Neural dynamics in brain networks during the resting state and visual word recognition

This thesis investigates the dynamics of information flow within brain networks during the resting state and visual word recognition. Functional connectivity within brain networks has become increasingly prominent across cognitive neuroscience and neuroimaging in recent years and conventional approaches for identifying instantaneous interactions within brain network and across the whole head are now commonplace. Magnetoencephalography (MEG) recordings have a very high temporal resolution which allows for the characterisation of delayed interactions between distant brain regions such as those caused by limited conduction speeds along white matter fibres. This thesis presents an approach to characterising such time-delayed interactions and critically, inferring the direction of information flow. This approach is used to demonstrate the existence of statistically significant differences in the information flow in each direction of a connection between two nodes in a resting state network. A Hidden Markov Model is then used to characterise dynamic changes in this directionality. Task driven directional connectivity is then investigated in the context of visual word recognition. A complex and rapidly evolving pattern of connectivity arises during visual word recognition, with specific connections modulated by the psycholinguistic properties of the stimulus. Critically, the influence from the Left Inferior Frontal Gyrus is shown to transfer more information to visual regions when reading a challenging stimulus.