Skip to content

CodeWreckPro/bio_synaptic

Repository files navigation

🧠 bio-synaptic

A state-of-the-art interactive simulator for Organoid Intelligence (OI) biocomputing — built to be both scientifically rigorous and accessible to anyone.

Grew brain cells on a chip. Taught them to play Pong. Built a dashboard for it.

License: MIT Built with Vite


What Is This?

bio-synaptic is an interactive web dashboard simulating what happens inside a real organoid intelligence biocomputer lab — where living human brain cells grown on silicon chips are used as biological processors.

Every number in this simulator is grounded in real neuroscience:

  • The voltage traces use real Hodgkin-Huxley and Izhikevich neuron models
  • Burst detection uses the MaxInterval algorithm (same as MEA-NAP / meaRtools)
  • The Pong training loop replicates the actual DishBrain Free Energy Principle feedback
  • The ethics panel follows the Baltimore Declaration on Organoid Intelligence
  • Benchmark figures come from published Nature and Frontiers papers

🆕 New to this topic? Read EXPLAINER.md — a plain-English guide from zero to expert, written for fresh grads and curious people.


Features

🌱 Stem-Cell Grow Room

  • L-system axon branching pathfinding on an HTML5 Canvas
  • Nerve Growth Factor (NGF) injection triggers growth surges
  • Live metrics: cell count, synaptic density, myelination factor

⚡ Electrophysiology MEA (8×8 Grid)

  • 64 interactive micro-electrode channels — click to stimulate
  • Real-time burst detection (MaxInterval): burst frequency, mean IBI, network synchrony
  • Rolling oscilloscope trace
  • Spike event counter (4-second rolling window)

🔬 Dual Neuron Physics Models (toggle in sidebar)

Model Equations Accuracy Cost
Izhikevich 2 ODEs Biologically plausible spike shapes Fast (200 steps/tick)
Hodgkin-Huxley 4 ODEs (Na⁺/K⁺/leak channels) Nobel Prize-level accuracy (1963) Full (1000 steps/tick)

Both models run on all 64 electrodes simultaneously using Euler integration inside the simulation loop.

🕹️ Cognitive Conditioning (Training Playground)

  • DishBrain Pong — biological feedback loop replicating the Cortical Labs experiment
  • Logic Gates (AND/OR/XOR) — synaptic logic learning
  • Learning efficiency modulated by dopamine, GABA, and synaptic density

🧪 Incubator Life-Support

  • Temperature, Glucose, O₂, Dopamine, GABA sliders
  • Cell death triggered by starvation
  • Chemical depletion from cell metabolism
  • Seizure detection when network synchrony > 70%

⚖️ Ethics Monitor (always-visible sidebar widget)

  • IIT-Φ proxy gauge — simplified Integrated Information Theory measurement
  • Sentience risk score (0–100) — calibrated from synchrony × cell count × synaptic density
  • Baltimore Declaration compliance checklist — 5 items, live-updating
  • Welfare event log — auto-logs threshold crossings
  • Welfare levels: Safe → Monitor → Review Required → Halt Protocol

📊 Silicon vs. Biotech Benchmarks

  • Power draw comparison (organoid ~10µW vs H100 cluster ~700W)
  • Synaptic density (3D biological vs 2D silicon)
  • Carbon footprint
  • Market projections

📖 About / Explainer Tab

  • 11-section interactive accordion explaining the science
  • Covers: organoids, MEA, DishBrain, AI vs biology, transformers vs neurons, backprop vs Hebbian learning, ethics, and where this is all going
  • Written for a 15-year-old with no background

Tech Stack

Layer Technology
Framework React 18 + TypeScript
Build Vite 6
Styling Vanilla CSS (glassmorphism, custom animations)
Fonts Google Fonts — Outfit + JetBrains Mono
Icons Lucide React
Physics Custom Hodgkin-Huxley + Izhikevich Euler integration
Deploy Vercel

No Tailwind. No external charting libraries. All visualisations are raw HTML5 Canvas or CSS.


Run Locally

git clone https://github.com/CodeWreckPro/bio_synaptic
cd bio-synaptic
npm install
npm run dev

Open http://localhost:5173


Project Structure

src/
├── hooks/
│   └── useSynapticSim.ts       # Core simulation engine
│                               #   — HH + Izhikevich models
│                               #   — MaxInterval burst detection
│                               #   — Ethics metrics (IIT-Φ proxy)
│                               #   — Raster plot spike events
├── components/
│   ├── GrowRoom.tsx            # Axon branching canvas
│   ├── ElectrophysiologyGrid.tsx  # MEA + oscilloscope + burst metrics
│   ├── TrainingPlayground.tsx  # Pong + logic gates
│   ├── IncubatorControls.tsx   # Life support dashboard
│   ├── Benchmarks.tsx          # Silicon vs synaptic comparisons
│   ├── EthicsPanel.tsx         # Baltimore Declaration compliance widget
│   └── About.tsx               # Plain-English explainer accordion
├── App.tsx                     # Lab shell — model toggle, sidebar, routing
├── index.css                   # Bio-cybernetic design system
└── main.tsx
EXPLAINER.md                    # Deep-dive guide for fresh grads

Scientific Basis

Feature Real-World Reference
DishBrain Pong training Kagan et al., Nature Electronics (2022)
Baltimore Declaration ethics Bhanu et al., Nature (2024)
FinalSpark neuroplatform finalspark.com, Frontiers in Neuroscience (2024)
Hodgkin-Huxley model Hodgkin & Huxley, Journal of Physiology (1952) — Nobel Prize 1963
Izhikevich model Izhikevich, IEEE Trans Neural Networks (2003)
MaxInterval burst detection Pasquale et al., Journal of Neuroscience Methods (2010)
IIT / Φ Tononi et al., BMC Neuroscience (2008)
Energy efficiency figures FinalSpark whitepaper (2024), Cortical Labs CL-1 datasheet

Credits

Conceived, researched, and built by @CodeWreckPro — with the assistance of Antigravity, an agentic AI coding assistant by the Google DeepMind Advanced Agentic Coding team.


License

MIT — do whatever you want with it. If you publish research using this simulator, a citation would be appreciated.

About

An interactive simulation environment for cultivating biological neural networks, studying organoid intelligence, and testing organoid-AI closed-loop learning.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages