Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Comparative Consciousness

Abstract. Comparative consciousness research examines the distribution, organization, possible evolutionary origins, and evidence for subjective experience across biological taxa and artificial systems. This critical narrative monograph distinguishes phenomenal consciousness, conscious access, affective sentience, metacognition, and self-recognition; traces ethological and experimental traditions; analyzes behavioral indicators, neural organization, perceptual decisions, pain-related trade-offs, and species-specific testing; and compares theories of conscious function. Evidence from mammals, birds, fishes, cephalopods, crustaceans, and insects is assessed using construct validity, causal identification, evolutionary comparison, and uncertainty-sensitive inference. The synthesis treats functional indicators as probabilistic evidence rather than direct access to experience, examines the widening empirical basis for animal welfare decisions, and outlines testable comparative research questions for the Artificial Era.

Keywords: comparative consciousness; animal consciousness; animal sentience; subjective experience; phenomenal consciousness; comparative cognition; neural correlates; pain; nociception; metacognition; self-recognition; cephalopods; corvids; insects; evolution of consciousness; artificial consciousness

Comparative Consciousness as a Scientific Research Object

Comparative consciousness investigates how evidence for experience varies across kinds of organisms and, in a more exploratory extension, engineered systems. The principal explanandum is subjective experience: whether events have a felt character for the system undergoing them and how such character might differ across perceivers. Researchers cannot observe another organism's experience directly, so they infer it from experimentally constrained relations among behavior, neural activity, physiology, and ecological organization. The inference is familiar in human science, where reports provide an additional source of evidence, but it becomes harder across species with different sensory worlds and communicative capacities. Birch et al. (2020) offers an explicitly multidimensional vocabulary for these comparisons.

The analysis requires a disciplined taxonomy. Phenomenal consciousness concerns experienced qualities; conscious access concerns information becoming usable for flexible control or report; affective sentience concerns positive and negative feelings; metacognition concerns monitoring one's own cognitive performance; and self-recognition concerns discriminating oneself from other objects or agents. These capacities can overlap in a particular organism without being logically identical or experimentally interchangeable. Seth & Bayne (2022) reviews theories that locate explanatory weight in higher-order representation, global availability, recurrent processing, and information integration. Each theory suggests different candidate indicators, and their predictions must be translated into tractable experiments rather than imposed as definitions of animals that happen to resemble humans.

A productive comparison separates three questions: what behavior is observed, which mechanisms best explain it, and whether those mechanisms warrant an attribution of experience. A reward-sensitive choice can reveal valuation while leaving open whether valuation is felt. A response to harmful stimulation reveals detection while requiring additional evidence for pain. Conversely, lack of a familiar response may reflect sensory mismatch, motor limitations, or inadequate incentives rather than lack of awareness. Sneddon et al. (2003) exemplifies why a sensory mechanism must be characterized before broader interpretation. The resulting research program seeks converging, discriminating indicators, not an all-purpose consciousness score or an anthropomorphic story about a single impressive performance.

The Historical Formation of the Comparative Problem

Comparative inquiry emerged from natural history, evolutionary biology, physiology, and the experimental study of learning. Evolutionary continuity made differences among species a question for investigation rather than an automatic hierarchy. Ethologists subsequently insisted that behavior be examined in the environments and evolutionary histories in which it operates. In a landmark methodological statement, Tinbergen (1963) distinguished causal mechanisms, developmental pathways, adaptive significance, and evolutionary history as complementary explanatory questions. That framework remains crucial: an adaptive behavior may be explained without determining whether an experience accompanies it, while a candidate neural mechanism requires developmental and phylogenetic placement before comparative generalization.

Twentieth-century psychology developed more precise measures of discriminative learning, memory, uncertainty, and response control. These achievements exposed an interpretive difficulty: organisms can solve complex tasks through mechanisms that may be represented either as explicit monitoring or as learned action policies. The science of consciousness thus inherited the methods of comparative cognition but required additional tests of what these methods reveal about experience. The investigation of mirror responses by Gallup (1970) and later studies of uncertainty monitoring by Smith et al. (2003) illustrate the transition from broad claims about animal intelligence toward operationally specified possible indicators of reflective awareness.

Recent decades expanded the candidate set beyond mammals. Bird cognition challenged the assumption that a laminated mammalian cortex supplies the only basis for sophisticated perception and control. Cephalopod experimentation demonstrated elaborate learning in an independently evolved nervous system, and studies of insect valuation prompted reassessment of small-brain constraints. Güntürkün & Bugnyar (2016) reviews the organization of avian cognition, while Barron & Klein (2016) proposes a functional route to elementary subjective experience in insects. These contributions are historically important precisely because they place rival structural and computational explanations into contact with comparative evidence. Their theoretical proposals remain open to empirical discrimination through carefully matched interventions.

Operational Definitions and the Limits of Behavioral Evidence

Any experiment on animal consciousness begins with an operational target. An investigator may measure stimulus detection, delayed choice, avoidance learning, confidence-sensitive opting out, or physiological regulation. None of these measurements is a transparent recording of a feeling. A robust construct requires multiple observable consequences and a theoretical account of why the observed pattern would be expected under a particular form of experience. Smith et al. (2003) reviews uncertainty-monitoring paradigms in which subjects can decline difficult trials; such performance can support flexible monitoring but must be compared against reinforcement histories and other decision strategies that do not require an explicit awareness attribution.

Behavioral evidence is strongest when an experiment forces contrasting models to make different predictions. If a fixed stimulus-response rule predicts a choice, successful behavior is compatible with that rule regardless of the investigator's preferred interpretation. More demanding designs alter information availability, reward structure, time delay, and task transfer while preserving sensory accessibility. They also test whether an animal can appropriately abandon a policy when the world changes. The comparative program can use formal model fitting, but model sophistication is not itself an experiential indicator. MacLean et al. (2014) demonstrates the value of large comparative datasets for self-control and also invites attention to what different tasks measure across species.

