Reverse-Engineering the Wetware: Spiking Networks, TD Errors, and the End of Matrix Math
Summary
An engineer-friendly exploration of how the brain processes information, contrasting predictive coding and top-down perception with backpropagation in AI. It explains Spike-Timing-Dependent Plasticity, dopamine-driven TD errors, and the hardware implications of neuromorphic chips, with pointers to open-source projects like Nengo and pymdp. It also discusses potential implications for AI tooling, research directions, and cross-domain learning between biology and machine learning.