Data-Based and Theory-Based Network Models of Perturbations to Neural Dynamics
Abstract
Much of neuroscience is centered on uncovering simple principles that constrain the behavior of the brain. When considering the formation of neural architectures, similar structures can be recreated following the principles of minimizing wiring and maximizing topological complexity. However, a similar understanding of neural dynamics on top of these structural connections has not yet been achieved. One promising strategy for identifying underlying principles of neural dynamics is quantifying and modeling the response of neural systems to perturbation. Here, we use a spectrum of data- and theory-based network models to characterize the response of neural systems to different types of perturbations. We report how functional networks change in the context of pathological epileptic activity and brain-computer interface control. We also specifically test one possible principle: that activity is constrained to spread along connections in both the context of brain-computer interfaces and direct electrical stimulation. In the first study, we demonstrate across a wide variety of functional connectivity metrics and frequency bands that epileptic activity increases amplitude-based functional interactions, an observation that can now be incorporated into future theory-based models. In a second study, we determine that modeling activity that is constrained to spread along connections suggests why certain connections are important for brain-computer interface learning; specifically, these connections support sustained activity in attention regions. In our third study, we demonstrate that modeling activity changes from direct electrical stimulation using white matter connectivity explains more variance than models with rewired connections. This model generates testable predictions about which individuals, regions, and time points would lead to successful applications of direct electrical stimulation. Overall, this work demonstrates the potential uses of a range of data- and theory-based models for uncovering simple guiding principles that determine the behavior of a system. It also uses one specific principle - that activity is constrained to spread along connections - to understand the role of specific connections that may support learning, and provide a method to optimize individually tailored stimulation therapies for a specific outcome.
Subject Area
Neurosciences
Recommended Citation
Stiso, Jennifer, "Data-Based and Theory-Based Network Models of Perturbations to Neural Dynamics" (2021). Dissertations available from ProQuest. AAI28316226.
https://repository.upenn.edu/dissertations/AAI28316226