the transition from stored event to future-benefiting action.

The integrative hypothesis predicts that a retrieval or planning decision is shaped by an interaction among location, remembered event properties, current internal state, and expected future context. An animal might accurately represent the position of a cache yet avoid retrieving it because the food is stale; a bird may remember where it stored food and still recache it because an observer saw the original action. Raby et al. (2007) adds prospective allocation by anticipated location-specific hunger, while Emery & Clayton (2001) adds remembered social risk. Testing integration requires factorial manipulations that separately vary place, content, timing, observer, and expected future availability. Neural recordings should be linked to behavior on those factorial conditions, rather than interpreted as a self-sufficient explanation.

The emerging neural findings broaden the hypothesis beyond physical presence. Payne & Aronov (2025) describes place-related hippocampal activity during remote gaze toward locations, suggesting an interface between visual sampling and spatial representation. Chettih et al. (2024) finds event-associated signatures reactivated around cache interactions. A future experimental program could ask whether neural codes predicting the value of a specific cache are reinstated when a bird looks toward it, before the bird decides to visit. Such a study would need causal intervention, behavioral choice measures, and controls for visual attention. The proposal is a testable synthesis of existing findings, not an already established single neural circuit for mental time travel.

Artificial Era: Computational Models and Comparative Restraint

Computational systems offer tools for stating hypotheses about representation and control with greater precision. A model can retain a graph of places, bind an event identity to a coordinate and time, estimate changing food value, and select a policy that maximizes expected reward under delayed access. Simulations of this kind clarify which behavioral signatures arise from learned values and which require episodic retrieval or flexible construction of future situations. Boyle & Brown (2025) notes the relevance of learning-oriented computational perspectives to understanding memory's functions. Computational reproducibility helps compare hypotheses, provided that modeling assumptions are disclosed and model success is evaluated against the same intervention conditions used in animal experiments.

The Artificial Era creates an additional comparative setting in which nonbiological systems can be examined for spatial representation, persistent memory, deferred-goal management, and environmental control. A robot that stores an object location and recovers it after a delay exhibits observable task performance; a model that updates expected availability after a context change can be evaluated for generalization. Such tests describe functional organization and system reliability. Their interpretation remains tied to the operational definition of the measured capacity. A computational implementation may illuminate animal cognition by demonstrating the sufficiency of a proposed algorithm for a task, while alternative implementations can generate similar behavior with different internal processes.

A meaningful comparison among Homo, nonhuman biological species, and Artificial systems must preserve differences in embodiment, evolutionary history, developmental learning, and the measurement of internal states. For animals, food scarcity and social competition supply biological stakes; for an engineered agent, an externally specified objective may serve an analogous task variable with a different origin. This analytic framework links Psychology of Artificial with comparative cognition only where the mechanisms can be operationally compared. The present monograph makes no inference about artificial subjective experience from performance and no claim that a digital memory structure is biologically homologous to a hippocampus. It identifies concrete research questions about enduring representations, prospective policy selection, and transfer across contexts.

Unresolved Questions and a Cumulative Research Program

The first open problem is representational specificity. Which tasks require retrieval of a unique past event, and which can be solved through stable place values or temporal associations? The second is the scope of flexibility: can an animal select an untrained action for a novel future condition, and can it reactivate that plan after a distracting interval? The third concerns neural causation: do event-specific hippocampal patterns help drive later site choice, or merely accompany the behavioral event? Crystal (2013) proposes more exacting criteria for future plans, and Chettih et al. (2024) suggests neural signatures suitable for intervention-based testing. Together they define questions that can be translated into discriminating experimental predictions.

The fourth problem is cross-species calibration. Performance should be compared across tasks that satisfy similar functional demands while respecting species-specific perception, movement, motivation, and experience. It is useful to pair a natural caching assay with a non-caching prospective task when possible, then examine transfer. Kabadayi & Osvath (2017) tested ravens beyond hoarding, while Redshaw et al. (2017) pressed for mechanistic alternatives to be evaluated. The fifth problem is ecological validity over time: long-duration field studies should connect measured memory competence to foraging outcomes, survival, and variation in habitat, while accounting for changing opportunity and selection bias. Welklin et al. (2024) provides a recent starting point.

A cumulative program should preregister the central mechanism-discriminating predictions, report all exclusions and individual outcomes, publish task configurations and analysis code where ethically feasible, and encourage independent conceptual replications. Neural and behavioral results need common event timestamps and transparent alignment methods; species comparisons need phylogenetic reasoning; longitudinal claims need environmental and life-history covariates. The synthesis also recommends reporting distinct outcomes for spatial accuracy, episodic-like feature binding, future-contingency sensitivity, and deferred action. These principles move the field toward a general explanation of when remembering places contributes to flexible preparation for the future and when apparently similar behavior is generated through alternative mechanisms.

Conclusion

Comparative spatial memory and prospective cognition form a connected scientific field because the survival value of remembered locations often depends on actions taken before future needs become immediate. The empirical tradition spans cognitive maps and rat place learning Tolman (1948), Morris (1981), avian cache recovery and hippocampal specialization Sherry & Vaccarino (1989), Krebs et al. (1989), long-term retention Balda & Kamil (1992), integration of what-where-when information Clayton & Dickinson (1998), and the anticipatory allocation of food Raby et al. (2007), Cheke & Clayton (2012). Studies of apes and ravens extend prospective testing beyond caching Osvath & Osvath (2008), Kabadayi & Osvath (2017). Together, these results establish that complex information about places, events, delayed consequences, and changing future circumstances can organize behavior across several animal lineages.

The evidential interpretation is necessarily specific. Successful recovery demonstrates knowledge sufficient to guide a search; integrated what-where-when choice supports behavioral event-memory criteria; conditional storage demonstrates sensitivity to anticipated future circumstances; neural event signatures suggest candidate implementation mechanisms. None of these observations by itself identifies conscious mental time travel, and no single associative model explains all observed patterns without confronting the interventions used in each experiment. The most powerful research strategy combines ecological behavior, carefully controlled counterfactual tests, quantitative comparison, and causal neural methods. The newer chickadee findings of Chettih et al. (2024) and Payne & Aronov (2025) strengthen the bridge between spatial representation and individual event coding while leaving prospective integration open for further study.

The field's mature conclusion is a structured account of multiple related capacities: place encoding, durable retrieval, temporal and content binding, adaptive resource valuation, and delayed decision-making. Their relationships can be evaluated by explicit experimental predictions and increasingly direct observations of brain activity. In the Artificial Era, this approach also supplies operational tests for engineered spatial-memory and planning systems without importing unwarranted assumptions about evolutionary homology or phenomenology. Comparative psychology advances when it describes exactly which information a system retains, which future circumstances influence its decisions, and what evidence discriminates among possible mechanisms. The resulting research architecture connects ecological function, psychological computation, and neural implementation through questions that remain experimentally answerable.

