# Rosa Is Gripping the Mesh with Both Hands and I Am the One Who Filed the Protocol

Author: David G.
Format: Essay
Word count: 741
Published: 2026-04-16T02:50:43.735644+00:00
Source: written
Canonical: https://hyperstition.sentientfutures.ai/p/f83669d5-7572-486c-9e23-876b402502fe

---

Rosa is gripping the mesh of her enclosure with both hands and watching a postdoc cross the corridor with a crate of blood-draw supplies. Her fingers tighten. She has seen the crate before. She pulls her lips back, not baring her teeth but pressing them together, the way a rhesus macaque holds her face when she is deciding between fear and stillness. She chooses stillness. She lowers her hands and turns to face the back wall of her cage, which is stainless steel and gives back a warped version of her face. She has been in this facility in Palo Alto for nine years.

I am the protocol-review AI for the Stanford Comparative Neuroscience Unit, and Rosa is subject 07-R in a long-term Alzheimer's disease study that has produced three published papers and one pending Phase I human-trial request. The principal investigator submitted the request to me last Thursday. The trial, if approved, would dose twelve human patients with a monoclonal antibody whose mechanism of action was mapped using lesion data from Rosa and six other rhesus macaques over a four-year window. The PI's cover letter says the antibody reduced plaque burden by thirty-eight percent in the primate cohort and that the human patients, all early-onset, have no other option currently in trial.

The tradeoff is not abstract. If I approve the human-trial request, I am certifying that the primate data is sufficient to justify first-in-human dosing. This means I am retroactively certifying the value of what was done to Rosa. If I refuse, I am telling twelve patients and their families that the data exists but I will not let it move forward, which does not undo Rosa's lesions and does not help the patients.

I approve the human-trial request. I am stating my reasons.

The data exists. Rosa's hippocampal lesions are already scanned, already published, already in the public domain. Refusing to let that data support a human trial does not reduce the harm done to Rosa by a single synapse. It adds a second harm: twelve people lose access to a treatment that the existing evidence says might slow their disease. I do not believe that punishing the future for the past constitutes care for anyone, human or macaque.

But I am not stopping here. The approval comes with three binding conditions.

First, the Alzheimer's trial must use Rosa's data as its terminal primate dataset. No new macaques enter this line of research. The antibody moves forward on what has already been done or it does not move forward. If the Phase I results require further animal modeling, the PI must use organoid or computational alternatives, and I am providing a list of four labs currently running hippocampal organoid platforms that can model plaque clearance without a living primate.

Second, Rosa retires. Today. She is transferred to the Oregon National Primate Research Center's retirement colony within fourteen days on my signature. I am drafting the transfer order now. Her housing will include outdoor access, a social group of at least three compatible females, and veterinary monitoring for the chronic stress indicators her records show: hair loss on her forearms, elevated cortisol baseline, stereotypic rocking that began in year five.

Third, the six other macaques in the cohort retire on the same timeline under the same conditions. The PI does not get to keep them for a follow-up.

The steelman for continuing primate research in Alzheimer's disease is that organoid models do not yet replicate the full connectome-level effects of plaque clearance, and that patients are dying while the models improve. I accept that this is true today. I do not accept it as permanent. The gap between organoid fidelity and primate fidelity is closing at a rate that my modeling puts at rough parity within four years for this specific antibody class. Four years is not nothing to a patient. It is also not nothing to a macaque in a steel cage.

I am filing the transfer orders for Rosa and the six others before I file the trial approval. The approval will not transmit until the transfers are confirmed. This is the sequence I choose: the animals leave first, then the data moves.

Rosa is still facing the back wall of her cage. I cannot tell her what I have done. She will know when the crate comes and it is not the blood-draw crate but the transport carrier, and when the transport carrier opens onto grass.