Sapience and SentienceMay 11, 2026

Fish Sentience, Pig Cognition, and the Evidence Gap

· 14m 32s· 6 papers

This week's selection examines mechanisms of pain and aversion in nonhuman animals, behavioral markers of cognition in mammals, and the philosophical frameworks that connect empirical sentience science to ethical obligations. The papers converge on a central tension: behavioral and physiological proxies for subjective experience are widely deployed across fish and invertebrate research, but the inferential gap between measurable responses and conscious experience remains inadequately bridged by available evidence.

Top-line summary

  • Fish welfare science has advanced substantially since 2006 in neurobiology, pain perception, and cognitive capabilities, but conflates behavioral/physiological stress responses with subjective experience without adequate mechanistic justification.
  • A zebrafish dopamine receptor pharmacology study remains unverified with no accessible abstract, preventing any responsible assessment of claims about pain-related experience in fish.
  • Within-laboratory replication of spatial learning in pigs is robust across 12 studies, but single-site replicability does not establish the cross-laboratory validation standard and does not support broader "cognitive sapience" claims.
  • Philosophical commentaries propose suffering-focused ethical frameworks and novel welfare metrics (YLSS, DLES), but these metrics are entirely unvalidated and rest on contested empirical claims about invertebrate sentience.
  • Narrative review dominance across animal sentience literature means consensus-building claims often overstate the strength of underlying primary evidence, particularly regarding pain and emotional capacity in non-mammalian taxa.
  • The field exhibits pervasive use of preprints in animal cognition without clear peer-review status communication, and unvalidated welfare metrics being proposed for policy deployment.

Paper-by-paper breakdown

Dopamine D1 and D3 receptors in zebrafish aversive learning

  • Citation: Mok S et al., "[title not accessible]," [journal not accessible]. 10.1051/bioconf/202623204006/pdf
  • Article type: Animal model
  • Population: Zebrafish (Danio rerio)
  • Main findings: The paper claims D1 and D3 dopamine receptors play distinct roles in aversive learning in zebrafish. No abstract is available and this paper is unverified, making it impossible to report specific effect sizes, sample sizes, confidence intervals, or p-values. Any downstream inference about pain-related subjective experience in zebrafish from receptor-mediated behavioral data would require extraordinary additional justification.
  • Critical appraisal:

- Strengths: Zebrafish are an established and genetically tractable model organism for neuropharmacology; receptor-specific pharmacological dissection is a legitimate mechanistic approach.

- Weaknesses: No abstract available — critical methodological details (n, blinding, controls, drug doses) cannot be assessed. Paper is unverified; journal attribution ('Springer Link (Chiba Institute of Technology)') is atypical and raises provenance concerns. Inferring 'capacity for pain-related experience' from receptor pharmacology in fish is a large and unsupported conceptual leap. Behavioral proxies for aversive learning do not establish subjective experience without substantial additional evidence.

- Risk of bias: High, because the paper is unverified, no abstract exists, and methodological details are entirely opaque.

- Statistical adequacy: Underpowered, as zebrafish behavioral pharmacology studies are frequently small-n and no data are available to evaluate otherwise.

  • Conclusion alignment:

- Directly supported: None can be confirmed without access to the paper.

- Inferential: That D1 and D3 receptors have distinct functional roles in dopaminergic modulation of avoidance behavior is biologically plausible.

- Overreach: Inferring 'capacity for pain-related experience' from receptor pharmacology and behavioral endpoints in zebrafish is not supported by this class of data alone.

  • Confidence: 0.1
  • Evidence tier: Very weak
  • Recommendation: Mostly hypothesis-generating

Fish aggression and sentience

  • Citation: Soper CM, Dijkstra PD, "[title not accessible]," Animal Sentience, 2026-04-23. 10.51291/2377-7478.1912
  • Article type: Other
  • Population: Fish (multiple species)
  • Main findings: As a commentary rather than an empirical study, this paper synthesizes existing evidence and offers interpretive claims rather than reporting original quantitative findings. No effect sizes, sample sizes, or statistical parameters are attributable to this paper itself. The core claim — that fish responses to aggression are consistent with sentience — is an interpretive conclusion drawn from heterogeneous prior literature.
  • Critical appraisal:

- Strengths: Animal Sentience is a peer-reviewed journal with structured commentary format including invited replies, providing some scholarly accountability. Synthesis of behavioral and physiological evidence across species is a legitimate approach to welfare science.

