The Episodic-Like Memory Experiment (Scrub Jays) – Nicola Clayton and Anthony Dickinson

For the better part of the twentieth century, human uniqueness in the cognitive domain was anchored in an unyielding philosophical and psychological dogma: the capacity to consciously mentally travel back in time to re-experience unique autobiographical episodes belonged exclusively to Homo sapiens. This view, codified by cognitive psychologist Endel Tulving in his pioneering work on human memory taxonomies, posited that non-human animals were perpetually trapped in an eternal present—bound by immediate sensory inputs, conditioned reflexes, and generalized semantic knowledge, but utterly devoid of the phenomenological machinery required to revisit their individual, temporally indexed pasts. Animals might remember what an object was or where to find sustenance, yet the subjective contextual richness of when that specific event unfolded was deemed biologically inaccessible to non-linguistic minds.

This long-standing paradigm suffered an empirical earthquake in 1998 with the publication of a landmark paper in Nature by behavioral ethologist Nicola S. Clayton and experimental psychologist Anthony Dickinson. Working with the western scrub-jay (Aphelocoma californica)—a food-storing corvid possessing remarkable spatial acuity—Clayton and Dickinson demonstrated that these birds could integrate memories of what food item was hidden, where it was stored, and how long ago the caching event occurred, systematically adjusting their recovery strategies according to the physical decay rates of perishable food items. Rather than relying on simple associative heuristics or olfactory signals, the jays demonstrated an integrated, trial-unique memory of historical events that fulfilled the behavioral triad of episodic memory.

Because cognitive science could not definitively interrogate the inner, subjective phenomenological qualia of an avian mind, Clayton and Dickinson cautiously christened this ability episodic-like memory. This semantic and operational concession did not blunt the paradigm’s revolutionary impact; instead, it dismantled the assumption that episodic memory was inextricably tied to the human neocortex and human language. In doing so, it inaugurated a sweeping renaissance in comparative psychology, neuroethology, and evolutionary biology, establishing an objective empirical methodology to probe the deep temporal cognition of diverse species across the phylogenetic tree.

1. Historical and Theoretical Foundations of Episodic Memory

1.1 Endel Tulving’s Definition and the Anthropocentric View

The contemporary scientific taxonomy of memory was formally articulated in 1972 when Endel Tulving introduced his foundational dichotomy separating semantic memory from episodic memory. Semantic memory, in Tulving’s initial formulation, encompasses generalized, context-free knowledge about the world—facts, vocabularies, concepts, and rules that exist independently of the personal circumstances surrounding their acquisition. One can know that Paris is the capital of France or that an apple is edible without recalling the discrete, temporally bound moment that information was first encoded. In stark contrast, episodic memory refers to an autobiographical memory system dedicated to the storage and conscious retrieval of personally experienced events situated in specific spatiotemporal contexts. To remember an episodic event is to resurrect not merely an isolated datum, but the holistic theater of an experience: the ambient setting, the sequence of occurrences, and the emotional resonance of that singular encounter.

Crucially, Tulving and many of his contemporaries asserted that episodic memory represents a uniquely human evolutionary adaptation, emerging late in hominin phylogeny alongside complex syntactic language, symbolic culture, and prolonged prospective planning. Tulving argued that episodic retrieval is fundamentally dependent upon autonoetic consciousness—a specific mode of subjective awareness that allows the remembering organism to consciously project itself into the past and re-experience the event as an active, self-aware observer. Under this anthropocentric view, non-human animals were categorized as cognitively restricted to procedural memory (motor habits and sensorimotor skills) and anoetic or noetic semantic systems (unconscious habits or conscious factual associations lacking personal self-projection). For decades, animal memory was dismissed as nothing more than the gradual accretion of associative strengths, reinforcement histories, and conditioning paradigms governed by Thorndikian or Skinnerian principles.

This theoretical stance created an impenetrable epistemological boundary around the animal kingdom. Because non-human animals lack natural human language, they cannot verbally testify to their internal phenomenological states or declare, “I am remembering the event that occurred yesterday.” In the absence of linguistic self-reporting, mainstream cognitive psychology historically conflated the inability to verbally declare an inner state with the absolute absence of the cognitive capacity itself. Animal memory experiments throughout the mid-twentieth century were overwhelmingly designed within behaviorist frameworks—T-mazes, Skinner boxes, and delayed matching-to-sample paradigms—which inadvertently reinforced the circular assumption that non-human creatures operated solely via mechanistic stimulus-response links, blind to the temporal architecture of their lives.

1.2 The Tripartite Model of Episodic Retrieval: What, Where, and When

To overcome the methodological impasse imposed by linguistic definitions of autobiographical memory, comparative researchers realized that episodic retrieval had to be operationalized strictly through observable behavioral metrics. The theoretical breakthrough arose from unpacking the fundamental structural architecture of an episodic memory trace. At its core, an episodic memory is not an amorphous sensory impression; it is an integrated, multidimensional cognitive record consisting of a behavioral triad: the identity of the event or entity (what), its spatial coordinate within the environment (where), and its placement along an elapsed temporal trajectory or explicit historical timeline (when).

This tripartite model transformed an intractable philosophical conundrum into a testable empirical challenge. If an animal could demonstrate, through goal-directed actions, that it retains a unified representation linking what occurred, where it occurred, and when it occurred in a single, unrepeated event, cognitive science could no longer dismiss the behavior as simple semantic association. Operationalizing episodic recall in this manner bypassed the requirement for verbal declarations of conscious re-experiencing. An organism would not need to verbally describe its morning foraging expedition; it merely had to behave in a way that revealed its recovery choices were simultaneously determined by the item’s identity, its precise location, and the elapsed time since it was deposited.

A critical theoretical hurdle in this formulation was the absolute necessity of demonstrating binding. It was not enough for an animal to hold an isolated spatial memory of a location and a separate, independent memory that a certain food item existed, or a vague sense of time having elapsed. In a true episodic trace, these dimensions are intrinsically bound into a coherent, relational gestalt. If the spatial, temporal, and item-identity memory tracks operated as disjointed associative primitives, the animal would fail when contingencies required the temporal coordinate to modulate the spatial value of a specific item. Demonstrating that an animal possesses bound representations requires experimental paradigms where behavioral success is contingent upon the simultaneous, unified processing of all three axes, precluding any explanation based solely on independent semantic facts or disjointed procedural rules.

1.3 The Mental Time Travel Hypothesis and Autonoetic Consciousness

Building upon Tulving’s framework, evolutionary psychologists Thomas Suddendorf and Michael Corballis formalized the mental time travel hypothesis, postulating that episodic memory is merely one half of a broader, bidirectional cognitive faculty. According to Suddendorf and Corballis, the primary evolutionary utility of looking backward is not passive nostalgia, but the prospective simulation of potential futures. Mental time travel allows an organism to project itself mentally backward to reconstruct past occurrences (episodic memory) and project itself forward to simulate, plan for, and navigate impending scenarios (episodic foresight or episodic future thinking). Both vectors of this temporal projection were hypothesized to be entirely absent in non-human animals, who were presumed to operate exclusively within the bounds of immediate physiological drives.

Central to this critique is the epistemological concept of autonoetic consciousness—the introspective, “self-knowing” awareness that accompanies episodic remembering. When a human recalls a wedding, a trauma, or a mundane breakfast, they do not simply access a database of events; they experience a distinctive qualitative state characterized by the felt conviction that they, their personal self, previously lived through that event. Suddendorf, Corballis, and Tulving maintained that autonoesis is the sine qua non of episodic memory. They argued that without empirical proof of this subjective qualia—the feeling of “re-living” the past—any animal behavior, no matter how complex its what-where-when mapping, could theoretically be explained by highly sophisticated semantic knowledge systems, akin to an advanced computational system reading an unfeeling chronological index.

This theoretical stance created an acute division within cognitive science. On one side stood the phenomenological purists, who held that because subjective qualia are fundamentally private, third-person scientific methodology can never verify autonoetic consciousness in non-human subjects, rendering true animal episodic memory an epistemological impossibility. On the other side stood the behavioral functionalists and comparative ethologists, who argued that an insistence on untestable subjective states violates methodological parsimony and introduces an arbitrary evolutionary discontinuity. If an animal exhibits the exact functional, structural, and behavioral markers of episodic reconstruction within its ecological niche, demanding introspective verbal confirmation is nothing more than an anthropocentric firewall designed to preserve human cognitive exceptionalism by definitional fiat.

2. Ethological Context: Caching Behavior in Western Scrub-Jays (Aphelocoma californica)

2.1 Natural History and Food-Storing Ecology of Corvids

To understand the breakthrough in episodic memory research, one must examine the evolutionary pressures governing the family Corvidae (crows, ravens, magpies, and jays), and specifically the western scrub-jay (Aphelocoma californica, now recognized in distinct lineages including Aphelocoma woodhouseii). These birds are non-migratory scatter-hoarders inhabiting the oak woodlands, scrub chaparrals, and arid pine forests of western North America. In these environments, food abundance fluctuates violently between the bountiful autumn harvest and the severe, resource-depleted winter and early spring. To survive prolonged seasonal food deficits, scrub-jays have evolved an obligate caching strategy, harvesting tens of thousands of individual food items—ranging from acorns and pine nuts to small vertebrates and invertebrates—and scattering them across hundreds of square kilometers.

