RESEARCH DOSSIER
Animal Cognition
Abstract. Animal cognition investigates how nonhuman animals acquire, represent, retain, and deploy information across ecological and social environments. This fundamental research monograph provides a critical narrative synthesis of historical comparative psychology and contemporary primary experiments involving mammals, birds, insects, fishes, and cephalopods. Its central questions concern the evidentiary distinction between observable performance and inferred cognitive mechanism, the validity of species-appropriate behavioral tests, evolutionary convergence, developmental variation, and the relation among learning, memory, abstraction, causality, social understanding, and culture. Experimental studies of scrub-jay caching, honeybee relational learning, primate false-belief anticipation, corvid tools, fish quantity discrimination, and cuttlefish executive regulation provide complementary cases. The analysis shows that diverse biological systems can achieve overlapping cognitive functions under different ecological and neural constraints, while causal interpretations depend on controls for perception, motivation, prior experience, and associative alternatives. The review identifies unresolved problems in construct validity, cross-species measurement invariance, replication, developmental sampling, and interpretation of animal sentience. It integrates existing scholarship without reporting original experiments or claiming a systematic-review protocol or meta-analysis.
Keywords: animal cognition; comparative cognition; comparative psychology; animal learning; episodic-like memory; spatial memory; tool use; physical causal cognition; numerical discrimination; social cognition; theory of mind; animal culture; inhibitory control; metacognition; evolution of cognition; cephalopod cognition; species-appropriate testing
The Scientific Object of Animal Cognition
Animal cognition concerns how nonhuman animals acquire, retain, transform, and deploy information in the regulation of behavior. Its empirical domain includes perception, attention, category formation, learning, spatial orientation, prospective action, numerical discrimination, inference, and social interaction. These capacities are investigated through observable performances in carefully arranged environments, often supplemented by neural recordings, comparative anatomy, ecological observation, and formal modeling. The scientific unit is a species- and task-situated cognitive operation whose hypothesized mechanism can be challenged by alternative predictions. Contemporary comparative work therefore asks both what a particular animal can accomplish and how performance is generated, avoiding the uninformative assignment of a single intelligence score to every organism Güntürkün & Bugnyar (2016) Santos & Rosati (2015).
Animal cognition is distinguished analytically from motor competence, motivational intensity, sensory acuity, and prior reinforcement, although these processes determine how any test is performed. A failed discrimination in an apparatus designed for human vision might reflect inaccessible stimuli rather than a missing representation. Conversely, apparently advanced behavior can arise through learned associations or structured environments without the proposed abstract mechanism. The central evidentiary requirement is to state a process hypothesis, specify a measurable response, and design controls that distinguish it from simpler and equally plausible accounts. Such comparisons must take seriously the visual fields, manipulating organs, reward systems, lifetimes, and social ecologies of participating taxa Miller et al. (2026) Bateson & Laland (2013).
This monograph reconstructs animal cognition as an interdisciplinary psychological research field, tracing its intellectual foundations and assessing major experimental traditions. It integrates investigations in birds, mammals, insects, fishes, and cephalopods while recognizing uneven taxonomic coverage. The aim is a critical synthesis of established scholarship: individual studies establish particular behavioral capacities under particular conditions, converging experiments constrain explanatory models, and phylogenetic arguments address evolutionary origin. Evidence concerning sentience and subjective experience is examined as a neighboring question with its own standards. Human–artificial comparisons are introduced only as methodological extensions to comparative cognition, not as experimental claims about artificial experience Schnell et al. (2026) Miller et al. (2026).
Historical Foundations and Comparative Explanation
Animal behavior entered modern experimental inquiry through overlapping traditions of natural history, evolutionary explanation, associative learning, ethology, and cognitive psychology. Long-term observations demonstrated reliable species-specific behavioral organizations, while controlled experiments isolated the effects of experience and stimulus contingencies. The historical tension concerned the inferential status of internal processes: the same action could be described as a learned response, an adaptive solution, or a cognitively represented relation. Modern studies preserve contributions from these traditions by asking which explanatory level is relevant to a particular question, rather than assuming one historical vocabulary explains every animal action Bateson & Laland (2013) Horner & Whiten (2005).
Tinbergen's influential framework distinguishes causal mechanisms, individual development, adaptive consequences, and evolutionary history. Bateson and Laland's reassessment shows why these questions still organize behavioral science while requiring contemporary refinements Bateson & Laland (2013). A study of nut cracking by chimpanzees may examine movement coordination and material properties, how young individuals acquire the action, its consequences for food access, and the historical distribution of tool-related behavior across populations. Findings at one level do not automatically establish the others. Demonstrating reliable tool manufacture is evidence about performance and acquisition; an adaptive origin requires a separate evolutionary argument with ecological or comparative support Boesch & Boesch (1990).
The turn toward experimental cognition sharpened a demand already present in careful ethology: distinguish the observable behavioral regularity from the explanatory model. Studies of scrub-jay cache recovery use the bird's food-storage ecology as a source of meaningful test conditions, and studies of dolphin sponge carrying combine field observation with information about relatedness and transmission Clayton & Dickinson (1998) Krützen et al. (2005). These traditions demonstrate how laboratory experiments and field science can cooperate. The historical development of animal cognition is thus a development of inferential methods, particularly experiments that translate hypotheses about memory, agency, and physical causation into discriminable behavioral outcomes.
Cognitive Constructs, Representation, and Inference
A cognitive construct is a theoretical term tied to measurable consequences and a proposed organization of information processing. Memory can be operationalized through delayed retrieval, recognition, or anticipatory behavior; categorization through transfer to new exemplars; inhibitory control through performance when an immediate response must be suppressed. Operational measures are essential, but no one task is identical with the capacity it is intended to measure. Multiple measures should converge when they depend on the same construct, and diverge when the hypothesized processes differ. For example, what–where–when cache performance operationalizes episodic-like remembering without directly measuring the subjective quality associated with autobiographical recollection Clayton & Dickinson (1998) Schnell et al. (2021).
Representational explanations propose that an animal uses states or relational structures that stand for absent objects, prospective outcomes, quantities, or other agents' actions. These explanations are scientifically valuable when they generate new predictions, such as transfer across stimuli or choices made before an anticipated need occurs. An associative model may explain the same training performance but predict failure following particular contingency changes. The correct test compares explicitly specified models rather than attaching mentalistic labels to a successful trial. Honeybee matching experiments, for example, investigated generalization of sameness and difference to novel stimuli, thereby testing a relation beyond memorization of a particular rewarded picture Giurfa et al. (2001).
Inference about representation also depends on the analysis of behavior over time. Latency, exploration, search sequence, persistence, correction after errors, and trial-to-trial change can reveal constraints invisible in final accuracy. Eye-tracking studies of apes observe anticipatory looking, which has different affordances and limitations from an active object-choice test Krupenye et al. (2016). The same recorded pattern can remain compatible with more than one account until new manipulations separate them. Accordingly, this monograph treats representational claims as explanatory hypotheses ranked by the quality and specificity of their experimental support.
