RESEARCH DOSSIER
Homology, Convergence, and the Evolution of Cognition
Abstract. Homology, convergence, and the evolution of cognition examines how comparable cognitive operations emerge, persist, and diversify across animal phylogenies. This fundamental research monograph defines the evidentiary difference between ancestral continuity, independently acquired functional similarity, and task-generated resemblance. It integrates historical comparative biology, Tinbergen's explanatory framework, phylogenetic modeling, cross-species construct validation, original experimental findings, comparative neuroanatomy, and ecological hypotheses. Case studies cover corvid memory and planning, great-ape social cognition, honeybee relational transfer, cuttlefish delayed reward, inhibitory control across mammals and birds, and the neural organization of avian pallia. The critical synthesis concludes that behavioral convergence, neural functional resemblance, developmental homology, and genealogical inheritance are distinct hypotheses requiring different tests. Its methodological agenda addresses phylogenetic sampling, measurement invariance, mechanistic alternatives, replication and causal inference. Comparisons with engineered artificial cognition are developed as explicit functional analogies with separate provenance criteria. The study is a critical integrative narrative synthesis, without claims of a preregistered systematic review, meta-analysis, new experimental data, or human clinical findings.
Keywords: evolution of cognition; homology; convergent evolution; homoplasy; comparative cognition; phylogenetic comparative methods; cognitive evolution; avian cognition; corvid planning; cephalopod cognition; honeybee relational learning; neural evolution; construct validity; evolutionary psychology; artificial cognition
The Research Object: Evolutionary Identities of Cognitive Capacities
Homology, convergence, and the evolution of cognition concerns how scientists establish the historical identities of information-processing capacities across lineages. A capacity is provisionally identified by a reproducible operation—discrimination, memory-guided retrieval, prospective choice, causal intervention, or inference about another agent—rather than by a broad label such as intelligence. The scientific question is whether corresponding capacities share inherited organization, originated independently under comparable pressures, or only look alike because an apparatus rewards the same response. Identifying an evolutionary history requires evidence about phylogenetic position, developmental organization, neural implementation, and the actual computational demands of behavior. Similarity is an observation; homology or convergence is a historical explanation requiring independent support Hall (2013) Fitch et al. (2010) MacLean et al. (2012).
This monograph conducts a critical integrative narrative synthesis of evolutionary theory, comparative experiments, neural evidence, and phylogenetic methodology. It is organized around the inferential sequence from trait definition to controlled testing, ancestral reconstruction, mechanistic comparison, and explanatory evaluation. The material includes primary studies with corvids, great apes, honeybees, and cuttlefish, supported by methodological and theoretical scholarship. A review is described as a narrative synthesis because no preregistered search strategy, exhaustive screening log, or statistical meta-analysis is claimed. Its conclusions express the evidentiary reach of particular research designs rather than assigning an all-purpose rank to living species MacLean et al. (2012) Fitch et al. (2010).
The central synthesis is that cognitive evolution operates on nested traits. A sensory circuit can be deeply homologous even when the complex behavioral capacity it supports emerged independently. Two animals may use dissimilar circuits for comparable task functions, or closely related species may deploy an inherited mechanism differently in distinct ecologies. Homology and convergence consequently must be asserted at a specified level—gene network, cell type, developmental circuit, representational operation, learned policy, or observable performance. Moving between levels without new evidence generates apparent evolutionary certainty where the empirical question remains open Güntürkün & Bugnyar (2016) Hall (2013) Stacho et al. (2020).
Conceptual Definitions: Homology, Analogy, Homoplasy, and Convergence
Homology designates similarity attributable to descent from a shared ancestor that possessed the relevant feature or its precursor. Convergence designates similarity acquired independently in distinct evolutionary histories; homoplasy encompasses independently arising similarities and evolutionary reversals, while analogy historically names comparable functions without implying shared origin. Parallel evolution is sometimes reserved for independent similar change against closely related developmental starting conditions, although terminology and diagnostic criteria vary across biological fields. The classifications remain hypotheses about character histories, which must be made explicit and evaluated against phylogenetic evidence Hall (2013).
Behavioral traits present distinctive difficulty because they are temporally organized activities rather than preserved anatomical structures. A bird's caching decision, an ape's anticipatory gaze, and a cuttlefish's reward choice each depend on organism-specific perception, motivation, and motor output. Shared function can therefore be coded at a coarse grain while the intermediate mechanism differs considerably. The converse also occurs: a broadly homologous neural subsystem can support divergent functions after changes in connectivity, development, or ecological demands. Behavioral homology does not require invariant genes or identical development throughout every descendant lineage Hall (2013) Fitch et al. (2010).
Operational clarity begins by specifying a target trait at several levels: behavioral outcome, information required, computation hypothesized, neural or developmental realization, and position on a phylogenetic tree. Evidence then supports separate statements such as 'these lineages perform analogous discriminations,' 'their pallial structures share deep ancestry,' or 'their particular flexible planning strategies plausibly evolved independently.' Such formulations prevent functional comparisons from silently becoming claims about anatomical identity, private experience, or historical origins Güntürkün & Bugnyar (2016) Güntürkün et al. (2024) MacLean et al. (2012).
From Comparative Anatomy to Comparative Cognitive Science
The history of the problem extends from comparative anatomy into evolutionary biology and experimental psychology. Richard Owen's nineteenth-century usage of homology was framed before evolutionary descent became the organizing explanation; Darwinian theory later connected common structural plans with inheritance and divergence. Comparative behavioral scientists accumulated parallel traditions of naturalistic observation, conditioning experiments, and ethological analysis. Modern work inherits this history when it attempts to infer mechanisms of memory, social inference, or planning from tasks performed by species separated by large evolutionary distances Hall (2013) Tinbergen (1963).
Tinbergen's 1963 synthesis distinguished causation, development, survival value, and evolutionary history as complementary questions about behavior. Its enduring methodological contribution is that an immediate behavioral mechanism, a developmental account, an ecological function, and a phylogenetic origin answer different questions. A controlled experiment may establish how one animal selects an action under a specified condition while contributing little direct evidence about where the capacity originated. Conversely, a phylogenetic association may suggest a historical pattern without isolating the cognitive algorithm responsible for an individual choice Tinbergen (1963) Bateson & Laland (2013).
Comparative cognition acquired new experimental traction as investigators transferred structured tasks across taxa and combined them with evolutionary modeling. Scrub-jay caching studies linked memory to identifiable information about food, place, and temporal interval; studies of honeybees tested relational transfer beyond trained displays; research on birds and mammals challenged assumptions about cortex-dependent intelligence. Each line also exposed the need to distinguish task achievement from the inferred representational explanation. Historical development is therefore a succession of more discriminating tests rather than a simple discovery that animals possess increasingly humanlike minds Clayton & Dickinson (1998) Giurfa et al. (2001) Güntürkün & Bugnyar (2016).
