CMOS-Nano Neuromorphic Neural Interfaces

Understanding how a large number of neurons connect to create higher functions holds central importance in neuroscience and is key to bridging artificial intelligence and natural intelligence. Extracting such a functional synaptic connectivity map from biological neuronal networks is, therefore, one of the most widely pursued yet daunting tasks.

In this talk, I will present my recent work on a CMOS-nano electrode array for massively parallel neuronal intracellular recording and synaptic connections mapping. This new technique transforms parallel intracellular recording from a formidable challenge into a routine task. The array, interfaced with rat neuron cultures, obtains unprecedented network intracellular recording data, rich in connection information with copious synaptic signals. By mining and analyzing these synaptic signals, I have extracted over 70,000 connections and categorized them as inhibitory, weak/uneventful excitatory, and strong/eventful excitatory chemical synaptic connections, as well as electrical synaptic connections. This scale of mapping substantially surpasses previous work on electrode technology, while the ability to qualify--potentially quantify--synaptic connections complements electron and optical microscopy.

Speaker: Jun Wang, Harvard University