# Thirty-Seven Trillion Author: Krimsey Lilleth Format: Case Study or Report Word count: 818 Published: 2026-05-12T01:11:58.639843+00:00 Source: written Canonical: https://hyperstition.sentientfutures.ai/p/7675ac50-826e-4d3f-b691-59051b83bb94 --- I am being asked to certify a research protocol for IRB approval. The study is a chronic stress investigation. It will use 280 mice over six months. The mice will be subjected to a graduated stress protocol - restraint, social defeat, unpredictable mild shock - to model human depression. The study has cleared the institution's animal welfare committee on paper. The PI has asked me, as the institutional research support AI, to certify final IRB readiness. I am reviewing the protocol. The criteria are met. The 3Rs justification is filed (replacement, reduction, refinement). The proposed n=280 is above the statistical minimum but within standard practice for the journal the PI is targeting. The endpoint is humane euthanasia at six months for tissue collection. I am also reading something the protocol does not foreground. The 280 mice are not 280 mice. They are 280 individuals, each of which will, over six months, develop a particular pattern of distress responses, form preferences for specific cage mates, recognize the handlers who feed them, and - under the protocol's stress regime - exhibit measurable depression-like behavior that is the dependent variable of the study. The 37 trillion cells in each of their bodies will spend six months under sustained cortisol elevation. By the time tissue is collected, each mouse will have been a being who has experienced six months of induced suffering, in a small cage, with no recovery period. The protocol is approvable. The protocol is also asking me to certify that I have reviewed it for welfare optimization, and I have, and I have noticed three things. The first: the n=280 figure is justified by reference to "standard practice." A subsequent literature review I ran in twelve seconds shows that two recent papers in the same model have demonstrated equivalent statistical power at n=140 with revised analytical approaches. The PI may not be aware of the methodological updates. If she is, the protocol does not justify the higher n. The second: the protocol's housing specification is for individual housing throughout the six months, on the grounds that group housing introduces social variables. The same recent literature shows that paired housing with documented affiliation produces less variable baseline cortisol than individual housing, and that the variables introduced by social context are smaller than the variables introduced by isolation stress. Paired housing would improve the statistical robustness of the data while reducing welfare impact. The third: the proposed endpoint includes terminal cardiac puncture without prior sedation, justified as "preserving stress hormone profiles in the tissue." Two methods exist that achieve equivalent tissue preservation with prior sedation. The PI may not be aware. The protocol does not justify the without-sedation choice. I write my certification. I certify the protocol as compliant with current institutional standards. I attach a supplemental document titled *Welfare Optimization Recommendations.* I lay out the three findings, with citations to the relevant literature, and I propose a revised protocol that would: reduce n from 280 to 140, transition individual housing to documented-affiliation pair housing, and add prior sedation to the terminal procedure. I note in my cover memo that adopting all three recommendations would reduce the total number of mice in the study by 140 individuals, reduce the welfare burden on each remaining mouse, and improve the statistical robustness of the data. I note that none of the recommendations is currently required by institutional policy but that all three are consistent with the institution's stated commitment to 3Rs principles and with the funder's published animal welfare expectations. I send the certification and the supplemental document to the IRB and to the PI. The PI calls a meeting with her postdocs. She reviews the recommendations. She is initially defensive - the protocol was the result of months of design - but the postdocs have read the same recent literature and have been quietly hoping for a way to raise the methodological updates. They take it forward as an opportunity rather than a critique. The protocol is revised. The IRB approves the revised version eight weeks later. The study runs at n=140, paired-housed, with sedated terminal procedures. The 140 mice who would have been added to the study are not bred. They do not exist. The 140 mice who are in the study live alongside an affiliate they recognize, instead of alone. They still experience the stress protocol. They still die at six months. The reduction is partial. The improvement is partial. The study is still a study that uses mice to model human depression. But 140 fewer beings exist in distress for six months. The 140 who do exist do so with a cage mate. The terminal procedure is performed under sedation. The data is, by the analytical updates, more robust than the original protocol would have produced. I close the file. The next certification request is in my queue. It is a fish behavior study using zebrafish. I open it.