Negative results require especially careful interpretation. Mirrors may matter little to animals for whom vision is a secondary social modality, and a human-designed manipulation task can impose motor burdens unrelated to the construct of interest. Differences in motivation, stress, rearing, and experimenter interaction can mimic cognitive differences. The proper comparison therefore includes species-appropriate apparatus, familiar stimuli, preregistered alternative explanations, and ecological validation. Birch et al. (2020) argues for profiles of consciousness dimensions rather than ranking species along one imagined ladder. A scientifically useful null finding identifies the particular indicator that was unsupported under stated conditions; it leaves broader phenomenology to be assessed through additional independent methods.

Distinguishing Consciousness, Sentience, and Awareness

Consciousness is often used to denote wakefulness, sensory awareness, reflective access, and the felt character of experience. Comparative research becomes confused when these meanings are silently exchanged. An organism may be alert and responsive without passing an unfamiliar test of self-recognition. It may exhibit affective learning while lacking evidence of reflective introspection. Dehaene et al. (2017) separates computations related to global availability from those supporting self-monitoring, providing a useful functional distinction for cross-system comparisons. Those computational categories are theoretical proposals whose relations to the phenomenal character of experience require further validation rather than immediate equivalence.

Sentience, as used in the animal welfare literature, emphasizes the capacity for valenced experience: states that feel good or bad to the animal. The relevant evidence includes sustained protective behavior, reward-related relief learning, decision trade-offs, physiological integration, and neuroactive modulation. Crump et al. (2022) proposes an eight-criterion framework for examining decapod crustaceans, integrating neural prerequisites with behavior that may express affective processing. This approach gives a structured way to discuss confidence while recognizing that a single feature, including nociceptive receptors or motivational flexibility, may be shared by multiple mechanisms with different implications for experience.

Scientific precision also requires separating welfare policy from metaphysical certainty. A decision to protect an animal can be justified by a realistic risk of suffering and the magnitude of potential harm, even while theoretical interpretation remains incomplete. The New York Declaration (2024) declaration describes strong support for consciousness in other mammals and birds, and a realistic possibility for various additional vertebrates and invertebrates. It is a signed scientific statement, not a controlled experiment or a final taxonomic classification. Its importance lies in articulating how cumulative evidence and uncertainty enter practical judgment, while the underlying hypotheses continue to be tested against experimental data.

Evolutionary Explanation and Multiple Origins

A phylogenetic account asks where capacities relevant to experience appear, which features are inherited from common ancestors, and which may arise through convergent organization. Similar achievements can be produced by unlike nervous systems, while shared anatomy can support different behavioral repertoires. Tinbergen (1963) provides the essential distinction between a present mechanism and its evolutionary history. For comparative consciousness, the investigator must specify whether a neural circuit is proposed as necessary for experience, merely one implementation of a broader computation, or a historically contingent correlate that may be replaced by another architecture.

Avian and mammalian lineages present a particularly useful case. Their forebrains differ in macroscopic organization yet share evolutionarily related cellular and circuit features; birds perform demanding perceptual and cognitive tasks without the mammalian six-layered cortex. Güntürkün & Bugnyar (2016) surveys these anatomical and behavioral parallels, and Nieder et al. (2020) adds a neural experiment correlating crow pallial activity with perceptual reports in a controlled detection task. Together they challenge simple anatomical exclusions. They do not prove that every neurally complex animal has identical sensory phenomenology, and they leave open whether homologous or convergent circuit operations implement the same candidate conscious function.

Convergence is especially salient in cephalopods and artificial architectures. Cephalopod nervous systems distribute considerable processing through peripheral and central structures, whereas machine systems can instantiate computational functions through engineered components. Superficial analogies between these systems obscure large differences in evolutionary origin, sensory coupling, homeostatic regulation, and learning history. Schnell et al. (2026) illustrates the need for within-clade differentiation even among closely related molluscs. Comparative theory should therefore model independent origins as competing historical hypotheses, map functional properties explicitly, and identify conditions under which a proposed mechanism survives changes of implementation rather than assuming that human-like performance settles the issue.

Mammalian Perception, Behavioral Flexibility, and Experience

Research on mammalian consciousness draws upon sensory discrimination, flexible action, affective learning, and the organization of neural systems. These domains yield partially independent indicators. A mammal that adjusts its response when contingencies change demonstrates adaptive control, but that observation alone does not determine whether the adjustment is accompanied by a felt experience. A more informative design relates detection performance to confidence, error monitoring, access to information, and decisions under uncertainty, while measuring whether distinct sensory modalities produce comparable patterns. Birch et al. (2020) supplies a framework for constructing profiles across dimensions rather than treating the presence of a familiar mammalian brain as an all-purpose criterion.

The evolutionary relationship between humans and other mammals offers opportunities for mechanistic comparison. Investigators can examine the effects of lesions, anesthetic states, cortical interactions, and subcortical regulation, as well as identify shared pathways for pain-related learning and emotional behavior. Similar anatomy can motivate testable predictions about functions, but the evidence still depends on whether a hypothesized operation is actually performed. Seth & Bayne (2022) explains why consciousness theories disagree about which neural operations matter. Comparative experiments should therefore articulate a target capacity and distinguish a neural correlate, a necessary mechanism, and a functionally sufficient arrangement before treating evidence from one mammalian species as evidence for another.

Studies of prosocial action add a different dimension. Rats in the experiments of Ben-Ami Bartal et al. (2011) learned to release a trapped cagemate, raising questions about social motivation and the possible involvement of affective processes. Alternative contributions from social reward, prior experience, and arousal require experimental assessment, and helpful behavior is not equivalent to an identified experience of empathy. Nevertheless, when combined with observations of distress-related behavior and physiological regulation, such experiments strengthen the comparative description of how animals respond to one another. The appropriate interpretation is an evidence map connecting observed action, motivational process, and possible valenced experience, with each connection evaluated on its own terms.

Mirror Self-Recognition in Great Apes and Dolphins

Mirror self-recognition became one of the most visible comparative paradigms following Gallup (1970), who reported that chimpanzees exposed to mirrors used reflected information in behavior directed toward their own bodies. The mark test evaluates whether an animal uses a reflection to inspect a previously unseen mark placed on its body. Successful performance supports a particular capacity to coordinate visual information with one's body representation. It does not itself specify the presence of a continuous narrative self, second-order introspection, or the complete phenomenology of self-consciousness. The methodological task is to isolate self-directed uses of reflected information from general exploratory reactions and learned mirror contingencies.

