Investigating speech processing with analyses of effective
connectivity in the normal and damaged brain.
Doctoral thesis, UCL (University College London).
The experiments presented in this thesis were designed to elucidate the distributed neuronal networks that support speech perception. In contrast to much of the published literature on this topic, I have chosen to do this by building and testing biologically informed models of how neuronal populations might interact to process speech stimuli. I took this approach because I believe that to gain a deeper understanding of brain function it is important to move from a topographic to a mechanistic level of description; in other words to move from asking, “Where in the brain are stimuli processed?” and move towards asking, “How does the brain process stimuli?”. The technique I have used to do this is called Dynamic Causal Modelling (DCM). This technique involves building models of how networks of neuronal populations might interact with each other to generate signals that are observable via neuroimaging methods, and to test these models against real data recorded from subjects performing tasks in the scanner. The first experiment applies dynamic causal modelling to magnetoencephalography (MEG) data to investigate the processing of meaningful speech sounds (phonemes). The second experiment applies this technique to functional magnetic resonance imaging (fMRI) data to investigate processing of the auditory, phonemic and lexicosemantic information present in speech. The results of both experiments are used to specify a novel, hierarchically organised network model of how the brain perceives and understands the sounds and meaning of speech. The final two experiments examine the functional consequences (at the behavioural and neuronal levels) of structural damage to the brain. This is done using dynamic causal modelling of fMRI data recorded from aphasic patients while they listen to speech. The results of these experiments show that structural damage to the auditory processing network identified in normal subjects results in widespread dysfunction throughout the entire speech processing network, and a significant decrease in the ability to understand the meaning of speech.
|Title:||Investigating speech processing with analyses of effective connectivity in the normal and damaged brain|
|Additional information:||Permission for digitisation not received|
|UCL classification:||UCL > School of Life and Medical Sciences > Faculty of Brain Sciences > Institute of Neurology|
Archive Staff Only