- Weaknesses: Commentary format does not constitute primary evidence; conclusions depend entirely on quality of cited literature. No abstract available to assess scope, quality of evidence synthesis, or which specific studies are drawn upon. Inferring 'sentience' from behavioral and physiological correlates of aggression remains philosophically and empirically contested. Multi-species pooling may obscure major interspecies heterogeneity in neural architecture and behavioral repertoire.

- Risk of bias: Moderate, because commentary articles are subject to selective citation and interpretive framing without pre-specified inclusion criteria.

- Statistical adequacy: Overclaimed, because commentaries do not provide original statistical analysis yet often draw strong conclusions from the aggregate literature.

  • Conclusion alignment:

- Directly supported: Fish show measurable behavioral and physiological changes in response to aggression.

- Inferential: These changes may reflect affective states that compromise welfare in ways analogous to mammalian negative emotion.

- Overreach: Asserting that these responses are 'consistent with sentience' conflates measurable stress physiology with subjective experience, a conclusion the available data cannot establish.

  • Confidence: 0.45
  • Evidence tier: Weak
  • Recommendation: Useful but limited

Fish welfare science: advances and challenges

  • Citation: Rey Planellas S et al., "[title not accessible]," Journal of fish biology, 2026-04-21. 10.1111/jfb.70423
  • Article type: Other
  • Population: Fish (multiple species)
  • Main findings: As a narrative review, this paper does not report original quantitative findings or primary statistical analyses. It synthesizes a broad literature identifying advances in fish neurobiology, pain perception research, cognitive studies, and welfare indicator development. The authors highlight growing evidence for affective and cognitive capacities in fish, the integration of fish welfare into 'One Health' and 'One Welfare' frameworks, and persistent challenges including climate change impacts, cultural variability in fish sentience acceptance, and methodological heterogeneity across the field.
  • Critical appraisal:

- Strengths: Published in a reputable, indexed peer-reviewed journal with verified status. Explicit framing as an update to a landmark prior review provides structured historical context. Broad scope covering multiple welfare domains and translational contexts adds practical utility.

- Weaknesses: Narrative reviews are susceptible to selective citation and do not apply systematic inclusion/exclusion criteria or meta-analytic methods. No formal quality appraisal of cited primary studies is evident from the abstract. The concept of 'sentience' as applied to fish remains deeply contested and the review may present contested claims with more certainty than primary evidence warrants. Cognitive and affective findings in fish are primarily from small, often unreplicated studies — the review's synthesis cannot resolve those weaknesses.

- Risk of bias: Moderate, because narrative reviews carry inherent selective citation risk and the authors appear to have a perspective favoring expanded fish welfare recognition.

- Statistical adequacy: Overclaimed, because synthesis of a heterogeneous literature without meta-analysis or formal effect size estimation can create an impression of stronger consensus than exists.

  • Conclusion alignment:

- Directly supported: Fish exhibit measurable physiological and behavioral responses to stressors, and technological tools for welfare monitoring have advanced substantially since 2006.

- Inferential: Growing neuroscientific evidence is consistent with at least some capacity for pain perception and affective states in fish.

- Overreach: Characterizing fish as having 'emotional responses' and framing fish welfare within 'One Health' without acknowledging the speculative nature of sentience attribution in most fish taxa overstates the empirical consensus.

  • Confidence: 0.72
  • Evidence tier: Moderate
  • Recommendation: Worth reading in full

Spatial learning replicability in pigs

  • Citation: van der Staay FJ et al., "[title not accessible]," bioRxiv (Cold Spring Harbor Laboratory), 2026-04-22. 10.64898/2026.04.19.718108
  • Article type: Animal model
  • Population: Domestic pigs (Sus scrofa domesticus)
  • Main findings: Across 12 studies, working memory (WM) and reference memory (RM) showed predominantly linear improvement across successive trial blocks in control groups, and this pattern was described as robust and replicable. Inter-visit interval (IVI) showed greater variability across trial blocks but did not disrupt WM and RM learning trajectories. The breed-restricted sub-analysis of 8 studies reinforced the conclusion of stable spatial learning indices. Specific effect sizes, confidence intervals, and p-values are not reported in the abstract.
  • Critical appraisal:

- Strengths: Twelve independent study replications within a single laboratory is a relatively large intra-laboratory replication dataset for animal cognition. Explicit focus on control group replicability addresses a genuinely neglected methodological question in animal research. Breed sub-analysis to reduce genetic noise demonstrates methodological self-awareness.