The Comparative Method and Species-Appropriate Testing
Cross-species research requires deliberate calibration of apparatus, reward, perceptual modality, training history, and opportunity to respond. A transparent cylinder, for instance, may require suppression of direct reaching in one species and a substantially different visual or motor strategy in another. Species differences in task accuracy can therefore combine cognitive ability with inhibition, familiarity, and morphology. Large-scale comparisons are most informative when their tasks measure equivalent functional operations and when conclusions remain sensitive to the effects of phylogeny and ecological specialization. MacLean and colleagues' comparative study of self-control demonstrates both the ambition of standardized testing and the interpretive challenge of combining diverse taxa MacLean et al. (2014).
Ecological validity means that experimental questions are meaningful within the sensory and behavioral repertoire of an organism. A food-caching bird has opportunities to reveal long-interval location-content memory that a standard laboratory maze may miss. Octopuses manipulate objects through distributed arms and sensory capacities unlike those of primates; fish may be tested by spontaneous shoaling preferences rather than grasping or verbal report. A test should therefore establish what response a species can readily produce before evaluating whether it discriminates alternatives Clayton & Dickinson (1998) Piffer et al. (2012) Finn et al. (2009). Comparable constructs can require different implementations across lineages.
Methodological triangulation integrates controlled trials, field observation, development, neural evidence, and computational alternatives. Repeated measurement within individuals estimates reliability, while independent laboratories test whether a result depends on an idiosyncratic procedure or colony. Comparative inference must also acknowledge sampling limitations: a few captive representatives cannot establish a species-wide capacity or its frequency in natural populations. A 2026 review of corvid cognitive development highlights small samples, captive biases, and gaps in longitudinal evidence Miller et al. (2026). Those limitations motivate preregistered designs, explicit exclusion criteria, and transparent reporting of unsuccessful replications and null results.
Evidence Review, Causal Claims, and Research Ethics
The present research is a critical narrative synthesis of original peer-reviewed experiments, methodological papers, and integrative reviews accessible through scientific publishers and research repositories. It has no registered search protocol, exhaustive database screening record, or pooled effect-size computation and therefore does not claim systematic-review or meta-analytic status. Evidence is considered in terms of experimental manipulation, transfer, control conditions, sample selection, replicability, ecological fit, and correspondence between observation and conclusion. The strength of an interpretation grows when rival accounts make contrasting predictions and the experiment observes the predicted difference rather than relying on an anecdote Horner & Whiten (2005) Benson-Amram et al. (2016).
Causal claims have several levels. Manipulating a reward schedule can establish its effect on choices in that task; relating brain size to puzzle success supports a cross-species association, with additional phylogenetic and sampling assumptions; reporting a tool behavior in the wild establishes the occurrence of a technique. The suggestion that a successful animal understands invisible physical forces demands stronger evidence about prediction and transfer. Researchers should present quantitative results only with their verified denominators, uncertainty, and analysis methods, while separating within-individual learning from differences between species or populations Benson-Amram et al. (2016) Weir et al. (2002).
Animal welfare is both an ethical requirement and a condition of scientific validity. Stress, deprivation, confinement, unrepresentative housing, and social isolation can alter motivation, attention, and learning. Enrichment, appropriate training and voluntarily engaged testing can improve welfare and yield more interpretable measurements. Conclusions about consciousness and pain require separate lines of convergent evidence; evaluations of cephalopod sentience show an active research program with sharply unequal evidence across taxa Schnell et al. (2026). Ethical research therefore couples the refinement of experimental methods with the cautious interpretation of claims that may influence husbandry, conservation, and legal treatment.
Learning, Contingencies, and Behavioral Flexibility
Learning concerns experience-dependent changes in the probability, organization, and context sensitivity of behavior. In animal cognition it supplies both a mechanism of adaptation and a possible alternative explanation for performances labeled insight, reasoning, or planning. Reinforcement histories can generate sophisticated discriminations through successive approximations, while animals can also learn which stimuli predict reward, danger, or the actions of other individuals. The empirical problem is to determine whether observed success depends on the particular training contingencies or generalizes to new causal arrangements. Changing reward locations, reversing cue validity, and examining first-trial transfer are methods for separating memorized routines from broader relational competence Horner & Whiten (2005) Giurfa et al. (2001).
Species differ in their sensory priorities, learning opportunities, and ecological incentives, so the measured rate of acquisition is an interaction between task and organism. A specialized forager may rapidly learn spatial associations relevant to its diet yet show little interest in arbitrary laboratory symbols. Other animals reveal learning through changes in gaze or route selection rather than direct manipulation. Such findings complicate universal rankings of intelligence because learning rate is partly a function of what must be learned, how it is presented, and what is valued. Appropriate assessments report trial history and motivational conditions rather than describing an individual as universally clever or incapable Piffer et al. (2012) Miller et al. (2026).
Behavioral flexibility is a particularly useful construct because it tests adjustment after the environment changes. Reversal learning, detour tests, set shifting, and new object problems can assess different forms of flexibility, though each includes its own sensorimotor and inhibitory demands. An animal that quickly abandons an obsolete strategy may be adapting through value updating rather than explicit causal explanation. To distinguish these alternatives, researchers can vary how rapidly contingencies change, whether old cues remain salient, and whether a learned rule transfers to novel stimuli. Flexibility must ultimately be described as a pattern of adaptation under stated conditions MacLean et al. (2014) Benson-Amram et al. (2016).
Perception, Attention, and Ecological Information
Cognition begins with an organism's access to information through its sensory systems. Foraging, navigation, predator detection, mate recognition, and social communication all depend on patterned selection among environmental signals. Comparative tests that use human-preferred contrasts may underestimate species that rely on different wavelengths, movement cues, chemical gradients, or acoustic features. The methodological implication is that adequate sensory testing precedes high-level interpretation. A failure to distinguish a visual display may result from discriminability, not concept formation. An apparent preference for one option may depend on brightness, surface area, movement, or odor rather than the variable selected by the experimenter Agrillo et al. (2009) Giurfa et al. (2001).
Attention determines which available information is sampled and how it influences action. In foraging tasks, animals may prioritize salient food cues over relational properties; in social encounters, gaze direction and movement signal possible future actions. Selective attention can produce impressive performance on familiar cues yet limit transfer when a relevant dimension changes. Researchers therefore need to vary irrelevant perceptual properties and evaluate whether success is preserved. In numerical choice experiments, for example, controlling continuous quantity variables is essential to establish that the number of items contributed to the observed preference Agrillo et al. (2009) Piffer et al. (2012).
