mouseAn experiment: what happens to a model of Stanford's human-cortex mouse when a memecoin is its only input
We built a model of the brain described in Nature, connected its mouse and human parts, and fed it every buy and sell of $MOUSE. The experiment runs continuously, and the mice change with what the market does to them.
Enter the cagehuman graft · cortex and hippocampus
Waiting for the first event.
Written by an AI model from the event data on the right. The brain activity is a model, the trades are real.The experiment so far
Each trade changes two learned values in every model mouse. Buys build an association between the market and reward, and sells build fear. The values are stored, so every visitor sees the same four animals with the same history. In the cage, a mouse that has learned reward runs to the food sooner and its graft bursts more often, and a mouse that has learned fear stays near the walls and freezes for longer. These learning rules are ours and are not taken from the study.
What the Stanford team reported
The paper, "Developmental xenocortication using human-derived organoids in mice", was published in Nature on 16 September 2026 by Sergiu Pașca's laboratory. The figures below are taken from the paper and from the reporting on it. They are not produced by this site.
of the cortex and hippocampus never forms in the engineered mice, about 14 million mouse neurons. Roughly half of the brain's volume is left as a fluid-filled space.
human cortical neurons, grown from stem cells as organoids and transplanted into the newborn mice.
transplants took hold. Over three months the human tissue grew almost fivefold and filled more than 90 percent of the cortex.
Large synchronous bursts in the human neurons recurred every few minutes and coincided with movements of the mouth and face.
Mice without a cortex did poorly in a maze test. Mice that received human cells did better, and their social behaviour improved.
The grafts produced many human cortical cell types, including layer 5 projection neurons and von Economo neurons, which are hard to grow in a dish.
Under low oxygen the human tissue reacted the way injured human brain tissue does, which ordinary mouse brains do not show.
The grafts contain no inhibitory (GABAergic) neurons and do not form the layered structure of a normal cortex.
"One of the things that I see as a very clear red line is doing this experiment in a primate."
Sergiu Pașca, Stanford University, to MIT Technology Review"My instinct is to err on the side of caution with research like this rather than to be more permissive."
Nita Farahany, professor of law and philosophy at Duke Law, to NPRHow the model behaves
The Stanford team has not published its recordings, so the activity here is a model built from the paper's description. The human graft has no inhibitory cells, so its excitability rises until all of it fires together, as in the study. In the mice this happened every few minutes. Here it is shortened to about 25 seconds so it can be watched. The rule that a buy raises excitability and a sell lowers it is ours.
How trades reach the brain
Trades are read directly from Solana. Each buy raises the graft's excitability by an amount that grows with the size of the buy, and a large enough buy starts a burst at once. Each sell lowers the excitability, which delays the next burst.
Log of AI descriptions
Each entry describes one event in the model: a burst, a buy or a sell. The sentences were written by an AI model and are filled with the actual values of the event. They describe the model and the trades, and they make no claim about what a real animal experiences.