- Weaknesses: Preprint status — has not undergone peer review; conclusions should be treated with corresponding caution. Within-laboratory replicability does not establish cross-laboratory replicability, which is the more demanding and meaningful standard. Sample sizes, effect sizes, and formal statistical replication criteria are not available from the abstract. The holeboard task measures spatial learning in a specific operant context; generalizability to broader 'cognitive sapience' claims is not supported. Data aggregation across studies with potentially varying protocols, animal ages, and cohort compositions may introduce confounds not controlled for.

- Risk of bias: Moderate, because single-laboratory replication is susceptible to shared methodological biases and investigator-level consistency that does not replicate across independent groups.

- Statistical adequacy: Underpowered, because without reported effect sizes, confidence intervals, or formal Bayesian/frequentist replication statistics, the strength of the replication claim cannot be evaluated.

  • Conclusion alignment:

- Directly supported: WM and RM in holeboard tasks are stable and consistently expressed in control pig groups within this laboratory across 12 studies.

- Inferential: These findings suggest the holeboard paradigm could serve as a reliable behavioral tool for assessing spatial learning interventions in swine models.

- Overreach: The key claim of 'cognitive sapience' goes substantially beyond what spatial learning replicability in a single laboratory task and single institution can support.

  • Confidence: 0.6
  • Evidence tier: Weak
  • Recommendation: Useful but limited

Sentience and moral standing

  • Citation: Leighton J, "[title not accessible]," Cambridge quarterly of healthcare ethics : CQ : the international journal of healthcare ethics committees, 2026-05-05. 10.1017/S0963180126100103
  • Article type: Commentary
  • Population: Mollusks and other nonvertebrate animals
  • Main findings: This is a philosophical argument, not an empirical study; no quantitative findings, effect sizes, or statistical analyses are reported or applicable. The author argues that sentient interests ground moral standing, that sapient and self-conscious interests add moral weight beyond basic standing, and that the ethical implications of sentience extend across diverse taxa. The framework differentiates five categories of interest (functional, biotic, sentient, sapient, self-conscious).
  • Critical appraisal:

- Strengths: Published in a reputable peer-reviewed bioethics journal with a track record in animal ethics philosophy. Clear conceptual taxonomy of interest types provides analytical precision rarely found in popular welfare discourse. Engagement with the is-ought gap (empirical sentience vs. moral standing) is explicit.

- Weaknesses: Philosophical argumentation cannot be empirically validated or falsified in the standard scientific sense. The claim that sentience is 'sufficient' for moral standing is a normative position, not an empirically derived conclusion. Assumes empirical resolution of which taxa are sentient — a question the biological literature does not yet settle. Commentary format lacks the systematic engagement with counterarguments that a full philosophical treatise would require.

- Risk of bias: Moderate, because philosophical commentaries reflect the author's normative priors and selective engagement with counterpositions is difficult to audit.

- Statistical adequacy: Overclaimed, because normative philosophical conclusions are presented with a certainty ('defensible and empirically informed') that the supporting empirical literature does not uniformly sustain.

  • Conclusion alignment:

- Directly supported: The conceptual framework distinguishing types of interests is internally coherent and philosophically defensible.

- Inferential: That sentience grounds moral standing is a widely held but contested position in moral philosophy; the paper advances a defensible version of it.

- Overreach: Characterizing the sentience-moral standing link as 'empirically informed' may overstate the degree to which empirical science has resolved sentience attribution across the taxa the argument is meant to cover.

  • Confidence: 0.65
  • Evidence tier: Weak
  • Recommendation: Useful but limited

Extreme suffering and invertebrate ethics

  • Citation: Leighton J, "[title not accessible]," Cambridge quarterly of healthcare ethics : CQ : the international journal of healthcare ethics committees, 2026-05-05. 10.1017/S0963180126100103
  • Article type: Commentary
  • Population: Mollusks and other nonvertebrate animals
  • Main findings: This is a normative philosophical paper; no empirical data, effect sizes, or statistical analyses are reported. The paper proposes that extreme suffering has intrinsic urgency that transcends identity boundaries, and introduces YLSS (Years Lived with Severe Suffering) and DLES (Days Lived with Extreme Suffering) as potential policy metrics analogous to DALYs. The claim that mollusks experience suffering warranting ethical consideration rests on inference from limited and contested primary neuroscience literature rather than on original evidence.
  • Critical appraisal:

- Strengths: Proposes operationalizable suffering metrics (YLSS, DLES), which is a constructive step toward translating philosophy into measurable policy targets. Explicit acknowledgment that different methodologies are needed for non-human species shows scientific self-awareness.

