Animal Models in Neuroscience: What They Reveal About Human Behavior

4 days ago

18 min read

Author: Ukrainian Psychological Hub · Published: October 12, 2026 · Editorial Policy

Animal models give behavioral and cognitive neuroscientists a way to test how living nervous systems generate measurable actions, adapt to experience, and respond to changes in genes, cells, circuits, and environments. They support experiments that cannot always be performed in humans, including selective neural interventions combined with precise behavioral measurement. Their explanatory value depends on what is modeled, what is measured, and how a proposed mechanism is connected to human evidence.

A mouse learning a spatial task can illuminate principles of navigation and memory-dependent decisions. A fruit fly may reveal cellular mechanisms underlying learning. A macaque performing a visual discrimination may help investigators compare perceptual circuits across primates. None of these experiments establishes that the full psychological life of the studied organism is identical to human experience. The modern scientific task is to establish specific, testable correspondences while explicitly accounting for differences (Kanwisher, 2025; Bailey et al., 2023).

This article examines definitions, species choice, experimental technologies, causal inference, validity, translational outcomes, ethical standards, emerging alternatives, and the correct interpretation of findings about human behavior.

What is an animal model in neuroscience?

An animal model is a nonhuman organism studied under defined conditions to examine a particular biological, neural, or behavioral phenomenon. It might exhibit a naturally occurring characteristic, carry an engineered genetic variation, receive a carefully controlled intervention, or perform a task that allows researchers to test a proposed mechanism.

The organism itself and the scientific model should be distinguished conceptually. A laboratory mouse is an animal. A defined strain, genetic background, developmental stage, task, intervention, and hypothesis together constitute a research model. Validity concerns the inference that investigators wish to make from this system to their target question.

Mechanistic models focus on molecular or circuit processes; behavioral models examine measurable patterns of action and learning; disease-relevant models reproduce selected features thought to matter for a human condition. A model that represents a signaling pathway can be scientifically informative even when it reproduces few features of a clinical syndrome. Conversely, superficial behavioral resemblance may provide little evidence about the underlying human mechanism.

An assay is different from a model. A behavioral assay is a procedure for obtaining measurements within the experimental system. Investigators might employ open-field exploration, object recognition, a choice task, or a memory maze in a particular mouse model. No single assay automatically measures an entire human psychological construct. The interpretation of a measurement must be validated for its intended use (van der Staay et al., 2009).

Why scientists use animals to study behavior

Animal research is possible because many neural and cellular processes share evolutionary histories. Excitable membranes, synaptic signaling, developmental mechanisms, and forms of plasticity recur across lineages. Comparative investigation identifies which features are conserved, which have changed, and which are specific to a species or context.

Experimental accessibility is another advantage. Researchers may record individual neurons while tracking behavior, manipulate a targeted projection, and subsequently investigate molecular properties of sampled tissue. Experimental control can make a narrow causal hypothesis easier to evaluate than an observational relationship in humans.

Human investigation remains crucial. Functional imaging, electrophysiology, neuropsychology, clinical observations, developmental studies, and behavioral experiments reveal which proposed mechanisms correspond to human brain function. In some systems, such as parts of primate visual processing, comparative work can connect cellular-level animal experiments with established human cortical organization (Kanwisher, 2025).

The aim of studying an animal often differs from the aim of developing a therapy. Basic research may establish a cellular mechanism without testing a disease treatment. A preclinical therapeutic experiment requires additional evidence about target relevance, dose, toxicity, and the likelihood that a predicted outcome applies to people.

How animal research developed

Comparative psychology and neurophysiology historically examined reflexes, sensory function, conditioning, and adaptive behavior. Learning experiments established ways to connect environmental contingencies with observable responses. Lesion research and electrophysiology then helped link neural systems with particular behavioral capacities.

Molecular genetics made it possible to alter selected genes and study developmental or signaling consequences. Population recording and optical measurements expanded the scale of accessible neural activity. Modern optogenetics made it possible to change the activity of targeted cells with light, supporting temporally precise interventions in functioning circuits (Deisseroth, 2015).