Tinbergen's Explanatory Architecture and Cognitive Evolution
Mechanism asks what immediate processes produce a response: sensory detection, memory retrieval, action valuation, learned associations, or neural dynamics. Ontogeny asks how those processes arise and change with maturation, experience, and social exposure. Adaptive significance concerns contributions to survival or reproduction under historical or current environments, while phylogeny addresses descent and evolutionary transformation. The four questions interact because evolutionary selection can act on developmental systems that produce context-dependent computations. A claim of convergence becomes compelling only when the relevant phylogenetic and mechanistic propositions are separately articulated Tinbergen (1963) Bateson & Laland (2013).
Consider apparent future planning. An experiment with ravens can show that a subject chooses a tool usable after a delay, but the immediate mechanism might involve remembered reward contingencies, prospective representation, or a combination. Developmental tests would examine how training and previous tool experience shape the strategy. Ecological tests would ask what long-run advantage flexible delayed action confers. Phylogenetic analysis would consider whether related birds possess comparable capacities and what the common ancestors plausibly did. One successful result cannot answer all four questions Kabadayi & Osvath (2017) Redshaw et al. (2017).
The same analytical separation governs the use of neural evidence. Similar neuronal densities or recurrent networks may make complex working memory feasible in different taxa, while independent evolutionary origins of a particular behavior require additional historical evidence. The distinction protects conclusions from both overstatement and underinterpretation: controlled demonstrations legitimately establish bounded behavioral effects, and convergent neuroscience legitimately reveals implementational possibilities. The appropriate scientific synthesis integrates these claims without erasing their evidentiary boundaries Olkowicz et al. (2016) Stacho et al. (2020) Güntürkün et al. (2024).
Phylogenetic Nonindependence and Ancestral-State Inference
Species are related observations, not exchangeable samples from an independent population. Closely related species may resemble one another because they inherit common developmental systems and environments, so ordinary correlations between a cognitive score and brain size can exaggerate independent evidence. Felsenstein's independent-contrasts framework formalized the use of phylogenetic trees and branch lengths to compare evolutionary changes under an explicit model. Comparative cognition can use related approaches to examine whether observed traits track shared ancestry, ecological variables, or predicted transitions Felsenstein (1985) MacLean et al. (2012).
Ancestral-state reconstruction assigns likelihoods to trait states at internal nodes using a chosen tree, coding scheme, and evolutionary model. Such results are conditional on what counts as a trait and which living taxa were sampled; most ancestral animals cannot be tested directly in modern behavioral paradigms. Loss, repeated gain, correlated evolution, and unobserved intermediates can yield similar present distributions. Cross-taxon datasets should therefore include sensitivity analyses for tree topology, uncertainty in branch lengths, state thresholds, and competing rates of evolutionary change Felsenstein (1985) MacLean et al. (2012).
Phylogenetic targeting improves experimental design before data collection. Researchers may select a close relative that differs in a proposed ecological pressure, then recruit distant taxa predicted to face analogous demands; a broad but uneven convenience sample is less informative. Performance must be matched with uncertainty estimates and descriptions of sample sizes, motivational conditions, and test exposure. Evolutionary similarity claims become stronger when patterns replicate under more than one trait operationalization and survive plausible alternative trees and measurement schemes MacLean et al. (2012) MacLean et al. (2014).
Homologous Substrates and Changing Cognitive Functions
The question 'Is this cognitive process homologous?' requires specifying which process and which ancestral scale. Vertebrate forebrains inherit organizational features while individual lineages elaborated different pallial structures. Bird and mammal pallia share evolutionary ancestry, yet their spatial architecture and distribution of cell populations are not interchangeable. High-level cognitive capacities such as flexible planning may rely on distributed, partially homologous neural resources and independently specialized functional networks. It is therefore possible to identify structural homology at one level and functional convergence at another Güntürkün & Bugnyar (2016) Stacho et al. (2020).
Developmental conservation is informative but insufficient by itself. Conserved gene-expression domains or projection patterns can trace cell populations across divergent brains, whereas equivalent mature connectivity can arise from different developmental routes. The 2020 reconstruction of local circuits in avian forebrain tissue found repeated column-like microcircuit organization and layered connections, providing a specific comparison with mammalian circuits. It established structural features in the studied material; the inference that any particular psychological experience or mental representation is identical across birds and mammals requires separate evidence Stacho et al. (2020) Hall (2013).
An evolutionary explanation must allow ancestral continuity with developmental reorganization. If a lineage shifts the thresholds governing learning, changes sensory input, and reorganizes a homologous circuit, the resulting behavior may differ without making every underlying component newly invented. Similarly, a common neurotransmitter function or pallial ancestry does not settle whether the memory operation expressed in two lineages has the same evolutionary origin. Homology is thus best represented as a structured claim linking a character definition to historical evidence, with the level of explanation explicitly recorded Hall (2013) Güntürkün et al. (2024).
Comparative Task Design and Construct Equivalence
Cross-species behavioral comparison begins with a construct, an operation, and a testable distinction among mechanisms. A task that requires a primate to reach around an opaque obstacle may demand a wholly different visuomotor solution when given to a bird or a cuttlefish. Raw success scores mix learning, perception, motivation, handling ability, neophobia, and prior exposure. A valid comparative design makes the target informational relation accessible to every species while documenting necessary species-specific changes. Those adaptations must preserve the logical contrast that identifies the proposed capacity rather than simply reproducing the outward apparatus MacLean et al. (2012) Fitch et al. (2010).
Functional equivalence means that two tasks place organisms under comparable computational demands, not that they present physically identical stimuli. Different colors, odors, response devices, or reward types may be essential for ecological and sensory fairness. The difficulty is that each adaptation can alter the available cues. Researchers should therefore distinguish a common formal problem from local realizations, test perceptual discrimination independently, and report training protocols as part of the evidence. Matching apparatus dimensions without matching the informational challenge produces misleading evolutionary rankings MacLean et al. (2012) Schnell et al. (2021).
Comparative evidence gains strength through task batteries targeting converging constructs with partly independent nuisance variables. If both caching choices and newly learned transfer tasks support prospective regulation in one species, a single reinforcement explanation becomes less comprehensive. If performance appears only in one familiar task, a specialized ecological competence may be the better-supported description. Measurement invariance across species remains an empirical hypothesis, particularly when differences in rearing, sampling, and motivation interact with phylogenetic history Kabadayi & Osvath (2017) Raby et al. (2007) Fitch et al. (2010).
