Modern computation is at an architectural crossroads.While Large Language Models (LLMs) achieve remarkable capabilities by scaling billions of parameters across massive cloud-based digital infrastructure, they operate under a fundamentally different set of physical and thermodynamic constraints than biological systems.
Silicon data centers optimize for high-frequency, synchronous, digital processing across vast distances, bound heavily by thermal dissipation and power-delivery limits. Biological brains, by contrast, navigate strict volumetric, wiring-cost, and severe metabolic constraints—solving immense computational and optimization problems on an operational budget that is a mere millionth of the energy consumed by a data center running an LLM.
The goal of this one-day symposium is to articulate the fundamental differences in how information is processed in “wet” brains versus artificial systems. We aim to flesh out the core principles that underlie biological computation—such as sparse connectivity, localized feedback, continuous-time analog dynamics, and structural hierarchy—and ask a critical, forward-looking question: how can we leverage these biological principles to build next-generation neuromorphic hardware?
Mind the Gap brings together an interdisciplinary group of scientists spanning statistical physics, network science, neuroanatomy, and neuromorphic engineering to explore these fundamental differences and propose a way forward toward a new paradigm of energy-efficient, bio plausible computation.