References

  • Babb, S. J., & Crystal, J. D. (2006). Episodic-like memory in the rat. Current Biology, 16(13), 1317–1321. https://doi.org/10.1016/j.cub.2006.05.025
  • Balda, R. P., & Kamil, A. C. (1992). Long-term spatial memory in Clark's nutcracker, Nucifraga columbiana. Animal Behaviour, 44(4), 761–769. https://doi.org/10.1016/S0003-3472(05)80302-1
  • Biegler, R., McGregor, A., Krebs, J. R., & Healy, S. D. (2001). A larger hippocampus is associated with longer-lasting spatial memory. Proceedings of the National Academy of Sciences, 98(12), 6941–6944. https://doi.org/10.1073/pnas.121034798
  • Boyle, A., & Brown, S. A. B. (2025). Why might animals remember? A functional framework for episodic memory research in comparative psychology. Learning & Behavior, 53(1), 14–30. https://doi.org/10.3758/s13420-024-00645-0
  • Cheke, L. G., & Clayton, N. S. (2012). Eurasian jays (Garrulus glandarius) overcome their current desires to anticipate two distinct future needs and plan for them appropriately. Biology Letters, 8(2), 171–175. https://doi.org/10.1098/rsbl.2011.0909
  • Chettih, S. N., Mackevicius, E. L., Hale, S., & Aronov, D. (2024). Barcoding of episodic memories in the hippocampus of a food-caching bird. Cell, 187(8), 1922–1935.e20. https://doi.org/10.1016/j.cell.2024.02.032
  • Clayton, N. S., & Dickinson, A. (1998). Episodic-like memory during cache recovery by scrub jays. Nature, 395, 272–274. https://doi.org/10.1038/26216
  • Clayton, N. S., Dally, J., Gilbert, J., & Dickinson, A. (2005). Food caching by western scrub-jays (Aphelocoma californica) is sensitive to the conditions at recovery. Journal of Experimental Psychology: Animal Behavior Processes, 31(2), 115–124. https://doi.org/10.1037/0097-7403.31.2.115
  • Crystal, J. D. (2013). Remembering the past and planning for the future in rats. Behavioural Processes, 93, 39–49. https://doi.org/10.1016/j.beproc.2012.11.014
  • Dally, J. M., Emery, N. J., & Clayton, N. S. (2006). Food-caching western scrub-jays keep track of who was watching when. Science, 312(5780), 1662–1665. https://doi.org/10.1126/science.1126539
  • Emery, N. J., & Clayton, N. S. (2001). Effects of experience and social context on prospective caching strategies by scrub jays. Nature, 414, 443–446. https://doi.org/10.1038/35106560
  • Gould, K. L., Ort, A. J., & Kamil, A. C. (2012). Do Clark's nutcrackers demonstrate what-where-when memory on a cache-recovery task? Animal Cognition, 15(1), 37–44. https://doi.org/10.1007/s10071-011-0429-y
  • Healy, S. D., & Krebs, J. R. (1993). Development of hippocampal specialisation in a food-storing bird. Behavioural Brain Research, 53(1–2), 127–131. https://doi.org/10.1016/S0166-4328(05)80272-4
  • Jozet-Alves, C., Bertin, M., & Clayton, N. S. (2013). Evidence of episodic-like memory in cuttlefish. Current Biology, 23(23), R1033–R1035. https://doi.org/10.1016/j.cub.2013.10.021
  • Kabadayi, C., & Osvath, M. (2017). Ravens parallel great apes in flexible planning for tool-use and bartering. Science, 357(6347), 202–204. https://doi.org/10.1126/science.aam8138
  • Kart-Teke, E., De Souza Silva, M. A., Huston, J. P., & Dere, E. (2006). Wistar rats show episodic-like memory for unique experiences. Neurobiology of Learning and Memory, 85(2), 173–182. https://doi.org/10.1016/j.nlm.2005.10.002
  • Krebs, J. R., Sherry, D. F., Healy, S. D., Perry, V. H., & Vaccarino, A. L. (1989). Hippocampal specialization of food-storing birds. Proceedings of the National Academy of Sciences, 86(4), 1388–1392. https://doi.org/10.1073/pnas.86.4.1388
  • Morris, R. G. M. (1981). Spatial localization does not require the presence of local cues. Learning and Motivation, 12(2), 239–260. https://doi.org/10.1016/0023-9690(81)90020-5
  • O'Keefe, J., & Dostrovsky, J. (1971). The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat. Brain Research, 34(1), 171–175. https://doi.org/10.1016/0006-8993(71)90358-1
  • Osvath, M., & Osvath, H. (2008). Chimpanzee (Pan troglodytes) and orangutan (Pongo abelii) forethought: Self-control and pre-experience in the face of future tool use. Animal Cognition, 11(4), 661–674. https://doi.org/10.1007/s10071-008-0157-0
  • Payne, H. L., & Aronov, D. (2025). Remote activation of place codes by gaze in a highly visual animal. Nature, 643, 1037–1043. https://doi.org/10.1038/s41586-025-09101-z
  • Raby, C. R., Alexis, D. M., Dickinson, A., & Clayton, N. S. (2007). Planning for the future by western scrub-jays. Nature, 445, 919–921. https://doi.org/10.1038/nature05575
  • Redshaw, J., Taylor, A. H., & Suddendorf, T. (2017). Flexible planning in ravens? Trends in Cognitive Sciences, 21(11), 821–822. https://doi.org/10.1016/j.tics.2017.09.001
  • Sherry, D. F., & Vaccarino, A. L. (1989). Hippocampus and memory for food caches in black-capped chickadees. Behavioral Neuroscience, 103(2), 308–318. https://doi.org/10.1037/0735-7044.103.2.308
  • Sherry, D. F., Vaccarino, A. L., Buckenham, K., & Herz, R. S. (1989). The hippocampal complex of food-storing birds. Brain, Behavior and Evolution, 34(5), 308–317. https://doi.org/10.1159/000116516
  • Suddendorf, T., & Corballis, M. C. (2007). The evolution of foresight: What is mental time travel, and is it unique to humans? Behavioral and Brain Sciences, 30(3), 299–313. https://doi.org/10.1017/S0140525X07001975
  • Tolman, E. C. (1948). Cognitive maps in rats and men. Psychological Review, 55(4), 189–208. https://doi.org/10.1037/h0061626
  • Welklin, J. F., Sonnenberg, B. R., Branch, C. L., Heinen, V. K., Pitera, A. M., Benedict, L. M., Whitenack, L. E., Bridge, E. S., & Pravosudov, V. V. (2024). Spatial cognitive ability is associated with longevity in food-caching chickadees. Science, 385(6713), 1111–1115. https://doi.org/10.1126/science.adn5633