Scatter-hoarding stands in stark ecological contrast to larder-hoarding, in which an animal stockpiles all collected resources into a single, fortified cache site, such as a subterranean burrow or a tree hollow. While larder-hoarding minimizes the memory burden on the hoarder, it introduces a catastrophic vulnerability: if a competitor discovers the central cache, the hoarder faces absolute starvation. Scatter-hoarding disperses this risk by hiding food in thousands of micro-sites, with only a few items per cache. However, this risk-mitigation strategy transfers the evolutionary burden entirely onto the bird’s cognitive architecture. A scatter-hoarder that cannot accurately locate its scattered caches will waste massive amounts of metabolic energy searching for them and inevitably perish.

Natural selection has thus acted as an unforgiving filter on corvid memory systems. Over millions of years, the scrub-jay genome was shaped to support hyper-specialized spatial, temporal, and associative memory faculties. A single scrub-jay may hide up to 5,000 acorns in a single autumn season, maintaining accurate spatial maps of these caches across retention intervals spanning six to nine months. Furthermore, these caches are recovered in a landscape radically transformed by winter weather, deciduous leaf loss, and snow cover. The bird cannot rely on simple sensory cues such as smell or direct vision, as caches are typically buried beneath soil, litter, or gravel; retrieval is mediated almost entirely by internal cognitive representations of the environment.

2.2 Spatial Cognition and Hippocampal Adaptations in Food-Caching Birds

The profound cognitive demands of scatter-hoarding are directly reflected in the comparative neuroanatomy of avian brains. For much of the twentieth century, the avian telencephalon was erroneously viewed as an overgrown, primitive striatum incapable of complex mammalian-like cognition. This misconception was decisively refuted by modern neurobiology, which recognized the avian pallium as homologous and functionally equivalent to the mammalian neocortex. Within this telencephalic architecture, the avian hippocampal formation serves as the primary neurocomputational engine for spatial navigation, memory consolidation, and relational learning.

Extensive comparative research conducted by John Krebs, David Sherry, and Susan Healy revealed a direct correlation between the degree of caching specialization and the relative volume and neuronal density of the hippocampal complex. Food-storing corvids, such as Clark’s nutcrackers (Nucifraga columbiana) and scrub-jays, possess significantly larger hippocampi—both in absolute terms and relative to total brain and body mass—than closely related, non-caching or opportunistic passerines. This structural expansion is driven not merely by passive genetic programming, but by extensive neurogenesis triggered by the experiential act of caching. Birds deprived of caching experience during development display blunted hippocampal growth, whereas birds engaged in active storing exhibit profound dendritic branching, synaptic plasticity, and the integration of newly generated neurons into existing functional circuits.

This hypertrophied hippocampal system supports advanced cognitive mapping abilities. Scrub-jays do not rely on simplistic egocentric path integration; they construct allocentric spatial representations that utilize distal visual landmarks, geometric boundaries, and solar-compass orientation to pinpoint cache coordinates within millimeters. Controlled laboratory manipulations have demonstrated that when local landmarks immediately surrounding a cache are subtly displaced, jays recalculate their search trajectories based on the spatial relationships of remaining distal cues, proving that their spatial recall operates via an integrated, relational map rather than rigid sensorimotor motor programs.

2.3 Perishability Dynamics in Natural Foraging Environments

While spatial precision is indispensable for cache retrieval, spatial memory alone is ecologically insufficient when an animal caches biologically diverse food types. In their natural habitat, scrub-jays do not store only durable, non-perishable seeds; they opportunistically cache high-protein invertebrates, caterpillars, beetle larvae, and small pieces of meat. These distinct commodities possess radically divergent decay kinetics. An acorn or pine seed, protected by a hard shell and stabilized by complex carbohydrates, can remain chemically stable and nutritionally viable beneath cold soil for hundreds of days. Conversely, a soft-bodied waxworm, cricket, or freshly captured vertebrate carcass decomposes within days or even hours, transforming from an intensely valuable, energy-dense reward into a putrid, toxic mass ridden with pathogenic bacteria.

This ecological reality introduces severe fitness costs. A jay that recovers a perishable item too late wastes critical foraging time and risks severe illness or death from ingesting decaying matter. Conversely, a jay that spends its initial retrieval window recovering durable seeds while neglecting perishable protein caches allows a high-value resource to rot unnecessarily. Natural selection therefore exerted powerful pressure on food-storing corvids to compute not only where an item was buried, but also what biological category of food resided in that specific coordinate and how long ago it was deposited relative to its specific degradation profile.

This temporal-decay calculus represents an evolutionary pre-adaptation for episodic-like cognition. The natural foraging ecology of the scrub-jay demands an ongoing, dynamic synthesis of item identity, precise spatial coordinates, and temporal duration. The bird’s survival hinges upon its capacity to deploy an internal timeline that informs its spatial retrieval decisions, prioritizing perishable resources when fresh, and abandoning them entirely once their temporal expiration threshold has been crossed. It was this precise ethological adaptation that Nicola Clayton recognized as the key to empirically unlocking episodic memory in an animal model.

3. The Clayton and Dickinson Breakthrough (1998): Experimental Architecture

3.1 The Seminal 1998 Nature Paper Overview

In the late 1990s, Nicola Clayton, an expert in corvid behavioral ecology, joined forces with Anthony Dickinson, a preeminent animal learning theorist at the University of Cambridge. Their collaboration was marked by a rare synthesis of deep ethological intuition and rigorous psychological experimental design. Clayton understood the natural behavioral repertoire and cognitive nuances of scrub-jays, while Dickinson brought the uncompromising analytical scrutiny of the British associative learning tradition. Together, they formulated an experimental paradigm capable of definitively testing whether scrub-jays remembered the what, where, and when of individual caching events while systematically ruling out conventional associative or sensory explanations.

The results of this collaboration were published in their seminal 1998 Nature paper entitled “Episodic-like memory during cache recovery by scrub jays”. The hypothesis was straightforward yet audacious: if scrub-jays encode the episodic components of their caching events as an integrated memory representation, their retrieval preferences should vary flexibly as a function of the interaction between the food type cached, the cache location, and the duration of time that had elapsed between caching and recovery. If the birds lacked temporal indexing or merely possessed a generalized, semantic knowledge that “worms decay,” they would fail to dynamically modulate their search patterns when tested across novel, trial-unique temporal intervals.

The publication of the paper sent shockwaves through comparative psychology and cognitive science. For the first time, an animal species had demonstrated the behavioral hallmarks of episodic recall in a rigorously controlled laboratory environment. The paper directly challenged the anthropocentric consensus championed by Tulving and Suddendorf, demonstrating that complex temporal cognition was not an exclusive evolutionary hallmark of the human lineage. By translating the philosophical construct of episodic memory into an ecologically valid behavioral assay, Clayton and Dickinson forced the scientific community to re-evaluate the cognitive architecture of non-human animals.

3.2 Food Types: Peanuts vs. Waxworms (Perishable vs. Non-Perishable)

The foundational genius of the Clayton-Dickinson paradigm lay in the operationalization of time through the differential decay schedules of two distinct food commodities: wax moth larvae (waxworms, Galleria mellonella) and roasted peanuts. Both foods were highly palatable to scrub-jays, but baseline preference testing established a clear and unequivocal hierarchy: when both items were fresh, western scrub-jays exhibited an overwhelming preference for waxworms over peanuts. Waxworms are soft, moisture-rich, energy-dense protein packets that birds will consistently seek out and consume before touching dry, fibrous peanuts.

However, these two food types possess fundamentally divergent temporal preservation dynamics. Peanuts are virtually non-perishable; they remain fresh, crunchy, and palatable for months under standard laboratory conditions. Waxworms, by contrast, are highly perishable. In the experiment’s operational protocol, when fresh waxworms were left in room-temperature sand for extended periods, they decayed into a black, liquefied, unpalatable state that the jays refused to consume. This physical reality created a natural value inversion driven entirely by elapsed time:

• At short temporal intervals: Waxworms are fresh and represent a vastly superior reward compared to peanuts.
• At long temporal intervals: Waxworms are rotten, repulsive, and inedible, rendering peanuts the only viable and preferred food source.

To ensure that the experimental subjects understood this decay contingency without introducing confounding non-episodic variables, Clayton and Dickinson subjected the jays to a pre-experimental training phase. During this phase, birds were allowed to cache fresh waxworms and discover them in their degraded, inedible state after long delays, or in their delicious, fresh state after short delays. Crucially, this training instilled the declarative or semantic rule that waxworms spoil over time, but it did not dictate which specific caches in the upcoming test phases contained fresh or rotten food. The birds were thus equipped with an understanding of the temporal dynamics of waxworm decay, setting the stage for testing whether they could apply this knowledge to episodic representations of unique past events.

