Structure as Computation: Developmental Generation of Minimal Neural Circuits

This repository contains the official implementation of the paper:

"Structure as Computation: Developmental Generation of Minimal Neural Circuits"
Zhou Duan (Independent Researcher), 2026.

🧠 Overview

This code simulates cortical neurogenesis from a single stem cell using gene regulatory rules derived from mouse single-cell transcriptomic data. The developmental process generates a minimal 85-neuron circuit that achieves rapid learning on MNIST and CIFAR-10 without any architectural modification.

🧬 Network Structure

The pre-generated network (grown_neural_network.json) contains a developmentally generated topology with the following cellular composition:

Cell Type	Count	Proportion
Neuronal progenitor	4,046	80.9%
Oligodendrocyte progenitor	431	8.6%
Stem cell	315	6.3%
Undefined	123	2.5%
Neuron (mature)	85	1.7%
Total	5,000	100%

The 85 mature neurons form 200,400 synaptic connections (average degree: 4,715).

This structure emerges from a developmental simulation of cortical neurogenesis, as described in the paper.

📦 Repository Structure

developmental-nn/
├── train_mnist.py # Training script for MNIST
├── train_cifar10.py # Training script for CIFAR-10
├── requirements.txt # Python dependencies
└── grown_neural_network.json # Pre-generated 85-neuron circuit (optional)

📦 Quick Start

1. Extract grown_neural_network.zip
2. Install dependencies: pip install -r requirements.txt
3. Train on MNIST: python train_mnist.py
4. Train on CIFAR-10: python train_cifar10.py