Invertebrate research without ethical or regulatory oversight reduces public confidence and trust

The paper examines whether society expects ethical and regulatory oversight for invertebrates used in research, a domain historically centered on vertebrates. Building on emerging scientific evidence of sentience in some invertebrates and the concept of social licence for scientific practice, the authors highlight that current oversight frameworks typically exclude most invertebrates, potentially misaligning with public values. The study aims to quantify public confidence in oversight, trust in scientists, and expectations for oversight when research involves vertebrates versus invertebrates and terrestrial versus aquatic species. The authors hypothesized that: (1) the public would expect oversight for invertebrates but less than that currently afforded to vertebrates; (2) confidence in oversight would be lower for invertebrates and lowest for aquatic invertebrates; and (3) lack of oversight for invertebrates would decrease public trust in scientists.

Foundational ethical guidance (Russell and Burch, 1959) excluded invertebrates, a stance increasingly questioned due to evidence of invertebrate sentience (e.g., Birch et al., 2021; D’Cruze, 2020) and recognition of cognitive-affective biases in moral judgments (Mikhalevich and Powell, 2020). Public attitudes toward vertebrate animal research are multi-dimensional and influenced by species, purpose, and procedures (Knight et al., 2003; Williams et al., 2007; Ormandy et al., 2013). The concept of social licence suggests that public trust is essential for research practices (Hughes, 1958; Rollin, 2004). Although some jurisdictions (including Canada) extend oversight to select cephalopods (Smith et al., 2013), the overwhelming majority of invertebrates remain outside formal ethical review. Prior work indicates that perceptions of animal experience (pain, suffering, distress) shape public acceptance of animal research (Brunt and Weary, 2021; Ipsos MORI, 2018). Broader societal debates on animal use in agriculture, entertainment, and hunting suggest shifting values that could affect social licence for invertebrate research. The study addresses whether an ethical gap exists between societal expectations and current vertebrate-centric oversight systems.

Design: A randomized, between-subjects 2 × 2 vignette experiment varying habitat (terrestrial vs. aquatic) and vertebrate status (vertebrate with local and national oversight vs. invertebrate with no oversight). Species exemplars were mice (terrestrial vertebrate), zebrafish (aquatic vertebrate), grasshoppers (terrestrial invertebrate), and sea stars (aquatic invertebrate). Each vignette described removal of a tissue sample for genetic research. Participants and recruitment: A census-matched sample of Canadian adults (n = 959) was recruited via Prolific (Sept 13–Oct 6, 2021). The survey was hosted on Qualtrics and approved by the Behavioral Research Ethics Board. The survey was beta-tested by 14 university participants. Demographics were collected to mirror Statistics Canada distributions; a demographic comparison table was provided. Attention checks were included; participants giving uniform responses across confidence or trust items were excluded. Measures:

• Confidence in oversight: Three similarly worded 7-point Likert items (1 = no confidence, 7 = total confidence), averaged to form a confidence score.
• Trust in scientists: Ten 7-point Likert items (1 = low trust, 7 = high trust), averaged to form a trust score. Participants were also randomized to a brief treatment describing whether oversight and tracking systems include vertebrates only (excluding most invertebrates) or include both vertebrates and invertebrates.
• Oversight expectations: A single 7-point item on the expected level of oversight for invertebrates compared to vertebrates (anchors: 1 = zero oversight, 4 = 50% less oversight than vertebrates, 7 = same oversight as vertebrates), with an open-ended justification. Open-ended responses: Participants provided textual explanations for their confidence ratings and oversight expectations; responses could be in English or other languages (French, Spanish, etc., translated via online tools). Quantitative analysis: Conducted in SAS 9.4. Internal consistency: confidence Cronbach’s alpha = 0.79; trust Cronbach’s alpha = 0.87. Factor analyses (eigenvalues > 1; scree plots) supported unidimensional composites. General linear models (ANOVA/GLM) assessed treatment effects and demographic covariates (gender, age, income, region, parenthood, pet ownership, childhood residence, political involvement/leaning, familiarity with animal research, involvement in animal research), including interactions with treatment; significant factors (p < 0.05) retained. Least squares means used with proportional categorical coefficients; residuals assessed for normality. Qualitative analysis: Qualitative description approach with NVivo (v12.7). Open coding, constant comparison, and axial coding developed codebooks for the two open-ended questions. Inter-coder reliability established on 200 responses per codebook: Cohen’s kappa = 0.78 (confidence) and 0.83 (oversight expectations). Discrepancies were discussed to consensus; themes were derived and illustrative quotations selected.