3.3 The Two-Interval Temporal Paradigm (4 Hours vs. 124 Hours)

To interrogate the temporal dimension of memory, Clayton and Dickinson established a rigorous two-interval testing framework comparing a short retention interval of 4 hours against a long retention interval of 124 hours. This precise temporal contrast was chosen to balance ecological realism with experimental control. Within 4 hours, a buried waxworm remains fresh, active, and fully palatable. By 124 hours (approximately five days), an untreated waxworm decomposes into an unpalatable, blackened state. The experimental paradigm was structured such that the jays were exposed to distinct caching sessions separated in time, creating a scenario where cache retrieval required a complex temporal calculation.

The experimental architecture was systematically counterbalanced across subjects and trial conditions. In a typical caching sequence:

• A bird might be permitted to cache one food type (e.g., peanuts) into a specific, distinctively marked caching tray.
• Following a calculated delay, the bird would be presented with a second, distinct tray and allowed to cache the alternative food type (e.g., waxworms).
• After an additional delay, the bird was returned to the testing room and presented with both trays simultaneously for recovery, with the elapsed time since caching calibrated to either 4 hours or 124 hours for the waxworm caches.

This design created two distinct recovery conditions. In the 4-Hour Condition, waxworms had been cached recently (4 hours prior), meaning they were still in their fresh, prime state; therefore, the jays were predicted to search predominantly in the waxworm caching sites. In the 124-Hour Condition, the waxworms had been cached five days prior and were now entirely degraded; therefore, the jays were predicted to reverse their preference and search predominantly in the peanut caching sites, despite their baseline physiological preference for worms over nuts. By systematically alternating the order of presentation and caching sequences, the experimenters ensured that total retention time and presentation order were meticulously counterbalanced, isolating the elapsed time of specific caching episodes as the sole causal variable.

3.4 Visuospatial Caching Trays and Controlled Laboratory Design

To provide high-resolution spatial landmarks and prevent spatial ambiguity, Clayton and Dickinson designed specialized visuospatial caching trays. These apparatuses consisted of plastic ice-cube trays containing discrete, partitioned compartments, typically arranged in a two-by-eight or three-by-six grid. Each compartment was filled with finely sifted, clean sand, allowing the scrub-jays to execute their natural caching behaviors—digging a small depression with their beaks, inserting the food item, and covering the site with sand grains and small pebbles using characteristic sweeping head movements.

To transform the identical plastic trays into unique, identifiable spatial environments, the researchers affixed complex visual landmarks directly to the tray frames. Distinctive configurations of colorful Lego bricks, geometric plastic shapes, and patterned tape surrounded the compartments. This spatial scaffolding allowed the birds to encode the position of individual cache sites relative to both localized proximal landmarks (e.g., “the compartment directly adjacent to the red two-by-four Lego block”) and distal contextual cues within the testing room (e.g., door frames, ceiling fixtures, and room walls). The testing arena was partitioned into distinct spatial quadrants, ensuring that trays placed on the left or right side of the testing cage remained spatially segregated and distinct.

Laboratory conditions were strictly standardized to eliminate extraneous environmental noise. Illumination was maintained on a rigid 14:10 light-dark cycle, temperature and humidity were electronically controlled to ensure consistent food preservation conditions, and acoustic isolation minimized unexpected external distractions. Most importantly, the experimental sand substrate completely blinded any visual cues to the cache contents post-storage. Once the bird swept sand over its cache, the surface appeared uniform, flat, and devoid of visual traces. Any subsequent recovery attempt directed toward a specific compartment could not be driven by direct visual perception of the buried item; it had to be guided entirely by the bird’s internal memory map of the caching episode.

4. Deconstructing the What-Where-When Paradigm

4.1 Encoding ‘What’: Differential Palatability and Decay Properties

The first structural pillar of the episodic triad is the encoding of what: the qualitative identity, energetic value, and physical properties of the cached object. In the Clayton and Dickinson paradigm, scrub-jays demonstrated flawless tracking of item identity across distinct spatial compartments. When birds were allowed to inspect the recovery trays, their search behavior was not an indiscriminate, random exploration of previously manipulated sites. Instead, they directed their initial, decisive beak-probes toward specific compartments based entirely on the type of food that had been deposited there during the encoding phase.

Crucially, this ‘what’ representation was not a static, hardwired preference. If the scrub-jays were operating merely on an unalterable sensory preference for worms, they would have relentlessly probed the worm sites across all experimental conditions. Instead, their pursuit of the ‘what’ component was dynamically modulated by the item’s decay status. In the 4-hour condition, the birds expressed their natural palatability preference, dedicating over 75% of their initial recovery searches to the compartments where waxworms were buried. They treated the peanut sites with relative indifference, bypassing them to retrieve the high-value invertebrate rewards.

Conversely, in the 124-hour condition, the birds dramatically suppressed their interest in the waxworm sites. Despite waxworms having higher caloric and protein densities when fresh, the jays actively avoided those compartments, redirecting their primary recovery searches toward the peanut compartments. This finding proved that the scrub-jays did not retain an isolated memory of “food is at location X.” Rather, they remembered that waxworms were at location X and peanuts were at location Y, and they simultaneously integrated this identity record with an internal valuation calculation that accurately predicted the current, degraded state of the item.

4.2 Encoding ‘Where’: Spatial Localization within Caching Arrays

The spatial coordinate—the where component—was tracked by the scrub-jays with pinpoint, millimeter-level accuracy. Within the multi-compartment ice-cube trays, adjacent caching cells were separated by mere centimeters. Despite this physical proximity, the jays did not engage in broad, sweeping exploratory digging across the sand substrate. When presented with the trays during the recovery phase, the birds flew directly to the specific compartments where items had been deposited, executing targeted, non-random searches while ignoring identical, immediately adjacent empty cells.

This high-precision localization was quantified through detailed behavioral metrics, including the number of looks, false probes, and the exact sequence of compartments inspected. The birds consistently made their initial recovery probe directly into the target cache cell or an immediately adjacent compartment, demonstrating a focused spatial localization curve. Empty compartments that had never contained food during the encoding session were completely bypassed, showing that the birds retained an exact topographic map of the tray rather than a vague impression of the tray’s general attractiveness.

Furthermore, this spatial encoding proved remarkably resilient to retroactive interference. Even when birds were subjected to intervening caching sessions involving different trays and distinct food items between the initial encoding and final recovery, their spatial recall for the primary target sites remained intact. The scrub-jays maintained distinct, non-overlapping spatial representations for the peanut trays and the waxworm trays, demonstrating that their spatial memory stores were capable of maintaining multiple, simultaneous coordinate sets without catastrophic cross-talk or spatial degradation.

4.3 Encoding ‘When’: Temporal Differentiation and Interval Timing

The definitive linchpin of the Clayton-Dickinson experiment was the demonstration of the when component. Prior to this study, critics contended that animals were fundamentally incapable of temporal indexing—that while an animal might know what something is and where it is, it possesses no concept of when it happened. Clayton and Dickinson demolished this critique by demonstrating that the birds’ retrieval decisions were dictated by the duration of time that had elapsed since the specific caching episode occurred.

The jays’ behavioral trajectory systematically shifted across the temporal horizon:

• At 4 hours post-caching: Search allocation was overwhelmingly concentrated on the waxworm compartments. The memory of the caching event was fresh, and the birds anticipated succulent larvae.
• At 124 hours post-caching: The birds’ search strategy completely inverted. Search allocation shifted to the peanut compartments. The jays remembered that worms were cached there, but they understood that 124 hours of elapsed time had rendered those worms unpalatable.

To prove that this was driven by a true integrated ‘when’ representation and not a simple circadian clock mechanism, the researchers carefully calibrated the timing of the trials. The recovery sessions were conducted at consistent times of day to ensure that the birds could not rely on internal circadian phases (such as morning vs. evening physiological cues) to infer cache states. The temporal differentiation was driven entirely by an estimation of elapsed duration—an internal interval metric tracking how long ago the encoding event had transpired. By coupling this elapsed time with the known biological decay properties of the worm, the jay computed the current viability of the cache, providing undeniable empirical evidence that the ‘when’ coordinate was an active, functional component of the retrieved memory.

5. Experimental Controls and Methodological Rigor

5.1 Controlling for Olfactory and Visual Degradation Cues

A primary challenge to any cognitive claim in non-human animal research is the “Clever Hans” effect—the possibility that the animal is responding to subtle, immediate sensory cues rather than retrieving an internal mental representation. In the case of food-caching corvids, the most glaring potential confound was olfaction. Could the scrub-jays simply be smelling the decomposing waxworms beneath the sand? Rotten insect larvae emit volatile sulfurous compounds, short-chain fatty acids, and amines. If the birds were simply sniffing the compartments, detecting the odor of decay, and choosing to avoid the rotten-smelling cells, their behavior would reflect immediate sensory discrimination, completely devoid of episodic memory.

Clayton and Dickinson neutralized this confound through a flawless control manipulation: the cache recovery trials were conducted in the complete absence of food. Prior to the recovery phase, the experimenters removed the caching trays from the testing arena out of the bird’s sight. The researchers completely excavated the cached food items, discarded them, and refilled the compartments with entirely fresh, pristine, unsoiled sand. Thus, when the scrub-jay was reintroduced to the arena for the recovery test, the trays were completely empty and physically clean. There were no buried waxworms, no buried peanuts, and no localized chemical or olfactory traces of decomposition.