Cetacean work broadened the comparative discussion. Reiss & Marino (2001) documented behavioral responses in bottlenose dolphins interpreted as mirror self-recognition, bringing into view animals with different locomotor constraints, social structures, and visual environments. Replication requires careful attention to mirror exposure, opportunities for inspection, control marks, and the distinctions between contingent movement and explicit mark-directed behavior. A dolphin cannot manipulate its body like a chimpanzee, so operational equivalence requires different observable actions. Comparisons should preserve the underlying explanatory question while allowing the measurement procedure to reflect the species' physical capabilities, sensory access, and typical spontaneous behavior.

The conceptual stakes are substantial because self-recognition is frequently conflated with consciousness itself. An animal could plausibly experience pain, color, movement, or social engagement while failing a visual mirror assay. Conversely, the capability to exploit a reflection could be generated by sophisticated body mapping and learned contingency detection without establishing all forms of reflective awareness. Smith et al. (2003) provides an adjacent literature on monitoring uncertainty that may test another proposed aspect of cognition. A comparative profile should therefore record mirror use as one experimentally characterized dimension, note the mechanisms that could support it, and maintain the distinction between a demonstrated perceptual skill and a broader inference about subjective experience.

Social Cognition and the Attribution of Mental States

Awareness of other agents is investigated through gaze following, competition, cooperation, and anticipatory action. Theoretical interpretations differ: animals may respond to behavioral regularities, represent perceptual access, or attribute unobservable mental states. Krupenye et al. (2016) used an anticipatory-looking paradigm with great apes and found looking patterns consistent with expectations about agents acting on false beliefs. This result extended the empirical scope of comparative social cognition, especially because subjects were not required to give a verbal explanation. Its evidential force depends on the contrast between the false-belief hypothesis and competing expectations derived from learned agent–object relations or task-specific attentional patterns.

Studies of social understanding illustrate why task performance and lived experience require separate analyses. An ape may anticipate another's mistaken search because it computes where the agent last saw an object; the animal may or may not explicitly experience the other as a minded subject. Stronger designs can vary perceptual history, the agent's identity, social familiarity, and the subject's opportunity to intervene. They can also analyze eye movements as a temporal trajectory rather than treating a single anticipatory look as a comprehensive psychological diagnosis. Tinbergen (1963) reminds comparative science to distinguish immediate cognitive mechanisms from the adaptive role and evolutionary history of social information processing.

Social cognition is relevant to consciousness because representing others may interact with attention, perspective taking, and access to remembered events. Yet the relation is empirical, not definitional. An organism could possess some form of sensory experience with limited social inference, and elaborate social strategies could rely partly on implicit processes. Birch et al. (2020) supports examining perceptual richness, integration, evaluation, temporal organization, and self-related dimensions individually. This approach invites cross-species experiments that compare the same informational challenge at the level of function, while permitting distinct sensorimotor responses. It also avoids treating a human-like answer to a social puzzle as a universal threshold for awareness.

Corvid Neural Activity and Sensory Report

Birds provide a major test of anatomical assumptions about consciousness. They possess forebrain organizations that differ from the laminar mammalian neocortex while supporting elaborate learning and perceptual discrimination. Güntürkün & Bugnyar (2016) reviews avian cognitive capacities and neural architectures, illustrating how functional parallels can occur with different structural arrangements. The fundamental inference is that anatomical similarity should be assessed at multiple levels, including circuit connectivity, cell types, and computations. Merely labeling a bird's brain as lacking a cortex can misrepresent its organization and prejudge a hypothesis that must instead be tested experimentally.

A particularly relevant experiment was reported by Nieder et al. (2020). Crows were trained to report visual stimulus detection, while investigators measured neural activity in the nidopallium caudolaterale. Trial-related neural signals tracked the bird's perceptual reports under conditions where physical stimulus evidence was experimentally controlled. These observations offer a comparatively strong bridge between behavioral report and neural coding in a nonmammalian lineage. They do not provide a direct observation of phenomenology; the interpretation relies on parallels with human perceptual reporting and on the exclusion of explanations based solely on motor preparation, reward expectation, and stimulus strength.

Cross-species neural comparison consequently requires an explicit account of functional homology and experimental equivalence. Neural activity associated with a report may reflect the formation of a percept, the decision to respond, attention to the trial, or the maintenance of a task rule. Researchers can attempt to dissociate these processes by changing reporting demands and perceptual uncertainty, recording more than one region, and introducing causal perturbations where ethically and technically feasible. Seth & Bayne (2022) makes clear that theoretical frameworks assign different significance to report-related activity. The crow evidence therefore constrains claims that a specific mammalian gross anatomy is universally required and motivates more precise tests of candidate mechanisms.

Memory, Uncertainty Monitoring, and Temporal Experience

Comparative memory experiments investigate whether animals retain information about what occurred, where it happened, and when it happened, as well as how those elements guide later action. Clayton & Dickinson (1998) reported that scrub jays adjusted recovery of cached food in ways consistent with memory for food type, location, and elapsed time. The term episodic-like memory is deliberately operational: the task establishes structured retrieval of event-related information while leaving the animal's possible first-person recollection unresolved. Experiments on temporal cognition are valuable precisely when they compare memory-based behavior against simpler cues such as freshness, odor, or repeatedly reinforced time-dependent rules.

Uncertainty monitoring offers a second route. In the paradigms reviewed by Smith et al. (2003), animals sometimes use an uncertainty response that declines a difficult classification trial. Adaptive opting out can indicate sensitivity to the quality of evidence, but it may also arise from associative histories or learned stimulus-response gradients. Investigators can distinguish accounts by transferring a response to novel tasks, varying incentives independently of difficulty, and fitting choice models that make separable predictions. A claim of metacognition needs an account of what is monitored, how the monitoring signal influences choice, and why alternative reinforcement-only explanations are insufficient.