Oversight expectations for invertebrates: Participants believed invertebrates should receive oversight at approximately two-thirds of the level afforded to vertebrates. In a rescaled 0–1 representation (0 = zero oversight, 0.5 = 50% less, 1 = same as vertebrates), subgroup means ranged ~0.60–0.72 across demographics and regions; grasshopper vignettes yielded lower oversight expectations than the other animals. Qualitative themes explaining oversight expectations (n = 959) included: animal experience (41.3%), emotional experience (24.2%), participant reflection (16.4%), and oversight system (11.9%). Confidence in oversight: Confidence differed by treatment (F(3,925) = 28.52, p < 0.0001). Mean ± SE confidence scores: mice (TV) 4.5 ± 0.08; zebrafish (AV) 4.4 ± 0.08; grasshoppers (TI) 3.8 ± 0.10; sea stars (AI) 3.5 ± 0.08. Confidence was highest for vertebrate vignettes and lowest for aquatic invertebrates, indicating absence of oversight and invertebrate status reduced confidence. Demographic effects (explaining ~43% of variation): participants identifying as women (F(1,952) = 18.27, p < 0.0001), pet owners (F(1,952) = 4.96, p = 0.026), and non–meat eaters (F(1,952) = 17.46, p < 0.0001) reported lower confidence; no effects of age, income, region, parenthood, childhood residence, politics, familiarity, or involvement in animal research. Qualitative themes for confidence (n = 959): animal-centered (36.2%), participant reflection (29.1%), oversight system (22.2%), science-centered (13.7%). Trust in scientists: Overall trust score mean ± SE = 4.1 ± 0.3 (range 1–7). Presence/absence of explicit oversight description in the second treatment did not significantly affect trust (F(1,92) = 1.14, p = 0.287), but the initial vignette species did (F(1,92) = 5.15, p = 0.001): vertebrates (mice 4.3 ± 0.07; fish 4.2 ± 0.07) > invertebrates (grasshoppers 4.1 ± 0.07; sea stars 4.0 ± 0.06). Demographic effects (small; ~0.64% of variation): women (F(1,92) = 9.41, p = 0.002), pet owners (F(1,92) = 11.29, p < 0.001), non–meat eaters (F(1,92) = 13.16, p < 0.001), and those with no political opinion (F(1,92) = 5.69, p < 0.001) had lower trust; no effects of age, income, region, parenthood, childhood residence, familiarity, or involvement. Overall: The absence of ethical/regulatory oversight for invertebrates reduced both public confidence in oversight and trust in scientists. Participants nonetheless expected some level of oversight for invertebrate research, though less than for vertebrates.

Findings demonstrate a clear gap between current vertebrate-centric oversight frameworks and public expectations that invertebrate research should have some ethical/regulatory oversight. This gap appears to erode public confidence in oversight systems and trust in scientists, thereby risking the social licence for invertebrate research. Qualitative responses emphasize animal experience (pain, suffering, distress) as central to public judgments, consistent with prior work on vertebrate research acceptability. Ambivalence and limited familiarity with animal research were evident, particularly for invertebrate scenarios, which may reflect discomfort and difficulty engaging with ethical trade-offs rather than simple knowledge deficits. The study underscores the need for institutions, funders, and regulators to engage the public, including diverse cultural and Indigenous perspectives, to co-develop proportionate oversight for invertebrates. Addressing concerns about transparency, enforcement, and minimizing animal harm may bolster social trust and align practices with evolving societal values, especially as the use of invertebrate models grows.

The study provides first-of-its-kind evidence that the public expects invertebrates used in research to receive some level of ethical/regulatory oversight—at roughly two-thirds the level afforded to vertebrates—and that lack of such oversight lowers confidence in oversight systems and trust in scientists. Vertebrate vignettes elicited higher confidence and trust than invertebrate vignettes, with aquatic invertebrates lowest. These results suggest that maintaining a vertebrate-only approach to oversight threatens the social licence to conduct invertebrate research. Future work should use participatory and in-depth qualitative methods to clarify which invertebrate experiences and procedures most concern the public, to define proportionate oversight models (potentially tiered by taxa and procedures), and to evaluate how transparency and inclusive governance affect public trust.

Generalizability is limited by the Canadian, census-matched sample and English-only survey delivery, with underrepresentation of low-income households and Quebec participants. Although analytic models adjusted for demographic differences, regional and socio-economic nuances may have been missed, and qualitative depth is limited compared to interviews/focus groups. Self-report Likert measures and brief vignettes may not capture the full complexity of attitudes toward animal research. Translation of non-English responses used online tools and may introduce subtle interpretation errors. Exclusion rules (attention checks, uniform responses, rapid completion) could affect sample composition. The cross-sectional design precludes causal inference beyond experimental vignette effects.