Despite the complete physical absence of the food and its odors, the scrub-jays executed the exact same behavioral patterns. At 4 hours, they relentlessly searched the empty compartments that had previously contained waxworms; at 124 hours, they searched the empty compartments that had previously contained peanuts. Furthermore, to eliminate human observer bias, all behavioral video scoring was conducted blind: independent observers scored the birds’ recovery attempts without knowing which compartments had held worms or peanuts, or whether the bird was being tested under the 4-hour or 124-hour condition. This proved conclusively that the birds’ search decisions were guided entirely by endogenous, internally generated cognitive representations, completely decoupled from immediate external sensory cues.

5.2 Eliminating Relative Familiarity as a Confound

A sophisticated theoretical critique emerging from mainstream animal learning theory centered on the concept of relative trace strength or familiarity. In memory psychology, an older memory trace is naturally weaker, more degraded, and less familiar than a recently formed memory trace. Skeptics argued that the scrub-jays did not need to represent “time elapsed” or “when an event occurred.” Instead, they could simply be using the raw strength of the memory trace as a perceptual proxy:

• Strong, vivid memory trace = recent event → search for worms.
• Weak, faded memory trace = distant event → search for peanuts.

Under this trace-strength hypothesis, the bird does not possess an integrated episodic memory; it merely possesses a procedural rule: “If the spatial memory trace of this tray is vivid, dig for worms; if the trace is faint, dig for peanuts.”

Clayton and Dickinson dismantled this familiarity argument through an ingenious series of follow-up experiments incorporating differential caching schedules. They designed paradigms where peanuts and waxworms were cached simultaneously or in reverse orders, such that the birds were tested on caches of identical physical age, or where the non-perishable item was associated with the fresher trace. In these balanced conditions, if the birds were relying on trace strength alone, they would have committed systematic errors, invariably picking the food associated with the strongest trace regardless of its biological decay profile.

The jays completely defied the trace-strength prediction. Even when a peanut cache was fresher than a worm cache, or when both caches were formed at distinct intervals that disrupted a simple “fading trace” heuristic, the birds dynamically integrated the specific identity of the food with its specific spatial location and its unique, individual elapsed time. They did not default to a generic “strong vs. weak” behavioral toggle. Their retrieval patterns demonstrated that the memory representation preserved the unique event characteristics rather than a passive, non-specific index of temporal decay.

5.3 Disentangling Satiation and Motivation States

Another critical variable that required rigorous isolation was the internal motivational state of the animal, specifically sensory-specific satiety. In behavioral neuroscience, it is well established that an animal that has recently consumed a large quantity of a specific food type experiences a temporary drop in the reward value of that particular food, while remaining motivated to consume alternative foods. If the experimental schedule had inadvertently allowed the birds to consume waxworms prior to certain testing intervals, the subsequent avoidance of waxworm sites could have been driven by simple satiation rather than a cognitive realization of decay.

Clayton and Dickinson controlled for this by enforcing strict dietary and caloric regimens. The birds were maintained on a precise baseline nutritional schedule throughout the experimental months, ensuring that their overall caloric demand and foraging drive remained stable. During the caching trials, the birds were allowed to store items, but their immediate consumption was meticulously tracked and limited to prevent satiety artifacts. Pre-feeding controls were established: birds were deliberately pre-fed either peanuts or waxworms prior to specific recovery trials to measure how acute satiety affected their choices.

The data demonstrated that while sensory-specific satiety did modulate immediate food consumption, it operated through an entirely distinct behavioral channel from episodic cache recovery. When jays were not sated, their avoidance of 124-hour waxworms was absolute, despite their high hunger levels and intense motivation to eat fresh waxworms. The avoidance of decayed worm sites was clearly driven by the expectation of unpalatable, rotten food—a cognitive assessment based on elapsed time—rather than a physiological lack of appetite for waxworms. The cross-over design confirmed that the birds’ motivational states were completely reversible and that their memory choices reflected cognitive evaluations of cache viability.

6. Distinguishing Episodic-Like Memory from Semantic and Associative Learning

6.1 Associative Learning Models and Reinforcement History

Throughout the history of experimental psychology, behavioral phenomena that appear complex and cognitively advanced have routinely been explained by parsimonious models of associative conditioning, most notably the Rescorla-Wagner model of classical conditioning. When Clayton and Dickinson presented their findings, traditional behaviorists immediately sought to explain the scrub-jays’ behavior through simple associative linkages: stimulus-stimulus (S-S) or stimulus-response (S-R) chains forged through repetitive reinforcement history.

Under a strict associative learning framework, an organism learns the value of a conditioned stimulus (CS) through gradual, iterative pairings with an unconditioned stimulus (US) that delivers a reward or punishment. Skeptics argued that the scrub-jays had simply acquired a standard conditional discrimination:

If the birds were merely executing a conditional associative chain acquired through dozens of repetitive reinforcement trials, there would be no need to invoke rich, trial-unique episodic representations.

However, Clayton and Dickinson’s methodology fundamentally decoupled their findings from associative conditioning. The crucial differentiator was that the testing trials were trial-unique and conducted in extinction (unrewarded). The birds did not undergo hundreds of training trials to forge a rigid conditional discrimination for a specific tray layout; rather, each test tray layout was unique, the spatial placement of items was varied dynamically, and the recovery tests were executed without rewarding the birds with food. A standard associative model cannot explain why an animal, on a single, novel trial, would instantly combine a single encoding event with a five-day temporal delay to direct its very first behavioral action toward a specific empty spatial coordinate. The behavior was not an accreted associative habit; it was the targeted retrieval of a unique historical episode.

6.2 Rule-Learning vs. Unique Historical Events

To deepen the distinction between semantic knowledge and episodic retrieval, it is essential to analyze the interplay between general rules and event-specific memories. In human cognition, we possess semantic rules (e.g., “milk left unrefrigerated for two weeks will spoil”) that we apply to unique episodic memories (e.g., “I opened this specific bottle of milk yesterday afternoon”). The semantic rule is general, abstract, and timeless; the episodic memory is concrete, anchored, and temporally indexed. Neither system replaces the other; rather, semantic rules operate on episodic memories to guide rational decision-making.

Clayton and Dickinson demonstrated this exact cognitive architecture in scrub-jays. The pre-experimental training phase provided the birds with a general semantic rule: “Waxworms decay over long temporal intervals, whereas peanuts remain stable.” This rule by itself is completely useless during a recovery trial unless the bird can bind it to a unique historical event:

• Where did I cache worms this morning?
• Which tray contained peanuts five days ago?

Without a specific, trial-unique episodic record of that single caching episode, the semantic rule has no target data upon which to execute its operational logic.

The definitive proof that scrub-jays were relying on unique event memories rather than rigid procedural programming came from experiments where the rules were updated after the caching event had already taken place. In subsequent studies, Clayton and colleagues allowed jays to cache food, and during the retention interval (while the food was hidden), they exposed the birds to new information regarding the perishability of that food type. The jays spontaneously updated their retrieval strategies when tested later, despite having encoded the memory under the old assumption. This capacity for retrospective memory updating is a definitive hallmark of declarative, episodic representations in humans, proving that the jays’ internal records are flexible, accessible, and distinct from rigid, habit-based motor programs.

6.3 The Rationale for the ‘Episodic-Like’ Qualification

Given the striking parallels between the scrub-jays’ cognitive performance and human episodic retrieval, why did Nicola Clayton and Anthony Dickinson insist on using the hyphenated qualification “episodic-like”? The decision was an act of deliberate philosophical and methodological modesty, designed to insulate their empirical findings from destructive semantic debates regarding subjective phenomenology.

As established by Endel Tulving, the human definition of episodic memory was fundamentally intertwined with autonoetic consciousness—the subjective, introspective qualia of mentally projecting oneself backward in time. By defining their subjects’ behavior as “episodic-like,” Clayton and Dickinson explicitly acknowledged the intractable epistemological barrier: science cannot objectively measure the internal, phenomenological qualia of a non-verbal animal. A researcher can track a bird’s gaze, measure its peck velocity, and record its spatial choices with microsecond precision, but no machine can peer into the avian mind to confirm whether the bird subjectively feels the nostalgic, autonoetic sensation of “re-living” the caching event.

The “episodic-like” moniker established an operational, cross-taxa standard for comparative cognitive science:

• It focused exclusively on behavioral criteria: the demonstration of an integrated, trial-unique memory of what, where, and when.
• It avoided the trap of anthropomorphism, refusing to make unprovable assertions about animal subjective states.
• It simultaneously thwarted anthropocentric dismissal, forcing critics to acknowledge that the animals were executing all the functional computational tasks of episodic memory.

This operational bridge allowed the scientific community to study the evolutionary roots and neurobiological substrates of episodic memory without becoming paralyzed by the philosophical “problem of other minds.”

7. Expanding the Temporal Paradigm: Which and How Long Ago

7.1 The ‘Which’ Component: Integrating Contextual Information

Following their 1998 breakthrough, Clayton, Dickinson, and their colleagues embarked on an ambitious program to dissect the deeper cognitive dimensions of corvid episodic-like memory. One of the first frontiers was expanding the behavioral triad into a quaternary model by incorporating the which component: the integration of complex contextual, source, and episodic-background cues into the memory representation.