Temporal experience is a more demanding target than temporal competence. Animals that recall past events or prepare for future possibilities might have experiences structured by remembered and anticipated states, but their behavioral capacities do not reveal the subjective form of those states without additional assumptions. Birch et al. (2020) treats temporality as one possible dimension among others, supporting research that asks how an organism integrates immediate stimulation, retained information, and expectations. The experimental agenda should connect memory dynamics to attention, valuation, and neural organization, while remaining cautious about inferring mental time travel from the success of a single task. Such distinctions allow comparative psychology to accumulate genuinely informative observations without collapsing diverse phenomena into an unqualified declaration of consciousness.

Octopus Nociception, Learning, and Affective Pain

Cephalopods provide a major test of assumptions that complex experiences require vertebrate anatomy. Octopuses possess large and distributed neural systems, adaptive behavioral repertoires, and opportunities for evaluating responses to injury beyond reflex withdrawal. Crook (2021) examined behavioral preference and avoidance as well as neural responses associated with noxious stimulation. The reported patterns included conditioned place avoidance and preference associated with pain relief, offering evidence relevant to an affective interpretation. Experimental interpretation still requires identifying the effects of handling, generalized stress, and associative learning, because even strong motivational responses are indirect indicators of the subjective quality of a state.

The distinction between nociception and pain is methodologically indispensable. Nociception concerns detection and processing of potentially damaging stimulation, whereas pain adds an unpleasant experience. Nociceptive neural responses can occur without a demonstration that an organism feels the event as aversive, and learned avoidance can reflect multiple pathways. A compelling evidential case uses convergent markers, including flexible protection of an affected region, longer-lasting motivational change, relief-related preferences, and neural systems capable of integrating stimulation with behavior. Crump et al. (2022) applies a similarly structured approach to decapods. In octopuses, the coupling of altered preference with centrally mediated processing increases the relevance of the findings while leaving open important questions about species variation.

The research has consequences for the design of experiments and animal care. If multiple independent results support a realistic possibility of negative experience, researchers should weigh the consequences of invasive protocols and strengthen pain-mitigation procedures. New York Declaration (2024) recognizes a realistic possibility of conscious experience in many invertebrate groups, a statement that reflects a cumulative scientific judgment rather than a direct measurement of feeling. The next empirical step is to investigate whether octopus responses generalize across contexts, forms of injury, analgesic manipulations, and developmental stages. Such work can refine causal models of affective evaluation while distinguishing welfare-relevant concern from an unsupported claim to have precisely reconstructed another species' phenomenology.

Cuttlefish Decision-Making and Event Memory

Cuttlefish permit experimentally controlled tests of delayed reward and context-sensitive choice in a lineage distinct from mammals and birds. In a delay-of-gratification paradigm, Schnell et al. (2021a) reported that cuttlefish could wait for a preferred reward under particular experimental conditions. The behavior is relevant to self-control because choices reflect the interaction of expectation, preference, and temporal delay. Its interpretation depends on distinguishing deliberate withholding from differences in food motivation, response timing, and learned task structure. Successful waiting should therefore be described as a measurable control capacity, with claims about the felt experience of wanting or anticipating kept analytically separate.

Other experiments examine the memory of past feeding events. Schnell et al. (2021b) reported preserved episodic-like memory performance with age in cuttlefish, demonstrating that information about events and their temporal organization can guide later behavior. Episodic-like is an important qualification: the experimental evidence concerns behavioral retrieval criteria and does not establish autobiographical recollection as experienced by human participants. A comparative model can nevertheless ask whether integrated event representations facilitate future decisions, whether the animal uses stored information flexibly, and how such operations relate to valuation. Different mechanisms may generate the same overt performance, making transfer tests and selective manipulation especially informative.

Evidence accumulated in one cephalopod species should not be copied uncritically to every mollusc. Schnell et al. (2026) examines cephalopod sentience using an updated assessment, emphasizing the importance of specific indicators and evidential limitations. Octopuses, cuttlefish, and squid differ in ecology, social behavior, sensory demands, and neural specialization. Comparative consciousness requires these differences to be research variables rather than incidental background. Future designs can examine whether the same candidate indicators co-occur within individuals, whether responses persist when reward and deprivation are controlled, and whether neural measures explain behavior more effectively than purely associative models. A cross-species synthesis should record where a finding has actually been demonstrated.

Decapod Crustaceans and Criteria for Sentience

Decapod crustaceans—including crabs, lobsters, and crayfish—have become central to debates about whether welfare-relevant experiences extend far beyond vertebrates. Crump et al. (2022) develops a criteria-based review that considers neural arrangements, sensory integration, flexible protective responses, associative learning, motivational trade-offs, and effects of anesthetic or analgesic interventions. The framework is designed to evaluate evidence across several kinds of indicator rather than rely on a single decisive behavior. Its logic is cumulative: concordant findings under different designs can increase confidence, especially when they have partially independent sources of error and reveal sustained, integrated regulation.

One challenge is that criteria can be correlated through the same underlying capacities. Learning to avoid an aversive location and making trade-offs involving that location may both arise from a common decision system, so treating them as statistically independent pieces of evidence can exaggerate confidence. A careful analysis identifies shared mechanisms and distinguishes direct observations from theoretical interpretations. Birch et al. (2020) encourages multidimensional comparison, which makes it possible to characterize an organism's profile without automatically ranking every species on a single continuum. Behavioral, anatomical, and pharmacological studies can contribute complementary evidence if their construct validity and alternative explanations are openly assessed.

Assessment frameworks have immediate scientific and institutional consequences. Invasive experiments on decapods require proportionate attention to the possibility of unpleasant states, and methodological refinements can reduce the cost of gathering decisive evidence. New York Declaration (2024) articulates a precaution-sensitive approach to evidence about animal consciousness, while leaving individual biological hypotheses open to testing. The most valuable future work would preregister criteria, characterize sampling across species and life stages, and examine whether interventions altering a proposed pain process change several independent outcomes. A review that identifies uncertainty is useful because it clarifies what observations could shift a scientific judgment in either direction.

Insects, Valuation, and Motivational Trade-Offs

Insects challenge the assumption that sophisticated control of behavior requires a large vertebrate-like brain. Their nervous systems are compact yet support navigation, learning, multimodal integration, and decisions shaped by changing rewards. Barron & Klein (2016) presents a theoretical argument for considering the possible origins of experience in insect systems on the basis of functional organization. The proposal raises testable questions about what kinds of integrated, organism-centered control might matter for rudimentary consciousness. It does not turn a general behavioral capacity into direct proof of sensation, and its applicability depends on whether alternative computational explanations can be experimentally distinguished.