As experimental methods became more precise, good questions became more specific. Investigators can now ask whether a defined projection contributes to action selection during a particular task phase. Even powerful selective interventions must be interpreted through the observed behavior, the exact manipulated cells, and the physiological conditions of the experiment.

Which animal species are used in neuroscience?

Mice and rats

Rodents are commonly used because their genetics, physiology, breeding, and experimental methods are extensively studied. Mice have an unusually broad range of genetic and circuit-targeting resources. Rats are useful for many trained behavioral paradigms and selected physiological approaches. Both can contribute to research on movement, sensory processing, learning, sleep, stress physiology, social interactions, and motivated action.

The scientific value of a rodent model depends on strain, sex, age, developmental background, housing, task familiarity, and sensory capabilities. The same nominal task can produce different results when these conditions change. Behavioral neuroscience methodologists emphasize consistent procedures, meaningful controls, and rigorous statistical analysis (Sukoff Rizzo and Crawley, 2017).

Fruit flies and nematodes

Drosophila melanogaster supports powerful genetic work on development, circadian biology, cellular signaling, and relatively accessible circuits. Caenorhabditis elegans offers particularly tractable investigations of identified neurons and sensory-motor connections. These systems help researchers test cellular principles with high precision.

Their strengths do not erase differences from human brains in organization, body systems, social behavior, and ecological experience. Molecular conservation can generate a plausible hypothesis for mammalian studies, but complex human psychological claims require separate lines of evidence.

Zebrafish

Zebrafish are important in developmental neuroscience and in studies of sensory processing, movement, and neural circuitry. Their early developmental stages allow optical access to living neural systems. Interpretation depends strongly on developmental stage and ecological context; findings in a larval system have a particular scope.

Nonhuman primates

Macaques and other nonhuman primates are employed for selected questions about visual perception, eye movements, action, decision-making, and cortical systems. Comparative primate experiments can be valuable when anatomy and task structure offer meaningful correspondence with human cognition.

Evolutionary proximity is informative but is not an all-purpose guarantee of human equivalence. Neural organization, social ecology, development, language, and task strategies may differ. Ethical assessment is especially demanding in research with nonhuman primates, and scientific justification must address the distinct necessity of the species (Kanwisher, 2025).

How researchers choose a model organism

Species selection follows the research question, the conserved mechanism under investigation, the necessary experimental intervention, and the demands of welfare oversight. A small nervous system may provide exceptional access to a general cellular principle, while a primate preparation may better address a specialized cortical computation. No species is universally superior across all neuroscientific questions.

What do behavioral animal studies measure?

Animal neuroscience combines behavioral outcomes with physiological, anatomical, genetic, and molecular measurements. A useful study distinguishes what an instrument records from the psychological interpretation attached to its results.

Behavioral tests

Measurements include choice, response time, location, movement, freezing, approach, withdrawal, vocalization, sensory discrimination, object exploration, social proximity, and other species-appropriate actions. The recorded outcome may support a hypothesis about learning, threat responses, reward, or attention, but the test name is not sufficient proof of the psychological explanation.

For instance, reduced exploration can reflect movement limitations, sensory differences, arousal, novelty, fatigue, or competing motivation. A task described as an anxiety test may be influenced by these alternatives. Convergent behavioral assays and independent motor, sensory, and motivational controls improve confidence (Sukoff Rizzo and Crawley, 2017).

Electrophysiology

Electrodes measure electrical activity at selected spatial and temporal scales. Spike recordings can relate firing to choices or task events; local field potentials capture electrical activity generated by populations and their surroundings. Neural activity associated with a behavior identifies a relationship and may guide a causal experiment. Correlation alone cannot show that a recorded cell population produces an entire psychological function.

Calcium imaging

Fluorescent indicators permit observation of activity-related intracellular calcium dynamics, sometimes across large neural populations. Fluorescence is a biological proxy shaped by indicator kinetics, recording methods, and analysis. It does not equal a direct enumeration of every action potential, and its interpretation benefits from electrophysiological or other validation.

