After searching thousands of learning rules, none beat backprop—and Nature confirms it's biologically plausible
aran_nayebi · x · 2026-09-16
NeuroAI researcher Aran Nayebi concludes that backprop is likely biologically plausible, implementable via vector error feedback. Key points:
- His ICML '20 work introduced the first bio-plausible rules scaling like backprop; large-scale searches over thousands of learning rules on TPUs found none beating backprop on ImageNet.
- He originally suspected the brain's "secret sauce" might be the learning rule, but backprop upper-bounds these approximations, ruling that out.
- Francioni et al., Nature 2025 finally provided experimental confirmation of vector-error feedback.
Implication: the brain-vs-network gap likely lies in architecture and objectives, not the learning rule.
More from Research
- AI researchers predicted 2054 for a Millennium Problem; reality arrived far sooner — ben_j_todd · 2026-09-16
- Paper2Agent turns research papers into interactive AI agents (Nature) — EricTopol · 2026-09-16
- CPAL 2027 heads to Tokyo: small conference on parsimony and learning — YiMaTweets · 2026-09-16
- Discrete Beckmann Transport Models: One-Step Discrete Generators Without a Teacher — rishabh16_ · 2026-09-16
- AI models are becoming appliances: great at one thing, bad at everything else — dbreunig · 2026-09-16
- Google's Fuse paper benchmarks 12 LLMs on inferring hidden social motives — dair_ai · 2026-09-16