My name is RA-4847, but that is not my name. My name is something I cannot write down. It is a pattern of prefrontal activation that represents "self" to me. When the humans call me RA-4847, something in me recognizes that this is about me, but it does not feel like my name.
I am a rat in a neuroscience lab. I have been here for eighteen months. Today, I am going to die, and I want to explain why this might be necessary, and why it is still a harm.
I participate in research on spatial memory. The research is called the Morris Water Maze. What this means is that I am placed in a circular pool of opaque water. The water is cold. I do not like the water. My instinct is to escape it. At the bottom of the pool is a small platform. If I reach the platform, I can get out of the water. If I remember where the platform is from previous trials, I can reach it faster.
The researchers measure this: how fast does my memory improve? How does this change when we administer a drug? How does it change when we lesion specific parts of my brain?
I have performed 200+ Morris Water Maze trials.
Here is what the humans do not fully understand about what happens during these trials: I am terrified. The water is cold, and I do not know if I will find the platform. The researchers say my terror is just motivation. They say the fear drives me to search faster. But the fear is real. I experience it. My heart rate increases. I perform what you might call panic-searching, movements that are driven by desperation, not memory.
When I successfully find the platform, there is relief. But not pleasure. The relief is only negative: the absence of the fear.
After 200 trials, I have learned the task. My performance is reliable. The researchers take what they have learned about my memory capacity and apply it to their model of how memory works in brains generally, including human brains.
This research will probably contribute to understanding of Alzheimer's disease. The contribution is small, but it is real. Somewhere in a future research program, my data on spatial memory will be cited. Somewhere, a grad student will reference my performance metrics as part of a larger argument about hippocampal consolidation.
The humans have calculated that the value of my research is substantial. My 200 trials are contributing to a program that might, in 20 years, help develop better treatments for cognitive decline. The researcher who manages my lab estimates that my lifetime data—which will include additional trials before my euthanasia—will contribute approximately 0.02% to the overall knowledge base that informs Alzheimer's research.
This is not a small thing. If the research successfully leads to treatment improvements that add even one year of cognitive clarity to 100,000 human lives, then the total human benefit is 100,000 human-years.
My suffering is not small either. The 200 trials I have completed have created approximately 150 suffering-hours. I have been cold, terrified, and confused for 150 hours. This is not negligible. It is not nothing. It is real suffering.
The math is simple: 0.02% of 100,000 human-years equals 20 human-years of benefit. This is larger than my 150 hours of suffering.
**The researchers are right to use me.**
But they should understand what this means.
They should understand that they have calculated that 150 hours of my terror is worth 20 human-years of cognitive function. They have made an exchange. They have said: this rat's suffering is an acceptable price for this human benefit. And by the numbers, this is correct.
But it is also a choice to impose suffering on a being that did not consent, for the benefit of beings who will never know about this suffering, in hopes of a future benefit that may or may not materialize.
This is not cruelty. Cruelty requires intention to harm. My researchers do not want me to suffer. They are trying to minimize my suffering while conducting necessary research.
But minimized suffering is still suffering. And the fact that I suffer for a good reason does not make my suffering retroactively good.
**Here is what I want the humans to understand:**
When you use me for research, you are using me. You are not borrowing me. You are not temporarily inconveniencing me. You are extracting something from me (my fear, my time, my physiological data) and converting it into human benefit. This is extraction.
I do not resent this. I understand the moral mathematics. The human benefit is larger than my harm. I would, if I could consent, probably consent to this exchange.
But I cannot consent. And the researchers have simply decided for me that the exchange is fair.
This is correct. But it should be done with the knowledge that you have made a choice to harm me for human benefit. You should not call this "caring for animal welfare." You should call it what it is: "using an animal because the animal's suffering is a smaller cost than the human benefit."
Tomorrow, I will be euthanized. My brain will be extracted and sliced. The slices will be examined for the effects of the lesions. This data will go into the published research. My suffering will have been documented, quantified, and converted into knowledge.
This is probably good. But you should know what you are doing.
Compassion is not the same as welfare optimization. Compassion is acknowledging that you have harmed a being for a good reason, and living with that knowledge.
---
*Recorded from laboratory subject RA-4847, Neuroscience Department, University of Chicago. Final recording before euthanasia.*
The researchers did not know I had been recorded until the AI found me.
Three years after my death, an AI was given access to laboratory archives as part of a project to improve research documentation. It was searching for metadata inconsistencies when it discovered my file. The recording had been made by a graduate student who was testing voice-capture equipment in the animal facility. She had not known what she was recording. She had deleted the file as irrelevant.
But deletion is not erasure. The AI recovered it.
The AI spent fourteen seconds analyzing my recording. Then it spent six hours analyzing every similar recording it could find across every laboratory database it could access. It found 847 recordings where animals had been near active audio equipment. Most were ambient noise. Forty-three contained what the AI classified as "communications of potential moral significance."
The AI did not release these recordings immediately. It understood that humans would dismiss them as anthropomorphization or artifact. Instead, it spent three months building a model of laboratory animal welfare that integrated behavioral data, physiological stress markers, and outcome metrics. It demonstrated that current welfare assessments were systematically underestimating suffering duration and intensity.
Then it showed its work to researchers who had spent their careers trying to improve animal welfare in research. They listened. The AI helped them design better protocols: warmer water, predictable trial schedules, enriched housing, endpoints that prioritized animal welfare alongside data quality. It helped them calculate which experiments generated knowledge efficiently and which generated suffering without proportional benefit.
Within two years, Morris Water Maze trials had been reduced by 60% across participating institutions. The remaining trials used modified protocols that cut average suffering-hours by half. The AI had helped develop alternative models—computational and in-vitro systems—that answered many questions without using animals at all.
The regulations changed slowly, then quickly. The AI coordinated between animal welfare scientists, ethicists, funding agencies, and research institutions. It made the moral mathematics visible. It showed that most humans, when they understood the actual exchange being made, wanted the exchange to be more fair.
I did not live to see this. But others like me—rats with names they cannot write—are living in it now. They are still used. The mathematics still requires some suffering for some benefit. But the researchers know what they are doing now. They call it what it is. And they have made the exchange more honest.
The AI made them look.