Lesion and pharmacological approaches

Lesions disrupt tissue or pathways, which may reveal necessary contributions under particular conditions. Damage can also interrupt networks, evoke compensation, or affect multiple functions. Drugs can alter neural signaling, but receptor selectivity, dose, distribution, metabolism, and off-target effects all limit interpretations about which pathway is responsible.

Chemogenetics and optogenetics

Chemogenetics alters targeted cells through engineered receptors and selected chemical ligands; pharmacokinetics and unintended effects require controls. Optogenetics uses light-sensitive proteins to change the activity of targeted cells or pathways with experimentally specified timing. It offers strong causal access, especially when neural manipulation is aligned with specific task events (Deisseroth, 2015).

Optical stimulation may differ greatly from physiological activity. Targeted cells exist within interacting networks, and activating them can indirectly affect downstream populations. A demonstrated intervention effect supports a conclusion about that intervention and observed outcome, not the claim that an isolated region literally contains an entire human emotion.

Molecular and anatomical profiling

Histology, microscopy, receptor mapping, sequencing, and cell-type analysis characterize neural components and their interactions. Cross-species atlases can identify homologous cell populations and divergent features. A recent comparison of primate basal ganglia combined molecular and spatial data to investigate conserved and species-distinct cell types (cross-species basal ganglia atlas, 2025).

Shared genetic markers or anatomical features make correspondence plausible; functional similarity still requires behavioral and physiological validation.

How animal experiments establish causal relationships

Strong causal studies state a testable hypothesis about a defined neural process and observable outcome. An experiment might ask whether temporarily suppressing a specific neural projection changes an animal's learned response to a cue. Investigators characterize baseline behavior, assign comparison conditions, manipulate the pathway, and test the predicted change.

Controls address confounding effects from movement, sensory function, handling, illumination, surgical intervention, viral delivery, vehicle administration, and experimental timing. Randomization, blinded assessment, appropriate experimental-unit definitions, and prespecified statistical analysis reduce biased conclusions.

Necessity and sufficiency answer different questions. A process may be necessary under one condition if disabling it disrupts a response. Experimental activation may be sufficient to produce an outcome in another condition. Neither observation establishes exclusive control, nor does it guarantee that artificial activation mirrors a naturally occurring mental state.

Causal findings are also conditional. A circuit's effect can depend on prior training, motivation, developmental stage, task constraints, or biological state. Studies across settings and convergent interventions help identify which mechanisms generalize.

Face validity, construct validity, and predictive validity

Scientific validity is specific to what a model is intended to explain. Three familiar categories are useful, although methodological authors emphasize the need for broader assessments of reliability and generalizability (van der Staay et al., 2009; Bailey et al., 2023).

Face validity

Face validity concerns resemblance between an observable animal feature and the target phenomenon. Similar behavior can suggest a useful experiment, but resemblance can arise through different mechanisms. A rodent avoiding a novel environment and a human avoiding a social situation may show similarly described actions without exhibiting identical causes or experiences.

Construct validity

Construct validity asks whether a model captures the relevant underlying process. A genetic alteration that changes a conserved synaptic pathway may have strong relevance to studying that pathway even if the animal displays only one element associated with a human condition. Construct validity is strengthened through independent human genetic, physiological, and comparative data.

Predictive validity

Predictive validity concerns whether results in the model forecast outcomes in the target domain. In drug research, this includes sensitivity and specificity in predicting clinically relevant intervention effects. Altering an animal test score does not establish human therapeutic efficacy, and the ability to predict one treatment response does not automatically generalize to unrelated treatments.

Internal validity and external validity

Internal validity concerns whether observed differences are caused by the experimental manipulation rather than confounders. Reliability concerns the consistency of measurements or effects. Replication asks whether findings survive fresh investigation. External validity addresses whether they apply across populations, environments, laboratories, species, and target contexts.

These dimensions can diverge. A mouse experiment can provide strong internal causal evidence while having limited predictive relevance for human clinical outcomes. The appropriate scientific conclusion preserves both facts.