Discriminating Cognitive Mechanisms from Behavioral Outcomes
A behavioral outcome underdetermines its generating process. Selecting a reward after a delay may arise from flexible future-oriented representation, conditioned preference, learned stimulus chaining, or a response rule triggered by a familiar context. Each theory predicts different changes under novel cues, altered contingencies, interrupted training, or transformations of the expected outcome. Experimental contrasts should be chosen for their capacity to discriminate these predictions, and analysis should report failures as well as positive transfers. Claims about mechanism become credible when rival explanations are tested rather than ruled out by descriptive vocabulary Redshaw et al. (2017) Horner & Whiten (2005).
In tool-use studies, transparent and opaque apparatuses provide distinct information about causal structure. Horner and Whiten's comparison of chimpanzees and children tested how causal visibility changes reproduction of demonstrated actions, illustrating that imitation and emulation are separate candidate strategies. Such findings concern the particular populations, experimental materials, and available demonstrations. They establish that contextual cues and causal affordances can change the action selected, thereby constraining any evolutionary argument that infers stable species-general representations from one performance pattern Horner & Whiten (2005).
The evidentiary standard can be represented as a hierarchy of increasingly demanding claims: the organism completes a task; its responses track a controlled variable; it transfers an operation to a new configuration; alternative processes are selectively excluded; the supported process occurs across multiple contexts; and its evolutionary origin is independently reconstructed. These are evidentiary steps rather than an assumed staircase of intelligence. A strong result at one step remains important even when the next cannot yet be established Fitch et al. (2010) MacLean et al. (2012).
Measurement Invariance, Species Sampling, and Statistical Inference
Cross-species scores can possess superficially common units while measuring different mixtures of latent abilities. A detour task, for example, can be affected by inhibitory control, exploration, locomotor skill, and the animal's attraction to visible food. Comparing average scores requires tests that performance differences are not primarily driven by such nuisance variables. Species-level replication, within-species sampling, age and experience distributions, and sufficient trial numbers are indispensable for evaluating latent construct comparability MacLean et al. (2014) MacLean et al. (2012).
The 2014 comparative self-control collaboration tested 567 individuals from 36 species with two task formats and analyzed associations with brain measures. The substantial coordination across sites extended comparative sample coverage, and results supported a positive association between absolute brain volume and task performance in that dataset. The result is still a relationship between measured performances and specified predictors, not a universal equation of brain volume with intelligence, nor a causal demonstration that expanding brain volume independently generates inhibition. Ecological selection and task-specific processes remain competing elements of explanation MacLean et al. (2014).
Studies of carnivores illustrate why one brain-performance relationship cannot substitute for another. Benson-Amram and colleagues used baited puzzle boxes with zoo animals from 39 species and found an association between relative brain size and success. Their task differed from the control tasks examined by MacLean and colleagues, and their population consisted of captive carnivores exposed to a particular mechanical challenge. Differences between predictor results across studies are scientifically valuable: the task, taxonomic sampling frame, neural measure, and behavioral construct jointly define what an association means Benson-Amram et al. (2016) MacLean et al. (2014).
Controlled Transfer, Intervention, and the Limits of Recognition Tests
Transfer experiments help separate learning a particular surface feature from learning a relation. Honeybees trained on matching or nonmatching visual patterns were subsequently tested with new stimulus materials, including across sensory modalities, in research by Giurfa and colleagues. Successful transfer strengthens an interpretation in terms of relational discrimination. It does not, by itself, establish introspective knowledge, a linguistic concept, or a single neural mechanism shared with humans; further interventions are needed to locate information storage and decision rules Giurfa et al. (2001).
In social cognition, anticipatory looking can provide a sensitive index of expectations about another agent's behavior. Experiments with great apes by Krupenye and colleagues reported gaze patterns consistent with anticipating an actor's action according to a false belief. The design contributed important evidence because the agent's knowledge differed from the actual state of the environment. Whether the animals represent propositional beliefs or track more economical agent–object relations is a separate process-level question that motivates additional discriminating tests Krupenye et al. (2016) Fitch et al. (2010).
A complete comparative program uses converging observables alongside targeted interventions. Sensory occlusion, counterbalanced outcome locations, changes in rewards, social visibility, and test delays can rule out different alternatives. Neural recordings add temporal specificity, while selective disruption can establish a causal role for a circuit in performance under the particular conditions tested. These approaches should be combined with preregistered predictions when possible, rather than treating a statistically significant score as sufficient evidence for a richly characterized mental state Nieder et al. (2020) Stacho et al. (2020).
Cross-Lineage Neuroscience and the Problem of Mapping Levels
Neural comparison has three distinct targets: whether structures are homologous, whether computational organizations are functionally similar, and whether neural activity contributes causally to a psychological operation. The avian pallium and mammalian cerebral cortex are historically related forebrain systems whose cellular arrangement differs. Research has identified similarities in neural connectivity and information-processing capacities without making all detailed circuit arrangements identical. Cognitive convergence can therefore coexist with deeper neurodevelopmental homology Güntürkün & Bugnyar (2016) Stacho et al. (2020).
Counting neurons offers one quantitative alternative to total brain weight. Olkowicz and colleagues reported high neuron densities and primate-like forebrain neuron numbers in several birds, supporting an account in which small brains can possess substantial processing resources. The result describes measured neuroanatomical composition, not a validated cross-species metric of global intelligence. Neuron allocation, network topology, axonal conduction, metabolic demand, ecological specialization, and task demands all matter for the relationship between structure and function Olkowicz et al. (2016) Güntürkün et al. (2024).
Nieder, Wagener, and Rinnert reported neuronal activity in crows that tracked reported detection of near-threshold visual stimuli. This finding identifies a candidate neural correlate of sensory report under a particular perceptual task. It sharpens comparative tests of conscious-access-like neural signals, yet its implications for subjective experience remain interpretive. The study should be read with other neural and behavioral work rather than transformed into evidence of identical conscious architectures across birds and primates Nieder et al. (2020) Güntürkün & Bugnyar (2016).
Reproducibility, Publication Bias, and the Comparative Evidence Base
Comparative cognition experiments face familiar challenges of modest sample sizes, subject availability, selective species reporting, and heterogeneous laboratory procedures. Within-species replication can fail because age, prior exposure, husbandry, reward value, or sensory conditions change the behavioral opportunity. Across-species replication is more demanding: the original test may cease to measure the same operation in a new ecological or morphological system. Transparent protocols should document these differences instead of interpreting every failed transfer as an evolutionary absence MacLean et al. (2012) Fitch et al. (2010).
