| name | diffusing-blame-dale-principle-credit-assignment |
| description | Error Diffusion (ED) methodology for biologically plausible credit assignment under Dale's principle. Dual-stream excitatory/inhibitory architecture with modulo error routing. Achieves 96.7% MNIST and 61.7% CIFAR-10 under strict Dale's constraint. Integrates with PPO for RL. Trigger words: Dale's principle, error diffusion, excitatory-inhibitory, biologically plausible learning, credit assignment, dual-stream network. |
Diffusing Blame: Task-Dependent Credit Assignment in Biologically Plausible Dual-Stream Networks
arXiv: 2606.31700v1 | Date: 2026-06-30 | Venue: ALIFE2026
Authors: Yutaro Yamada, Luca Grillotti, Rujikorn Charakorn, Sebastian Risi, David Ha, Robert Tjarko Lange
Core Methodology
Problem Statement
Biological neural circuits obey Dale's principle: each neuron's synapses are uniformly excitatory or inhibitory. Artificial networks respecting this constraint must coordinate separate E/I populations, fundamentally changing credit assignment. Previous biologically plausible rules struggle to scale beyond MNIST under strict Dale's constraint.
Error Diffusion (ED) Framework
- Dual-stream architecture: Separate excitatory and inhibitory populations
- Global error routing: Error signals routed to all layers without:
- Transporting transposed forward weights (no weight transport)
- Random feedback matrices
- Modulo error routing: Extension enabling multi-class classification (beyond binary)
Three Domain-Specific Innovations
- Layer-specific sigmoid widths — adaptive gain per layer
- Batch-centered class error signals — normalized error propagation
- Asymmetric initialization — E/I balance-aware weight init
Key Results
| Benchmark | Accuracy | Notes |
|---|
| MNIST | 96.7% | State-of-the-art under strict Dale's principle |
| CIFAR-10 | 61.7% | First strong baseline under Dale's constraint |
| Brax (RL) | Competitive | ED-PPO vs Direct Feedback Alignment |
| Craftax | Competitive | Open-ended exploration task |
Critical Findings
Task-Dependent Credit Assignment
- Ablation analysis reveals reversal of innovation importance between MNIST and CIFAR-10
- Exposes task-dependent bottlenecks invisible to single-benchmark evaluation
- Implication: Multi-benchmark evaluation essential for biologically plausible learning research
RL Integration
- ED + PPO achieves competitive performance
- Demonstrates scalability to continuous control
- Benchmark: Direct Feedback Alignment (backprop-free baseline)
Pitfalls & Considerations
- CIFAR-10 gap: 61.7% is baseline, not SOTA — representation learning possible but limited
- Dale's constraint cost: Significant performance drop vs unconstrained networks
- Innovation interaction effects: Layer-specific gains + asymmetric init interact non-trivially
- RL scaling: Craftax results preliminary; full RL benchmark suite needed
Related Work
- Direct Feedback Alignment (Lillicrap et al.)
- Target Propagation
- Feedback Alignment family
- Predictive Coding networks
Activation Keywords
Dale's principle, error diffusion, excitatory-inhibitory, dual-stream, biologically plausible learning, credit assignment, modulo error routing, PPO integration