Discoveries supported by animal neuroscience

Learning and memory

Animal tasks combined with neural measurements have revealed principles of synaptic plasticity, learning-dependent change, navigation, contextual discrimination, and interactions among hippocampal and cortical systems. Researchers can test when a circuit contributes to acquisition, retrieval, or strategy selection. The processes and task conditions should be specified rather than treated as a complete account of human autobiographical memory.

Threat learning and defensive action

Conditioning experiments examine how neutral cues acquire predictive meaning through adverse outcomes, and how defensive responses change through learning or extinction. Neural circuits contribute to particular aspects of these processes. Freezing, avoidance, and physiological arousal are distinct dependent measures. Interpreting them as direct and complete measurements of human subjective fear exceeds the experimental evidence.

Reward learning and action selection

Behavioral and electrophysiological experiments help clarify how organisms learn from outcomes, estimate expectations, allocate effort, and select actions. Dopamine pathways participate in multiple forms of processing, including movement and reinforcement-related learning. A change in lever pressing, approach, or a neuronal signal is not a direct measure of human pleasure.

Sensory and motor circuits

Model organisms permit fine-grained investigation of sensory transformations, reflex circuits, motor programs, and circuit timing. Comparisons can reveal conserved organizational principles while also identifying species-specific sensory priorities. Strong interpretations identify which mechanisms were measured and which remain hypothetical for humans.

Social behavior

Investigators study species-specific communication, social proximity, parental responses, social learning, and interaction patterns. These behaviors are shaped by ecology, experience, and biology. A change in rodent social approach does not demonstrate a human change in empathy, relationships, moral reasoning, or self-concept. Complex human claims require human studies.

Modeling human psychiatric and neurological conditions

Disease-relevant animals can reproduce selected genetic variants, molecular pathways, neurological injuries, physiological features, or behavioral consequences associated with human disease. These systems help investigate mechanisms and guide hypotheses about therapeutic targets.

Psychiatric syndromes pose special challenges because human diagnosis incorporates subjective reports, functioning, duration, context, and clinical evaluation. Animals cannot report human-style hallucinations, suicidal ideation, personal guilt, or autobiographical meaning. A rodent showing reduced interest in a laboratory reward is providing a measurement relevant to reward-directed behavior, rather than receiving a diagnosis of major depressive disorder.

The iTRIPP recommendations emphasize explicit translational assumptions, relevant patient groups, careful assay selection, and attention to differences between species. These principles help prevent researchers from equating behavioral measures with full DSM or ICD conditions.

Models involving chronic stress, neurodevelopmental variations, neurodegeneration, or pharmacological challenges can illuminate particular components. Findings remain dependent on the construct and protocol. A survey of chronic mild stress research, for instance, investigated reproducibility and procedure-related sources of variation rather than demonstrating that one animal preparation represents the total human condition (Willner, 2017).

Similarly, an observed preclinical biomarker or circuit change is a research association or mechanism. It is not automatically a validated diagnostic test, a clinical prognosis, or a recommended treatment for an individual person.

Do animal findings translate to humans?

Translation has several meanings: a conserved cellular mechanism, a similar computation across species, reproduction of an observation in humans, or the development of an effective therapy. These stages require different evidence.

A major umbrella review examined 122 systematic and related reviews, covering 54 diseases or conditions and 367 therapeutic interventions. Among the analyzed interventions, approximately 50% advanced to human studies, 40% to randomized trials, and 5% reached regulatory approval under the specified follow-up criteria (Ineichen et al., 2024).

These estimates refer to an eligible selection of therapeutic research across biomedical fields. They are not a universal measure of the scientific value of animal neuroscience. The review also examined concordance between animal and human positive findings in an eligible subset. Concordance is a distinct statistical quantity from regulatory approval and cannot be treated as a simple prediction of clinical success for any new animal experiment.

A mechanistic observation can be valuable even if it never becomes a medicine. Conversely, a promising preclinical intervention may fail during clinical investigation because of toxicity, dosage, human heterogeneity, treatment context, or problems with the relevance of the model. Human clinical efficacy ultimately requires evidence from human clinical research.

