Human and animal intelligence constitute interconnected expressions of biological cognition shaped by evolution, ecological demands, nervous-system organization, learning, and social interaction. Their scientific investigation addresses one of the most fundamental questions in artificial intelligence and cognitive science: What makes a system intelligent, and how can intelligence be recognized across organisms with different bodies, senses, evolutionary histories, and ways of interacting with the world?
Humans are biological animals, and human intelligence is therefore part of the broader natural history of animal cognition. Nevertheless, humans exhibit an unusually developed combination of symbolic language, cumulative culture, abstract reasoning, explicit teaching, long-term cooperation, and technological innovation.
Nonhuman animals demonstrate diverse cognitive abilities, including associative learning, flexible problem-solving, spatial memory, tool use, numerical discrimination, communication, social learning, and adaptive decision-making. These abilities are distributed unevenly across species and cannot be adequately represented by placing all organisms along a single scale of intellectual superiority.
Comparative cognition investigates the mechanisms, evolutionary origins, ecological functions, and developmental processes underlying these abilities. It distinguishes observable behaviour from the cognitive mechanisms hypothesized to produce that behaviour and applies controlled experiments to evaluate competing explanations.
This research is foundational to artificial intelligence because it reveals that intelligence is not necessarily dependent on human language, human anatomy, or a single computational architecture. Instead, intelligent behaviour can emerge through different combinations of perception, learning, memory, representation, action, and environmental adaptation.
1. DEEP DEFINITION
1.1. What Is Human and Animal Intelligence?
Human and animal intelligence refers to the capacity of biological organisms to acquire, process, retain, and use information in ways that support flexible behaviour, problem-solving, learning, prediction, and adaptation to relevant environmental conditions.
This definition is operational rather than a claim that every intelligent organism possesses the same mental processes.
At a broad level, biological intelligence can involve:
- Detecting and discriminating environmental information.
- Learning relationships between events, actions, and outcomes.
- Remembering relevant experiences.
- Adjusting behaviour when environmental conditions change.
- Selecting actions according to needs, opportunities, and constraints.
- Generalizing learned information to unfamiliar situations.
- Navigating physical and social environments.
- Coordinating behaviour with other organisms.
- Using information to anticipate possible future outcomes.
These capabilities do not necessarily occur together or to the same degree.
A species may exhibit exceptional spatial memory without demonstrating human-like language. Another may possess sophisticated social coordination while performing modestly on an unfamiliar object-manipulation task.
Consequently, intelligence should be understood as a collection of interacting capacities whose expression depends on biological organization and ecological context [6].
1.2. Human Intelligence as a Biological Form of Cognition
Human intelligence includes the cognitive capacities that support learning, reasoning, communication, problem-solving, planning, and purposeful adaptation.
It also involves mechanisms supporting abstraction, symbolic representation, executive control, social understanding, and the transmission of knowledge across generations.
Among the most consequential human cognitive developments are:
Symbolic language: Humans can communicate about objects, events, possibilities, and abstractions that are absent from the immediate environment.
Compositional reasoning: Human thought can combine concepts and relationships into structures capable of expressing many novel propositions.
Cumulative culture: Knowledge and technology can be preserved, modified, and improved across generations rather than repeatedly rediscovered by each individual.
Explicit instruction: Humans routinely transmit complex skills through explanation, demonstration, correction, and organized education.
Collective problem-solving: Human groups can distribute knowledge, develop specialized roles, and coordinate action through institutions and shared representations.
Metacognition: Humans can explicitly assess aspects of their own knowledge, uncertainty, strategies, and reasoning.
These capacities interact rather than functioning as completely independent faculties.
For example, scientific research depends on language, memory, symbolic representation, social cooperation, technological tools, and methods for evaluating evidence.
Human intelligence is therefore not merely a property of isolated individual reasoning. Its practical capabilities are strongly influenced by cultural environments, educational opportunities, external representations, and accumulated collective knowledge.
1.3. Animal Intelligence Beyond Human-Centered Definitions
Nonhuman animal intelligence encompasses cognitive abilities that support adaptive behaviour across a wide range of ecological circumstances.
Examples include:
Corvids: Crows and related birds demonstrate sophisticated behaviours involving tool use, problem-solving, caching, and social information.
Great apes: Chimpanzees and other great apes exhibit complex social cognition, flexible tool use, observational learning, and cooperative or competitive strategies.
Cetaceans: Dolphins and other cetaceans possess elaborate communication systems, social relationships, and forms of socially transmitted behaviour.
Elephants: Elephants exhibit complex social behaviour, learning, memory, and cooperative interactions.
Dogs: Domestic dogs demonstrate substantial sensitivity to human communicative cues, learned signals, and social environments.
Rodents: Rats and other rodents support extensive research on learning, spatial cognition, memory, and behavioural flexibility.
Insects: Bees and other insects demonstrate learning, navigation, communication, and experimentally measurable numerical or categorical discrimination.
Cephalopods: Octopuses and related organisms exhibit flexible exploration, learning, and problem-solving despite nervous systems organized differently from those of vertebrates.
These examples demonstrate that sophisticated cognition is not confined to primates.
However, the presence of a particular behaviour does not automatically prove the existence of a particular internal mental representation.
Tool use, for example, may involve different combinations of learned associations, causal understanding, perception, and motor control.
Comparative cognition seeks to determine which mechanisms best explain the observed behaviour rather than treating superficial behavioural resemblance as proof of identical intelligence [6].
1.4. Intelligence, Adaptation, and Ecological Context
Biological intelligence develops and operates within an organism’s ecological environment.
An animal’s relevant cognitive problems are influenced by its habitat, sensory capabilities, social relationships, food resources, predators, reproductive demands, and movement requirements.
A food-caching bird may benefit from accurately remembering numerous storage locations.
A migratory animal may depend on sophisticated navigation mechanisms.
A social primate may benefit from recognizing individuals, relationships, and competitive intentions.
An insect may solve navigation problems using sensory information that humans do not naturally perceive in the same way.
These capacities should be assessed against the ecological problems they help organisms address.
