Quantifying Nonlinear Neural Dynamics During Development and Learning

In the last century, we have witnessed astounding technical advances in neuroscience studies. These have shown that the nervous system represents information and initiates actions through ensemble spiking activity of its neurons. However, biological processes underlying spike transmission across neural circuits are difficult to study in vivo. This is because they are highly nonlinear dynamical processes and are often time-varying during development and learning. As the amount of experimental data made publicly accessible has gradually increased in recent years, it is now possible to reconsider many of the longstanding questions in neuroscience, such as when and where neural development and learning occur. However, comparative studies which investigate changes in the neural dynamic in different parts of the nervous system do not seem to exist.The proposed study aims to develop a novel modeling methodology to identify the time-varying properties of nonlinear neural dynamics by observing neural spike train inputs and outputs only. The first objective is to estimate the time-varying nonlinear dynamic models of local neural circuits during development and learning using the new electrophysiological recordings now made publicly accessible. These data-driven models would provide valuable information to the development of treatments, and neural prosthetic devices to address neurologic diseases.The proposed project will also provide a new computationally efficient tool for neurobiologists to study local neural circuit changes in real-time during different experiments by adaptive signal processing techniques. For post-experiment analysis when statistical validation of the changes is required, the variation in model parameters will be approximated by weighted sum of orthogonal basis functions. The changes in functional input-output properties of local neural circuits will be correlated with observable changes in functional properties of the nervous system and with variations in performance and/or task variables. The final step of the project is to attempt to reconstruct time-variant connectivity models between brain regions through developmental or learning rule(s) which define how to modify the system models with input and output spike trains only. It is expected that such rule(s) would help us understand the underlying mechanisms of development and learning of the brain.