Evidence of convergent cognition is most credible where independent laboratories reproduce a structured behavioral phenomenon and connect it to neurobiological or phylogenetic results. Classic examples have generated productive criticism. Kabadayi and Osvath's raven planning experiments were followed by a published commentary that asked whether leaner associative explanations could account for the observations. This exchange illustrates normal scientific refinement: the successful delayed-choice finding and its strongest psychological interpretation are distinct propositions Kabadayi & Osvath (2017) Redshaw et al. (2017).
A research synthesis should classify evidence by design rather than reputation. Single-laboratory demonstrations establish local capacities; independent replications establish robustness; multi-task and cross-context transfer constrain the operation; interventions support mechanism; and phylogenetic models address historical pathways. Reporting effect sizes, confidence intervals, subject attrition, and null results would allow stronger future comparisons than categorical labels alone. No single technique eliminates the need to integrate observation, experiment, developmental history, and phylogeny Felsenstein (1985) MacLean et al. (2012) Bateson & Laland (2013).
Corvid–Ape Comparisons as a Test of Cognitive Convergence
Corvids and apes have become a central comparative case because they are distantly related vertebrates with conspicuous differences in pallial architecture, yet both groups can succeed in some problems involving tool manipulation, memory, and social information. Emery and Clayton assembled early evidence for a flexible cognitive repertoire in crows and their relatives and proposed that similar socioecological challenges can favor analogous functional toolkits. The evolutionary assertion must be specified at the level of the tested operation rather than the totality of intelligence: similarities in one task do not establish a common package of capacities across all members of the two lineages Emery & Clayton (2004) Güntürkün & Bugnyar (2016).
Convergence implies independent evolutionary acquisition of a character. For distant lineages it is plausible that certain complex functions arose separately, but deep vertebrate neural homologies and broad ancestral learning capacities complicate a simple independent-origin story. A bird and an ape can both learn from reward contingencies using inherited vertebrate components, even if their domain-specific flexible strategies were separately elaborated. Tree-based evidence should therefore distinguish ancestral ingredients from derived assemblies, treating the latter as testable historical hypotheses Hall (2013) Fitch et al. (2010).
The strongest comparative claims integrate tasks with converging ecological profiles. Corvid caching provides opportunities to test spatial memory and future contingencies, while primates are frequently tested in visually guided social and object problems. The tasks carry different backgrounds of evolutionary selection and familiar experience. Comparisons become informative when researchers specify the common computational challenge and experimentally control the distinct pathways by which each species encounters it. This approach turns the corvid–ape comparison into a program of mechanism discovery rather than a contest for species prestige Emery & Clayton (2004) MacLean et al. (2012).
Episodic-Like Memory, Caching, and the Interpretation of Recollection
Clayton and Dickinson's 1998 experiments investigated memory for the contents, locations, and temporal status of food caches in scrub jays. Birds modified retrieval choices as preferred perishable food became unsuitable after longer delays. The manipulation demonstrates integration of what, where, and when in recovery behavior, supporting the operational category episodic-like memory. It does not establish that subjective recollection is identical to human autobiographical remembering. The operational qualification is scientifically substantive because behavioral criteria and phenomenological properties require different evidentiary routes Clayton & Dickinson (1998).
Emery and Clayton subsequently investigated whether birds adjust cache protection in light of previous pilfering experience and observation by another bird. Jays with pilfering history showed distinctive recaching when another bird had watched the initial cache. These results connect memory of social circumstances to adaptive behavior and provide a testable basis for hypotheses about social knowledge. Alternative explanations include acquired rules based on past social contingencies; the study's evidence concerns who did what under a particular cache-recovery sequence rather than directly revealing another animal's private beliefs Emery & Clayton (2001).
At an evolutionary level, caching and retrieval strategies illustrate the difference between an ecological specialization and an abstract memory faculty. Selection may favor remembering perishable food in particular food-storing lineages, while analogous temporal organization could appear in other animals for unrelated reasons. To evaluate generality, researchers can change reward types, use non-caching tasks, and examine species with different caching ecologies. A homologous ancestral memory substrate and an independently derived cache-control policy are mutually compatible explanations at different analytical levels Clayton & Dickinson (1998) Emery & Clayton (2001) Hall (2013).
Future-Oriented Choice and the Evolution of Planning
Raby and colleagues manipulated when and where western scrub jays could expect breakfast and reported caching choices that provided for anticipated future needs. The design is influential because the future availability of food was experimentally arranged, giving researchers more than a naturalistic observation of cache storage. It supports prospective behavioral organization under defined conditions while leaving room to test how much the behavior depends on abstract simulation, reward-learning history, or general ecological predispositions Raby et al. (2007).
Kabadayi and Osvath expanded the comparative planning question by studying ravens choosing tools and barter objects for later use outside food caching. Their experiments incorporated delays and choices relevant to future reward, providing evidence that a raven's temporal organization can generalize beyond the specific ecological act of caching. The result does not alone specify the representation by which the bird evaluates its future state. Redshaw, Taylor, and Suddendorf highlighted the need to compare these findings against economical alternative accounts, a reminder that behavioral generalization is necessary yet not sufficient for a unique mechanism claim Kabadayi & Osvath (2017) Redshaw et al. (2017).
The evolutionary origin of planning is best framed as the history of component capacities: remembering a prior contingency, maintaining information during a delay, selecting an instrument, forecasting a future opportunity, and suppressing an immediate option. Some ingredients may be ancient or widely distributed; their particular integration may have evolved in multiple lineages. Historical inference requires broad taxon sampling and explicit predictions about which components are likely shared by ancestry and which were assembled under separate ecological pressures MacLean et al. (2012) Fitch et al. (2010) Kabadayi & Osvath (2017).
Avian Brain Organization, Neuron Allocation, and Network Convergence
Neuroscience has substantially changed how comparative psychologists interpret avian cognition. Birds do not possess a mammalian six-layered neocortex, but they possess pallial networks that support sophisticated learning and action control. Güntürkün and Bugnyar reviewed neurobiological evidence for homologous forebrain organization alongside different macroscopic anatomy, rejecting cortex-centered assumptions that cognitive complexity requires one particular cortical arrangement. The relevant distinction is between evolutionarily homologous forebrain origins and the possibly convergent computational organization of flexible behavior Güntürkün & Bugnyar (2016).
Olkowicz and colleagues measured neuron numbers in avian brains and found large populations of pallial neurons in several parrots and corvids despite comparatively small brain volumes. The work supplied anatomical evidence that absolute brain mass cannot by itself capture available neural processing capacity. Stacho and colleagues later described repeated local circuit motifs in the avian pallium that are relevant to comparisons of information-processing architecture. Neither finding implies that avian neurons are computationally interchangeable with mammalian cortical neurons; both enlarge the candidate mechanisms available for future causal testing Olkowicz et al. (2016) Stacho et al. (2020).
