UC San Diego

We sense, move and think by dynamical interactions between neurons and brain regions. This thesis focuses on 1) how neural activities are organized in a network level to 2) achieve essential computation for cognition. With tools from dynamical systems, probability theory, statistical learning and information theory. I approached the above two questions as below:

In Chapter 2, I developed a Dynamical Differential Covariance (DDC) to efficiently infer directed functional connectivity through neural activity recordings. It was validated and compared favorably with other causality inference methods on networks with false positive motifs and multiscale neural simulations where the ground truth connectivity was known. When applied to resting-state fMRI recordings, DDC consistently detected regional interactions with strong structural connectivity in over 1,000 individual subjects obtained by diffusion MRI (dMRI). The topological measures of DDC connectivity showed better behavioral relevance. For example, the global efficiency of DDC network is inversely correlated with reaction time in multiple psychological tasks.

In Chapter 3, I linked the brain's anatomical connectivity to its functional connectivity with temporal precision in the circuit of hippocampal formation. I proposed the notion of predicting ahead to compensate for inter-regional transmission delays in the regime of precisely regulated temporal coding. The hypothesis was supported by analyzing the spike coupling between simultaneously recorded neurons in hippocampal subregions. Further modeling work also recapitulated characteristic place cell features.