While major progress has been made over the past decades, the dynamic behavior of the brain at large is still poorly understood. A major charter in the neuroscience community for the next decade is to create a dynamic map of brain activity – an initiative that most recently received the full support of the White House. Doing so will require the most advanced imaging capabilities operating at multiple scales of resolution – from 10’s of microns to the complete brain.
Recent advances in microscopic sensing, ULP processing, and communications are leading to brain-machine interfaces that may be able to observe thousands if not millions of active neurons in vivo. These “nanomorphic” circuits truly push the limit of nanometer scale semiconductor devices.
One of the unintended side effects of a deeper understanding of the operation of the brain is that in reverse it may lead to novel architectures and models for nanoscale information-processing systems. While the brain may not be considered “general purpose” in terms of its computational capabilities, it performs a set of functions such as feature extraction, classification, synthesis, recognition, learning, and higher-order decision-making amazingly well and efficiently. Some properties of neural processing match the features that are inherent to emerging nanoscale semiconductor fabrics: it thrives on randomness and variability, processing is performed in the continuous or discrete domains, and massive parallelism, major redundancy, and adaptivity are of essence.
Computational paradigms inspired by neural information processing hence may lead to energy-efficient, low-cost, dense, and/or reliable implementations of the functions the brain excels at.
In this presentation, we will explore both sides of this neuroscience-information technology interaction. One thing is for sure – the joint future will be exciting.