A conceptual representation of ecologically situated cognitive performance is:
Cᵢ = Eₑ∼Dᵢ [Uᵢ(πᵢ, e)]
Where:
- Cᵢ represents expected task-related performance for organism or agent i.
- Dᵢ represents a distribution of relevant environmental conditions.
- Uᵢ measures performance under an explicitly specified ecological or experimental objective.
- πᵢ represents the organism’s behavioural policy.
- e represents an encountered environmental condition.
This is an analytical framework, not a universally accepted numerical definition of biological intelligence.
Its importance lies in illustrating that performance depends on the environment and the task used for evaluation.
A species cannot be declared globally less intelligent merely because it performs poorly on a task requiring unfamiliar sensory, motor, or motivational abilities.
1.5. General Intelligence and Specialized Cognitive Abilities
An important question in intelligence research concerns the relationship between broadly transferable cognitive ability and specialized task competence.
In human psychometrics, Charles Spearman’s work contributed to the development of the general intelligence factor, commonly denoted by g [5].
In a simplified one-factor measurement model:
Xᵢⱼ = μⱼ + λⱼgᵢ + εᵢⱼ
Where:
- Xᵢⱼ represents individual i’s performance on cognitive measure j.
- μⱼ represents the measure’s baseline or intercept.
- λⱼ represents its loading on the common factor.
- gᵢ represents the individual’s latent general factor.
- εᵢⱼ represents residual variation.
Such models summarize patterns of correlation among cognitive test results.
They do not establish that all cognition is reducible to one biological mechanism.
Nor does the existence of a general factor in human test batteries establish that the same factor can be measured equivalently across unrelated animal species.
Animal cognition research often finds that abilities vary according to ecological specialization, experimental design, developmental experience, and the cognitive mechanisms involved.
Some cross-species studies investigate broad cognitive predictors, but the existence, comparability, and interpretation of a universal animal intelligence factor remain contested.
A distinction is therefore necessary between:
- General cognitive performance within a defined test battery.
- Specialized abilities supporting particular ecological tasks.
- Transfer of learned knowledge across different contexts.
- Human psychometric intelligence constructs.
- Claims about intelligence across entire species.
These concepts overlap but are not interchangeable.
1.6. Intelligence, Learning, Cognition, and Consciousness
Although frequently discussed together, these terms refer to different scientific concepts.
| Concept | Central meaning | Important distinction |
|---|---|---|
| Intelligence | Flexible and effective use of information for adaptive tasks | Not identical to processing speed or memory capacity |
| Cognition | Processes involved in acquiring, representing, transforming, and using information | Includes processes that may be automatic or unconscious |
| Learning | Experience-dependent change in knowledge, behaviour, or underlying response tendencies | Not all learned behaviour demonstrates broad intelligence |
| Memory | Encoding, retention, and retrieval of information | Strong memory does not necessarily imply strong reasoning |
| Reasoning | Drawing inferences or evaluating relationships between information | May occur through different mechanisms |
| Consciousness | Subjective experience or awareness, depending on the theoretical definition | Cannot be established solely from successful task performance |
| Instinct | Behaviour influenced by evolved biological predispositions | Can interact with learning and experience |
| Adaptation | A trait or process associated with adjustment to environmental demands | Adaptive outcomes do not necessarily require conscious planning |
An animal can demonstrate sophisticated learning without providing conclusive evidence about the nature of its subjective experience.
Similarly, a highly effective behaviour may result from evolved specialization rather than flexible, general-purpose reasoning.
The scientific study of animal intelligence therefore requires conceptual discipline: observable competence, inferred cognitive mechanism, and subjective experience should be examined separately.
1.7. Evolutionary Continuity and Cognitive Diversity
Modern evolutionary biology recognizes that humans and other animals share common ancestry.
Human cognition did not originate independently of biological evolution.
However, evolutionary continuity does not mean that every species possesses identical cognitive abilities or that intelligence develops along one predetermined path.
Different evolutionary lineages can produce similar functional solutions through different biological mechanisms.
This phenomenon is known as convergent evolution.
Research on corvids and apes has highlighted similarities in complex problem-solving and social cognition despite substantial differences in evolutionary history and brain organization [9].
These findings suggest that some cognitive abilities can emerge independently when different organisms encounter partially comparable adaptive challenges.
At the same time, similar external behaviour does not guarantee identical internal mechanisms.
The strongest scientific explanation must therefore integrate behavioural evidence with evolutionary relationships, neurobiology, development, and controlled experimental analysis.
1.8. Historical Development of Comparative Intelligence Research
The study of human and animal intelligence draws on centuries of philosophical inquiry and more than a century of systematic behavioural experimentation.
Charles Darwin and evolutionary continuity
In The Descent of Man, and Selection in Relation to Sex (1871), Charles Darwin examined similarities and differences between human and nonhuman mental capacities, helping establish an evolutionary approach to cognition [1].
Darwin’s work was historically influential, although some of its terminology and claims reflect nineteenth-century assumptions that are not accepted as contemporary scientific standards.
Edward Thorndike and experimental animal learning
In 1898, Edward L. Thorndike systematically investigated animal learning using controlled problem-solving situations, including puzzle-box experiments.
His work contributed to the experimental study of how behaviour changes through experience and its consequences [2].
Edward Tolman and cognitive representations
In 1948, Edward C. Tolman’s influential discussion of cognitive maps challenged exclusively stimulus-response interpretations of learning.
His work supported the investigation of internal representations in spatial navigation and behaviour [3].
Niko Tinbergen and ethological explanation
In 1963, Niko Tinbergen described four complementary questions concerning behaviour: immediate causation, development, adaptive function, and evolutionary history [4].
This framework remains important because observing what an animal does is not sufficient to explain why the behaviour occurs or how it developed.
Psychometrics and human cognitive differences
Spearman’s early work on intelligence measurement introduced influential methods for examining common variation across human cognitive tasks [5].
Subsequent research developed increasingly sophisticated approaches to assessing cognitive abilities while also exposing methodological limitations and cultural influences.
Modern comparative cognition
Research synthesized by Sara Shettleworth and others integrated experimental psychology, behavioural ecology, cognitive science, and evolutionary biology [6].
Modern comparative studies investigate physical cognition, social cognition, numerical abilities, memory, communication, and flexible problem-solving across diverse species.
