Bitcoin has no shortage of unusual traders, but a simulated fruit-fly brain might actually be the strangest one to date.
Coinbase’s software engineer Alex Wormuth gave a digital simulation based on a fruit fly’s neural wiring $100 to trade Bitcoin.
The system’s trading decisions are based on neural activity. One day after launching the experiment, Wormuth reported a rather surprising early result: the fly was up $1.
The experiment, which has been dubbed “Stonkfly,” connects an extremely complex simulation of an adult male fruit fly’s nervous system to Bitcoin price data and Coinbase. It is based on the recently released MaleCNS v1.0 connectome.
The important caveat is that there is no literal living fly sitting in front of a trading terminal (if this is not abundantly clear).
How trading works
Stonkfly takes public BTC-USDC market data from Coinbase and converts it into an RGB visual representation. That image is then fed into thousands of simulated sensory inputs that correspond to neurons in the visual system of the insect.
If the portfolio makes money, Stonkfly stimulates 15 identified PAM11 dopamine neurons. If it loses money, two PPL101 neurons associated with aversive signaling get activated.
Of course, profitable learning has not been demonstrated. A $1 gain could be simply market movement combined with arbitrary trading decisions.
Fly-brain experiments
There has been a flurry of fly-brain experiments after a major neuroscience milestone was announced earlier this month.
With the help of AI, scientists from Google Research and several other institutions produced a complete connectome of the brain and central nervous system of an adult male fruit fly. The map contains 166,700 neurons and captures the wiring extending through the brain, optic lobes, and ventral nerve cord.
The fly can be made to do a surprising number of things. Wormuth previously created DOOMFLY, which connects the same MaleCNS model to the classic video game Doom.
Other developers have already connected fly-connectome simulations to Beat Saber, Super Mario 64, Minecraft, Pong, and so on.
Of course, the researchers did not recreate an entire living brain. Neurotransmitters, chemical modulation, gene expression, and other extremely sophisticated processes cannot be fully reproduced with the aforementioned simulation.