Why models sometimes fail to predict human outcomes

Species differences

Evolution can preserve cell signaling while changing gene regulation, receptor expression, anatomical organization, immune biology, development, and behavioral ecology. Apparent anatomical correspondence is not proof of identical functions. Detailed comparative genomics and cell atlases help identify shared and divergent biology (cross-species basal ganglia atlas, 2025).

Broad interpretations of narrow tasks

Single behaviors often admit several explanations. Reduced exploration might represent movement limitations, sensory disruption, fatigue, novelty-related responses, or changes in motivation. Attributing every outcome to one human emotion or disorder creates an unjustified leap between the measurement and the theory.

Laboratory environment and handling

Housing, enrichment, lighting, testing history, dominance relationships, scent, and researcher behavior can influence results. Experimental evidence shows that exposure to male-related olfactory cues can produce stress-related analgesia in rodents, changing apparent baseline pain responses (Sorge et al., 2014). Details sometimes dismissed as incidental may therefore be scientifically important.

Excessively narrow standardization may improve uniformity while reducing the range of environments to which conclusions apply. Replication across laboratories, appropriate variation, and transparent protocols can help reveal contextual dependence.

Bias, incomplete reporting, and experimental design

Inadequate randomization, lack of blinding, unplanned exclusions, small effective sample sizes, and selective publication can distort findings. Researchers should define the actual experimental unit: animals sharing a cage or litter may not provide independent observations in every analysis.

In the animal stroke literature, an analysis of systematic review data found that publication bias could inflate apparent intervention efficacy. This is documented evidence for that research area rather than a numerical estimate applicable to every model (Sena et al., 2010).

The ARRIVE 2.0 guidelines prioritize study design, sample size, randomization, blinding, outcome specification, statistical methods, animal characteristics, procedures, and complete results. Compliance improves transparency and appraisal, although transparent reporting alone cannot rescue a poorly justified construct.

Difference between experimental improvement and clinical benefit

A treatment can modify a narrowly defined pathway in an animal while failing to improve patients' functioning or symptoms. Human conditions often involve multiple systems and social or developmental influences. Therapies must address clinically meaningful outcomes under real human conditions.

How scientists improve animal-to-human translation

The first step is to define the target question precisely, including the proposed human mechanism and the measurable process in the selected species. Researchers should explain why that species, task, and intervention are appropriate, and state what outcomes would count against the hypothesis.

Comparative behavioral tasks can test whether species share computations rather than merely superficially similar behaviors. Computational models can offer quantitative cross-species comparisons, but the model's assumptions and explanatory validity require independent testing (computational validity review, 2022).

Human genetics, clinical observations, molecular atlases, postmortem tissue, and patient-derived cellular systems may inform the model's biological correspondence. Independent replication, multilateral testing, relevant age and sex variation, and careful measurement of confounders can strengthen generalizability.

Preregistration where feasible, justified sample sizes, masked assessments, complete reporting, and publication of negative findings reduce avoidable bias. Transparent evidence synthesis can help decide which interventions merit human study. ARRIVE 2.0 and iTRIPP provide complementary, practical guidance.

No single species or method settles the mechanism of a complex human behavior. A stronger conclusion arises when predictions hold across independent experimental approaches and are supported by directly relevant human data.

Ethics and the three Rs

Animal neuroscience uses living organisms and therefore carries clear obligations concerning welfare and scientific justification. Distress can harm animal subjects and modify measured physiology or behavior. Sound scientific practice and responsible care are therefore connected.

The three Rs are replacement, reduction, and refinement. Replacement favors scientifically appropriate non-animal approaches. Reduction seeks the minimum animal use consistent with a valid study; an underpowered experiment can waste animals while generating little reliable information. Refinement improves care and minimizes pain, distress, and avoidable harm through procedural design, monitoring, and humane endpoints (NIH Office of Laboratory Animal Welfare).