In a complex natural environment, a scrub-jay does not cache in a sterile, uniform void. It caches across multiple distinct patches of forest, on different days, under varying weather conditions, and within differing geographical contexts. To model this, the researchers introduced multiple visual contexts into the laboratory setting. Trays were fitted with distinctly patterned borders (e.g., black-and-white stripes versus polka dots), placed in different physical rooms, or associated with distinct ambient auditory backgrounds. The birds were then presented with complex caching regimens where they had to remember not merely what was cached where, but which specific caching event had occurred in which specific environmental context.

The findings demonstrated that scrub-jays seamlessly organized their memories into hierarchical, episode-specific structures. The jays did not confuse items cached in “Tray A (Striped)” with items cached in “Tray B (Polka-dot),” even when the spatial coordinates within the trays were identical. The birds demonstrated sophisticated source memory: they remembered the specific environmental and contextual circumstances surrounding the original encoding event. This proved that the memory trace was not a fragile, isolated associative string, but a rich, multi-featured cognitive record that preserved the broader episodic context of the experience.

7.2 Differentiating Elapsed Time from Specific Temporal Milestones

A critical theoretical question that emerged from the temporal paradigm was: How do scrub-jays track time? In cognitive psychology and neurobiology, temporal tracking can operate through at least two fundamentally different computational mechanisms:

1. Time of Day (Phase Timing): Utilizing an internal circadian clock to anchor events to specific physiological or external temporal markers (e.g., “This event happened at dawn” or “This event happened at dusk”).
2. Time Elapsed (Interval Timing): Utilizing an internal, stopwatch-like accumulator that computes the exact duration of elapsed time since the event concluded, completely independent of the circadian cycle.

To adjudicate between these two mechanisms, Clayton and colleagues designed an experiment that systematically dissociated time of day from elapsed duration. They manipulated the caching and recovery schedules such that caching sessions occurred at different times of the day (e.g., 09:00 vs. 15:00), while holding the elapsed retention interval constant at either 4 hours or 28 hours (both being multiples that matched or mismatched the circadian phase). If the jays were relying on a circadian phase-marker, they would be easily confused when the recovery trial occurred at the same time of day as caching, regardless of whether 24 or 48 hours had passed.

The results provided decisive evidence in favor of an elapsed time mechanism. The scrub-jays did not rely on time-of-day matching. Instead, their retrieval choices were governed by an internal metric that measured the total accumulated duration since the caching episode. The birds tracked “how long ago” an event occurred rather than merely “at what point in the day” it occurred. This proved that corvid episodic-like memory possesses an intrinsic, duration-based interval timing engine, allowing the animal to locate a past episode accurately along an internal temporal continuum.

7.3 Complex Decay Schedules and Ripening Experiments

To definitively shatter any remaining criticism that scrub-jays were simply operating on a passive aversion to older memory traces, Clayton and Dickinson engineered an extraordinary reversal of their original paradigm: the ripening experiment. In nature, certain food items, such as green fruit or raw seeds, are unpalatable when fresh but become soft, sweet, and highly nutritious after a period of aging or ripening. If the jays’ avoidance of old caches in the original 1998 experiment was merely due to a generalized cognitive aversion to aged, fading memories, they would inevitably fail if tasked with tracking food that improved over time.

The researchers introduced a novel food item into the paradigm: fresh, hard dog food pebbles that were relatively unpalatable to the jays, but which became soft, rich, and intensely preferred when allowed to soak and “ripen” in moist sand over a long interval. Conversely, they paired this with waxworms, which spoiled over that same long interval. This created two mirror-image, competing temporal trajectories:

• Waxworms: Delicious at 4 hours → Rotten and unpalatable at 124 hours.
• Dog Food (Ripening item): Hard and unpalatable at 4 hours → Soft and delicious at 124 hours.

The scrub-jays mastered this complex computational challenge with breathtaking cognitive flexibility. When tested at the 4-hour interval, they bypassed the unripened dog food and dug exclusively for the fresh waxworms. But when tested at the 124-hour interval, the birds executed a complete, coordinated behavioral shift: they ignored the decayed waxworms and actively sought out the aged, ripened dog food. This experiment provided conclusive, unassailable proof that the jays were not operating on passive memory trace decay or a simplistic heuristic of “avoid old caches.” Instead, they maintained independent, item-specific temporal functions, dynamically calculating the fluctuating palatability states of multiple food commodities as a function of elapsed time.

8. Social Cognition and Tactical Caching: The Observer Effect

8.1 Pilferage Risk and Recaching Behaviors

The cognitive brilliance of the western scrub-jay is not confined to the physical and temporal dynamics of food decay; it is equally expressed in the cutthroat arena of social competition. In the wild, scrub-jays inhabit complex social landscapes where pilferage is rampant. Up to 30% of all caches made by a scrub-jay are stolen by watching conspecifics (competitor jays) within hours of storage. The risk of pilferage introduces an intense evolutionary arms race, transforming food-storing into a high-stakes chess match of stealth, observation, and deception.

To explore how episodic-like memory intersects with social cognition, Nicola Clayton, together with Nathan Emery, conducted a series of seminal experiments examining the observer effect. They tested scrub-jays in conditions where they cached food either in complete privacy or in full view of a conspecific bird perched in an adjacent cage. Later, the caching bird was returned to the arena to recover its caches, this time in total privacy. The researchers observed a remarkable tactical response: birds that had been observed while caching returned to their caches and systematically re-cached the food—digging it up from the sites where the competitor had watched them hide it and moving it to new, secret, unobserved locations.

Crucially, this tactical recaching behavior was not a random, panicked reaction. The birds specifically targeted the caches that had been observed, while leaving alone the caches that they had made in private or behind visual barriers. The jays remembered not only what was cached, where it was cached, and when it was cached, but also who was watching during that specific, historical caching episode. This proved that social identity and social context are fully integrated into the scrub-jay’s episodic memory architecture.

8.2 Theory of Mind and Experience Projection in Scrub-Jays

The social caching experiments reached a profound theoretical peak when Emery and Clayton investigated the psychological drivers behind this deceptive maneuvering, uncovering a phenomenon commonly referred to in comparative cognition as “thief knows thief”. The researchers discovered that not all scrub-jays recache their food after being observed. Rather, recaching behavior was exhibited exclusively by jays that had prior personal experience as pilferers—birds that had previously stolen the caches of other jays.

Birds that had been raised in the laboratory and had never pilfered another bird’s cache showed no inclination to recache their food when observed by a competitor, despite having identical visual experiences of being watched. Only when these naive birds were given the opportunity to steal other birds’ caches did they suddenly begin recaching their own hidden food when they themselves were subsequently observed. Having learned firsthand how visual observation facilitates theft, the newly minted thieves projected this experience onto watching conspecifics, anticipating that the observer would attempt to steal their food.

This remarkable finding provides compelling evidence for experience projection—a fundamental precursor to Theory of Mind (the capacity to attribute mental states, knowledge, and intentions to others). The experienced jay understands that an observer possesses visual knowledge of the cache site (“He saw where I put it, so he knows where it is”), and uses its own episodic memories of past pilferage to predict and counter the future behavior of a rival. This represents a seamless, bidirectional integration of episodic memory (retrieving past experiences of personal theft) and episodic foresight (projecting into the future to execute tactical deception).

8.3 Auditory and Visual Masking Tactics during Cache Deployment

The sophistication of scrub-jay tactical caching extends beyond post-hoc recaching; it actively shapes the very mechanics of how caches are initially deployed in real time. In a subsequent series of experiments, Clayton and colleagues demonstrated that jays actively deploy both visual and auditory masking techniques to suppress information transmission to observing competitors during the caching act itself.

When jays were forced to cache in the presence of a competitor, they systematically selected caching sites that were visually occluded—hiding food behind physical barriers or in deep, shadowed substrates where the competitor’s line of sight was blocked. If only open trays were available, the jays positioned their own bodies directly between the observer and the cache site, utilizing their torso and wings as a physical shield to obscure their beak movements from the rival’s view.

Even more extraordinarily, the jays demonstrated an acute awareness of acoustic information. In experiments where jays were provided with two caching substrates—one consisting of quiet, soft sand and the other consisting of loud, rattling gravel—the birds’ substrate choice was dictated by the social environment. When caching in private, they utilized both substrates indiscriminately. But when an observing competitor was present behind an opaque barrier (meaning the rival could hear but could not see), the caching jays selectively cached in the silent sand substrate, deliberately avoiding the noisy gravel to prevent auditory eavesdropping. This behavioral modulation requires high-level executive control: the jay must inhibit its natural motor routines and select specific physical actions based on a mental representation of what the competitor can hear and infer.

9. Neurobiological Correlates of Avian Episodic-Like Memory

9.1 Avian Hippocampal Formation and Parahippocampal Structures

The empirical demonstration of episodic-like memory in corvids ignited intense interest in the neurobiological mechanisms underlying these complex behaviors. For decades, mammalian neuroscience, pioneered by Brenda Milner’s studies of patient H.M., had identified the mammalian hippocampus—specifically the trilaminar archicortex comprising the dentate gyrus, CA3, and CA1 subfields—as the indispensable biological hub of episodic memory encoding, consolidation, and retrieval. Avian brains, however, lack the laminated six-layered neocortex and the classic morphological architecture of the mammalian Ammon’s horn.

