Electronic Thesis and Dissertation Repository

Resting State Network Dynamics Across Wakefulness and Sleep

Evan Houldin, The University of Western Ontario

Abstract

The function of sleep is a longstanding mystery of the brain. By contrast, the function of resting state networks (RSNs) is one of its most recent mysteries. The relationship between RSNs and neuronal activity has been unclear since RSNs were discovered during the advent of functional magnetic resonance imaging (fMRI). Somewhat paradoxically, investigating these enigmatic phenomena in parallel can help to illuminate the function of both. The three studies described as part of this thesis all involve an evaluation of RSN dynamics across wakefulness and sleep. They are all based on the same dataset, derived from an experimental paradigm in which healthy, non sleep-deprived participants (N=36, 21 female) slept in an MRI scanner, as their brain activity was recorded using simultaneous electroencephalography (EEG)-fMRI. An independent component analysis (ICA) was performed in the first study. Spatial boundaries of components in each sleep stage were compared with those of wakefulness, in the first attempt to catalogue RSNs across all healthy alternate functional modes of the brain. Against expectations, all non-wake-RSN components were positively identified as noise. This indicated that sleep is supported by much the same RSN architecture as wakefulness, despite the unique operations performed during sleep. In the second study, between-RSN functional connectivity (FC) dynamics were evaluated across wakefulness and sleep, in order to determine whether they reflect known cortical neurophysiological dynamics. This was confirmed, highlighting the connection between RSNs and neuronal activity. Moreover, the dynamic pattern suggested that one of the functions of sleep may be to homeostatically counterbalance wakefulness RSN FC. A further pattern, indicating increased FC of “higher-order” RSNs (e.g., default mode network), suggested that slow wave sleep might manifest an altered, rather than a reduced state of awareness, in contrast to historical depictions. Finally, the third study correlated frequency-banded oscillatory activity, as measured by EEG, with RSN activity, as measured with fMRI. This was done in order to track changes in representations of frequency-banded neuronal activity in each RSN across stages. It was discovered that the pattern of frequency band representation dynamics reflects the aforementioned cortical neurophysiological dynamics, further strengthening the connection between RSNs and neuronal activity.