Social cognition and cultural intelligence
Research comparing young human children with great apes has supported the importance of human social-cognitive capacities in understanding cultural learning and cooperation [7].
Animal culture and social transmission
Studies of chimpanzee populations documented geographically variable behavioural traditions and strengthened investigation of nonhuman cultural processes [8].
Convergent cognitive evolution
Research on corvids and primates demonstrated the value of studying similar cognitive capacities in distantly related evolutionary lineages [9].
Large-scale comparative experimentation
Cross-species experiments have examined behavioural self-control and its relationship to brain characteristics, social ecology, and evolutionary history [15].
Together, these developments transformed animal intelligence research from anecdotal comparisons into a multidisciplinary scientific field based on controlled observation, competing hypotheses, statistical inference, and evolutionary explanation.
2. TECHNICAL MECHANICS
2.1. The Biological Information-Processing Cycle
Biological cognition involves continuous interaction between an organism’s nervous system, body, and environment.
A simplified information-processing cycle includes:
Sensory acquisition: Detecting relevant environmental information.
Perceptual processing: Organizing sensory signals into representations suitable for behavioural control.
Attention: Prioritizing information according to current conditions and task demands.
Memory: Encoding, maintaining, and retrieving relevant information.
Evaluation: Integrating present information with learned associations, expectations, and biological needs.
Action selection: Choosing behavioural responses.
Motor execution: Producing actions that affect the environment.
Feedback processing: Updating behaviour and internal representations according to observed consequences.
A general mathematical abstraction is:
aₜ ∼ π(a | oₜ, mₜ, gₜ)
Where:
- aₜ represents the selected action.
- oₜ represents current sensory information.
- mₜ represents available memory or internal state.
- gₜ represents a current objective, drive, or goal-related condition.
- π represents the behavioural policy.
The organism’s internal state may change according to:
mₜ₊₁ = F(mₜ, oₜ, aₜ, rₜ₊₁)
Here, F is an abstract update function and rₜ₊₁ represents outcome-related feedback.
These equations describe a computational perspective on behaviour.
They do not imply that an animal explicitly calculates these mathematical expressions or that every behaviour depends on a consciously represented goal.
2.2. Associative Learning and Prediction
Associative learning is a foundational mechanism studied in humans and many nonhuman animals.
It concerns experience-dependent changes in responses to relationships among stimuli, actions, and outcomes.
The Rescorla–Wagner model provides a historically influential mathematical account of certain forms of Pavlovian conditioning [16].
A simplified learning rule is:
ΔVᵢ = αᵢβ(λ − ΣⱼVⱼ)
Where:
- ΔVᵢ represents the change in associative strength for stimulus i.
- αᵢ represents stimulus salience or associability.
- β represents a learning-rate parameter associated with the outcome.
- λ represents the maximum associative strength supported by the outcome.
- ΣⱼVⱼ represents the combined associative strength of stimuli present.
Learning is driven by the difference between an outcome’s supported associative strength and the value already predicted by the present stimuli.
This prediction-error structure has influenced broader theories of learning.
However, associative learning models explain only selected components of cognition.
They do not, by themselves, account for the full range of symbolic reasoning, complex social understanding, or cumulative cultural learning.
2.3. Spatial Cognition and Cognitive Maps
Many animals must remember where resources, threats, or important locations are situated.
Spatial cognition involves the representation and use of environmental relationships.
Examples include:
- Remembering food-storage locations.
- Navigating between familiar landmarks.
- Identifying alternative routes.
- Returning to nesting or shelter sites.
- Estimating direction and distance.
- Updating movement according to changing environmental conditions.
Tolman’s cognitive-map framework helped establish that navigation may involve representations extending beyond simple learned movement sequences [3].
A conceptual spatial representation may be expressed as a graph:
G = (V, E)
Where:
- V represents locations or environmental states.
- E represents navigable connections between them.
A planning system may evaluate routes according to distance, risk, energy expenditure, or other task-specific criteria.
This graph representation is an analytical analogy; actual biological navigation may involve distributed neural representations, path integration, sensory landmarks, and other mechanisms.
Not every successful navigation strategy requires an explicit map-like representation.
Experimental tests must therefore distinguish flexible spatial knowledge from alternative strategies such as memorized routes or local cue following.
2.4. Working Memory and Temporary Information Storage
Working memory allows organisms to maintain information temporarily while performing an activity.
It supports behaviours such as remembering locations, comparing alternatives, following sequences, and adjusting actions according to recent observations.
A simplified capacity-limited representation can be written as:
Mₜ = {x₁, x₂, …, xₖ}
Where Mₜ represents information currently available for a task and k denotes the number of represented elements in the chosen model.
This expression should not be interpreted as a universal biological memory-capacity equation.
The effective capacity of working memory depends on the task, stimulus characteristics, representation format, training, and measurement method.
Research by Inoue and Matsuzawa found that young chimpanzees demonstrated exceptional performance in a particular numerical working-memory paradigm, including conditions in which they outperformed tested human adults [13].
This finding does not establish that chimpanzees possess globally superior memory.
It demonstrates that particular species or individuals can excel in cognitive tasks requiring specialized combinations of attention, training, speed, and memory.
2.5. Problem-Solving and Cognitive Flexibility
Problem-solving involves selecting or discovering actions that transform a current situation into a desired outcome.
Flexible problem-solving is especially important when previously successful behaviour becomes ineffective.
A typical experimental sequence may involve:
- Presenting an unfamiliar problem.
- Establishing available actions.
- Observing initial responses.
- Recording successful and unsuccessful attempts.
- Modifying the problem conditions.
- Testing whether the organism adapts its strategy.
Researchers may distinguish between:
Trial-and-error learning: Behaviour improves through repeated attempts and feedback.
Insight-like performance: A solution appears through a relatively rapid reorganization of behaviour, although the underlying mechanism requires investigation.
Transfer learning: Knowledge acquired in one situation improves performance in another.
Reversal learning: An organism adjusts when previously rewarded responses become unrewarded.
Causal reasoning: Behaviour reflects sensitivity to relationships between causes and effects rather than only superficial associations.