A 2024 theoretical synthesis proposed a combination of dense associative pallial neurons, prefrontal-like functional specialization, dopaminergic organization, and working-memory dynamics as contributors to sophisticated bird cognition. These are hypotheses about potentially convergent neural solutions whose broad empirical merit depends on extension across species and direct intervention. They exemplify a useful explanatory strategy: compare measurable circuit properties, specify cognitive operations, and infer adaptive history only after testing the mapping between the two Güntürkün et al. (2024) Nieder et al. (2020).
Great Ape Social Cognition and Limits of Ancestral Reconstruction
Comparisons among great apes and humans are informative because shared ancestry offers stronger prior grounds for considering homologous mechanisms. Krupenye and colleagues tested whether apes' anticipatory looking tracked where an agent holding a false belief would search. The reported behavior extends evidence about sensitivity to another's informational perspective. The precise representational account remains open because expectation can be encoded at different levels, from structured behavioral regularities to richer mental-state representations. Shared family membership does not eliminate this underdetermination Krupenye et al. (2016) Fitch et al. (2010).
Fitch, Huber, and Bugnyar proposed constructing cognitive phylogenies by decomposing complex faculties such as language and social cognition into constituent mechanisms. This approach is essential for avoiding whole-trait narratives. Humans and apes may share multiple social-cognitive ingredients while human communicative and cultural systems involve further changes in coordination, learning, and representational structure. Phylogenetic hypotheses about particular components can be evaluated with appropriate comparisons among apes, monkeys, and more distant outgroups Fitch et al. (2010).
Ancestral-state inference remains probabilistic when modern tasks sample only a small subset of species and developmental environments. A great-ape response occurring in one laboratory cannot be directly projected onto a deep ancestor without character coding and evolutionary models. The appropriate claim often concerns a distribution of plausibility: some shared cognitive operations are consistent with inheritance from the common ancestor, while alternative patterns of gain, loss, and developmental reorganization require further testing. Scientific precision resides in specifying these alternatives, not in supplying a singular origin story MacLean et al. (2012) Felsenstein (1985).
Tool Use, Social Transmission, and Cultural Similarities
Tool use across birds and primates offers vivid similarities but also different paths toward physical problem solving. Horner and Whiten manipulated the causal transparency of a tool task to examine imitation and emulation in children and chimpanzees. Their evidence indicates that available causal information changes what demonstrators' actions learners reproduce. The result demonstrates why replicating an observed action and understanding the causal relation of parts are separate constructs. Taxonomic comparisons should evaluate these components independently before treating any form of tool success as a unitary cognitive trait Horner & Whiten (2005).
Whiten and colleagues' 1999 comparative synthesis of long-term chimpanzee field sites documented geographic variation in behavioral repertoires consistent with cultural traditions. Such site differences are historically important evidence for socially transmitted behavioral diversity, but explaining a particular tradition requires evaluating ecological availability, genetic variation, and mechanisms of social acquisition. The mere presence of similar techniques at two sites or in two species does not resolve whether they spread by common ancestry, social learning, individual invention, or shared task constraints Whiten et al. (1999).
Cultural evolution can produce rapid transmission and repeated reinvention on timescales different from genetic phylogeny. A tool-making procedure may be behaviorally homologous within a socially connected tradition, analogous across unrelated populations, and supported by older inherited cognitive capacities in both cases. Distinguishing these histories requires observation of learning trajectories, intervention on opportunities to copy, and appropriately modeled transmission networks. Evolutionary analysis must therefore incorporate the additional historical structure of traditions when interpreting similarities in cognitively mediated behavior Whiten et al. (1999) Fitch et al. (2010).
Cephalopod Cognition and Distant Evolutionary Solutions
Cephalopods provide a powerful contrast with vertebrates because their evolutionary history, body plans, and nervous-system arrangements differ substantially from those of birds and mammals. Octopuses, cuttlefish, and squid nevertheless show sophisticated perception, learning, memory, and behaviorally flexible responses. Schnell and Clayton's review emphasizes both the diversity of these capacities and the difficulty of determining which apparently complex behaviors require rich internal representations. Similar task functions across distant clades are evidence for possible functional convergence, while differences among cephalopod species make within-group comparison equally necessary Schnell & Clayton (2019).
In delay-maintenance experiments, Schnell and colleagues tested cuttlefish choosing between an immediately accessible reward and a preferred reward available after waiting. The subjects tolerated delays and variation in waiting was associated with reversal-learning performance in the studied animals. This supports a bounded form of prospective reward regulation and invites comparisons with vertebrate inhibitory control. Yet intertemporal decisions depend on species-specific reward ecology, hunger, perceptual information, and learned contingencies, so the existence of similar waiting behavior does not establish an identical executive-control mechanism or shared origin Schnell et al. (2021) MacLean et al. (2014).
Understanding cephalopod convergence requires identifying what was independently assembled in the molluscan lineage and what inherited neural features are evolutionarily deeper. Comparative analyses can examine distributed arm control, central learning systems, visual discrimination, and predatory behavior as interacting specializations. A phylogenetic claim should be tied to explicitly coded traits and contextualized with closely related mollusks, rather than resting solely on a contrast with humans. The strongest account will join ecological opportunity, embodied constraints, developmental evidence, and neural organization Schnell & Clayton (2019) Coombs & Trestman (2025).
Insect Relational Learning and Cognition in Compact Neural Systems
Honeybees are informative because their neural systems are substantially smaller and differently organized than vertebrate forebrains, while they can solve problems involving relational discrimination. In the experiments by Giurfa and colleagues, bees learned matching-to-sample and nonmatching-to-sample choices and transferred the learned relations to new stimuli. The finding goes beyond memorizing a single reinforced display and supports the presence of a generalizable relational operation under the tested conditions. It does not establish that the operation is represented linguistically or consciously, nor that every insect deploys the same process Giurfa et al. (2001).
The evolutionary significance of insect relational learning is distinct from the demonstration itself. Transfer in a bee and matching in a primate can be functionally comparable, but the underlying sensory coding, internal state, reward learning, and circuit architecture may differ. Deeply conserved molecular components of learning need not imply homology of a particular high-level relational operation. A rigorous convergence hypothesis would specify candidate ancestral capacities and compare related insect taxa to test whether the observed behavior is widespread, recently derived, or conditionally expressed Giurfa et al. (2001) Hall (2013).
Insect studies also clarify the value of ecological task construction. Bees naturally navigate visual patterns, odors, and flower rewards, so well-calibrated stimuli can make learning visible where a human-centered manual task would fail. Yet ecologically appropriate tests require independent controls for brightness, odor traces, side preferences, and reinforcing cues. The same combination of task adaptation and logical equivalence should guide comparisons among bees, birds, mammals, and artificial learning systems if the claim concerns an operational relation rather than superficial movement Giurfa et al. (2001) MacLean et al. (2012).