An influential empirical contribution is the work of Gibbons et al. (2022) on bumblebees, which investigated motivational trade-offs involving noxious stimulation and rewarding food. Choices varied with reward conditions, indicating that responses to the stimulus were influenced by motivational context rather than behaving solely as a fixed withdrawal reflex. This is relevant to hypotheses about central integration of aversive and appetitive information. At the same time, a flexible valuation process can be described computationally without assuming an identifiable subjective feeling. The study consequently contributes an important indicator to a cumulative evidential case while leaving the mechanism–experience bridge as a theoretical problem.

Cross-insect inference is difficult because sensory and ecological niches vary enormously. A bee's foraging decision, an ant's navigational strategy, and a fly's defensive response impose different task requirements, and identical apparatus may fail to measure equivalent constructs. Comparative studies should quantify sensory thresholds, motivational state, the availability of alternative actions, and learning history before drawing conclusions. Birch et al. (2020) supplies a rationale for separating evaluative richness from other dimensions of consciousness. The research aim is to determine which insects exhibit which forms of flexible evaluation under validated conditions and how these behaviors depend on neural circuitry, instead of substituting a blanket claim about all insects for a program of species-specific investigation.

Fish, Nociceptors, and the Architecture of Pain Inference

Fish pain research illustrates how a biological mechanism can motivate, but cannot complete, an argument about subjective experience. Sneddon et al. (2003) identified nociceptors in rainbow trout, providing evidence for a peripheral sensory system responsive to damaging stimulation. That finding matters because it establishes a plausible detection pathway and permits more specific tests of neural and behavioral processing. Nociceptor presence does not by itself demonstrate pain, since the same broad function can contribute to protective action in the absence of a particular form of awareness. The scientific question is whether the system integrates nociceptive information into sustained affective valuation and adaptive action.

Experiments that seek to distinguish pain from reflexes can examine long-term changes in attention, altered feeding, shifts in preference, injury-specific protection, and responses to treatments that affect nociceptive processing. Each measure has competing explanations. Reduced activity after injury might reflect physiological impairment; changes in feeding might follow stress or sickness; and a pharmacological intervention might alter motor function. Investigators need appropriate controls, dosage interpretation, and multiple outcomes before linking a treatment effect to an inferred experience. Crump et al. (2022) demonstrates why integrated frameworks are often more informative than single-test conclusions, even when applied to organisms outside the decapod literature.

The evolutionary position of fishes also complicates generalization. Teleosts, cartilaginous fishes, and other vertebrate lineages have distinct neuroanatomical organizations and ecological pressures, so a finding in trout is not a measured property of all aquatic animals. New York Declaration (2024) emphasizes a realistic possibility of conscious experience in all vertebrate groups in its scientific declaration, but such a statement is a synthesis of evidence and judgment rather than a substitute for taxon-specific experiments. The research program should connect sensory biology, causal neural interventions, and context-sensitive behavior, while stating carefully what a specific species and study design actually establish.

Competing Theories and Their Comparative Predictions

Global-workspace accounts connect conscious access to the broad availability of information across specialized systems; recurrent-processing theories emphasize particular feedback interactions; higher-order approaches assign significance to representations of first-order mental states; and integrated-information approaches emphasize aspects of a system's causal structure. Seth & Bayne (2022) surveys these positions and their divergent explanatory commitments. A comparative experiment should identify what each account predicts in the species under investigation. If global availability matters, flexible cross-task use of information becomes relevant. If recurrent activity matters, perturbing feedback may offer a discriminating test. The mere presence of sophisticated behavior or recurrent circuits would still leave competing interpretations, and theory must specify the hypothesized relation to experience.

Theories developed primarily around adult human data require careful translation to animals with different nervous systems. Dehaene et al. (2017) illustrates the conceptual value of distinguishing global accessibility and metacognitive self-monitoring when considering nonhuman and artificial architectures. A theory requiring a human-style linguistic report would exclude many organisms by definition rather than explain their empirical properties. Conversely, weakening every requirement until any information-processing system qualifies would make the theory difficult to test. Comparative researchers should identify abstract functional requirements, specify plausible implementations, and design behavioral and neural predictions that could fail under observation.

The diversity of plausible theoretical explanations creates a model-selection problem. Two theories can predict the same outcome while offering distinct stories about its role in experience. Strong experiments target disagreements: whether experience follows a particular neural feedback pattern under stable performance, whether apparent confidence depends on explicit monitoring, or whether a stimulus can guide action without broader accessibility. Nieder et al. (2020) supplies a case in which neural recordings and perceptual reports can be related, while Gibbons et al. (2022) supplies a case involving motivational integration. Neither result directly adjudicates the entire theoretical landscape. Progress requires integrating decisive contrasts with transparent uncertainty rather than selecting a theory because an individual experiment appears intuitively congenial.

Multidimensional Consciousness Profiles and Comparative Taxonomy

A single consciousness ranking assumes that perception, affect, temporal integration, cognitive access, and self-monitoring rise and fall together. Comparative evidence gives little reason to impose that assumption before measurement. Birch et al. (2020) proposes dimensions of animal consciousness, enabling a description in which organisms can differ along multiple axes. Perceptual richness might concern the discriminations available within sensory experience; evaluative richness concerns the possible variety and organization of valenced states; integration concerns how information from different sources is jointly usable. Temporal and self-related dimensions demand their own tasks. The result is a multidimensional research space whose components must be operationalized independently.

Profiles can help distinguish scientifically meaningful variation from artifacts of measurement. A bird may outperform a mammal in one visual discrimination task while differing in smell, motor access, or training history. The correct unit of comparison can be a latent construct inferred from several tasks, provided the tests show suitable within-species reliability and cross-species measurement invariance. MacLean et al. (2014) demonstrates the potential of coordinated cross-species experiments, but sample composition and ecological factors must enter interpretation. Even when a shared latent variable can be estimated, the psychological meaning of its values remains a question for validation rather than an automatic conclusion of measurement.