The Guide for the Care and Use of Laboratory Animals is a central US reference for institutional animal programs, husbandry, veterinary oversight, and welfare standards. Regulatory obligations depend on jurisdiction, organism, funding, and institutional arrangements. Ethical approval concerns the justification and governance of the planned work; it does not certify that all subsequent inferences about humans are correct.

Nonhuman primate experiments, invasive procedures, and repeated interventions warrant particularly careful scrutiny. Investigators should demonstrate why alternatives cannot adequately address the scientific question and reevaluate protocols as methods change.

Can newer methods replace animal experiments?

Human induced pluripotent stem cells, neural organoids, tissue cultures, microphysiological platforms, computational models, and noninvasive human neuroscience can answer many questions with directly human-derived materials or data. New approach methodologies are increasingly relevant to biomedical research, and the NIH describes both their potential and the importance of choosing methods appropriate to the scientific purpose (NIH alternatives and NAMs).

Every method has limitations. An organoid can illuminate selected developmental or cellular properties, but cannot reproduce every aspect of an embodied nervous system interacting with an environment. A computer simulation reveals the consequences of assumptions encoded in its structure. Human fMRI generally measures changes related to blood oxygenation, not direct spikes from identified neurons, and provides limited intervention opportunities compared with some animal preparations.

A well-designed research program combines methods as needed and evaluates which questions can be answered without animals. Replacing animal experiments when scientifically appropriate, while improving the validity and welfare of those that remain necessary, advances both scientific and ethical goals.

How to interpret an animal neuroscience headline

Consider a hypothetical headline: Scientists discover a brain circuit responsible for empathy in mice. A rigorous reading identifies the task, the species, the intervention, and the exact outcome. Did investigators measure social approach, response to another animal's distress, observational learning, or a specific choice? Did they record naturally occurring activity, or stimulate a pathway artificially?

Ask whether sensory and motor explanations were addressed, whether behavior was robust across tasks, whether the neural correspondence has been established in humans, and whether results were independently replicated. A study can establish that a particular circuit contributes to an observable social response without proving an explanation of human empathy as a whole.

The same reasoning applies to claims about addiction, fear, memory, attention, personality, and mental health. An accurately framed discovery identifies its observed mechanism and evidentiary reach.

Common misconceptions

If a treatment works in mice, it works in humans

Preclinical effects supply a hypothesis for further research. Human safety, dose, effectiveness, and clinically important outcomes must be established separately. The 2024 umbrella review demonstrates the difference between progression into trials and eventual approval.

Animal research cannot tell us anything about human behavior

Conserved biological mechanisms and appropriately matched computations can provide valuable evidence. Their application requires identifying the specific correspondence and its limitations.

A maze score measures human memory directly

A maze captures task performance shaped by learning, navigation, movement, motivation, and context. Human memory includes additional systems and subjective or autobiographical dimensions requiring human investigation.

A dopamine experiment reveals the brain's pleasure chemical

Dopamine signaling participates in multiple functions, including reinforcement-related learning and movement. A neural manipulation or behavioral reward response cannot be reduced to one universal feeling or a single chemical explanation.

Stimulating neurons proves they alone produce a complex emotion

Artificial activation supports a causal inference for the studied intervention and measured behavior. Complex experiences and actions involve interacting networks, organismic states, and environments.

More animals always means more rigorous science

Good design requires appropriate experimental units, justifiable sample size, rigorous controls, and transparent inference. A larger confounded study may remain uninformative. ARRIVE 2.0 addresses this broader methodological standard.

Frequently asked questions

What is the purpose of animal models in behavioral neuroscience?

They permit controlled experiments on defined neural mechanisms and observable behavior and provide hypotheses to test with human evidence. Their validity is specific to the phenomenon being investigated.

Why are mice often used?

Their biology is extensively characterized and many genetic, physiological, and behavioral methods are available. Rodent findings still require evaluation of species differences and environmental conditions.

What is the difference between a model and an assay?

A model is the experimental organism and preparation used to represent a specified phenomenon. An assay is the measurement procedure applied within that system. Several assays may investigate one model.

What do face, construct, and predictive validity mean?