Planning: Actions are organized according to anticipated future consequences.
A crucial methodological problem is that several mechanisms may produce similar outcomes.
An animal that opens an unfamiliar container might have inferred a causal mechanism, learned through exploration, or discovered a successful movement through reinforcement.
To distinguish these explanations, researchers must design tasks in which the competing hypotheses predict different patterns of behaviour.
2.6. Numerical Cognition Across Species
Numerical cognition refers to the ability to discriminate, estimate, represent, or manipulate quantities.
Evidence of numerical abilities has been reported across several animal groups.
Such capacities may include:
- Discriminating between sets of different sizes.
- Selecting larger or smaller quantities.
- Estimating approximate numerosity.
- Learning ordinal relationships.
- Recognizing patterns involving absence or zero-like quantities.
Human numerical cognition includes these foundations but also supports symbolic arithmetic, formal mathematics, and elaborate external numerical representations.
These capabilities should not be equated merely because both involve quantities.
For example, research on honey bees demonstrated numerical ordering behaviour involving empty sets following experimental training [14].
This result provides evidence of a specific numerical discrimination ability.
It does not establish that bees understand zero in all the ways humans use the mathematical concept.
The distinction between behavioural discrimination, conceptual representation, and symbolic mathematics is central to rigorous comparative analysis.
2.7. Tool Use and Physical Cognition
Tool use occurs when an organism manipulates an external object to achieve an effect.
Tool-related behaviour may require perception, motor control, learning, object discrimination, and sensitivity to physical relationships.
Examples occur in certain primates, birds, and other animals.
Research on corvids has highlighted their capacity for sophisticated tool-related behaviour and flexible physical problem-solving [9].
However, the mere presence of tool use should not automatically be interpreted as proof of human-like causal reasoning.
Researchers investigate whether an animal can:
- Select an appropriate tool.
- Modify a tool to meet task requirements.
- Generalize to unfamiliar tool configurations.
- Recognize functional properties rather than visual similarity alone.
- Adjust behaviour when the physical mechanism changes.
The evidence is strongest when experimental designs separate learned associations from alternative causal explanations.
2.8. Social Cognition and Theory of Mind
Social cognition concerns how organisms perceive and respond to other individuals.
It may involve recognition, relationship memory, dominance information, cooperation, competition, communication, and sensitivity to what another individual can perceive.
An especially debated question is whether nonhuman animals represent the mental states of others.
Premack and Woodruff’s influential 1978 paper introduced a major research discussion about whether chimpanzees possess a theory of mind [12].
In this context, theory of mind involves attributing states such as beliefs, intentions, perceptions, or knowledge to another individual.
For example, an animal may behave differently depending on whether a competitor can see a food source.
However, such behaviour does not necessarily establish that the animal explicitly represents the competitor’s beliefs.
Alternative explanations may involve sensitivity to gaze direction, past behaviour, environmental cues, or learned behavioural contingencies.
Researchers have therefore emphasized the need to distinguish sophisticated behavioural prediction from particular claims about internal mental-state representation [18].
The extent and character of theory-of-mind abilities across nonhuman species remain areas of scientific investigation.
2.9. Human Social Cognition and Cultural Learning
Human cognition is strongly shaped by interaction with other people.
Children learn not only through individual exploration but also through shared attention, imitation, instruction, language, and participation in cultural practices.
The cultural intelligence hypothesis proposes that distinctive human social-cognitive abilities contribute substantially to cultural learning and knowledge transmission.
Herrmann and colleagues compared aspects of physical and social cognition in young human children and great apes.
Their findings supported the importance of specialized human social-cognitive skills rather than an interpretation based solely on uniformly superior performance in every cognitive domain [7].
These results are influential but should not be interpreted as resolving every question about human cognitive uniqueness.
Performance depends on task construction, participant experience, measurement choices, and the specific abilities being tested.
2.10. Social Learning and Animal Culture
Social learning occurs when an organism’s behaviour is influenced by observing or interacting with others.
Possible mechanisms include:
Local enhancement: Attention is attracted to a location associated with another individual’s activity.
Stimulus enhancement: Observing another individual increases attention to a particular object.
Emulation: An observer learns about the outcome of another individual’s actions.
Imitation: An observer reproduces aspects of another individual’s behaviour.
Teaching: An experienced individual modifies its behaviour in ways that facilitate another individual’s learning, under an explicitly defined scientific criterion.
Cultural transmission involves the spread or maintenance of behaviours through social learning.
Research on chimpanzee populations documented substantial variation in behavioural traditions and helped establish the importance of nonhuman animal culture [8].
However, cultural processes differ in their complexity, stability, fidelity, and cumulative effects.
Human societies demonstrate especially extensive accumulation of technological and symbolic knowledge.
A distinction is therefore necessary between evidence of socially transmitted traditions and evidence of open-ended cumulative cultural development.
2.11. Measuring Cognitive Performance Experimentally
A scientifically informative intelligence test should measure the intended cognitive process while minimizing irrelevant influences.
A basic success metric is:
Success Rate = Successful Trials / Valid Trials
For reaction-time tasks:
Mean Reaction Time = Σᵢtᵢ / N
Where tᵢ is the response time for trial i and N is the number of valid trials.
For signal-detection experiments, sensitivity may be represented using:
d′ = Φ⁻¹(H) − Φ⁻¹(F)
Where:
- H is the hit rate.
- F is the false-alarm rate.
- Φ⁻¹ is the inverse standard-normal cumulative distribution function.
This measure can help distinguish sensitivity from response bias under the relevant signal-detection assumptions.
Extreme hit or false-alarm rates require an appropriate correction before applying the inverse transformation.
No single performance metric is adequate for every species or task.
Accuracy may be influenced by motivation, visual discrimination, motor demands, fear, reward preferences, and familiarity with the experimental apparatus.
Consequently, behavioural performance should be interpreted as evidence about a particular task under specified conditions rather than a direct measurement of an organism’s total intelligence.
2.12. Statistical Comparison Across Species
Cross-species intelligence studies must account for variation among individuals, tasks, and species.