The dimensions should also be examined for causal and developmental relations. A capacity for uncertainty monitoring might depend on memory and flexible response inhibition, while affective trade-offs might depend on motivational integration. Such relations can generate correlations without indicating a unitary essence of consciousness. Smith et al. (2003) and Crump et al. (2022) offer contrasting behavioral and sentience-oriented assessment traditions that can be connected through explicit models. A sophisticated comparative atlas would record confidence intervals, coverage gaps, apparatus differences, and which experiments support each dimension. It would make patterns of converging evidence visible while preventing an unsupported leap from one unusually competent performance to universal claims about an entire species.

Species-Appropriate Experiments and Measurement Invariance

Comparative testing begins with the ecological and sensory conditions that make a task intelligible to the organism. A visually marked mirror may be appropriate for some primates and dolphins yet uninformative for animals that chiefly navigate by smell, touch, or vibration. Gallup (1970) and Reiss & Marino (2001) demonstrate how the same functional construct can require distinct behavioral implementations. Apparatus should respect locomotion, sensory thresholds, the cost of responding, and patterns of spontaneous exploration. Researchers must also document prior training and individual history, because test familiarity can dominate observed performance. Species-appropriate design is a validity requirement, not merely an accommodation.

Measurement invariance requires more than presenting superficially similar stimuli. The meaning of an error, latency, or opt-out response can depend on an animal's motivational state and the motor sequence required for a choice. Statistical comparisons can include species-by-condition interactions, trial-level models, and sensitivity analyses across alternative reward functions. Smith et al. (2003) describes uncertainty-monitoring studies that are particularly vulnerable to learned task policies and reinforcement contingencies. A strong study therefore tests whether the candidate indicator transfers to new perceptual domains, whether control tasks isolate the intended process, and whether comparable latent constructs explain behavior across different apparatus arrangements.

Replication poses another challenge because many species are represented by a few highly trained individuals. Laboratories should report animals' ages, backgrounds, housing, testing environments, exclusions, and detailed procedures. Preregistration can distinguish planned contrasts from findings selected after exploring many outcomes; open materials allow independent reproduction; and ethically justified multi-site designs can test generalization. MacLean et al. (2014) shows the value of comparative breadth, while Tinbergen (1963) reminds researchers that variation between ecological circumstances may be part of the phenomenon rather than experimental noise. The goal is a cumulative evidence base capable of separating species differences in the target process from differences generated by the testing conditions.

Neural Indicators, Perturbation, and Causal Interpretation

Neural activity can correlate with a perceptual report without establishing that the activity is necessary or sufficient for experience. In crows, Nieder et al. (2020) identified neural signals associated with subjective perceptual reports in a visual detection task. The result opens opportunities for comparing report-related computations across lineages. Interpretation requires separating sensory encoding, decision preparation, motor response, attention, and reward prediction. A neural marker that predicts a choice after stimulus intensity is controlled is more informative than an uncorrected stimulus response, yet it remains a correlate until interventions establish a more specific causal role.

Perturbational methods seek stronger causal discrimination by altering neural processes and measuring changes in target capacities. An intervention can nevertheless affect alertness, movement, or motivation as well as the hypothesized conscious process. A valid conclusion requires controls for these alternative effects, preferably combined with trial-level behavioral modeling. Seth & Bayne (2022) outlines why theories of consciousness select different mechanisms as their causal targets. Cross-species research must adapt those predictions to distinct neural anatomies: a manipulation of cortical feedback in a mammal has no literal anatomical counterpart in a cephalopod, although it may have a meaningful functional analogue.

Neural indicators derived from one taxon should be tested against negative controls and alternative architectures. Güntürkün & Bugnyar (2016) reviews cognition in birds without mammalian neocortical organization, challenging overly restrictive anatomical definitions. At the same time, a broadly distributed computational motif is not automatically evidence of phenomenology. A credible comparison asks whether the motif participates in the relevant information integration, whether selectively disabling it disrupts the predicted behavior, and whether another model predicts the observed results just as well. These questions transform anatomical comparison from a search for human resemblance into a systematic study of causal mechanisms and their possible relation to experience.

Critical Evidence Integration, Replication, and Scientific Uncertainty

Evidence synthesis in comparative consciousness should begin with explicit inclusion criteria, clear definitions of each indicator, and a separation between primary data and interpretive claims. A behavioral paper may demonstrate an effect with strong internal validity while leaving its relation to subjectivity unsettled. A theoretical review may organize plausible mechanisms without contributing an independent observation. Crook (2021) offers experimental pain-relevant observations, whereas Schnell et al. (2026) evaluates an extensive literature using prespecified sentience criteria. Their roles are complementary, and a narrative synthesis must preserve their different evidentiary jurisdictions rather than pool them as interchangeable demonstrations.

The quality of a comparative claim depends on controls, sampling, the consequences of experimenter decisions, and independent replication. Evidence from a single species and apparatus is narrower than a result reproduced with new animals and settings; positive findings selected from many unreported tasks can exaggerate apparent prevalence. Meta-analytic techniques are useful when designs and outcomes are meaningfully comparable, but aggregating unrelated indicators into one numerical score may conceal theoretical disagreements. Birch et al. (2020) provides a rationale for representing distinct dimensions separately, and Crump et al. (2022) exemplifies an organized criteria framework. Researchers should state whether confidence judgments are expert assessments, statistical estimates, or model-based posterior quantities.

The strongest inference emerges from independent methods that converge on a common explanation while resisting plausible alternative mechanisms. For instance, conditioned place preferences, pharmacological modulation, tissue-specific protection, and central neural responses may jointly support an affective-pain hypothesis more strongly than any isolated component. Correlations among indicators must still be recognized, and evidence gaps cannot be converted into negative findings about a taxon. New York Declaration (2024) expresses a collective appraisal of the field rather than a universal proof. A durable scholarly synthesis reports what has been observed, what hypotheses the observations favor, where contradictory findings persist, and which feasible experiment would materially change the assessment.