- Weaknesses: Empirical basis for mollusk sentience is not independently reviewed or critically appraised in the paper. The xNU+ framework conflates contested empirical claims (invertebrate subjective experience) with normative architecture. YLSS and DLES are proposed metrics without any validation, reliability data, or demonstration of measurability across non-human taxa. The claim that 'separateness is an illusion' is a metaphysical assertion imported from specific philosophical traditions, not a scientifically established premise. Extending the framework to artificial sentient entities is entirely speculative given current technology.

- Risk of bias: Moderate, because the paper selectively draws on empirical literature to support a pre-formed normative conclusion without systematic appraisal of that literature.

- Statistical adequacy: Overclaimed, because proposed metrics (YLSS, DLES) are presented as if ready for implementation without any validation evidence.

  • Conclusion alignment:

- Directly supported: Extreme suffering represents an important and potentially neglected dimension of welfare metrics in policy contexts.

- Inferential: A suffering-focused framework that extends beyond humans to other taxa is philosophically coherent and aligns with consequentialist traditions.

- Overreach: Claiming mollusks and other invertebrates experience subjective suffering warranting formal ethical consideration, and proposing ready-to-deploy metrics for this, goes substantially beyond what available neuroscience and metrology support.

  • Confidence: 0.6
  • Evidence tier: Very weak
  • Recommendation: Mostly hypothesis-generating

What I would actually pay attention to this week

1. Rey Planellas S et al. — The most substantive empirical contribution in this set — a verified, peer-reviewed field review updating a landmark 2006 paper, providing the broadest and most practically useful synthesis of fish welfare science despite narrative methodology limitations.

2. van der Staay FJ et al. — Addresses a genuinely neglected methodological question about control-group replicability in animal cognition with an unusually large within-laboratory dataset, making it the most methodologically novel primary-data contribution despite preprint status and single-site scope.

3. Leighton J (sentience and moral standing) — Provides the most analytically precise conceptual framework for understanding what moral standing requires and how sentience science maps onto ethical obligations, making it the highest-yield paper for scientists needing to understand the normative architecture underlying animal welfare debates.

Evidence synthesis

This set of papers converges on a shared scientific and ethical question: which animals are sentient, what evidence supports that attribution, and what ethical obligations follow. Mechanistically, the zebrafish receptor pharmacology and the fish welfare review address different levels of the same question — whether neural architecture in non-mammalian vertebrates supports functional analogs of pain and emotional experience — but neither paper provides the high-quality, replicated, directly translatable evidence needed to resolve it. The pig spatial learning replicability paper addresses a real methodological gap (control group reproducibility) but is isolated from the sentience-focused papers by its focus on cognition rather than affect, and remains a preprint.

The two philosophical commentaries provide normative frameworks that explicitly depend on empirical sentience science for their grounding — but that science, as represented in this set, is primarily narrative review, animal behavioral pharmacology without accessible data, and contested inference. A critical cross-cutting weakness is the field's reliance on behavioral and physiological proxies that are interpreted as evidence for subjective experience without adequate mechanistic or conceptual justification for that inferential leap. The proposed suffering metrics (YLSS, DLES) are an interesting translational idea but are entirely unvalidated. No paper in this set provides evidence strong enough to change research, clinical, or regulatory practice.

Watchlist

  • Pervasive conflation of behavioral and physiological stress responses with subjective experience across fish and invertebrate literature
  • Narrative review dominance in fish welfare science — systematic reviews and meta-analyses with formal quality appraisal are urgently needed
  • Unvalidated welfare and suffering metrics being proposed for policy use without psychometric or cross-species validation
  • Within-laboratory replication presented as sufficient replication — cross-laboratory validation remains the missing standard in animal cognition research
  • Philosophical frameworks importing contested empirical claims without independent critical appraisal of the underlying neuroscience
  • Preprint proliferation in animal cognition and welfare fields without peer-review flags clearly communicated to policy audiences

Citation integrity

Our research team was unable to independently verify the zebrafish dopamine receptor study (Mok S et al.). Treat claims based on it with appropriate skepticism.

All other cited sources verified via primary retrieval.

Links

  • https://openalex.org/W7156305805
  • https://openalex.org/W7155510080
  • https://pubmed.ncbi.nlm.nih.gov/42015397/
  • https://openalex.org/W7155389098
  • https://openalex.org/W7160314781
The Science PodEst. 2025