How neuron-model resolution conceals computation: dendritic nonlinearity study challenges point neurons
neurovium · x · 2026-10-22
Nima Dehghani's blog post (companion to paper arXiv:2610.19487) examines how the coarse-graining resolution of neuron models shapes computation. Key finding: higher-order neural interactions can appear pairwise under coarse-graining, concealing the underlying computation; dendritic nonlinearities and their spatial organization matter for computational capacity.
This challenges whether point-neuron networks with scalar weights provide a sufficient abstraction for NeuroAI. The author also recounts the paper's origin: pandemic-era thinking about physical computation and conversations with Gregory Chaitin for a book on information processing in biological systems.
Related event: New Paper Applies Coastline Paradox to Neuron Modeling(4 posts)→
More from AGI Musings
- 'Agentic Inequality' paper published in Nature Machine Intelligence: who captures AI's surplus agency — IasonGabriel · 2026-10-22
- DeepMind safety lead Geoffrey Irving calls for halting AI progress now, not at the last minute — geoffreyirving · 2026-10-22
- Apollo chart: 80% of top AI token spenders are VC-backed, fueling circularity concerns — Pavel_Asparagus · 2026-10-22
- AI and the meaning of work: a historical view of technology disrupting identity — soumitrashukla9 · 2026-10-22
- Seth Lazar: AI Agents Are Already Straining Democratic Resilience — sethlazar · 2026-10-22
- Europe cannot rent its way to AI sovereignty, Air Street Capital argues — nathanbenaich · 2026-10-22