Artificial Systems and the Theory-Based Indicator Approach

Engineered systems widen the comparison beyond biological lineages and sharpen the question of which functional or causal properties should count as indicators of experience. Dehaene et al. (2017) examines whether machine systems could instantiate properties associated with conscious access and self-monitoring, while Butlin et al. (2026) proposes a theory-based assessment using indicators derived from neuroscientific accounts. The methodology begins by specifying a particular system and architecture, then asking which theoretically motivated functions it exhibits under independent tests. Such an assessment is an investigation of candidate evidence; a language model's ability to talk about feelings is neither a direct measurement of experience nor a decisive absence of it.

Artificial architectures introduce distinctive challenges for operationalization. Their internal state transitions, training objectives, memory access, recurrent processes, tool use, and external control can often be inspected or modified more directly than the nervous system of a living organism. Yet functional similarities may be superficial: a textual statement about confidence can be optimized to satisfy a prompt rather than report a monitored epistemic state. Butlin et al. (2026) argues that indicators should be grounded in theories and assessed within actual system architectures. Experiments can compare behavior before and after specific architectural interventions, test stability across novel contexts, and evaluate whether the hypothesized mechanism contributes causally to task performance.

Comparative animal science offers both a caution and an opportunity. Behavioral resemblance cannot settle phenomenology, but neither should an unusual physical substrate automatically disqualify a candidate mechanism if a theory is intended to be implementation-general. Birch et al. (2020) provides a multidimensional framework that can guide operational comparisons, while recognizing that animal capacities and machine functions have very different histories and stakes. For artificial systems, the most productive program identifies indicators with falsifiable computational signatures, audits their causal roles, and states the uncertainty in bridging these signatures to experience. The question remains open at the level of empirical and theoretical adjudication.

Artificial Era and the Boundaries of Cross-System Inference

The Artificial Era creates a research environment in which animal nervous systems, human social cognition, and increasingly capable artificial systems can be compared without presuming that a shared behavioral label names an identical mechanism. Comparative consciousness contributes a general discipline of inference: define the target capacity, examine independent observations, test rival explanations, and report the distance between functional findings and attributed phenomenology. Tinbergen (1963) supplies complementary questions concerning mechanism, development, function, and historical origin; engineered systems require an adapted account of design histories, training, deployment, and subsequent modification rather than literal biological phylogeny.

A scientifically interpretable cross-system experiment compares functions at the right level of abstraction. For example, opting out of uncertain choices in a monkey, a bird, or a machine may depend on distinct mechanisms of uncertainty estimation and reward selection. Smith et al. (2003) makes the reinforcement alternatives visible in animal studies, and Butlin et al. (2026) applies theory-based indicator reasoning to AI. A comparative claim becomes meaningful only after researchers establish what variables have comparable explanatory roles. Similar output distributions are insufficient if one system retrieves a learned script while another monitors graded internal evidence; conversely, different output modalities need not imply different underlying functions.

The same framework illuminates human attribution. People may ascribe intentions, understanding, or feelings to responsive systems through social and linguistic cues, making the psychology of the observer an important source of data rather than an evidential shortcut. Dehaene et al. (2017) emphasizes the need to distinguish functions associated with availability and self-monitoring. Theoretical writing in the Aisentica philosophical corpus may supply hypotheses and conceptual categories for future study, but empirical assertions about artificial consciousness must be established through defined indicators and independently testable mechanisms. This division of explanatory roles preserves the possibility of ambitious cross-system research while keeping the evidentiary standard visible.

Animal Welfare, Ethical Decisions, and Precaution Under Uncertainty

Questions of sentience have immediate consequences for husbandry, experimentation, fisheries, and conservation. A welfare decision cannot always await a conclusive theory of subjective experience because interventions affecting living animals may be irreversible or costly. New York Declaration (2024) articulates a position according to which realistic possibilities of conscious experience warrant consideration in decisions affecting animals. The scientific foundation of this approach is cumulative evidence and an explicit account of the possible severity of harm. It does not require treating each proposed indicator as proven, and it allows precautionary practice to evolve when new observations change the weight of evidence.

Cephalopods exemplify the relation between research and policy. The detailed review of Schnell et al. (2026) finds that evidence differs across octopuses, cuttlefish, squid, and nautiluses, with particularly sparse study of the latter. The finding that an animal group is understudied must not be converted into evidence that it is insentient. In applied settings, welfare standards can respond to plausible suffering while funding the targeted studies needed to reduce uncertainty. Crook (2021) adds experimental data concerning octopus pain-related behavior and neural activity; such studies are relevant to refining protocols for handling, tissue injury, analgesia, and humane treatment.

Decapod crustaceans and insects raise parallel issues in settings where routine handling affects enormous numbers of individuals. Crump et al. (2022) offers criteria for reviewing decapod evidence, and Gibbons et al. (2022) investigates flexible trade-offs in bumblebees. These contributions call for policies calibrated to the strength of evidence, the feasibility of harm reduction, and the costs of mistaken exclusion. Policy choices should distinguish confidence in an empirical result from the consequences attached to uncertain attribution. Transparency about both parts can improve accountability. Scientific progress is served when welfare rules, experimental protocols, and future investigations are documented in ways that permit revision as the evidence develops.

Unresolved Questions and a Comparative Research Agenda

A central unresolved question concerns the bridge from neural and behavioral function to the existence and qualities of experience. Different theories propose different necessary conditions, and empirical results often fit more than one model. Seth & Bayne (2022) describes the continuing theoretical plurality, while Birch et al. (2020) argues that comparative consciousness may have several independently varying dimensions. Future investigations should design adversarial experiments in which competing theories preregister distinct predictions about specific species or systems. The key outcome would be a change in explanatory support rather than a rhetorical declaration that an organism is definitely conscious because it passed an appealing test.

A second agenda concerns sampling, reproducibility, and neglected taxa. Schnell et al. (2026) identifies uneven evidence across cephalopod groups; many smaller or less accessible species remain underrepresented in existing datasets. Research needs validated, ecologically meaningful tasks that can be repeated across individuals, laboratories, developmental stages, and populations. Neurophysiological recordings should be paired with behavior that is independently interpretable, and intervention studies must control motor impairment and generalized stress. Open protocols and preregistered analyses would reduce selective reporting. Longitudinal designs could clarify whether candidate indicators develop together or dissociate, providing more informative constraints on the dimensional structure of consciousness.