For a binary success outcome, a simplified hierarchical logistic model may be written as:
logit[P(Yᵢⱼₖ = 1)] = β₀ + β_species(j) + β_task(k) + β_interaction(j,k) + uᵢ
Where:
- Yᵢⱼₖ indicates whether individual i of species j succeeds on task k.
- β₀ is an intercept.
- β_species(j) represents a species-related effect.
- β_task(k) represents task difficulty or task-related effects.
- β_interaction(j,k) represents species-by-task variation.
- uᵢ represents individual-level variation.
Additional variables may account for age, prior training, motivational conditions, test environment, and other measurable influences.
When comparing species, shared evolutionary ancestry may introduce statistical dependence.
Phylogenetic comparative methods can address this issue using evolutionary relationships among taxa.
Studies of animal self-control have employed large comparative datasets and phylogenetic approaches to investigate relationships among cognition, ecology, and brain characteristics [15].
However, such findings apply to the measured capacities and examined species rather than proving a universal ranking of intelligence.
2.13. Brain Organization and Cognitive Performance
Cognition depends on biological mechanisms, including neural structure, connectivity, sensory systems, learning processes, and bodily interaction with the environment.
Human cognition depends substantially on distributed neural systems supporting perception, memory, language, attention, executive control, and social processing.
Other animals possess nervous systems with evolutionary similarities and important differences.
Birds, for example, can exhibit sophisticated cognition despite neural organization that differs significantly from that of mammals.
Absolute brain size, relative brain size, neuron counts, connectivity, and regional specialization may each provide useful information under suitable conditions.
However, brain size alone is not a universal measure of intelligence.
Healy and Rowe identified important conceptual and methodological problems in interpreting comparative brain-size studies [17].
Different species vary in body size, sensory specialization, neural architecture, developmental history, and ecological demands.
A reliable account of cognitive evolution should therefore combine behavioural evidence with neurobiological, developmental, and evolutionary analysis.
2.14. Separating Capacity, Performance, and Motivation
One of the most important technical distinctions in comparative cognition concerns the difference between cognitive capacity and observed performance.
An organism may possess the necessary cognitive ability but fail an experimental task because it:
- Cannot adequately perceive the stimuli.
- Lacks the required motor capability.
- Does not understand the reward contingency.
- Is unfamiliar with the apparatus.
- Is stressed or distracted.
- Has insufficient motivation.
- Uses an alternative strategy not recognized by the scoring method.
Conversely, successful performance may result from simpler mechanisms than the experiment was intended to test.
A rigorous experimental design should therefore include appropriate control conditions and alternative explanations.
Failure on a task does not automatically prove absence of a cognitive ability, just as success does not automatically establish the most sophisticated possible cognitive explanation.
3. ARCHITECTURAL TEXT MAP
3.1. Biological Intelligence: Integrated Cognitive Architecture
The following text architecture presents a generalized biological information-processing framework.
It is a conceptual synthesis rather than a literal neural blueprint shared by every species.
LAYER 1 — ECOLOGICAL AND SOCIAL ENVIRONMENT
Physical conditions, resources, threats, social relationships, environmental uncertainty, and opportunities for action.
↓
LAYER 2 — SENSORY ACQUISITION
Vision, hearing, touch, smell, taste, proprioception, and other species-specific sensory systems.
↓
LAYER 3 — PERCEPTION AND ATTENTION
Stimulus discrimination, feature integration, relevance selection, and contextual interpretation.
↓
LAYER 4 — MEMORY AND LEARNING
Associative learning, spatial memory, working memory, recognition, experience-dependent adaptation, and retained information.
↓
LAYER 5 — COGNITIVE REPRESENTATION
Object relationships, spatial information, expected outcomes, quantities, social information, and task-relevant internal states.
↓
LAYER 6 — EVALUATION AND ACTION SELECTION
Biological needs, learned preferences, decision processes, behavioural flexibility, and possible future consequences.
↓
LAYER 7 — MOTOR AND COMMUNICATIVE EXECUTION
Movement, object manipulation, vocalization, signaling, social interaction, and environmental modification.
↓
LAYER 8 — ENVIRONMENTAL CONSEQUENCES
Changed physical conditions, rewards, costs, social responses, and new sensory information.
↓
LAYER 9 — FEEDBACK AND ADAPTATION
Learning updates, memory modification, behavioural adjustment, and future decision refinement.
3.2. Individual and Evolutionary Timescales
Biological intelligence operates across multiple timescales.
Immediate behavioural timescale
Sensory information influences perception, decision-making, and action.
↓
Learning timescale
Experience modifies behavioural strategies and internal representations.
↓
Developmental timescale
Maturation, experience, social interaction, and environmental conditions shape cognitive development.
↓
Cultural timescale
Socially transmitted behaviours and knowledge may persist or change across individuals and generations.
↓
Evolutionary timescale
Inherited variation, natural selection, genetic drift, and other evolutionary processes influence the distribution of biological traits across populations over generations.
These processes should not be conflated.
An animal learning a route during its lifetime is not the same process as a population evolving neural or sensory adaptations across many generations.
Similarly, social transmission is not equivalent to genetic inheritance.
3.3. Comparative Cognition Research Pipeline
A rigorous comparative study can be organized as follows.
RESEARCH QUESTION
Identify the cognitive ability or mechanism under investigation.
↓
HYPOTHESIS DEVELOPMENT
Specify competing explanations for the observed or predicted behaviour.
↓
SPECIES-APPROPRIATE TASK DESIGN
Adapt stimuli, apparatus, responses, and rewards to the organisms being studied.
↓
EXPERIMENTAL CONTROLS
Account for sensory limitations, motor demands, motivation, previous experience, and environmental effects.
↓
BEHAVIOURAL DATA COLLECTION
Measure accuracy, response times, learning trajectories, action sequences, and relevant contextual variables.
↓
STATISTICAL AND PHYLOGENETIC ANALYSIS
Estimate effects, uncertainty, individual variation, species differences, and evolutionary dependence where relevant.
↓
MECHANISTIC INTERPRETATION
Evaluate whether the observations support associative, representational, social, causal, or other explanations.
↓
REPLICATION AND GENERALIZATION
Test reliability across individuals, conditions, tasks, populations, and independent studies.