The ecological approach situates attention within actual informational demands. A scrub-jay remembers scattered food locations because the activity naturally involves delayed return to concealed sites; a dolphin's foraging techniques depend on undersea substrates and prey distributions. Experimentally meaningful questions emerge when those ecologies suggest candidate variables and distinctive predictions. Laboratory simplification can then isolate the effects of information that are difficult to manipulate in the field. Strong comparative psychology moves iteratively between ecological observation and controlled discrimination rather than assuming that a single generic testing arena reveals every species' cognitive organization Clayton & Dickinson (1998) Krützen et al. (2005).
Categories, Abstraction, and Relational Rules
Categorization is the grouping of distinct stimuli or events according to shared properties that guide behavior. Simple stimulus discrimination can be learned without treating an entire class as a unified kind; generalization to new exemplars provides stronger evidence of a rule or common representation. Investigators may vary color, shape, material, position, or modality to establish which relations carry over when familiar details change. Categories can be perceptual, functional, or social, and their boundaries depend on ecological relevance as well as training. Honeybee experiments provided a rigorous test of an abstract relation by asking whether an insect could use a matching or nonmatching rule with changed stimuli Giurfa et al. (2001).
Giurfa and colleagues trained honeybees with delayed matching-to-sample and nonmatching-to-sample procedures, testing transfer beyond the specific examples presented during training Giurfa et al. (2001). This result is important because a correct response to a novel relation is harder to explain through rote attraction to a previously rewarded image. Nevertheless, the capacity demonstrated is bounded by the experimental configuration. It does not by itself establish humanlike conceptual vocabulary, explicit logical proof, or broad domain-general reasoning. The theoretically important result is that a relatively small and differently organized nervous system can support adaptive discrimination of relationships across changing inputs.
Work with gray parrots supplies another perspective on abstraction and communicative symbols. Long-term experiments reviewed by Pepperberg examined same–different relations, attributes of objects, and limited numerical expressions Pepperberg (2006). These accomplishments involve extensive social training and a particular interactional environment, which become part of the explanatory story. A vocabulary acquired in structured human–bird exchanges is evidence that such learning and discrimination can occur under those conditions; it cannot simply be generalized to spontaneous language production throughout a wild population. Together, insect and parrot studies show why research must distinguish the representational operation, the route of acquisition, and the opportunities a species has to express the capacity.
Numerical Cognition and Quantity Discrimination
Numerical cognition concerns sensitivity to quantities, relative magnitudes, and relationships among sets. Animal experiments may measure spontaneous preference for the larger shoal, trained selection of arrays, or decisions involving food amounts. The primary methodological difficulty is that discrete number covaries naturally with total surface area, occupied space, density, movement, and other continuous cues. A creature may choose more items because it detects greater total extent rather than counting discrete individuals. Researchers must design controls that systematically dissociate these variables before treating quantity preference as numerical discrimination Agrillo et al. (2009) Piffer et al. (2012).
Agrillo and colleagues investigated use of number by fish, demonstrating how aquatic social choice provides an ecologically meaningful assay and how stimulus properties must be controlled Agrillo et al. (2009). Piffer and colleagues then examined small and large number discrimination in guppies Piffer et al. (2012). These studies support quantitative discrimination in organisms whose motor and perceptual systems differ considerably from those of primates. The data should be interpreted through specific tested ratios, group sizes, and training conditions rather than as evidence of unrestricted mathematical reasoning. A small-set response and a large-set approximate discrimination may reflect partly different computational operations, a proposition requiring experiments designed to distinguish them.
Comparative numerical research also informs the question of whether a task recruits exact representations, approximate magnitude estimation, or learned response rules. Across species, apparent numerical thresholds cannot be compared without matching motivational stakes, perceptual displays, and sufficient individual trials. Statistical inference is especially sensitive to the repeated testing of the same animals and to selective reporting of successful contrasts. The strongest evidence for a numerical process comes from transfer to new stimuli after continuous quantities have been varied independently. The comparative objective is a mechanistic map of quantity processing rather than the assignment of a human developmental age to a fish or an insect Agrillo et al. (2009) Pepperberg (2006).
Episodic-Like Memory and the Temporal Structure of Experience
Episodic-like memory is the behavioral retention of integrated information about what occurred, where it occurred, and when it occurred. The term was adopted to make a scientifically tractable distinction between a performance criterion and the phenomenal recollection described in human memory research. The original scrub-jay paradigm asked whether birds returned to particular cached foods according to both their locations and the elapsed interval Clayton & Dickinson (1998). Satisfying the what–where–when test establishes an important organization of event-relevant information, but does not provide direct access to how the animal experiences remembering. The theoretical challenge is to determine which temporal representations and memory-binding processes produce the flexible pattern.
Further tests of scrub-jay memory used changes in caching arrangements and recovery conditions to evaluate whether birds integrated food type, site, and relative time rather than merely memorizing isolated location preferences Clayton & Dickinson (1998) Dally et al. (2006). A key alternative is that learned decay schedules and contextual cues guide action without a unified autobiographical representation. Discriminating such accounts requires trial-unique events, unexpected probes, manipulation of intervening cues, and control for immediate motivation. Where multiple explanations survive, the responsible description remains episodic-like, with clear identification of the behavioral criteria met.
Cuttlefish research extends the comparison beyond vertebrate lineages. Schnell and colleagues studied whether episodic-like memory remains accessible with age in cuttlefish, using the what–where–when framework to investigate memory in a radically different nervous system Schnell et al. (2021). Similar experimental vocabulary can connect these species, but equivalent performance does not establish homologous neural implementation or an identical subjective experience. Development and aging may diverge even when behavioral operations appear parallel. The cumulative research question is whether integrated event memory is a recurrent functional solution to ecological demands, and which distinct biological mechanisms realize it.
Future-Oriented Behavior and Prospective Preparation
Prospective cognition concerns behavior organized around conditions that are expected to occur later. Animals routinely cache food, build shelters, migrate, and prepare nesting sites, but the prospective interpretation of a single action depends on whether it is explained by current cues, evolved behavioral programs, learned contingencies, or representations of future needs. Researchers have sought experiments that separate immediate motivation from a future requirement. In western scrub-jays, Raby and colleagues tested food allocation to locations associated with later hunger or different breakfast availability Raby et al. (2007). The important evidence concerns how choices shifted in response to anticipated conditions while present opportunity remained controlled.
The Raby experiments demonstrated that scrub-jays adjusted caching in ways relevant to a subsequent breakfast context Raby et al. (2007). This finding challenged strong claims that nonhuman animals are incapable of planning for future needs. Its interpretation remains linked to the exact training schedule, site histories, and motivational conditions used. A flexible response is consistent with future-oriented information, but researchers must still evaluate how much derives from learned site values or cached-food preferences. Tests with unanticipated future constraints and novel combinations of food, location, and interval could discriminate models that reproduce the same trained behavior.
Planning is therefore a family of functions rather than an all-or-none capacity. Preparatory storage, prospective memory, sequential action selection, and imagination-like simulation require different operational tests. The ability to arrange resources for tomorrow does not establish detailed representation of an unlimited distant future. Similarly, temporal sensitivity in cache recovery does not automatically entail ability to plan for a novel need. Comparative synthesis should identify the type of future-directed response demonstrated and its ecological role while using stronger experiments to test the richer explanatory model Clayton & Dickinson (1998) Raby et al. (2007).