Despite these morphological differences, modern neuroanatomy has revealed profound deep homologies between the avian hippocampal formation (HF) and the mammalian hippocampus. Located dorsomedially in the avian telencephalon, the avian HF displays conserved developmental origins, neurochemical markers, and circuit connectivity. Neurotoxic lesion studies in food-storing birds have demonstrated that localized damage to the avian HF results in catastrophic deficits in spatial memory and cache recovery accuracy, while leaving general motor abilities, feeding motivation, and simple conditioning intact.

Electrophysiological recordings in behaving corvids and homing pigeons have identified place cells and spatial-view cells within the avian HF that fire selectively when the bird occupies or attends to specific spatial locations within an environment, mimicking the functional properties of mammalian hippocampal place cells discovered by John O’Keefe. Furthermore, hippocampal lesions systematically disrupt the temporal component of episodic-like tasks. When the HF is damaged, birds retain the ability to discriminate familiar from unfamiliar objects, but they can no longer bind the spatial coordinate to the temporal decay schedule, confirming that the avian hippocampal complex is the computational engine responsible for synthesizing the tripartite what-where-when memory trace.

9.2 Neural Homology between Avian Nidopallium and Mammalian Prefrontal Cortex

While the hippocampal formation provides the spatial and temporal scaffolding for memory traces, episodic recall and tactical planning require high-level executive control, rule application, working memory, and behavioral inhibition. In the mammalian brain, these operations are orchestrated by the granular prefrontal cortex (PFC). Because birds completely lack a laminated neocortex, early neurobiologists assumed they were incapable of human-like executive function. This assumption was demolished by the functional identification of the Nidopallium Dorsolaterale (NCL).

The NCL is an endbrain structure situated in the caudolateral telencephalon of the avian brain. Although it evolved through convergent evolution rather than direct phylogenetic continuity with the mammalian PFC, it exhibits an astonishing array of structural, neurochemical, and functional homologies:

• Dense Dopaminergic Innervation: Like the mammalian PFC, the NCL receives massive dopaminergic projections from the midbrain ventral tegmental area (VTA) and substantia nigra, modulating working memory and delayed reward valuation.
• Executive Gating: Single-unit electrophysiological recordings conducted by Onur Güntürkün and colleagues have revealed that NCL neurons exhibit persistent delay activity during working memory tasks, maintaining representations of abstract rules, expected rewards, and spatial goals in the absence of sensory input.
• Top-Down Modulation: The NCL projects back to sensory and limbic structures, providing the top-down cognitive control required to suppress immediate impulses (such as digging for tempting waxworms after 124 hours) in favor of delayed, rational rewards (digging for stable peanuts).

In the context of the Clayton-Dickinson paradigm, the NCL serves as the central executive processor that queries the hippocampal memory store, evaluates the elapsed temporal interval against learned biological decay rules, and coordinates the motor output. When a scrub-jay decides to abandon a worm cache at 124 hours, or chooses to recache a food item in silence because an observer is listening, the NCL provides the neurocomputational machinery necessary to weigh alternative actions, simulate outcomes, and override automatic behavioral defaults.

9.3 Neuroplasticity and Seasonal Hippocampal Volume Variations

A striking neurobiological feature of food-storing corvids is the radical, dynamic structural neuroplasticity of their memory systems. Unlike the mammalian brain, where adult neurogenesis is largely restricted to the subgranular zone of the dentate gyrus and the subventricular zone, the avian brain exhibits widespread, high-volume neurogenesis throughout adulthood. In scatter-hoarding corvids, this neurogenesis is seasonally regulated and tightly synchronized with the ecological demands of food caching.

During the autumn months, when food-storing activity reaches its annual peak, the corvid hippocampal formation undergoes a remarkable structural reorganization:

• Hippocampal volume expands significantly, driven by a surge in the recruitment and survival of newly generated neurons.
• Upregulation of neurotrophic factors, particularly Brain-Derived Neurotrophic Factor (BDNF), enhances synaptic plasticity and promotes dendritic arborization.
• New neurons are integrated directly into existing functional circuits, expanding the network’s computational capacity precisely when the bird must encode thousands of new, spatially unique caches.

Once the winter and early spring pass and the hidden caches are recovered or exhausted, the surplus hippocampal neurons undergo programmed apoptosis (cell death), pruning the network back to its baseline volume for the summer breeding season. This seasonal cycle of neuroplasticity provides an elegant evolutionary solution to a critical metabolic trade-off. Avian flight demands strict weight minimization and an exceptionally efficient energy budget; maintaining an oversized, metabolically ravenous neural network year-round would impose severe physiological costs. By dynamically scaling its hippocampal architecture up during caching season and pruning it during periods of low cognitive demand, the scrub-jay optimizes its metabolic investment while preserving the high-level computational machinery required for episodic-like memory.

10. Comparative Cognition: Cross-Species Evaluations Post-Clayton and Dickinson

10.1 Replications and Extensions in Non-Human Primates

The demonstration of episodic-like memory in an avian species sent shockwaves through the primatological community. For decades, non-human primates—specifically chimpanzees, bonobos, and orangutans—had been presumed to be the sole rightful heirs to advanced cognitive faculties outside the human lineage. The realization that a bird possessed an integrated what-where-when memory sparked an urgent drive to replicate and extend the Clayton-Dickinson paradigm in non-human primates.

Researchers including Josep Call, Michael Tomasello, and Christopher Martin adapted the perishable-food paradigm for great apes. Using customized laboratory enclosures, chimpanzees (Pan troglodytes) and orangutans (Pongo abelii) were presented with choices between highly perishable, favored foods (such as frozen fruit juice or fresh banana slices) and durable, less-preferred foods (such as dry monkey chow pellets or hard dried fruit) hidden in distinct spatial locations across varying retention delays. These experiments successfully demonstrated that great apes can track the what, where, and when of hidden food items, strategically prioritizing perishable treats after short delays and shifting to durable foods when long delays had caused the fruit juice to melt or spoil.

Subsequent studies by Gema Martin-Ordas and colleagues expanded these findings to show that chimpanzees can recall unique events that occurred up to three years prior. In these paradigms, apes were confronted with a specific problem-solving task (such as retrieving a tool hidden in a complex testing room) that they had experienced only once, years earlier. The apes immediately remembered where the specific tool was hidden and who had placed it there, demonstrating long-term, trial-unique memory retention that rivaled human autobiographical recall. However, these primate studies also revealed an unexpected paradox: despite their evolutionary proximity to humans and their massive, laminated neocortices, primates did not consistently outperform corvids in these tasks. The hyper-specialized ecological demands of scatter-hoarding had driven corvids to evolve an episodic-like memory system that was every bit as sharp, precise, and integrated as that of humanity’s closest living relatives.

10.2 Rodent Models of What-Where-When Memory (Eichenbaum and Fortin)

While corvids and primates provided profound insights into the behavioral architecture of episodic recall, unraveling the precise micro-circuitry of memory required mammalian models amenable to dense electrophysiology, optogenetics, and pharmacological manipulation: the laboratory rat (Rattus norvegicus) and mouse (Mus musculus). Inspired directly by Clayton and Dickinson’s work, cognitive neuroscientist Howard Eichenbaum, along with Norbert Fortin and Jonathan Crystal, adapted the what-where-when paradigm for rodents.

Because rodents are predominantly macrosmatic (scent-driven) rather than visually driven, the researchers translated the visual ice-cube trays into complex olfactory-spatial arrays. Rats were presented with distinct scent cups (e.g., cinnamon, thyme, cocoa) arranged in specific geometric configurations within an open field. The temporal dimension was operationalized either through biological decay or through complex odor sequences where the order of presentation dictated the reward value of a specific spatial cup after defined retention intervals. Eichenbaum and Fortin demonstrated that normal rats could flawlessly integrate what scent was presented, where it was positioned, and when (or in what ordinal sequence) it had been encountered.

These rodent adaptations yielded revolutionary neurobiological breakthroughs:

• Circuit Dissociation: Selective neurotoxic lesions to the hippocampal CA1 and CA3 subfields demonstrated a clear functional dissociation. Lesions to CA1 selectively shattered the temporal “when” component, leaving the spatial “where” and object “what” recognition intact, whereas parahippocampal cortical lesions disrupted item recognition while sparing temporal sequences.
• Time Cells: Electrophysiological recordings in the rodent hippocampus revealed the existence of time cells—neurons that fire at specific successive moments during empty temporal intervals, providing an explicit internal clock that bridges discrete events across time.
• Translational Utility: These rodent what-where-when paradigms have become standard behavioral assays in translational biomedicine, utilized globally to assess cognitive decline, screen neuroprotective compounds, and model the early pathophysiology of Alzheimer’s disease and other neurodegenerative tauopathies.

10.3 Episodic-Like Phenomena in Cetaceans, Canids, and Cephalopods

The conceptual liberation ignited by the scrub-jay experiments rapidly transcended traditional laboratory models, prompting researchers to investigate episodic-like phenomena across wildly divergent animal taxa. If episodic memory is an adaptive ecological solution to navigational, foraging, and social challenges, its evolutionary emergence should not be confined to birds, primates, and rodents.