↓
SCIENTIFIC CONCLUSION
State supported findings, unresolved uncertainties, and limits on generalization.
3.4. The Relationship Between Biological and Artificial Intelligence
Biological cognition can inspire artificial intelligence without implying that biological organisms operate through the same mechanisms as digital computers.
A useful conceptual comparison is:
| Biological cognition | Artificial intelligence analogue |
|---|---|
| Sensory processing | Input acquisition and feature extraction |
| Attention | Relevance weighting or selective processing |
| Working memory | Temporary context or state representation |
| Long-term memory | Persistent knowledge or learned parameters |
| Associative learning | Experience-dependent parameter updates |
| Spatial cognition | Mapping, localization, and planning |
| Action selection | Policy or decision mechanism |
| Social learning | Learning from demonstrations or interactions |
| Feedback adaptation | Learning or control updates |
| Cognitive flexibility | Adaptation to changing tasks or conditions |
These correspondences are functional analogies.
They do not establish that artificial systems possess the same developmental histories, biological motivations, subjective experiences, or cognitive mechanisms as humans and other animals.
Biological intelligence research is most useful to AI when it supplies experimentally testable principles rather than unsupported claims of equivalence.
4. USE-CASES & LIMITATIONS
4.1. Artificial Intelligence and Cognitive Architecture Design
Research on human and animal intelligence has influenced approaches to machine learning, reinforcement learning, perception, navigation, memory, and agent design.
Biological systems demonstrate how behaviour can be organized through interactions among information processing, learning, environmental feedback, and action.
For example, associative learning research helped motivate computational models in which prediction errors influence updates.
Spatial cognition research has inspired investigation of environmental representations and navigation strategies.
Comparative studies of flexible behaviour can also inform AI evaluations that move beyond fixed benchmarks toward unfamiliar environments and changing task conditions.
Limitations: Biological inspiration does not automatically produce effective artificial algorithms. Functional resemblance does not establish shared mechanisms, and experimentally successful biological strategies may be difficult to reproduce in engineered systems.
4.2. Developmental Robotics and Embodied AI
Developmental robotics investigates how intelligent behaviour can emerge through interaction among learning mechanisms, physical embodiment, and environmental experience.
Human infants and young animals provide important examples of learning through exploration, manipulation, movement, and social interaction.
Researchers may investigate how artificial agents acquire sensorimotor coordination, object knowledge, action prediction, and adaptive control.
Limitations: Robots differ from biological organisms in their sensory systems, bodies, learning environments, motivations, and computational resources.
A robot successfully reproducing a behavioural pattern does not necessarily reproduce the biological cognitive mechanism responsible for that pattern.
4.3. Comparative Intelligence Benchmarks
Animal cognition research provides methodological lessons for evaluating artificial intelligence.
A benchmark should distinguish actual task competence from familiarity with particular stimuli or procedures.
It should also examine whether performance transfers to new settings.
For example, an artificial agent may excel at recognizing previously encountered visual patterns while failing when the relevant relationship is presented in an unfamiliar format.
Similarly, an animal may fail a laboratory task because the apparatus imposes unnatural sensory or motor demands.
Both cases demonstrate why evaluation conditions must be considered before drawing conclusions about generality.
Limitations: Benchmark results may overstate or understate competence because of task selection, training overlap, scoring assumptions, and environmental mismatch.
No finite test battery provides a complete measure of all possible intelligence.
4.4. Animal Welfare and Enrichment
Understanding animal cognition can improve the design of environments for animals under human care.
Cognitive research may inform enrichment activities, training methods, opportunities for exploration, and appropriate environmental complexity.
Knowledge of species-specific perception and learning can also help reduce confusion and unnecessary stress during human–animal interactions.
For example, training methods can be adapted to an animal’s sensory capabilities and learned associations.
Limitations: Cognitive performance should not be used as a simple ranking of moral worth.
A species’ welfare interests do not depend solely on its ability to solve human-designed problems.
Scientific uncertainty about subjective experience should also be distinguished from evidence concerning observable health, behaviour, stress, and welfare requirements.
4.5. Conservation and Wildlife Management
Animal cognition research can inform conservation strategies by improving understanding of how organisms respond to habitat changes, environmental threats, and human activity.
Relevant areas include navigation, migration, foraging, social learning, and responses to unfamiliar environments.
For example, understanding how animals learn routes or recognize environmental cues may assist habitat planning and conservation interventions.
Limitations: Behaviour observed under controlled conditions may differ from behaviour in natural ecosystems.
Interventions may also produce unintended consequences when they modify learned responses or established ecological relationships.
Conservation applications require field validation and attention to species-specific conditions.
4.6. Human Education and Learning Science
Comparative cognition contributes to the scientific study of learning mechanisms.
Research involving memory, reinforcement, attention, categorization, and behavioural flexibility can provide foundational insights relevant to educational psychology.
However, applying findings from animal experiments to human instruction requires caution.
Human learning is influenced by language, culture, social relationships, prior knowledge, motivation, and formal symbolic representations.
Limitations: An experimentally demonstrated learning mechanism in one species does not establish that the same instructional intervention will be effective for human learners.
Educational applications require direct testing in relevant human populations.
4.7. Human–Animal Communication and Interaction
Research on animal perception, learning, and social cognition can improve interactions between humans and domesticated or managed animals.
Understanding how animals respond to signals, gestures, routines, and reinforcement may improve training and communication.
Dogs, for example, provide an important model for investigating how domestication and experience can influence responsiveness to human social cues.
Limitations: Apparent understanding may reflect learned associations rather than human-like interpretation of language or intention.
Human observers may also misinterpret animal behaviour through anthropomorphic assumptions.
Effective communication strategies must be evaluated through controlled behavioural evidence.
4.8. Species-Specific Intelligence and the Problem of Universal Ranking
A fundamental limitation of comparative intelligence research is that species differ in their evolved specializations.
A human-designed task may favour human sensory abilities, movement patterns, social expectations, or symbolic knowledge.
A species with superior performance in one domain may perform less effectively in another.
For example, strong numerical working-memory performance in a particular chimpanzee experiment does not demonstrate overall intellectual superiority over humans [13].
