UC San Diego
Bridging cognition and neurobiology with large-scale cortical dynamics and multimodal brain data
- Author(s): Gao, Richard
- Advisor(s): Voytek, Bradley
- et al.
Human cognition depends on complex and coordinated activity of neural populations, which are enabled by a rich diversity in cellular and network properties across the brain. To study the neural basis of cognition, cognitive neuroscience combines the analysis of large-scale electrophysiological signals—neural field potentials with contributions from millions of neurons—with behavioral tasks to observe brain dynamics during cognitive processes, such as attention, decision-making, and language understanding. This approach has been successful in discovering neural correlates of cognition, which can differentiate between behavioral states and tracking disease progression within and across individuals. However, we currently lack the ability to analyze the physiological contributions underlying these brain signals, especially in humans, thus hindering advances towards a mechanistic understanding of how cognition is linked to neurobiology, as well as how cognitive functions deteriorate with pathological changes in the brain.
In this dissertation, I leverage multimodal brain data from humans, animals, and in-vitro models, as well as novel simulation and analysis techniques to investigate the biological variables underlying oscillatory and asynchronous neural dynamics, ultimately relating them to cognition. Chapter 1 reports the emergence of complex oscillatory activity in human induced pluripotent stem cell-derived brain organoids, a model for early neurodevelopment, and the co-evolution of the cellular and synaptic properties that support it. Chapter 2 presents a computational model that accounts for changes in the 1/f power law exponent of asynchronous (or scale-free) neural activity as a shift in the balance between synaptic excitation and inhibition, which is validated on rodent and non-human primate electrophysiological data. Chapter 3 develops a novel computational technique to infer neuronal timescales across the human cortex, which follows a gradient along the sensorimotor-to-association cortical hierarchy, is shaped by variations in inhibitory synaptic and transmembrane ion channel proteins, and dynamically lengthens during working memory maintenance and shortens over aging in the long term. Together, these works present a framework for integrating open-source datasets and existing findings from cellular, systems, and cognitive neuroscience to decipher large-scale human brain recordings, thus linking brain structure and function with neural dynamics for mechanistic investigations of cognition and brain disorders.