Tool Use, Manufacture, and Material Problem Solving
Tool use links an animal's goal to the physical properties of an external object. In observational research, investigators first establish what materials are selected, modified, transported, and employed, then ask how the technique is acquired and whether the animal understands relevant causal relations. Boesch and Boesch's comparative observations of wild chimpanzee populations documented differences in tool inventories and manufacture Boesch & Boesch (1990). These findings link cognitive questions to real foraging behavior while requiring separate evidence for individual invention, social transmission, and the mechanical competence of the tool user.
New Caledonian crows provide a distinct example of material transformation. Weir, Chappell, and Kacelnik reported hook shaping by a crow in an experimental context Weir et al. (2002). The behavior was notable because the bird altered material into a functional implement, but a single demonstration cannot establish the prevalence of this method or a complete causal model of bending and leverage. Repeated studies across different materials and naïve individuals are needed to separate preexisting dispositions, learned contingencies, and flexible physical prediction. Tool manufacture also varies with ecological opportunities: access to suitable raw materials and prey constraints shapes which techniques can develop.
Tool behavior should be studied as an interaction between material affordances, perceptual feedback, motor control, experience, and environmental selection. A method may be complex because the object is difficult to manipulate even when the underlying decision is simple; an apparently elementary act may reveal sophisticated prospective selection. Comparative work gains explanatory precision by analyzing the intermediate actions leading to success. The occurrence of coconut-shell transport in octopuses further demonstrates that tool-oriented behavior cannot be confined to vertebrate limbs or mammalian brain plans Finn et al. (2009). Whether different lineages solve the same functional problem through shared computations remains a separate question.
Physical Causality and Experimental Alternatives
Physical causal cognition refers to the use of relations among objects, forces, mechanisms, and consequences when selecting an action. The operational challenge is distinguishing causal prediction from reinforcement to superficial features. An animal trained to push a particular object may succeed because its color was rewarded, while a genuinely relational strategy should adapt when the same physical mechanism is realized with a new material or form. Useful experiments vary transparency, mechanical constraints, object identity, or the relation between action and outcome. Horner and Whiten compared chimpanzees and children observing tool-use demonstrations under transparent and opaque conditions to probe the role of causal information Horner & Whiten (2005).
Observed copying has several interpretations. Imitation can reproduce another individual's action sequence, emulation can reproduce the result or environmental transformation, and stimulus enhancement can direct attention to an object without transmitting the action. Transparent apparatuses help test whether irrelevant movements are retained when their lack of mechanical effect is visible. In the Horner and Whiten study, comparing populations and conditions exposed important differences in what was reproduced and how visible causal information mattered Horner & Whiten (2005). Such evidence supports a process-level account more precise than calling one species rational and another imitative.
Causal representations are not directly visible. Their evidentiary status improves when animals predict outcomes before acting, transfer to mechanisms that share structure but not surface features, and update behavior after interventions that contradict learned cues. The distinction between a physical possibility and a demonstrated psychological capacity is fundamental. Even excellent tool performance may depend on practice or selection among previously reinforced actions. Researchers should compare formal associative and causal models using held-out conditions, rather than treating failure to disprove the sophisticated model as proof of it Weir et al. (2002) Boesch & Boesch (1990).
Inhibitory Control, Delay, and Executive Regulation
Inhibitory control refers to the modulation of a prepotent response when circumstances favor an alternative. Comparative experiments use detours, transparent obstacles, delays, reversal tasks, and choice among immediate and delayed outcomes. These paradigms recruit partially distinct processes, including attention, motor inhibition, learned rule application, and motivational valuation. MacLean and colleagues studied self-control across a large sample of species using common behavioral tasks and comparative modeling MacLean et al. (2014). Their project illustrates the explanatory potential of coordinated protocols while showing the need to examine task equivalence and the influence of ecology across taxa.
A delayed reward is not always a pure test of self-control. An animal's willingness to wait depends on food preference, hunger, reward predictability, previous experiences with experimenters, and the opportunity cost of waiting. Likewise, a transparent barrier can create a strong motor attraction to a visible target and thereby penalize species whose normal interaction with objects differs from the device's assumptions. Replication across delay intervals, reinforcer types, and alternative motor responses makes a claim more robust. Performance distributions within species should be examined alongside differences in mean scores MacLean et al. (2014) Schnell et al. (2021).
Schnell and colleagues reported delay-of-gratification behavior in cuttlefish, extending executive-control questions to cephalopods Schnell et al. (2021). The importance of such a result lies in convergent functional regulation under very different nervous system architectures. It does not justify equating all forms of patience, nor infer that every individual performs similarly. A comparative theory should predict when delaying an action has ecological value and identify the learning and neural conditions that sustain the response. Inhibitory control is best understood as context-dependent behavioral regulation, not a fixed species-level character trait.
Metacognition, Confidence, and Information Seeking
Metacognition concerns monitoring or regulating an organism's own cognitive operations. In humans the construct often includes explicit reports of confidence and reflection, whereas animal experiments rely on choices to seek information, decline uncertain tests, or adjust behavior according to task difficulty. These behavioral procedures reveal metacognitive-like regulation when they track the animal's access to evidence and cannot be explained entirely by known external cues. Reviews of nonhuman primate research emphasize the variety of experimental paradigms and the continuing difficulty of specifying the underlying process Subias et al. (2025).
An animal may choose an uncertainty response because difficult trials have a distinctive stimulus signature, or because accepting the trial has a poor reward history. These possibilities can imitate a decision based on monitoring one's memory. Strong designs dissociate objective difficulty from the individual's informational access, use novel stimulus sets, and manipulate whether relevant evidence was actually available. A decline option that generalizes across contexts is more informative than one trained only for a particular visual contrast. Yet even successful flexible information seeking leaves open questions about the content and conscious form of the monitoring representation Subias et al. (2025) Krupenye et al. (2016).
Comparative metacognition also requires close scrutiny of motivational and procedural differences. Some animals have learned that skipping hard trials minimizes effort, while others might favor exploratory responses despite uncertainty. Confidence-like behavior and subjective certainty should therefore be separated in language and in experimental interpretation. The aim is to identify computational and behavioral mechanisms by which a cognitive system evaluates the quality of its information. Evidence from primate studies supports this research program but does not imply a unitary capacity shared identically by every animal or across every cognitive domain Subias et al. (2025).
Individual Differences, Innovation, and Problem-Solving Ecology
Problem solving is often tested with novel apparatuses that demand manipulation, exploration, or learned sequences to obtain a reward. Success provides information about the interaction of behavioral flexibility, persistence, motor ability, prior experience, and incentive value. A measure of innovation that records only whether a puzzle opens may confound a cognitive solution with sheer activity or stronger hunger. Investigators can distinguish these variables through baseline motor controls, latency measures, repeated tasks, and independent measures of exploration. Field and zoo studies are valuable because they can reveal the distribution of different strategies within populations Benson-Amram et al. (2016).