Likewise, a human’s exceptional symbolic language abilities do not imply superior performance in every navigation, perceptual, or ecological task.
Attempts to rank all living species along a single intellectual ladder are inconsistent with modern evolutionary approaches.
Hodos and Campbell provided an influential critique of the historical scala naturae conception, emphasizing that evolutionary relationships cannot be reduced to a simple hierarchy of living organisms [19].
A scientifically meaningful comparison must specify the domain, task, environment, population, and measurement assumptions involved.
4.9. Anthropomorphism and Anthropodenial
Two opposing interpretive errors can distort animal cognition research.
Anthropomorphism occurs when human-like psychological properties are attributed to other animals without adequate evidence.
An observer might conclude that an animal possesses complex moral reasoning merely because its behaviour resembles a human moral action.
Anthropodenial describes the opposite tendency to reject relevant similarities between humans and other animals without adequate justification.
For example, a researcher might dismiss evidence of animal problem-solving solely because the organism lacks human language.
Both errors can undermine scientific interpretation.
The appropriate method is to evaluate competing mechanisms, seek discriminating evidence, and avoid making claims stronger than the observations support.
4.10. The Difficulty of Inferring Internal Mental States
Behaviour is observable, but internal cognitive representations are generally inferred indirectly.
Two organisms may perform the same action through different mechanisms.
A successful social prediction may reflect mental-state reasoning, learned behavioural patterns, or sensitivity to environmental cues.
Research concerning chimpanzee theory of mind illustrates this problem.
Premack and Woodruff proposed influential questions about mental-state attribution [12].
Subsequent arguments have emphasized the possibility of alternative explanations for apparently mentalistic behaviour [18].
The evidence must therefore distinguish what an organism demonstrably does from what researchers hypothesize about how it accomplishes the task.
The strength of a mechanistic claim depends on the ability of experiments to differentiate competing hypotheses.
4.11. Brain Size as an Incomplete Intelligence Indicator
Brain size can provide useful evolutionary information, but simplistic comparisons may be misleading.
Absolute volume, relative volume, neuron counts, neural organization, and ecological specialization can vary substantially across taxa.
Moreover, cognitive functions emerge from interacting neural systems rather than from total brain size alone.
A large comparative study found associations between absolute brain volume and specific measures of behavioural self-control across the species examined [15].
However, such associations do not demonstrate that brain volume universally predicts every cognitive capacity.
Methodological critiques emphasize the need for careful interpretation and direct behavioural evidence [17].
No single anatomical measurement should be treated as a universally valid intelligence score.
4.12. Experimental Bias and Ecological Validity
Laboratory studies offer important control over experimental variables, but they can also create artificial demands.
An experiment may rely on visual information even when the tested species depends heavily on another sensory modality.
An apparatus may require unfamiliar manipulation.
A reward may be insufficiently motivating.
Captive experience may influence performance differently from experience in natural settings.
Important methodological safeguards include:
- Species-appropriate stimulus design.
- Control of sensory and motor demands.
- Evaluation of individual motivation.
- Documentation of prior training.
- Appropriate sample sizes.
- Replication across environments.
- Transparent reporting of unsuccessful trials.
- Consideration of developmental and social history.
Ecological validity concerns whether an experimental finding meaningfully relates to behaviour in relevant natural or practical circumstances.
A task can be internally well controlled while still measuring a limited or unfamiliar aspect of an animal’s cognitive abilities.
4.13. Evolutionary and Phylogenetic Confounding
Species do not represent statistically independent observations when they share evolutionary ancestry.
Closely related species may resemble one another because of inherited characteristics rather than independent adaptations to similar environments.
Failing to account for phylogeny can distort relationships between cognitive performance and ecological variables.
For example, an observed association between social group size and task performance might partly reflect shared evolutionary history.
Phylogenetic comparative methods can help address such dependence.
However, these methods also depend on the quality of evolutionary trees, trait measurements, model assumptions, and available species samples.
Evolutionary explanation should integrate multiple sources of evidence rather than relying on a single statistical association.
4.14. Animal Cognition and Artificial Consciousness Claims
Comparisons between biological and artificial intelligence sometimes lead to claims about consciousness.
Such claims require particular caution.
Biological intelligence research demonstrates that many nonhuman animals possess complex perceptual, learning, and behavioural capacities.
However, intelligence and consciousness are conceptually distinct.
A successful behavioural test does not, by itself, establish the existence or character of subjective experience.
Likewise, the ability of an artificial system to generate language, solve problems, or imitate intelligent behaviour does not automatically establish consciousness.
Scientific investigation of consciousness requires additional theoretical and empirical considerations.
The relationship between biological cognition, subjective experience, and machine consciousness remains an open interdisciplinary research problem.
4.15. Generalization and the Limits of Cross-Species Transfer
Researchers frequently investigate whether findings from one species can inform understanding of another.
Such comparisons can be valuable when organisms share relevant biological mechanisms or face similar functional challenges.
However, cross-species generalization is limited by evolutionary divergence, developmental differences, sensory systems, environmental conditions, and distinct cognitive specializations.
A result obtained from rodents may illuminate aspects of learning while failing to explain human symbolic language.
A result obtained from great apes may reveal aspects of social cognition without establishing the full developmental mechanisms underlying human cooperation.
A computational model inspired by a bird’s navigation system may be useful for engineering without reproducing the bird’s actual neural processing.
Generalization should therefore be supported by mechanistic evidence rather than assumed from superficial similarity.
4.16. The Relationship Between Intelligence and Adaptive Success
Intelligence can contribute to adaptation, but evolutionary success is not determined by intelligence alone.
Organisms survive and reproduce through combinations of morphological, physiological, behavioural, ecological, and reproductive traits.
Highly specialized behaviour may be effective in a stable ecological niche.
Broad cognitive flexibility may be advantageous when conditions change, but it can also involve energetic, developmental, or other costs.
Natural selection does not necessarily maximize generalized intelligence.
It acts on heritable variation in relation to reproductive consequences within particular environments.
Thus, evolutionary fitness and intelligence are related concepts but not interchangeable measures.
A cognitively flexible organism is not necessarily more evolutionarily successful than a specialized organism under every set of conditions.