Embodied Constraints and Macroevolutionary Patterns of Cognitive Complexity
Evolutionary cognition is organized through bodies as well as brains. Active perception, locomotion, manipulative structures, and the geometry of an organism's environment shape which information problems are recurrent and how actions can solve them. Coombs and Trestman's comparative framework connected multiple body, sensory, neural, and motor traits across major animal lineages and proposed that combinations supporting active visual sensing and visuomotor control help explain convergent increases in cognitive complexity. This is a theoretical and comparative synthesis with explicit trait coding, rather than an experiment demonstrating that one trait mechanically causes complex cognition Coombs & Trestman (2025).
Embodied opportunities supply testable hypotheses. An animal that navigates spatially complex environments may benefit from tracking objects across occlusions and coordinating movement with stable landmarks. Predatory animals interacting with elusive prey face different demands from a stationary filter feeder. However, an ecological correlation is not automatically a history of selection: tradeoffs, developmental constraints, genetic covariance, and phylogenetic nonindependence can produce similar distributions. Distinguishing evolutionary pressures requires comparative datasets and independently specified adaptive predictions Coombs & Trestman (2025) Felsenstein (1985).
Complex cognition may be achieved through distinct configurations rather than a single universally ordered ladder. A species can excel in spatial memory while showing limited success in physical manipulation, and another can perform demanding discrimination tasks without comparable social learning. Such mosaics make composite intelligence scores precarious and favor multiple axes of cognitive organization. The appropriate macroevolutionary question asks which combinations of neural, sensory, motor, and social traits coevolve under specified ecological histories MacLean et al. (2012) Coombs & Trestman (2025).
Brain Size, Neuron Number, and Comparative Predictors
Brain size has often served as a surrogate for cognitive complexity because gross measurements are available for many species. Yet the relation between brain volume and behavioral performance depends on the measure, scaling assumptions, and tested sample. In a multi-species experimental comparison MacLean and colleagues found absolute brain volume predictive of performance on two inhibitory-control tasks. Another study of carnivores found that relative brain size predicted performance in a novel puzzle-box task. The studies differ in both task demands and sampling frame, offering complementary rather than interchangeable predictors MacLean et al. (2014) Benson-Amram et al. (2016).
Neuron counts complicate volume-only accounts. Olkowicz and colleagues reported dense avian forebrains containing large neuron populations in selected birds, while avian studies of local pallial connectivity demonstrate neural organizations capable of sophisticated computation. The relevant determinants of performance may include neuronal allocation, projection structure, neuromodulation, and learning dynamics. Even when a structural correlate predicts observed task scores, it does not prove that a given high-level psychological construct is evolutionarily homologous across taxa Olkowicz et al. (2016) Stacho et al. (2020) Güntürkün et al. (2024).
Future comparative work should evaluate multivariate models against alternative causal graphs. Brain traits may influence learning capacity, but the ecology and development of a species can influence both brain evolution and performance in captivity. Models built on one taxonomic order should be tested on independent clades, with phylogenetic covariance and observation error explicitly incorporated. A robust evolutionary account identifies a recurrent relationship across multiple behaviors and ecological contexts, not merely a single high correlation between body-adjusted size and a convenient task Felsenstein (1985) MacLean et al. (2012).
The Human Cognitive Phenotype as a Comparative Problem
Human cognition combines language, cultural transmission, abstraction, prospective organization, and social coordination in distinctive ways. Yet comparison reveals components of these capacities in other animals, and phylogenetic arguments can reconstruct which ingredients may have existed before the human lineage. Fitch, Huber, and Bugnyar proposed treating complex social cognition and language as decomposable systems whose constituent processes have different evolutionary histories. This method permits differentiated hypotheses about ancestral continuity, independent elaboration, and the combination of mechanisms Fitch et al. (2010).
Evidence from primate social experiments and studies of cultural variation challenges a simplistic division between humans possessing cognition and other species lacking it. Great-ape anticipatory looking studies, chimpanzee cultural repertoires, and cross-taxon tool research each isolate particular operations. Their results also reveal distinct evidentiary questions: eye movements index expectation, site distributions index behavioral variation, and copying experiments test transmission conditions. Combining them is informative precisely because no single assay captures the complete architecture of human social cognition Krupenye et al. (2016) Whiten et al. (1999) Horner & Whiten (2005).
Comparative reconstruction should locate human distinctiveness at the level of combinations and transformations whose evidence can be operationalized. Candidate mechanisms include higher-order relational integration, recursive communication, norm-mediated learning, and high-fidelity cumulative culture. A proposal of uniquely human organization becomes productive when it predicts developmental patterns and cross-species contrasts rather than assuming uniqueness from differences in outward performance. Reconstructing these histories is necessarily model-dependent, because extinct populations cannot be directly tested with contemporary experimental tasks Fitch et al. (2010) MacLean et al. (2012).
Artificial Cognition and the Logic of Functional Comparison
Artificial systems introduce a new comparative condition: engineered agents can produce behaviors resembling biological cognitive operations without possessing the phylogenetic history of an animal lineage. When a machine and an animal complete a similar task, the resemblance is functional and operational; biological homology, which depends on descent, is inapplicable to an independently engineered architecture. Artificial examples can therefore help expose which task features are sufficient for a behavioral outcome, provided researchers distinguish simulation, mechanistic models, and empirical biological findings Fitch et al. (2010) Hall (2013).
A machine that anticipates a sequence of future rewards may use explicit search, learned value functions, or compressed predictive representations; a raven's delayed tool choice may depend on different computational and motivational systems. Comparative testing should establish a common abstract problem, describe the agent's available observations and actions, and analyze mechanisms on each side. Shared performance under a benchmark supports a narrow functional analogy and motivates controlled investigation. Claims about artificial subjective experience or an evolutionary lineage shared with Homo require evidence beyond performance overlap Kabadayi & Osvath (2017) MacLean et al. (2012).
Within the Artificial Era research horizon, the most productive theoretical contribution is a multilevel comparison of psychological organization. Biological history supplies criteria for evolutionary homology and convergence among animals; engineering provenance supplies criteria for architectural inheritance and independent design in artificial systems. A causal model may span these domains through shared functional variables, but its successful use must be demonstrated by intervention, transfer, and prediction. This distinction creates researchable bridges without treating every shared task score as a shared mental kind Hall (2013) Güntürkün et al. (2024).
