A geometric and dynamical theory of latent computations in biological neural networks
Fatih Dinç, Marta Blanco-Pozo, David Alexander Klindt, Francisco Acosta, Cameron Sylber, Yiqi Jiang, Sadegh Ebrahimi, Adam S. Shai, Hidenori Tanaka, Peng Yuan, Nina Miolane, Mark J. Schnitzer
The paper introduces latent processing units (LPUs) to explain how neural networks perform reliable computations despite variability, using six theorems to unify geometric and dynamical perspectives.
bioRxiv · Jul 15, 2026 · PreprintNeuroscience