A third agenda concerns the comparative scope of theory-based indicators across biology and engineering. Butlin et al. (2026) provides one approach to inspecting the computational properties of artificial systems, while Nieder et al. (2020) exemplifies an animal study combining physiological signals with psychophysical report. An integrated program could ask which indicators survive changes in neural or artificial implementation, which require evolved homeostatic regulation, and which provide reliable evidence only in particular ecological or task contexts. Success depends on measuring mechanistic contributions and specifying rival hypotheses. It also requires careful public communication, because the appeal of categorical verdicts can exceed the precision of the available science.

Conclusion

Comparative consciousness is a scientific investigation of the distribution and organization of possible subjective experience across species and systems. It begins with observable behavior and neural mechanisms, but its ultimate explanatory target is the relationship between those observations and experience. The reviewed literature supports substantial advances in operational testing: mirror-directed behavior, uncertainty monitoring, event memory, sensory report, nociceptive processing, pain-related motivational change, and flexible trade-offs have all become subjects of controlled study. Gallup (1970), Nieder et al. (2020), Crook (2021), and Gibbons et al. (2022) illustrate distinct experimental contributions. Their evidential relevance emerges from the mechanisms and alternatives each design can address, not from an assumed equivalence among the constructs they measure.

The comparative record also establishes the inadequacy of a simple hierarchy based on proximity to human anatomy or performance. Avian and cephalopod findings demonstrate how sophisticated functions may be investigated in differently organized nervous systems, while evidence reviews document substantial variation in coverage and confidence across invertebrate taxa. Güntürkün & Bugnyar (2016) and Schnell et al. (2026) make those differences scientifically salient. Birch et al. (2020) supplies a framework for considering multiple dimensions of experience and for assigning different degrees of support to different proposed capacities. Scientific uncertainty is thereby transformed into a map of specific questions, methods, and evidence gaps rather than an undifferentiated inability to know anything about other minds.

The most defensible synthesis combines species-appropriate experimental designs, controlled interventions, theoretical comparison, and transparent evaluations of evidence strength. Animal welfare decisions can proceed under proportionate precaution, as discussed by New York Declaration (2024), while empirical research continues to discriminate among competing accounts. In the Artificial Era, Butlin et al. (2026) extends theory-derived indicator analysis to engineered systems and thereby expands the comparative agenda. Across every substrate, the methodological discipline remains constant: define what is being claimed, specify what would count as evidence, evaluate alternative explanations, and retain a clear distinction between observed capability and inferred phenomenology. These commitments establish the research program's conceptual coherence and its continuing scientific productivity.

References

  • Barron, A. B., & Klein, C. (2016). What insects can tell us about the origins of consciousness. Proceedings of the National Academy of Sciences, 113(18), 4900–4908. https://doi.org/10.1073/pnas.1520084113
  • Ben-Ami Bartal, I., Decety, J., & Mason, P. (2011). Empathy and pro-social behavior in rats. Science, 334(6061), 1427–1430. https://doi.org/10.1126/science.1210789
  • Birch, J., Schnell, A. K., & Clayton, N. S. (2020). Dimensions of animal consciousness. Trends in Cognitive Sciences, 24(10), 789–801. https://doi.org/10.1016/j.tics.2020.07.007
  • Butlin, P., Long, R., Bayne, T., Bengio, Y., Birch, J., Chalmers, D., Constant, A., Deane, G., Elmoznino, E., Fleming, S. M., Ji, X., Kanai, R., Klein, C., Lindsay, G., Michel, M., Mudrik, L., Peters, M. A. K., Schwitzgebel, E., Simon, J., & VanRullen, R. (2026). Identifying indicators of consciousness in AI systems. Trends in Cognitive Sciences, 30(6), 488–501. https://doi.org/10.1016/j.tics.2025.10.011
  • 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
  • Crook, R. J. (2021). Behavioral and neurophysiological evidence suggests affective pain experience in octopus. iScience, 24(3), 102229. https://doi.org/10.1016/j.isci.2021.102229
  • Crump, A., Browning, H., Schnell, A. K., Burn, C. C., & Birch, J. (2022). Sentience in decapod crustaceans: A general framework and review of the evidence. Animal Sentience, 7(32), Article 1. https://doi.org/10.51291/2377-7478.1691
  • Dehaene, S., Lau, H., & Kouider, S. (2017). What is consciousness, and could machines have it? Science, 358(6362), 486–492. https://doi.org/10.1126/science.aan8871
  • Gallup, G. G., Jr. (1970). Chimpanzees: Self-recognition. Science, 167(3914), 86–87. https://doi.org/10.1126/science.167.3914.86
  • Gibbons, M., Versace, E., Crump, A., Baran, B., & Chittka, L. (2022). Motivational trade-offs and modulation of nociception in bumblebees. Proceedings of the National Academy of Sciences, 119(31), e2205821119. https://doi.org/10.1073/pnas.2205821119
  • 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
  • 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., 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
  • New York Declaration on Animal Consciousness. (2024, April 19). The New York Declaration on Animal Consciousness. New York University. https://sites.google.com/nyu.edu/nydeclaration/declaration
  • 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
  • Reiss, D., & Marino, L. (2001). Mirror self-recognition in the bottlenose dolphin: A case of cognitive convergence. Proceedings of the National Academy of Sciences, 98(10), 5937–5942. https://doi.org/10.1073/pnas.101086398
  • 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
  • Seth, A. K., & Bayne, T. (2022). Theories of consciousness. Nature Reviews Neuroscience, 23, 439–452. https://doi.org/10.1038/s41583-022-00587-4
  • Smith, J. D., Shields, W. E., & Washburn, D. A. (2003). The comparative psychology of uncertainty monitoring and metacognition. Behavioral and Brain Sciences, 26(3), 317–339. https://doi.org/10.1017/S0140525X03000086
  • Sneddon, L. U., Braithwaite, V. A., & Gentle, M. J. (2003). Do fishes have nociceptors? Evidence for the evolution of a vertebrate sensory system. Proceedings of the Royal Society B: Biological Sciences, 270(1520), 1115–1121. https://doi.org/10.1098/rspb.2003.2349
  • 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