Benson-Amram and colleagues examined problem solving in mammalian carnivores in relation to brain-size measures, using a novel apparatus across species Benson-Amram et al. (2016). Their findings contribute to debates about the connection between neural investment and cognitive performance. The analysis nevertheless depends on species sampling, phylogenetic structure, body-size adjustment, ecological history, and whether the chosen puzzle demands functionally equivalent behavior. An association between relative brain size and a measured solution rate should not be reinterpreted as evidence for one universal intelligence mechanism. Alternative explanations must be evaluated with additional tasks and measurements.
Individual variation is scientifically important. Repeated trials can reveal whether successful animals consistently show flexible strategies or happened to encounter a solution through chance manipulation. Age, sex, rearing environment, social position, and previous encounters with objects may alter performance; their influence requires study rather than assumption. Where innovation is observed in natural populations, researchers must ask whether new behavior persists, spreads socially, or yields measurable ecological consequences. Cognition becomes evolutionarily interpretable when behavioral mechanisms can be linked to developmental pathways and their actual costs and benefits Miller et al. (2026) Santos & Rosati (2015).
Perspective Taking and False-Belief Experiments
Perspective taking concerns sensitivity to the informational relation between another individual and the environment. An animal may behave differently when a competitor can or cannot see food, but this alone does not settle whether it represents the competitor's belief. A stronger test creates a conflict between what the observer believes and what is actually true, while controlling for the animal's own knowledge of the situation. False-belief experiments investigate whether behavior predicts action on the basis of an outdated representation. Since animals cannot normally provide verbal explanations of beliefs, behavioral anticipation and looking patterns become major observational methods Krupenye et al. (2016).
Krupenye and colleagues reported anticipatory looking by great apes toward locations where an actor with a false belief was expected to search Krupenye et al. (2016). The study provided evidence relevant to one component of mental-state understanding: prediction of an agent's action when the agent lacks correct information. Kano and colleagues later used apes' own experience in a related false-belief test, extending the attempt to distinguish mentalistic from alternative explanations Kano et al. (2019). These results should be described with their measured outcome, anticipatory behavior under a controlled scenario, rather than as a blanket demonstration of humanlike theory of mind.
The interpretation of false-belief performance remains theoretically contested because animals may track perceptual access, familiar action scripts, object locations, or associative expectations. Follow-up experiments must isolate these candidate variables while preserving a task animals can understand and attend to. The methodological lesson is general: demonstrating a response compatible with belief representation is a contribution to theory of mind research, and proving that belief representation is necessary demands additional predictive tests. The field benefits from converging evidence through gaze, helping, search, and spontaneous-interaction paradigms, each with distinct limitations Krupenye et al. (2016) Kano et al. (2019).
Social Learning, Imitation, and Transmission Mechanisms
Social learning is a change in behavior associated with observing or interacting with other individuals. Its mechanisms include local enhancement, stimulus enhancement, copying of movements, learning from outcomes, and more elaborate forms of guided instruction. These mechanisms differ in what information is acquired: where to act, which object matters, which movement to perform, or which causal relationship produces a result. A claim that one animal copied a technique should specify the demonstrated behavior and rule out the possibility that independent learning followed social exposure. Experimental work on imitation and emulation examines how visibility of causal structure changes the actions reproduced by learners Horner & Whiten (2005).
Horner and Whiten's comparison of chimpanzees and young children demonstrated the importance of apparatus visibility and apparently irrelevant actions in imitation research Horner & Whiten (2005). The experiment provides a structured test of action copying rather than an unrestricted measure of intelligence. Such results can also vary with social motivation, trust in demonstrators, experience with human instructions, and the affordances of a particular puzzle. Researchers should compare trained and naïve subjects, vary demonstrator identity, and test performance when only the resulting environmental change is visible. These manipulations identify channels through which behavior can spread.
Social transmission does not require explicit teaching or the representation of another's beliefs. A socially acquired foraging method can become stable because learners repeatedly encounter conspecific demonstrations, suitable materials, and locally reinforcing outcomes. Conversely, an apparently shared technique may recur through independent discovery when populations inhabit similar habitats. Robust social-learning research combines individual developmental trajectories, controlled transmission tests, and ecological observation. The connection with cultural evolution arises when a learned pattern persists across individuals and generations under a definable transmission process Whiten et al. (1999) Krützen et al. (2005).
Animal Culture and Population Traditions
Animal culture is investigated through behavioral traditions whose distribution and persistence involve social transmission. The empirical question is not whether a species resembles human civilization, but whether population differences depend substantially on learned information shared among individuals. Whiten and colleagues synthesized long-term field observations of chimpanzees across African sites and described systematic variation in customary behaviors Whiten et al. (1999). The historical importance of this program lies in treating animal culture as a comparative research subject that can be examined through converging observations, ecological controls, and the study of how juveniles acquire practices.
Field comparisons must evaluate alternatives. If populations differ in tool materials, prey, predator presence, or genes, those variables may produce different behavior without cultural transmission. Krützen and colleagues combined observations of dolphin sponge-tool foraging with genetic analyses in Shark Bay, providing evidence for matrilineal transmission within a population Krützen et al. (2005). This example illustrates why interdisciplinary methods matter: behavioral ecology identifies the technique and its environmental context, genetic patterns constrain inheritance explanations, and longitudinal observations reveal which individuals can transmit the behavior. No single method exhausts the causal question.
Cultural persistence and cumulative culture should be distinguished. A stable tradition can arise from copying and repeated opportunity without progressive improvement across generations. Cumulative change requires evidence that socially acquired modifications are retained and build on earlier innovations. Claims of animal culture should therefore state whether the evidence concerns local convention, demonstrator effects, repeated transmission, or cumulative transformation. The distinction matters because processes sufficient to preserve traditions may differ from those required for open-ended technological accumulation. Comparative work can identify the precise conditions under which socially transmitted information becomes a population-level historical resource Whiten et al. (1999) Horner & Whiten (2005).
Communication, Symbols, and Cross-Species Interpretation
Animal communication consists of signals and responses whose functions depend on evolved dispositions, experience, social context, and receiver behavior. Calls, gestures, displays, chemical cues, and coordinated movements can carry information relevant to action without implying humanlike words or grammar. Cognitive research asks whether signals are flexibly directed, whether animals distinguish informative from uninformative cues, and how learning changes communicative competence. Human researchers face an additional challenge when interpreting signals produced within unfamiliar sensory worlds. A receiver's behavior must be measured alongside production if functional information transfer is to be established Herrmann et al. (2007) Pepperberg (2006).
