Neural dynamics in auditory processing and sleep
Options
BORIS DOI
Abstract
Understanding how the intricate dynamics of brain activity support sensory and cognitive processing is a fundamental question in neuroscience. This thesis aims to elucidate the relationship between multiple timescales and frequencies of neural activity and auditory processing of different stimulus complexities and brain states. The ultimate goal is to understand how neural dynamics shape our perception and memory of auditory experiences, from simple sounds to complex melodies. The three projects utilize computational methods and machine learning to analyze intracranial electroencephalography (iEEG) during wakefulness, sleep, and auditory stimulation.
The first project investigates intrinsic neural timescales, the characteristic time of neural activity fluctuations, and how they relate to auditory processing of simple tones. Neural timescales are thought to help segment and integrate environmental stimuli; however, a direct relationship between intrinsic timescales and processing of sensory stimuli was still missing. Focusing on the auditory modality, I showed that timescales progressively lengthen along the auditory network, forming a hierarchy. Crucially, regions with longer intrinsic timescales exhibit longer response latencies to auditory stimuli, suggesting that intrinsic dynamics directly shape sensory processing and provide diverse temporal windows across the processing hierarchy.
The second project examines how intrinsic neural timescales change across the sleepwake cycle and throughout the cortex. I utilized an open dataset with thousands of iEEG channels and showed that two complementary measures of neural timescales globally lengthen during sleep compared to wake. Importantly, the two measures showed opposite hierarchical gradients across sensory and association regions. These state-dependent changes were linked to slow wave activity, revealing that neural dynamics are modulated by anatomical constraints and physiological brain states.
The third project addresses naturalistic auditory cognition by investigating music listening and recognition. While brain regions and frequency bands involved in music listening are relatively well characterized, the precise temporal dynamics of music recognition remain unknown. I showed that specific temporal lobe networks exhibited sustained power increases in low-frequency bands for recognized melodies, revealing how complex auditory memories are retrieved through spatially organized and temporally extended neural activity patterns. Additionally, I used supervised machine learning to show that stable neural representations enabled predicting whether a participant would recognise a melody or not several seconds before a behavioural report.
In summary, this thesis elucidates how neural dynamics support auditory processing and are modulated by brain states, demonstrating the power of computational methods in advancing neuroscience research.
The first project investigates intrinsic neural timescales, the characteristic time of neural activity fluctuations, and how they relate to auditory processing of simple tones. Neural timescales are thought to help segment and integrate environmental stimuli; however, a direct relationship between intrinsic timescales and processing of sensory stimuli was still missing. Focusing on the auditory modality, I showed that timescales progressively lengthen along the auditory network, forming a hierarchy. Crucially, regions with longer intrinsic timescales exhibit longer response latencies to auditory stimuli, suggesting that intrinsic dynamics directly shape sensory processing and provide diverse temporal windows across the processing hierarchy.
The second project examines how intrinsic neural timescales change across the sleepwake cycle and throughout the cortex. I utilized an open dataset with thousands of iEEG channels and showed that two complementary measures of neural timescales globally lengthen during sleep compared to wake. Importantly, the two measures showed opposite hierarchical gradients across sensory and association regions. These state-dependent changes were linked to slow wave activity, revealing that neural dynamics are modulated by anatomical constraints and physiological brain states.
The third project addresses naturalistic auditory cognition by investigating music listening and recognition. While brain regions and frequency bands involved in music listening are relatively well characterized, the precise temporal dynamics of music recognition remain unknown. I showed that specific temporal lobe networks exhibited sustained power increases in low-frequency bands for recognized melodies, revealing how complex auditory memories are retrieved through spatially organized and temporally extended neural activity patterns. Additionally, I used supervised machine learning to show that stable neural representations enabled predicting whether a participant would recognise a melody or not several seconds before a behavioural report.
In summary, this thesis elucidates how neural dynamics support auditory processing and are modulated by brain states, demonstrating the power of computational methods in advancing neuroscience research.
Year of graduation
2026
Theses Type
dissertation
Keyword(s)
neuroscience
•
iEEG
•
data science
Language(s)
en
Author(s)
Faculty/Graduate School
Access(Rights)
embargo
Primary OA Publication
true