Reconstructing Cognitive History with Multilevel Evidence
A defensible reconstruction of cognitive evolution combines trait definitions, phylogenies, behavioral comparisons, ecological hypotheses, and mechanistic evidence into a single inferential sequence. First, the analyst specifies the operation under study and identifies measurable indicators. Next, matched tests establish its distribution among sampled organisms, with uncertainty arising from performance and sampling. Developmental and neural studies examine whether similar outcomes recruit comparable causal structures. A phylogenetic model then evaluates ancestral continuity and independent gains or losses. Each stage constrains the others but remains separately auditable Felsenstein (1985) MacLean et al. (2012) Hall (2013).
Consider two lineages showing future-oriented action. Behavioral experiments may establish sensitivity to future rewards; cross-context transfer may show an operation broader than a specialized ecological routine; neural studies may locate distinct patterns of activity; phylogenetic models may favor independent elaborations over inheritance of the complete behavior. That evidentiary combination supports functional convergence in the defined operation, even if certain lower-level learning components are homologous. If the ancestral-state inference changes substantially with taxon sampling, the historical conclusion should remain qualified while the local experimental result stays intact Raby et al. (2007) Kabadayi & Osvath (2017) MacLean et al. (2012).
The same logic applies to apparent behavioral differences. A species that fails one task may nevertheless possess the relevant operation if the required sensory or motor channels were inaccessible. Demonstrating absence demands matched opportunities and positive-control tasks, and often a developmental assessment of prior learning. Comparative science is strongest when the inference maps from a measured property to a clearly delimited evolutionary claim, rather than assigning whole organisms to fixed categories of sophisticated or unsophisticated minds Giurfa et al. (2001) Schnell & Clayton (2019) Fitch et al. (2010).
Competing Evolutionary Models and Critical Scientific Discussion
Selection for social complexity, ecological foraging demands, predation, sensory-motor opportunities, and energetic constraints are influential explanations of cognitive diversification. The models can overlap: social relations create ecological information problems, and improved action control may alter the benefits of social learning. Identifying the selected trait requires more than showing that one ecological variable correlates with an observed score. Mechanistic plausibility, phylogenetic models, natural variation, and consequences for survival or reproduction jointly constrain adaptive explanations Coombs & Trestman (2025) Benson-Amram et al. (2016) MacLean et al. (2012).
Several scientific tensions remain open. First, a common task can be solved by multiple algorithms, making behavioral equivalence weaker than mechanistic identity. Second, homologous neurodevelopmental structures can diverge in computation, while distant lineages can show similar network principles. Third, composite cognitive scores are sensitive to task selection and missing taxa. Fourth, evolutionary models assume particular character codes and rates of change. These tensions are best treated as productive research questions with alternative models rather than resolved by a single favored narrative Hall (2013) Güntürkün et al. (2024) Felsenstein (1985).
Evidence must be weighted according to experimental access. Large multispecies datasets permit broader association testing but may standardize tasks in ways that compromise local ecological equivalence. Intensive small-sample studies can isolate a mechanism more effectively yet support limited generalization. Neural similarity supports specific implementation hypotheses but may not settle psychological representation. Together, these complementary sources can produce stronger conclusions when their uncertainties remain explicit and independent replication is pursued MacLean et al. (2014) Schnell et al. (2021) Stacho et al. (2020).
Evidence Matrix: What Major Paradigms Actually Establish
Experimental paradigms differ in their inferential jurisdiction. Scrub-jay cache retrieval establishes memory-guided choices sensitive to food, place, and elapsed time; raven tool and barter tasks establish flexible delayed choices under controlled future contingencies; honeybee matching tasks establish relational generalization; cuttlefish delay tasks establish waiting for a preferred outcome; ape anticipatory looking establishes sensitivity to a particular social-information structure. Each finding can motivate hypotheses about independently evolved cognitive organization. None establishes a fully shared representational mechanism merely by achieving a similarly described goal Clayton & Dickinson (1998) Kabadayi & Osvath (2017) Giurfa et al. (2001) Schnell et al. (2021) Krupenye et al. (2016).
Neuroanatomical and neural-recording studies answer different questions. Avian neuron-count measurements establish the allocation of neurons in sampled species; pallial microcircuit reconstruction documents particular connective motifs; corvid sensory-report recordings identify activity correlated with detection judgments. Combining these observations with behavioral evidence shows that complex behavior can be supported by multiple brain organizations. Demonstrating that a particular circuit causes a particular cognitive operation requires manipulations designed for that causal link and replication beyond its original preparation Olkowicz et al. (2016) Stacho et al. (2020) Nieder et al. (2020).
Phylogenetic comparative studies ask whether variation aligns with descent, ecology, and candidate brain or behavioral traits. Their evidentiary strength is conditional on sampling, measurement and model assumptions. A correlation between brain volume and task score supports a predictive association within the analyzed sample; it does not itself reveal the selective episode or developmental pathway that created it. This division of evidentiary labor should organize every future synthesis of homology and convergence in cognition Felsenstein (1985) MacLean et al. (2014) Benson-Amram et al. (2016).
Methodological Priorities for Future Comparative Research
Future studies should preregister which level of homology or convergence they intend to test. A protocol ought to define the trait, candidate mechanisms, phylogenetic sample, expected data pattern under independent origin, and alternatives such as inheritance followed by loss. If the planned evidence concerns performance alone, the historical hypothesis should be designated provisional. This practice will make competing evolutionary explanations testable rather than allowing broad character labels to shift after results become known MacLean et al. (2012) Fitch et al. (2010).
Research design should use ecologically calibrated common computational tasks with multiple response modalities, negative and positive controls, and documented differences in subject experience. Longitudinal studies can separate developmental emergence from immediate training, and independent laboratory replications can establish robustness. Cross-species tests should report individual distributions, sample sizes, uncertainty intervals, reward schedules, trial exclusions, and the complete range of studied taxa. Open materials and machine-readable character codings will improve cumulative ancestral reconstructions MacLean et al. (2012) Felsenstein (1985) Schnell & Clayton (2019).
Neuroscience can contribute through comparative cell-type mapping, circuitry, functional recordings, and reversible interventions while respecting the distinction between common ancestry and convergent computation. Model-based analyses should compare explicit generative accounts rather than treating one fitted function as uniquely explanatory. Engineering benchmarks can supply informative control models for behavioral sufficiency; biological evidence is still required to establish phylogenetic histories. A valuable long-term goal is a transparent atlas linking tasks, operations, circuits, ecologies, and evolutionary trees without collapsing their distinct levels Güntürkün et al. (2024) Stacho et al. (2020) Hall (2013).
Evolutionary Synthesis and the Artificial Era
Comparative cognition reveals a plural history of information processing in living organisms. Certain capacities persist through inheritance, others are transformed by development and ecology, and some sophisticated outcomes emerge repeatedly in independently elaborated lineages. This framework supports a structured understanding of cognition as an organization of perception, learning, memory, control, and action whose historical identity must be demonstrated. It also provides a disciplined basis for comparing biological organisms with engineered cognitive systems through operational functions and explicit implementation descriptions Hall (2013) Coombs & Trestman (2025) Fitch et al. (2010).
