Electrophysiological Markers of Sensory Plasticity and Connectomics in Ageing and Mild Cognitive Impairment

Abstract

Measures of perceptual learning provide a unique window on one of the brain's defining characteristics: its ability to adapt to the demands of the external world. Such measures provide insight into disease-specific disruptions of neuroplasticity that have immense implications for biomarker and intervention development. The aims of this thesis were to 1) investigate the brain's learning mechanisms using electroencephalography (EEG) and Dynamic Causal Modelling (DCM) and 2) explore the utility of these measures in distinguishing between healthy ageing, and potential latent neuropathology in those classified with amnestic Mild Cognitive Impairment (aMCI); an 'at risk' group for Alzheimer's disease. The studies of this thesis employ two independently developed EEG paradigms designed to index fundamentally distinct models of perceptual learning, namely the visual Long-Term Potentiation (LTP) paradigm as an index of Hebbian plasticity, and the roving Mismatch Negativity (MMN), as an index of Predictive Coding. In addressing aim 1, the modulation of effective connectivity induced by these paradigms was compared. Consistent with their respective learning models, the MMN was associated with a modulation of both forward and backward connectivity in a fronto-temporal network, while visual potentiation only modulated forward connections in an occipito-temporo-frontal network. These results are discussed with respect to the Free Energy Principle, which embodies both Hebbian and Predictive Coding mechanisms. In addressing aim 2, the degree to which each paradigm provides a non-invasive, objective index of neuroplasticity was explored in young, older and aMCI participants. Both paradigms revealed shifts in event related potentials (ERPs) and intrinsic and extrinsic cortical dynamics that are suggestive of an age-related refinement of learning mechanism and sensory predictive accuracy. aMCI was distinguished from healthy ageing by a specific disruption of the excitatory-inhibitory balance within sensory processing regions, and disrupted ERPs. Importantly, both paradigms afforded a unique perspective on this disruption. The results of this thesis not only shed light on how the brain learns, but also demonstrate the utility of EEG and computational modelling in developing clinically feasible indices of disrupted neuroplasticity. This may be of great value to biomarker and intervention development across a plethora of neurological and psychiatric disorders.