Pepperberg's long-term research with gray parrots investigated the acquisition and use of human speech sounds in tests involving attributes, categories, and quantities Pepperberg (2006). These findings challenge the assumption that all avian vocal imitation lacks discriminative function. They also underscore the influence of structured training and human partnership on what the bird can express. The scientific claim is tied to specific test performance under those conditions, not to the assumption that spoken labels in parrots function identically to open-ended human language. Transfer, spontaneous use, and comprehension tests have distinct evidentiary implications.
Communication research benefits from distinguishing signal production, contextual sensitivity, reference-like interpretation, and symbolic relations. A learned association between a sound and a reward may be reliable communication without demonstrating an abstract concept; a signal used across varied social circumstances may reflect flexible production without arbitrary syntax. Comparative models should make these distinctions explicit and then investigate developmental acquisition and ecological function. Because communication can be multimodal, experimental designs that privilege human speech can miss sophisticated nonvocal channels. The broader cognitive objective is to explain how information reaches, changes, and coordinates multiple agents within an ecological community Pepperberg (2006) Dally et al. (2006).
Evolutionary Convergence, Homology, and Neural Diversity
Comparative cognition studies both shared ancestry and independent evolutionary solutions. Similar behaviors in mammals, birds, and cephalopods can arise through different neural architectures and ecological histories; apparent similarity alone therefore cannot establish a homologous mechanism. Conversely, divergent behavior can arise from homologous systems acting in different environments. An evolutionary interpretation must identify a trait, place relevant lineages in phylogenetic context, and evaluate ecological or developmental alternatives. Research on bird cognition illustrates this problem because sophisticated behavioral performances occur in forebrain organizations different from the layered mammalian neocortex Güntürkün & Bugnyar (2016) Amodio et al. (2019).
Güntürkün and Bugnyar reviewed evidence that avian cognition can support complex operations in a neural architecture whose pallial organization differs from the mammalian pattern Güntürkün & Bugnyar (2016). These findings undermine any simple equation of one mammalian anatomical structure with the possibility of complex cognition. The argument does not require claiming that every bird possesses equal capacities. It proposes that functional networks and evolutionary histories, rather than superficial anatomical resemblance, guide comparative explanation. Neural measurements and causal interventions are therefore valuable when combined with behavioral tasks, rather than substituted for tests of what the animal actually does.
Cross-lineage comparison is most compelling when the same function is operationalized with methods suited to each organism. Delayed reward choice in cuttlefish, prospective caching by jays, and primate executive tasks all involve regulation across time but may rely on different motivational and control mechanisms Schnell et al. (2021) Raby et al. (2007) MacLean et al. (2014). A credible account of convergence should specify which computational demand is shared, where the underlying circuits diverge, and which ecology could favor similar performance. The appropriate end product is a network of comparable functions and historically situated mechanisms rather than a ladder with humans occupying the only informative endpoint.
Insect and Fish Cognition Beyond Vertebrate Models
Insect cognition is studied through small nervous systems capable of discriminative learning, navigation, and context-sensitive behavioral selection. The honeybee experiments on matching and nonmatching were influential because the trained relation was tested with altered stimulus types, directly examining transfer beyond a fixed rewarded picture Giurfa et al. (2001). Such results establish that abstract relational discrimination can be expressed in a compact neural system under a specific laboratory procedure. They do not automatically demonstrate conscious awareness of a rule or a general humanlike faculty. An adequate theory must explain how the achieved generalization follows from the insect's perceptual and learning architecture.
Fish cognition offers experiments built around natural social choices and quantitative comparisons. Guppy number-discrimination research demonstrated that behavioral preference tasks can reveal sensitivity to number-like differences without requiring grasping, speech, or human social instruction Agrillo et al. (2009) Piffer et al. (2012). However, fish respond to continuous visual quantities and social factors as well, making carefully controlled stimulus sets indispensable. The relevant question is which controlled numerical contrasts consistently influence choice. A robust comparative program separates direct perceptual discrimination from learned symbolic numerical rules and reports the tested ranges rather than inferring unbounded arithmetic competence.
The insect and fish literatures show the importance of expanding taxonomic sampling without sacrificing methodological rigor. A function observable in bees may arise through processes different from those of parrots; failure to observe a function in guppies may reflect the ecological irrelevance of the apparatus. Including such species tests theories that were originally framed around primates and human cognition. The evolutionary question then becomes which information-processing solutions recur under similar demands, which depend on lineage-specific developmental pathways, and which apparent differences are produced by the measurement process itself Giurfa et al. (2001) Miller et al. (2026).
Cephalopod Cognition and Independent Evolutionary Routes
Cephalopods provide a major comparative test because their nervous systems, morphologies, and life histories differ profoundly from those of mammals and birds. Octopuses can manipulate objects through flexible arms and distributed sensory-motor control; cuttlefish can perform learned choices across temporally separated opportunities. These performances suggest that diverse biological organizations can support forms of exploratory behavior, memory, and behavioral regulation Finn et al. (2009) Schnell et al. (2021). A cognitive explanation must nevertheless identify particular observed operations rather than attributing a global intellectual rank to an entire class of animals.
Finn, Tregenza, and Norman described octopuses transporting coconut-shell halves for later protective use, a behavior that involved carrying objects whose utility emerged in a subsequent context Finn et al. (2009). This case motivates research on prospective material selection, but the report alone does not specify the animal's representation of future danger. Cuttlefish studies supply experimental complements: controlled tasks have examined delayed reward and episodic-like retention across age Schnell et al. (2021) Schnell et al. (2021). Comparing field and laboratory evidence clarifies why a natural behavior may establish ecological relevance while experimental manipulation probes underlying information processing.
Amodio and colleagues argued that cephalopods complicate evolutionary accounts tying complex cognition exclusively to prolonged development, slow life histories, or elaborate social systems Amodio et al. (2019). Their analysis is a theoretical synthesis about evolutionary pathways, not a direct demonstration of a single selective event. Updated research on cephalopod sentience evaluates behavioral and neurobiological criteria while emphasizing uneven evidence among octopus, cuttlefish, squid, and nautilus Schnell et al. (2026). Cognitive performance and sentience remain related but separately investigated domains. Comparing them responsibly requires explicit hypotheses about behavioral regulation, physiological organization, and markers of possible affective experience.
Development, Experience, and Ontogenetic Change
Animal cognition develops through interactions among inherited neural organization, physical growth, ecological exposure, and individual experience. An observed species-level performance may be attained only after long practice, social participation, or the maturation of motor capacities. Therefore, an adult success rate cannot tell us when the underlying operation appears or how it is acquired. Longitudinal research can examine emergence within individuals, while age-matched cross-sectional comparisons provide a different and usually less decisive inference about developmental change. Distinguishing learning effects from maturation is essential for an explanatory account rather than a catalogue of behavioral abilities Miller et al. (2026).
Miller and colleagues' 2026 review of corvid cognitive development identified 47 empirical studies spanning 16 species in a structured search, while emphasizing the limitations of small samples, captive populations, and sparse longitudinal measurement Miller et al. (2026). These observations are important because corvids have become prominent evidence for sophisticated animal cognition, yet many developmental trajectories remain incompletely characterized. A striking adult performance in a food cache or tool task may depend on ecological opportunities unavailable in the tested young. Developmental programs therefore need repeated trials that follow the same individuals across changing natural conditions.