The Artificial Era extends the range of systems available for cognitive comparison and raises new questions about architectural descent, implementation constraints, and the construction of generalizable representations. In biological evolution, homology refers to genealogical continuity in descent; in engineered systems, one may study version inheritance, reused architectural components, and independently designed convergent solutions using different provenance criteria. The analogy becomes scientifically useful when each kind of continuity is operationalized on its own terms and when differences in training environments are controlled rather than ignored MacLean et al. (2012) Güntürkün et al. (2024).
This synthesis recommends a distinction among observable cognitive operation, representational interpretation, material mechanism, and historical origin. It allows researchers to compare function across biological and artificial systems without assuming sameness of subjective life, genetic history, or developmental organization. The intellectual benefit lies in the ability to design specific discriminating experiments: Which functional regularity transfers to new tasks, which implementation is causally sufficient, and which pattern is best explained by inherited constraints or independent construction? Those questions make comparative cognition relevant across the transition From Homo to Artificial Fitch et al. (2010) Hall (2013).
Conclusion
Homology and convergence are historical explanations for similarities that comparative cognition observes at multiple levels. Their responsible use requires exact trait definitions, valid cross-species comparisons, explicit consideration of sensory and motivational differences, and phylogenetically informed models of ancestral change. Original studies demonstrate robust examples of memory-guided retrieval, relational learning, future-directed choices, social expectations, and flexible control across distinct taxa, while modern neuroscience identifies diverse circuit configurations capable of supporting these performances Clayton & Dickinson (1998) Giurfa et al. (2001) Kabadayi & Osvath (2017) Güntürkün & Bugnyar (2016).
The major scientific finding of this synthesis is methodological as well as substantive: behavioral convergence, neural functional similarity, developmental conservation, and phylogenetic homology are related claims with different evidentiary requirements. The evidence supports a distributed evolutionary landscape in which shared ancestral ingredients coexist with independently derived assemblies. Success on comparable tasks invites inquiry into common functions, while carefully targeted interventions and evolutionary models determine how far the mechanistic and historical conclusion may proceed Hall (2013) MacLean et al. (2012) Stacho et al. (2020).
Research on cognitive evolution will advance through more representative taxonomic sampling, convergent test batteries, developmental longitudinal evidence, causal neural comparisons, independent replication, and transparent evolutionary modeling. These requirements also define the productive limits of human–artificial comparison in the Artificial Era: comparable operations can be studied experimentally, and distinct histories of emergence should be explained by their appropriate evidence. A coherent comparative psychology treats cognition as a scientifically investigable family of processes whose unity and diversity are established one operation, one mechanism, and one historical hypothesis at a time Fitch et al. (2010) Coombs & Trestman (2025) Güntürkün et al. (2024).
References
- 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
- 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
- Coombs, S., & Trestman, M. (2025). A multi-trait embodied framework for the evolution of brains and cognition across animal phyla. Behavioral and Brain Sciences, 48, e77. (Published online March 28, 2024.) https://doi.org/10.1017/S0140525X24000335
- Emery, N. J., & Clayton, N. S. (2001). Effects of experience and social context on prospective caching strategies by scrub jays. Nature, 414, 443–446. https://doi.org/10.1038/35106560
- 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
- Felsenstein, J. (1985). Phylogenies and the comparative method. The American Naturalist, 125(1), 1–15. https://doi.org/10.1086/284325
- Fitch, W. T., Huber, L., & Bugnyar, T. (2010). Social cognition and the evolution of language: Constructing cognitive phylogenies. Neuron, 65(6), 795–814. https://doi.org/10.1016/j.neuron.2010.03.011
- 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
- Güntürkün, O., Pusch, R., & Rose, J. (2024). Why birds are smart. Trends in Cognitive Sciences, 28(3), 197–209. https://doi.org/10.1016/j.tics.2023.11.002
- Hall, B. K. (2013). Homology, homoplasy, novelty, and behavior. Developmental Psychobiology, 55(1), 4–12. https://doi.org/10.1002/dev.21039
- 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
- Kabadayi, C., & Osvath, M. (2017). Ravens parallel great apes in flexible planning for tool-use and bartering. Science, 357(6347), 202–204. https://doi.org/10.1126/science.aam8138
- 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
- MacLean, E. L., Matthews, L. J., Hare, B. A., Nunn, C. L., Anderson, R. C., Aureli, F., Brannon, E. M., Call, J., Drea, C. M., Emery, N. J., Haun, D. B. M., Herrmann, E., Jacobs, L. F., Platt, M. L., Rosati, A. G., Sandel, A. A., Schroepfer, K. K., Seed, A. M., Tan, J., ... Wobber, V. (2012). How does cognition evolve? Phylogenetic comparative psychology. Animal Cognition, 15(2), 223–238. https://doi.org/10.1007/s10071-011-0448-8
- 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
- Nieder, A., Wagener, L., & Rinnert, P. (2020). A neural correlate of sensory consciousness in a corvid bird. Science, 369(6511), 1626–1629. https://doi.org/10.1126/science.abb1447
- Olkowicz, S., Kocourek, M., Lučan, R. K., Porteš, M., Fitch, W. T., Herculano-Houzel, S., & Němec, P. (2016). Birds have primate-like numbers of neurons in the forebrain. Proceedings of the National Academy of Sciences, 113(26), 7255–7260. https://doi.org/10.1073/pnas.1517131113
- Raby, C. R., Alexis, D. M., Dickinson, A., & Clayton, N. S. (2007). Planning for the future by western scrub-jays. Nature, 445, 919–921. https://doi.org/10.1038/nature05575
- Redshaw, J., Taylor, A. H., & Suddendorf, T. (2017). Flexible planning in ravens? Trends in Cognitive Sciences, 21(11), 821–822. https://doi.org/10.1016/j.tics.2017.09.001
- Schnell, A. K., & Clayton, N. S. (2019). Cephalopod cognition. Current Biology, 29(15), R726–R732. https://doi.org/10.1016/j.cub.2019.06.049
- 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
- Stacho, M., Herold, C., Rook, N., Wagner, H., Axer, M., Amunts, K., & Güntürkün, O. (2020). A cortex-like canonical circuit in the avian forebrain. Science, 369(6511), eabc5534. https://doi.org/10.1126/science.abc5534
- Tinbergen, N. (1963). On aims and methods of ethology. Zeitschrift für Tierpsychologie, 20(4), 410–433. https://doi.org/10.1111/j.1439-0310.1963.tb01161.x
- 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