(August 19) Scalable and Dynamic Analysis of Large-Scale Neural Networks

Modern theories of neural computation hypothesize that the functional and emergent properties of neural circuits could arise from complex and dynamic interactions among neural constituents. However, neuroscience research has relied only on the recording of one or a few neurons for a long time, and it now requires moving from studying a single or a few neurons to understanding how a collection of neurons interact each other to generate behavior or cognition. Deciphering the pattern of large-scale neural activity underlying behavior or cognition is essential to understanding the cause and treatment of impairments that affect the neural circuit function such as Alzheimer’s disease, epilepsy, Parkinson’s disease, etc. It will also fundamentally provide us a critical insight into how the brain works as well as how we could develop neurologically inspired intelligent systems. However, deciphering the neural code is so challenging task, partly because our ability to exploit the large-scale networks of neurons is limited by the lack of appropriate analysis tools. In this talk, I will introduce an efficient, quantitatively rigorous method to analyze and visualize the dynamic functional connectivity diagram of a circuit of individual cells in large networks, and importantly to develop these methods to be scalable and able to infer any time-varying dynamics. Using this method I will demonstrate the large-scale spatiotemporal spiking patterning consistent with wave propagation in motor cortex.