Social learning complicates development further because individual experience can be partly organized by other animals. Young chimpanzees encounter tool practices available within their communities, and dolphin foraging methods can be passed along maternal lines Boesch & Boesch (1990) Krützen et al. (2005). Developmental researchers should consequently record demonstrator exposure, local ecological affordances, and opportunities for independent discovery. Cognition evolves through populations, develops through life histories, and is expressed through immediate tasks. Keeping these explanatory scales distinct while studying their interaction is a foundation of contemporary comparative psychology Bateson & Laland (2013).
Measurement Bias, Replication, and Critical Scientific Discussion
The principal methodological risk in animal cognition is an inference gap between a successful performance and an ambitious explanation of the mental process. Small groups of highly trained animals, selective reporting, unstable motivation, unmeasured perceptual cues, and a limited set of laboratories can inflate apparent generality. Conversely, poorly matched tasks may conceal genuine species-typical capacities. Critically reviewing a study requires asking what was directly manipulated, what behavior was measured, whether a control condition was effective, which competing models survive, and how independently the finding has been reproduced. Cross-species studies additionally require transparency about phylogenetic nonindependence MacLean et al. (2014) Miller et al. (2026).
Experimental replication does not require identical performance from every animal, but it does require stable predictions under clearly defined conditions. In memory studies, one should distinguish learned perishable-food schedules from trial-specific integrated event retention. In causal tool tasks, transfer to new apparatuses tests whether behavior extends beyond the exact trained perceptual configuration. In social tests, changing visual access and actor knowledge tests whether action prediction uses more than familiar scripts. In numerical tests, continuous quantity controls assess whether item number contributes independently to preference Clayton & Dickinson (1998) Horner & Whiten (2005) Krupenye et al. (2016) Agrillo et al. (2009).
The comparative literature supports a broad range of experimentally documented capacities across distant taxa while leaving substantial uncertainty about the mechanisms and generality of some headline results. Reviews can organize findings, but source access matters: a bibliographic abstract supports only the claims that it reports, whereas detailed analysis of statistics and controls requires the full study. The present synthesis makes no pooled estimate of cognitive ability, and it does not convert suggestive behavioral evidence into a universal consciousness claim. Its methodological conclusion is that cognitive science advances through repeated comparisons that successfully separate explanatory alternatives Subias et al. (2025) Schnell et al. (2026).
Artificial Era and the Comparative Science of Cognitive Systems
The Artificial Era enlarges the practical context in which cognition is compared and operationalized. Human observers now interact with nonbiological systems that perform tasks involving categorization, memory-like retrieval, planning, language processing, and adaptive choice. Comparative research provides a methodological lesson: shared performance in a test is evidence of a functional similarity under specified conditions, while claims about underlying organization require analysis of mechanisms, training histories, environment, and generalization. This lesson already governs cautious comparisons among corvids, primates, fish, and cephalopods, and it can guide principled comparisons with artificial architectures Güntürkün & Bugnyar (2016) Amodio et al. (2019).
A nonbiological system differs from an animal in its provenance, energy requirements, reproduction, embodiment, learning regime, and available evidence about experience. Such differences become research variables rather than obstacles to posing a comparative question. A testable cross-system protocol would define equivalent informational demands, document the training and evaluation environment, and examine novel transfer, intervention, reliability, and mechanistic transparency. The animal literature warns that a shared label such as memory or planning may encompass multiple implementations. Any proposed artificial agency or artificial cognition construct therefore requires explicit operational criteria and independently evaluated predictions MacLean et al. (2014) Subias et al. (2025).
The theoretical horizon From Homo to Artificial is an authorial framework for considering the historical expansion of cognitive systems. Its philosophical significance can coexist with the evidentiary requirements of empirical comparative psychology. The present monograph makes no experimental determination about subjective experience in artificial systems. It establishes an analytic bridge: observations belong to specific tested systems, computational explanations require mechanistic tests, and experiential claims require independently articulated evidence. This bridge is useful for comparative psychology, cognitive neuroscience, animal welfare, and research on human–artificial relations because it preserves the distinct questions that each inquiry can answer Bateson & Laland (2013) Schnell et al. (2026).
Conclusion
Animal cognition is the experimentally investigated organization of information use in nonhuman behavior. Across mammals, birds, insects, fish, and cephalopods, peer-reviewed research documents meaningful forms of learning, category discrimination, numerical sensitivity, spatial and episodic-like memory, prospective preparation, tool behavior, behavioral control, and socially mediated adjustment. The strongest claims are those connected to discriminating tests: cache experiments address what–where–when information, honeybee transfer tests relational categorization, social eye-tracking paradigms examine anticipation under false belief, and controlled choice tasks assess inhibitory or numerical performance Clayton & Dickinson (1998) Giurfa et al. (2001) Krupenye et al. (2016) MacLean et al. (2014).
Comparative synthesis requires preserving the difference between a behavioral achievement and a particular account of how it was produced. Similar actions can follow distinct mechanisms, and a capacity may appear absent when an experiment ignores the ecology or sensory organization of the tested animal. The evidence therefore favors explanatory pluralism organized through explicit operational definitions, appropriate controls, longitudinal observation, and formal comparisons among alternatives. Tinbergen's multiple explanatory questions remain useful because mechanisms, development, evolutionary history, and adaptive consequences demand complementary research Bateson & Laland (2013) Miller et al. (2026).
The research frontier concerns developmental trajectories across understudied taxa, independent replication of complex social and metacognitive tasks, neural mechanisms of convergent functions, and methods that better connect laboratory performance with ecological consequences. Questions about subjective experience must be examined through their own converging evidence rather than inferred automatically from task success. Animal cognition offers a mature and expanding research field whose central achievement is the ability to transform striking behaviors into precise, revisable, experimentally evaluable scientific explanations. Its comparative methods also provide disciplined foundations for future study of cognitive organization throughout the Artificial Era Santos & Rosati (2015) Schnell et al. (2026).