In domestic dogs (Canis familiaris), Claudia Fugazza and colleagues utilized the “Do as I Do” paradigm to test event memory. Dogs were trained to imitate human actions on command. Crucially, the researchers introduced unexpected testing probes where the command to imitate was issued long after the human had performed an incidental action, in contexts where the dog had no expectation that memory retrieval would be required. The dogs successfully reproduced the complex actions after extended delays, proving that they encode incidental, trial-unique events that they had not been explicitly conditioned to remember—a foundational characteristic of true episodic memory.

Marine biologists demonstrated comparable abilities in bottlenose dolphins (Tursiops truncatus), showing that dolphins could report on their own recently executed behaviors (“repeat” or “do something new” commands) across variable delays, retaining an internal representation of their own immediate past actions. Perhaps most extraordinary was the discovery of episodic-like memory in the cuttlefish (Sepia officinalis), an invertebrate cephalopod whose nervous system diverged from the vertebrate lineage over 550 million years ago. Christelle Jozet-Alves and colleagues demonstrated that cuttlefish track what prey they had eaten (crab vs. grass shrimp), where they had found it, and how long ago the feeding event occurred, dynamically altering their hunting choices based on the replenishment rates of different prey species. The existence of what-where-when memory in a mollusk provides definitive, undeniable proof that episodic-like cognitive architectures have evolved independently multiple times across evolutionary history through profound evolutionary convergence.

11. Philosophical and Theoretical Debates on Phenomenological Experience

11.1 The Problem of Subjective Experience and Chronesthesia

Despite the overwhelming empirical success of the what-where-when paradigm, the philosophical debate surrounding animal episodic memory did not subside. Instead, it retreated into the deep epistemological fortress of subjective qualia. Endel Tulving introduced the concept of chronesthesia—defined as the conscious, introspective awareness of subjective time, the inner mental engine that allows a human being to consciously apprehend their own past and future as personal states of existence. Tulving maintained that chronesthesia is fundamentally distinct from the objective calculation of elapsed time.

The philosophical critique, championed by contemporary skeptics, argues that an animal can execute a flawless what-where-when calculation through purely computational, non-conscious heuristics:

• An advanced algorithmic system (or a modern smartphone) can track what photo was taken, where the GPS coordinate was, and when the timestamp occurred.
• The phone can use this database to organize photo albums, predict user behavior, and execute automated actions.
• Yet, nobody would argue that the smartphone possesses autonoetic consciousness, chronesthesia, or an internal phenomenological life.

Critics argued that Clayton and Dickinson’s scrub-jays were essentially biological equivalents of this database: extraordinary biological computers executing complex what-where-when algorithms without a single spark of subjective temporal awareness.

This epistemological barrier was directly addressed by the pioneer of cognitive ethology, Donald R. Griffin, who argued against the knee-jerk dismissal of animal consciousness. Griffin contended that cognitive psychology was trapped in an impossible double standard: assuming that complex, flexible behaviors in humans are driven by conscious awareness, while simultaneously demanding that identical behaviors in non-human animals must be assumed to be non-conscious mechanisms until impossible standards of proof are met. From a philosophical functionalist perspective, if an organism processes, integrates, updates, and acts upon complex, temporally indexed representations in the exact manner that conscious humans do, the assertion that the animal is a mindless automaton violates the philosophical principle of parsimony.

11.2 Behavioral Criteria vs. Experiential Reality in Animals

The core of the philosophical debate lies in the relationship between behavioral criteria and experiential reality. Can non-linguistic paradigms ever definitively confirm conscious recollective experience? The unvarnished epistemological answer is no; they cannot. However, this limitation is not unique to comparative cognition—it applies equally to the study of other human minds. We infer that our fellow human beings possess conscious episodic memories because their linguistic reports and behavioral choices mirror our own internal states. Because non-human animals lack human syntax, human theorists have often erected an unfalsifiable, anthropocentric barrier: defining the cognitive faculty in terms of a linguistic self-report that animals are biologically incapable of providing.

The danger of this restrictive definition is that it divorces cognitive science from evolutionary biology. If episodic memory is defined so narrowly that only an organism capable of declaring, “I remember that I personally experienced this event” qualifies, then by definition, pre-verbal human infants, humans suffering from specific expressive aphasias, and all non-human animals are classified as lacking episodic memory. This creates an absurd evolutionary discontinuity, implying that episodic memory materialized out of an evolutionary void alongside human speech, rather than evolving incrementally from pre-existing neurocognitive systems.

Furthermore, behavioral evidence increasingly demonstrates that scrub-jays do not respond as rigid, unconscious automatons. Their capacity to update cache valuations post-encoding, their tactical deployment of deceptive counter-strategies based on who is watching, and their flexible reversal of behavioral preferences during ripening experiments all point toward a rich, declarative representational architecture. While we cannot experience the qualia of an avian mind, to dismiss their rich, flexible behavioral adaptations as “unconscious heuristics” while labeling identical human behaviors as “conscious mental time travel” is an exercise in anthropocentric bias rather than objective scientific inquiry.

11.3 Redefining Tulving’s Evolutionary Continuity Principle

The ultimate legacy of the Clayton and Dickinson experiments within the philosophical sphere was the re-establishment of Charles Darwin’s foundational principle of evolutionary continuity: the realization that the difference in mind between humans and other higher animals is one of degree, not of kind. By dismantling the assumption that episodic memory was an unbridgeable cognitive chasm separating humanity from the rest of the biosphere, Clayton and Dickinson forced cognitive science to abandon qualitative dichotomies in favor of quantitative continua.

Rather than viewing episodic memory as a monolithic, all-or-nothing adaptation that appeared de novo in modern humans, contemporary cognitive science now conceptualizes it as a mosaic of interlocking neurocognitive components:

This continuum-based framework shifted the burden of proof onto anthropocentric theorists. It was no longer enough to assert that animals are trapped in the present; skeptics were now required to explain how a brain could execute multi-dimensional, trial-unique, temporally calibrated, and socially modulated retrieval decisions without possessing the functional representational equivalents of episodic memory. Beyond theoretical psychology, this paradigm shift carries profound ethical implications. Acknowledging that non-human animals possess rich, temporally indexed autobiographical memories and prospective foresight shatters the Cartesian view of animals as unfeeling machines, elevating our moral obligations regarding their welfare, conservation, and cognitive agency.

12. Legacy, Modern Developments, and Future Research Horizons

12.1 Nicola Clayton and Anthony Dickinson’s Enduring Scientific Impact

The 1998 publication of Clayton and Dickinson’s experiments marked a definitive watershed moment in the history of psychology and animal behavior. Over the subsequent quarter-century, their paper has garnered thousands of citations and inspired an entire generation of comparative neuroscientists, ethologists, and philosophers. By formulating a rigorous, falsifiable behavioral assay for episodic memory, they rescued the field from circular philosophical dead-ends and catalytically accelerated the broader “Avian Cognitive Renaissance.”

Following this breakthrough, Nicola Clayton established the world-renowned Comparative Cognition Laboratory at the University of Cambridge, transforming it into an epicenter for the study of corvid intelligence, social dynamics, and temporal cognition. Her ongoing work, alongside collaborators such as Nathan Emery and Christopher Bird, expanded our understanding of animal tool manufacture, causal reasoning, cooperative problem-solving, and prospective cognition. Anthony Dickinson continued his foundational work in animal learning theory, utilizing the scrub-jay findings to bridge the computational mathematics of associative learning with higher-order cognitive modeling.

Today, the Clayton-Dickinson paradigm is universally recognized as a classic of experimental design, featured prominently in standard textbooks of neuroscience, cognitive psychology, and animal behavior. Their work successfully challenged human cognitive exclusivity, demonstrating that the pinnacle of cognitive complexity could be achieved in an evolutionary lineage that diverged from mammals over 300 million years ago, housed within a brain no larger than a walnut.

12.2 Advanced Behavioral Paradigms and Neuroimaging in Corvids

The contemporary study of corvid cognition has evolved far beyond the ice-cube trays of the late 1990s, incorporating cutting-edge neuroimaging, automated psychophysics, and advanced behavioral assays. Researchers are now probing the deeper frontiers of the avian mind, including prospective planning and the testing of the famous Bischof-Köhler hypothesis—the assertion that non-human animals cannot anticipate future motivational states independent of their current physiological desires.

In a series of breathtaking experiments, Clayton and colleagues dismantled the Bischof-Köhler hypothesis. They demonstrated that scrub-jays and Eurasian jays (Garrulus glandarius) can plan for tomorrow’s breakfast: when birds were locked into rooms where they had learned they would be deprived of food the following morning, they proactively cached food in that specific room the evening before, despite being completely sated at the time of caching. The birds successfully overrode their immediate physiological state to provision for a future, anticipated hunger state in a specific spatial location. Modern automated touch-screen testing batteries are currently being deployed to measure the exact millisecond temporal dynamics of corvid working memory, executive gating, and metacognitive confidence.