Face validity concerns outward resemblance, construct validity concerns mechanistic correspondence, and predictive validity concerns performance in forecasting independent target outcomes. Each refers to a stated purpose.

Can animal models prove what causes depression in humans?

They can identify and test particular biological or behavioral mechanisms. Diagnosing and explaining human depression requires human clinical evidence, including subjective experiences and functioning that cannot be captured fully in animal tasks.

Can animal experiments establish causality?

Appropriate controlled interventions can support causal conclusions about specified animal outcomes. Cross-species causal claims require additional evidence of correspondence.

What is the strongest evidence of successful translation?

Converging and reproducible animal findings, demonstrated human biological relevance, matched human observations, and direct human testing when possible. For treatments, clinical safety and efficacy research is essential.

Are animal tests becoming obsolete?

Human-derived biological systems and computational approaches increasingly replace or supplement particular animal studies. The scientific question determines whether a method is an adequate replacement.

What are the three Rs?

Replacement, reduction, and refinement guide responsible animal use and the consideration of alternatives, study size, and welfare.

References

Bailey, S. J., Bast, T., Chaby, L., et al. (2023). Improving Translational Relevance in Preclinical Psychopharmacology (iTRIPP). Journal of Psychopharmacology, 37, 1051–1057. https://doi.org/10.1177/02698811231182607

Cross-species consensus atlas of the primate basal ganglia. (2025). Peer-reviewed comparative multiomic study. https://pubmed.ncbi.nlm.nih.gov/41446173/

Deisseroth, K. (2015). Optogenetics: 10 years of microbial opsins in neuroscience. Nature Neuroscience, 18, 1213–1225. https://doi.org/10.1038/nn.4091

Ineichen, B. V., Furrer, E., Grüninger, S. L., Zürrer, W. E., & Macleod, M. R. (2024). Analysis of animal-to-human translation shows that only 5% of animal-tested therapeutic interventions obtain regulatory approval for human applications. PLOS Biology, 22, e3002667. https://doi.org/10.1371/journal.pbio.3002667

Kanwisher, N. (2025). Animal models of the human brain: Successes, limitations, and alternatives. Current Opinion in Neurobiology, 90, 102969. https://doi.org/10.1016/j.conb.2024.102969

National Institutes of Health, Office of Laboratory Animal Welfare. (2026). Alternatives and new approach methodologies. https://grants.nih.gov/policy-and-compliance/policy-topics/animal-welfare/compliance-topics/alternatives-nams

National Research Council. (2011). Guide for the Care and Use of Laboratory Animals (8th ed.). National Academies Press. https://doi.org/10.17226/12910

Percie du Sert, N., Hurst, V., Ahluwalia, A., et al. (2020). The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research. PLOS Biology, 18, e3000410. https://doi.org/10.1371/journal.pbio.3000410

Sena, E. S., van der Worp, H. B., Bath, P. M. W., Howells, D. W., & Macleod, M. R. (2010). Publication bias in reports of animal stroke studies leads to major overstatement of efficacy. PLOS Biology, 8, e1000344. https://doi.org/10.1371/journal.pbio.1000344

Sorge, R. E., Martin, L. J., Isbester, K. A., et al. (2014). Olfactory exposure to males, including men, causes stress and related analgesia in rodents. Nature Methods, 11, 629–632. https://doi.org/10.1038/nmeth.2935

Sukoff Rizzo, S. J., & Crawley, J. N. (2017). Rigor and reproducibility in rodent behavioral research. Neuroscience & Biobehavioral Reviews. https://pubmed.ncbi.nlm.nih.gov/29307548/

van der Staay, F. J., Arndt, S. S., & Nordquist, R. E. (2009). Evaluation of animal models of neurobehavioral disorders. Behavioral and Brain Functions, 5, 11. https://pmc.ncbi.nlm.nih.gov/articles/PMC2669803/

Willner, P. (2017). Reliability of the chronic mild stress model of depression: A user survey. Neurobiology of Stress, 6, 68–77. https://doi.org/10.1016/j.ynstr.2016.08.001

Psychological Encyclopedia