4.17. Lessons for Evaluating Artificial General Intelligence
Comparative cognition offers several important lessons for evaluating claims about artificial general intelligence.
Intelligence is multidimensional. Strong performance in one domain does not establish broad competence.
Environment matters. Evaluation results depend on the conditions under which tasks are performed.
Transfer is essential. Flexible adaptation to unfamiliar problems provides different evidence from success on familiar benchmarks.
Mechanisms matter. Similar outputs can result from different information-processing processes.
Social abilities are significant. Communication, cooperation, learning from others, and shared knowledge contribute substantially to human cognitive achievement.
Embodiment can shape cognition. Sensory and motor capabilities influence what information is available and which actions are possible.
Measurement must be valid. A benchmark should test the intended ability without allowing irrelevant task demands to dominate the result.
Consciousness is a separate question. General-purpose performance does not automatically establish subjective experience.
These principles caution against treating a single benchmark score, model size, or impressive demonstration as conclusive evidence of general intelligence.
4.18. Scientific Principles for Responsible Comparative Intelligence Research
A robust research framework should satisfy several interconnected requirements.
Operational clarity: Define the cognitive capacity under investigation.
Mechanistic precision: Distinguish behavioural outcomes from hypotheses about internal processing.
Ecological relevance: Consider the organism’s natural environment and biological capabilities.
Experimental validity: Control for sensory, motivational, developmental, and motor confounds.
Statistical rigor: Account for individual variability, uncertainty, sample structure, and phylogenetic relationships.
Interpretive restraint: Avoid claims that exceed the available evidence.
Ethical responsibility: Design research in accordance with applicable animal-welfare and human-research requirements.
Reproducibility: Report methods and findings sufficiently clearly to support independent scrutiny.
The central conclusion is that human and animal intelligence are best understood as evolutionarily situated, biologically implemented, and environmentally expressed capacities for learning, representation, problem-solving, and adaptive behaviour.
Humans possess an exceptional combination of symbolic, social, and cumulative cultural capacities. Other animals demonstrate diverse and sometimes highly specialized cognitive achievements that cannot be adequately captured by human-centered tests alone.
Together, these findings provide a scientific foundation for understanding natural intelligence and a critical framework for designing and evaluating artificial intelligence systems.
5. RELATED ENCYCLOPEDIA TOPICS
Collective and Social Intelligence
Machine and Hybrid Intelligence
Embodied and Situated Intelligence
6. REFERENCES
- Darwin, C. (1871). The Descent of Man, and Selection in Relation to Sex. John Murray, London. Original work published in 1871; linked digital text represents a later edition.
- Thorndike, E. L. (1898). Animal Intelligence: An Experimental Study of the Associative Processes in Animals. The Psychological Review: Monograph Supplements, 2(4), i–109. DOI: 10.1037/h0092987.
- Tolman, E. C. (1948). Cognitive Maps in Rats and Men. Psychological Review, 55(4), 189–208. DOI: 10.1037/h0061626.
- Tinbergen, N. (1963). On Aims and Methods of Ethology. Zeitschrift für Tierpsychologie, 20(4), 410–433. DOI: 10.1111/j.1439-0310.1963.tb01161.x.
- Spearman, C. (1904). “General Intelligence,” Objectively Determined and Measured. The American Journal of Psychology, 15(2), 201–292. DOI: 10.2307/1412107.
- Shettleworth, S. J. (2010). Cognition, Evolution, and Behavior. Second Edition. Oxford University Press. DOI: 10.1093/oso/9780195319842.001.0001.
- Herrmann, E., Call, J., Hernández-Lloreda, M. V., Hare, B., & Tomasello, M. (2007). Humans Have Evolved Specialized Skills of Social Cognition: The Cultural Intelligence Hypothesis. Science, 317(5843), 1360–1366. DOI: 10.1126/science.1146282.
- 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. DOI: 10.1038/21415.
- Emery, N. J., & Clayton, N. S. (2004). The Mentality of Crows: Convergent Evolution of Intelligence in Corvids and Apes. Science, 306(5703), 1903–1907. DOI: 10.1126/science.1098410.
- Gallup, G. G., Jr. (1970). Chimpanzees: Self-Recognition. Science, 167(3914), 86–87. DOI: 10.1126/science.167.3914.86.
- Premack, D., & Woodruff, G. (1978). Does the Chimpanzee Have a Theory of Mind?. Behavioral and Brain Sciences, 1(4), 515–526. DOI: 10.1017/S0140525X00076512.
- Inoue, S., & Matsuzawa, T. (2007). Working Memory of Numerals in Chimpanzees. Current Biology, 17(23), R1004–R1005. DOI: 10.1016/j.cub.2007.10.027.
- Howard, S. R., Avarguès-Weber, A., Garcia, J. E., Greentree, A. D., & Dyer, A. G. (2018). Numerical Ordering of Zero in Honey Bees. Science, 360(6393), 1124–1126. DOI: 10.1126/science.aar4975.
- MacLean, E. L., et al. (2014). The Evolution of Self-Control. Proceedings of the National Academy of Sciences, 111(20), E2140–E2148. DOI: 10.1073/pnas.1323533111.
- Rescorla, R. A., & Wagner, A. R. (1972). A Theory of Pavlovian Conditioning: Variations in the Effectiveness of Reinforcement and Nonreinforcement. In A. H. Black & W. F. Prokasy (Eds.), Classical Conditioning II: Current Theory and Research, pp. 64–99. Appleton-Century-Crofts.
- Healy, S. D., & Rowe, C. (2007). A Critique of Comparative Studies of Brain Size. Proceedings of the Royal Society B: Biological Sciences, 274(1609), 453–464. DOI: 10.1098/rspb.2006.3748.
- Povinelli, D. J., Bering, J. M., & Giambrone, S. (2000). Toward a Science of Other Minds: Escaping the Argument by Analogy. Cognitive Science, 24(3), 509–541. DOI: 10.1207/s15516709cog2403_7.
- Hodos, W., & Campbell, C. B. G. (1969). Scala Naturae: Why There Is No Theory in Comparative Psychology. Psychological Review, 76(4), 337–350. DOI: 10.1037/h0027523.