Simultaneously, the frontier of corvid neuroscience is advancing through non-invasive neuroimaging and electrophysiology. Andreas Nieder and colleagues have achieved electrophysiological recordings in awake, performing corvids (specifically carrion crows, Corvus corone), revealing single neurons within the corvid endbrain that encode abstract numerical quantities, rule hierarchies, and subjective sensory experiences. Neuroimaging technologies, including functional Magnetic Resonance Imaging (fMRI) and positron emission tomography (PET) tailored for avian physiology, are beginning to map the functional connectome of the corvid brain in real time, visualizing the precise functional dialogue between the hippocampal formation and the nidopallium dorsolaterale during active episodic recall.

12.3 Evolutionary Synthesis: Convergent Evolution of High-Level Cognition

The definitive theoretical triumph emerging from the Clayton-Dickinson legacy is the broad acceptance of convergent evolution as the primary engine of high-level cognition across the tree of life. For over a century, classical neurology operated under the assumption that complex cognitive functions—such as episodic memory, abstract reasoning, and forward planning—were absolute structural monopolies of the six-layered mammalian neocortex. The scrub-jay shattered this architectural dogma.

Mammals and avians last shared a common ancestor approximately 320 million years ago: a primitive, lizard-like stem amniote possessing a rudimentary, non-laminated telencephalon. From this common ancestral baseline, mammals and birds embarked on radically divergent evolutionary trajectories:

• The Mammalian Path: The pallium expanded outward and folded into a laminar, six-layered neocortical sheet, characterized by vertical columnar architecture and dense horizontal connections.
• The Avian Path: The pallium organized into dense, clustered, nuclear groupings of hyper-compacted neurons, optimizing local processing speeds and maximizing neuronal packing density within an ultra-lightweight skeletal framework.

Despite these profoundly different neuroarchitectures, both lineages arrived at the exact same computational summits:

Cognitive Function	Mammalian Neuroanatomy	Avian Neuroanatomy
Relational Spatiotemporal Memory	Laminated Hippocampal Formation (CA1, CA3, Dentate Gyrus)	Avian Hippocampal Complex (Dorsomedial Telencephalon)
Executive Control & Working Memory	Granular Prefrontal Cortex (PFC)	Nidopallium Dorsolaterale (NCL)
Dopaminergic Modulation	VTA / Substantia Nigra → Neocortex	VTA / Substantia Nigra → Pallium
Behavioral Outcome	Autobiographical Memory & Future Planning	Episodic-Like Memory & Tactical Recaching

This evolutionary convergence demonstrates that high-level cognition is not the accidental byproduct of a single, idiosyncratic anatomical design. Rather, complex cognition represents an optimal computational solution driven by demanding ecological and social niches. When an organism—whether a human navigating complex tribal landscapes, a scrub-jay managing thousands of perishable caches across seasons, or an octopus hunting across a coral reef—faces severe evolutionary pressures requiring the synthesis of space, time, identity, and social dynamics, natural selection relentlessly shapes neural tissue toward the same computational mechanisms. Nicola Clayton and Anthony Dickinson did not merely discover that scrub-jays remember where they hid their food; they fundamentally altered our understanding of the evolution of the mind, proving that the capacity to reconstruct the past and navigate the future is a widespread, magnificent evolutionary tapestry woven across millions of years of life on Earth.

Conclusion

The 1998 scrub-jay experiments conducted by Nicola Clayton and Anthony Dickinson stand as an enduring monument to the power of ethologically informed experimental psychology. By designing a paradigm that honored the natural ecology, foraging adaptations, and sensory realities of the western scrub-jay, they broke through decades of anthropocentric dogma and fundamentally redefined the study of animal memory. Their operationalization of episodic retrieval into the bound, trial-unique triad of what, where, and when circumvented the intractable philosophical trap of linguistic self-reporting, transforming a theoretical stalemate into a thriving empirical discipline.

In proving that scrub-jays selectively recover food items based on their identity, location, and specific decay kinetics over elapsed time, Clayton and Dickinson forced cognitive science to confront the deep evolutionary roots of mental time travel. Subsequent expansions into social cognition, tactical deception, prospective planning, and neuroanatomy revealed that these corvids are not mindless stimulus-response machines, but sophisticated cognitive agents capable of representing past episodes, calculating future risks, and projecting the perspectives of competitors. Their findings definitively decoupled complex executive cognition from the mammalian neocortex, demonstrating that the avian nuclear pallium achieved equal computational heights through brilliant convergent evolution.

Ultimately, the legacy of the Clayton-Dickinson breakthrough extends far beyond the borders of ornithology and comparative psychology. It compels humanity to embrace a more humble, scientifically rigorous perspective on our place in nature. We are not isolated, solitary sovereigns residing in a cognitive fortress of mental time travel; rather, we are part of an ancient, diverse evolutionary continuum of minds that remember, anticipate, and make sense of their worlds. In the quiet, deliberate choices of a scrub-jay deciding whether to dig for a worm or a peanut across the gulf of time, we witness the profound, universal power of memory—an evolutionary bridge linking the past, the present, and the future across the living world.

References

• Clayton, N. S., & Dickinson, A. (1998). Episodic-like memory during cache recovery by scrub jays. Nature, 395(6699), 272–274. https://doi.org/10.1038/29267
• Clayton, N. S., & Dickinson, A. (1999). Scrub jays (Aphelocoma coerulescens) remember the relative time of caching as well as the location and content of their caches. Journal of Comparative Psychology, 113(4), 403–416. https://doi.org/10.1037/0735-7036.113.4.403
• Clayton, N. S., Yu, K. S., & Dickinson, A. (2001). Scrub jays (Aphelocoma californica) form episodic-like memories for “what”, “where”, and “when” of cache events. Journal of Experimental Psychology: Animal Behavior Processes, 27(1), 17–29. https://doi.org/10.1037/0097-7403.27.1.17
• Clayton, N. S., Bussey, T. J., & Dickinson, A. (2003). Can animals recall the past and plan for the future? Nature Reviews Neuroscience, 4(8), 685–691. https://doi.org/10.1038/nrn1180
• Correia, S. P., Dickinson, A., & Clayton, N. S. (2007). Western scrub-jays anticipate future needs independently of their current motivational state. Current Biology, 17(10), 856–861. https://doi.org/10.1016/j.cub.2007.03.063
• 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
• Eichenbaum, H., & Fortin, N. J. (2005). Episodic memory and the hippocampus: It’s about time. Current Directions in Psychological Science, 14(5), 248–252. https://doi.org/10.1111/j.0963-7214.2005.00376.x
• Emery, N. J., & Clayton, N. S. (2001). Effects of experience and social context on prospective caching strategies by scrub jays. Nature, 414(6862), 443–446. https://doi.org/10.1038/35106560
• Emery, N. J., & Clayton, N. S. (2004). The mentality of crows: Convergent evolution of intelligence in corvids and apes. Science, 306(5703), 1903–1907. https://doi.org/10.1126/science.1098410
• Fortin, N. J., Wright, S. P., & Eichenbaum, H. (2004). Recollection-like memory retrieval in rats is dependent on the hippocampus. Nature, 431(7005), 188–191. https://doi.org/10.1038/nature02853
• Fugazza, C., Pogány, Á., & Miklósi, Á. (2016). Recall of others’ actions after incidental encoding reveals episodic-like memory in dogs. Current Biology, 26(23), 3209–3213. https://doi.org/10.1016/j.cub.2016.09.057
• Griffin, D. R. (2001). Animal minds: Beyond cognition to consciousness. University of Chicago Press. https://doi.org/10.7208/chicago/9780226227122.001.0001
• Güntürkün, O., & Bugnyar, T. (2016). Cognition without cortex. Trends in Cognitive Sciences, 20(4), 291–303. https://doi.org/10.1016/j.tics.2016.02.001
• 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
• 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
• Martin-Ordas, G., Berntsen, D., & Call, J. (2013). Memory for thinking about the future in chimpanzees and orangutans. Current Biology, 23(15), 1438–1441. https://doi.org/10.1016/j.cub.2013.06.017
• Nieder, A. (2017). Inside the corvid brain—probing the physiology of cognition in crows. Current Opinion in Behavioral Sciences, 16, 8–14. https://doi.org/10.1016/j.cobeha.2017.02.005
• Raby, C. R., Alexis, D. M., Dickinson, A., & Clayton, N. S. (2007). Planning for the future by western scrub-jays. Nature, 445(7130), 919–921. https://doi.org/10.1038/nature05575
• Suddendorf, T., & Corballis, M. C. (1997). Mental time travel and the evolution of the human mind. Genetic, Social, and General Psychology Monographs, 123(2), 133–167. https://pubmed.ncbi.nlm.nih.gov/9204544/
• 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
• Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.), Organization of Memory (pp. 381–403). Academic Press.
• Tulving, E. (1985). Memory and consciousness. Canadian Psychology / Psychologie canadienne, 26(1), 1–12. https://doi.org/10.1037/h0080017
• Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of Psychology, 53(1), 1–25. https://doi.org/10.1146/annurev.psych.53.100901.135114
• van der Vaart, E., & Clayton, N. S. (2012). Caching and recovery in scrub-jays: A computational model of episodic-like memory and planning. Cognitive Science, 36(5), 903–923. https://doi.org/10.1111/j.1551-6709.2012.01243.x