Biological Processing Units: Leveraging an Insect Connectome to Pioneer Biofidelic Neural Architectures

2025-07-14

AI Summary

This study investigates whether biologically evolved neural architectures—specifically, the complete larval Drosophila connectome—can support artificial intelligence computation. Method: We first directly translate the full insect connectome into a fixed-topology bio-inspired processing unit (BPU), constructing a recurrent neural architecture with high biological fidelity; we then integrate graph neural networks (GNNs), convolutional neural networks (CNNs), and minimax search to enable multimodal task adaptation. Contribution/Results: We propose structured expansion and modality-specific ablation analysis to systematically validate BPU efficacy: MNIST accuracy reaches 98.0% and CIFAR-10 achieves 58.2%; GNN-BPU attains 60.1% move-selection accuracy on ChessBench; lightweight CNN-BPU outperforms same-parameter Transformers; and depth-6 minimax reasoning achieves 91.7% accuracy. These results demonstrate that evolutionarily optimized neural circuits possess substantial potential for AI applications.

Technology Category

• Machine Learning: Evolutionary Learning
• Search and Optimization: Evolutionary Computation
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Models for Web evolution

Abstract

The complete connectome of the Drosophila larva brain offers a unique opportunity to investigate whether biologically evolved circuits can support artificial intelligence. We convert this wiring diagram into a Biological Processing Unit (BPU), a fixed recurrent network derived directly from synaptic connectivity. Despite its modest size 3,000 neurons and 65,000 weights between them), the unmodified BPU achieves 98% accuracy on MNIST and 58% on CIFAR-10, surpassing size-matched MLPs. Scaling the BPU via structured connectome expansions further improves CIFAR-10 performance, while modality-specific ablations reveal the uneven contributions of different sensory subsystems. On the ChessBench dataset, a lightweight GNN-BPU model trained on only 10,000 games achieves 60% move accuracy, nearly 10x better than any size transformer. Moreover, CNN-BPU models with ~2M parameters outperform parameter-matched Transformers, and with a depth-6 minimax search at inference, reach 91.7% accuracy, exceeding even a 9M-parameter Transformer baseline. These results demonstrate the potential of biofidelic neural architectures to support complex cognitive tasks and motivate scaling to larger and more intelligent connectomes in future work.

Problem

• Investigates if insect brain circuits can support artificial intelligence
• Tests biofidelic neural networks on image and chess tasks
• Explores scaling connectome-based models for better performance

Innovation

• Convert Drosophila connectome into Biological Processing Unit
• Scale BPU via structured connectome expansions
• Combine BPU with lightweight GNN for chess