A fruit fly’s escape circuit meets DeFi. Can you keep its position alive?
You start with a pretend loan of 7,000 dollars, backed by 5 ETH. Your goal: survive three minutes of falling prices while using as little extra money as possible.
No wallet needed. No real money in practice.Collateral is the security deposit backing a loan. Its price can fall even though your debt—what you owe—stays the same.
Above 1.00: there is a safety buffer. A bigger number means more room for price drops.
This slider is a separate example; it does not change your game. The 78% rule is specific to this challenge.Think of this as a tiny alarm system. Two groups of cells sense danger, then another group sends an escape signal. You do not need to remember the scientific names to play.
One input channel to the alarm. In the game, more danger makes these simulated cells fire more often.
104 lobula columnar type 4 cells. In real flies, this pathway carries information about the expansion speed of an approaching object.
A second input channel. It receives the same game-danger signal, but connects through its own recorded biological wiring.
210 lobula plate / lobula columnar type 2 cells. In real flies, they detect outward expansion and contribute information about an approaching object’s apparent size.
The output signal. If it is strong enough and the loan needs protection, our pilot rule can add collateral or repay debt.
Two giant-fiber descending neurons. They receive visual input and help trigger a rapid escape takeoff in flies, sending signals toward body motor circuits.
An alarm, not a dopamine hit. Dopamine is a chemical messenger; this app does not simulate dopamine or reward feelings. The phrases above are friendly labels, not thoughts the fly is having. We substitute market danger for visual input and apply a designed financial rule to the simulated escape signal.
Hz = spikes per second of simulated neural time, averaged per neuron. A group at 50 Hz averages one spike per cell in a 20 ms sampling window. The graph keeps the latest 10 game seconds. Pause the game to inspect it.
Green and teal rising mean the threat sensors are responding. Orange rising means the escape alarm is firing. A brief burst is normal; one snapshot cannot show the whole pattern.
Start practice to record the first signals.
Biology: Ache et al., 2019 ↗ · Klapoetke et al., 2017 ↗
01 · Encode dangerFalling price lowers health factor: collateral × price × 0.78 ÷ debt. Below 1.22, threat input increases. Current HF: 1.11.
02 · Sound the alarmThe two threat sensors receive danger input. Their connections carry it to the escape alarm. Current alarm activity: 0.0 Hz.
03 · Apply the ruleWith pilot on, escape alarm activity above 10 Hz and health below 1.18 trigger 0.50 vETH collateral, or up to 350 vUSD repayment if collateral runs out. Actions have a four-second cooldown.
Try a controlled comparison: run the storm with an intact circuit, reset the same seed, silence both visual outputs, and replay. Compare survival and capital used. This tests a designed game controller, not a fly’s ability to trade.
Practice above, then inspect and protect your position in the official Chainlink lending challenge. These balances are independent of the game.
1. Get gasUse a little Sepolia ETH for transaction fees. Joining mints the virtual challenge tokens; no ETH collateral payment is sent.
2. Understand the tradeoffAdding collateral preserves your loan but commits capital. Repayment lowers risk and reduces the outstanding loan.
3. Confirm your actionEach transaction asks in MetaMask. The local neural pilot never controls your wallet; confidential CRE automation is not deployed.
Selling an asset does not always return its displayed price. Compare real WETH → USDC quotes, then see how much could remain for repayment after a slippage buffer.
Ethereum mainnet data · Read-only simulation · No wallet or gas needed. WETH is wrapped ETH. USDC is separate from the game’s virtual dollars and Sepolia tokens.