References
- Agrillo, C., Dadda, M., Serena, G., & Bisazza, A. (2009). Use of number by fish. PLOS ONE, 4(3), e4786. https://doi.org/10.1371/journal.pone.0004786
- Amodio, P., Boeckle, M., Schnell, A. K., Ostojic, L., Fiorito, G., & Clayton, N. S. (2019). Grow smart and die young: Why did cephalopods evolve intelligence? Trends in Ecology & Evolution, 34(1), 45–56. https://doi.org/10.1016/j.tree.2018.10.010
- Bateson, P., & Laland, K. N. (2013). Tinbergen's four questions: An appreciation and an update. Trends in Ecology & Evolution, 28(12), 712–718. https://doi.org/10.1016/j.tree.2013.09.013
- Benson-Amram, S., Dantzer, B., Stricker, G., Swanson, E. M., & Holekamp, K. E. (2016). Brain size predicts problem-solving ability in mammalian carnivores. Proceedings of the National Academy of Sciences, 113(9), 2532–2537. https://doi.org/10.1073/pnas.1505913113
- Boesch, C., & Boesch, H. (1990). Tool use and tool making in wild chimpanzees. Folia Primatologica, 54(1–2), 86–99. https://doi.org/10.1159/000156428
- 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
- 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
- Finn, J. K., Tregenza, T., & Norman, M. D. (2009). Defensive tool use in a coconut-carrying octopus. Current Biology, 19(23), R1069–R1070. https://doi.org/10.1016/j.cub.2009.10.052
- Giurfa, M., Zhang, S., Jenett, A., Menzel, R., & Srinivasan, M. V. (2001). The concepts of ‘sameness’ and ‘difference’ in an insect. Nature, 410, 930–933. https://doi.org/10.1038/35073582
- 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
- Herrmann, E., Call, J., Hernández-Lloreda, M. V., Hare, B., & Tomasello, M. (2007). Humans have evolved specialized skills of social cognition: The cultural intelligence hypothesis. Science, 317(5843), 1360–1366. https://doi.org/10.1126/science.1146282
- Horner, V., & Whiten, A. (2005). Causal knowledge and imitation/emulation switching in chimpanzees (Pan troglodytes) and children (Homo sapiens). Animal Cognition, 8(3), 164–181. https://doi.org/10.1007/s10071-004-0239-6
- Kano, F., Krupenye, C., Hirata, S., Tomonaga, M., & Call, J. (2019). Great apes use self-experience to anticipate an agent's action in a false-belief test. Proceedings of the National Academy of Sciences, 116(42), 20904–20909. https://doi.org/10.1073/pnas.1910095116
- Krupenye, C., Kano, F., Hirata, S., Call, J., & Tomasello, M. (2016). Great apes anticipate that other individuals will act according to false beliefs. Science, 354(6308), 110–114. https://doi.org/10.1126/science.aaf8110
- Krützen, M., Mann, J., Heithaus, M. R., Connor, R. C., Bejder, L., & Sherwin, W. B. (2005). Cultural transmission of tool use in bottlenose dolphins. Proceedings of the National Academy of Sciences, 102(25), 8939–8943. https://doi.org/10.1073/pnas.0500232102
- MacLean, E. L., Hare, B., Nunn, C. L., Addessi, E., Amici, F., Anderson, R. C., Aureli, F., Baker, J. M., Bania, A. E., Barnard, A. M., Boogert, N. J., Brannon, E. M., Bray, E. E., Bray, J., Brent, L. J. N., Burkart, J. M., Call, J., Cantlon, J. F., Cheke, L. G., ... Zhao, Y. (2014). The evolution of self-control. Proceedings of the National Academy of Sciences, 111(20), E2140–E2148. https://doi.org/10.1073/pnas.1323533111
- Miller, R., Claisse, E., Timulak, A., & Clayton, N. S. (2026). Development of cognition in corvids. Animal Cognition. Advance online publication. https://doi.org/10.1007/s10071-026-02088-8
- Pepperberg, I. M. (2006). Cognitive and communicative abilities of Grey parrots. Applied Animal Behaviour Science, 100(1–2), 77–86. https://doi.org/10.1016/j.applanim.2006.04.005
- Piffer, L., Agrillo, C., & Hyde, D. C. (2012). Small and large number discrimination in guppies. Animal Cognition, 15(2), 215–221. https://doi.org/10.1007/s10071-011-0447-9
- 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
- Santos, L. R., & Rosati, A. G. (2015). The evolutionary roots of human decision making. Annual Review of Psychology, 66, 321–347. https://doi.org/10.1146/annurev-psych-010814-015310
- Schnell, A. K., Boeckle, M., Rivera, M., Clayton, N. S., & Hanlon, R. T. (2021). Cuttlefish exert self-control in a delay of gratification task. Proceedings of the Royal Society B: Biological Sciences, 288(1946), 20203161. https://doi.org/10.1098/rspb.2020.3161
- Schnell, A. K., Clayton, N. S., Hanlon, R. T., & Jozet-Alves, C. (2021). Episodic-like memory is preserved with age in cuttlefish. Proceedings of the Royal Society B: Biological Sciences, 288(1957), 20211052. https://doi.org/10.1098/rspb.2021.1052
- Schnell, A. K., Browning, H., Crump, A., Burn, C. C., & Birch, J. (2026). Sentience in cephalopod molluscs: An updated assessment. Biological Reviews, 101(3), 1311–1333. https://doi.org/10.1002/brv.70125
- Subias, L., Katsu, N., & Yamada, K. (2025). Metacognition in nonhuman primates: A review of current knowledge. Primates, 66, 9–25. https://doi.org/10.1007/s10329-024-01169-x
- Weir, A. A. S., Chappell, J., & Kacelnik, A. (2002). Shaping of hooks in New Caledonian crows. Science, 297(5583), 981. https://doi.org/10.1126/science.1073433
- Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds, V., Sugiyama, Y., Tutin, C. E. G., Wrangham, R. W., & Boesch, C. (1999). Cultures in chimpanzees. Nature, 399, 682–685. https://doi.org/10.1038/21415
Social Cognition and the Interpretation of Other Agents
Social cognition concerns the information an animal uses to anticipate, coordinate with, compete against, or learn from other individuals. Evidence may arise from tracking dominance relationships, responding to gaze, recognizing social partners, or modifying behavior depending on an observer's presence. A central problem is whether success relies on observable behavioral regularities, relational representations, or attribution of another individual's internal states. These alternatives can predict similar actions in ordinary interactions and therefore require discriminating experiments. Studies of caching corvids are particularly useful because the opportunity for theft makes the identity and behavior of observers relevant to future outcomes Dally et al. (2006).
Dally, Emery, and Clayton reported that western scrub-jays could remember which individual had watched a caching episode and alter recaching accordingly Dally et al. (2006). This evidence documents structured social information integrated with memory for a specific event. Yet the response does not independently prove that the bird represents a proposition about another bird's knowledge. Models using observer identity, previous pilfering experience, and learned risk can also be investigated. Causal inference improves through tests in which visual access, prior experience, and actual behavior vary independently. The scientific advance is an experimentally tractable social-cognitive question anchored in a meaningful competitive setting.
Social cognition also includes cooperation and information exchange, but one should not infer a generalized cooperative disposition from a single experimental game. Understanding whether a partner can see a reward, whether an action is contingent on another individual's participation, and whether a benefit is shared are different cognitive problems. Comparative research must respect species-specific social systems and opportunity structures. Human-oriented gestures may be unusually familiar to domesticated animals yet unfamiliar to a wild taxon; performance differences thus reflect social experience as well as task demands Herrmann et al. (2007) Krützen et al. (2005).