Flybrain AI

I replaced Minecraft’s mob AI with a biologically accurate Fruit Fly (Drosophila) Spiking Neural Network (SNN)
I’ve been working on a custom Fabric mod called FlyBrainAi that completely replaces the vanilla, blocky Minecraft pathfinding with a biological Spiking Neural Network (SNN) inspired by the Drosophila melanogaster (fruit fly) connectome.
Instead of standard A* pathfinding and static Goal logic, mobs now operate purely on membrane potentials, leaky integrate-and-fire (LIF) neurons, and multi-sensory feedback loops.

Key Features:

• Saccades & Chaotic Kinematics: Mobs no longer walk in straight, robotic paths. They navigate using sharp micro-angle deviations (saccades), sudden jittering when idle due to simulated neural noise, and instant directional lunges.
  
• Realistic FOV & Occlusion Raycasting: Mobs have a restricted 120° visual field. If you sneak up directly behind them or hide behind blocks, they won't magically know where you are.
  
• Looming Stimulus & Panics: Close-range threats trigger an instantaneous escape jump (looming response), bypassing high-level decision-making for raw instinct.
  
• Auditory & Tactile Perception: Sprinting near mobs builds sensory voltage even outside their line of sight, forcing them to turn toward sound sources.
  
• Hereditary Memory & Neuro-Evolution: Every time a mob dies, its neural thresholds, decay rates, and fear memories are persisted (config/flybrain_dna.txt). Newly spawned mobs inherit mutated parent DNA, making future generations progressively harder to catch.
  
• Cordyceps Fungal Pathology: Has a chance to override the SNN logic entirely, turning passive evasion into aggressive, mind-controlled host behavior.
  

The result is an uncanny, erratic, and genuinely insect-like mob environment where entities react organically to motion, light, and danger.
Let me know what you think, or if you have ideas for further biological mechanics to add to the neural stack!