Semantic memory, EEG markers & cognitive decline

Abstract

This dissertation takes a theoretical approach to address the existing gap between current episodic memory theory and clinical phenotypes found on memory disorders. This issue is particularly relevant to Alzheimer's disease diagnose and to efforts on diagnosing Alzheimer's disease with electroencephalography (EEG) markers.

The dissertation uses well know computational methods and multivariate patter analysis to help start bridging the marker measures, theoretical models and the observable clinical memory disorders simultaneously.

I used secondary data from 73 healthy elderly participants that took a cross-modal old/new behavioural task while being recorded with a 128-channel EEG sensor. I will test for significant correlation between the representational similarity of the brain response and representational similarity of 110 clustered image-sound stimuli. Additionally, I carried out an event related spectrum perturbations analysis to gain information on the oscillatory nature of the semantically grouped responses.

I found a significant correlation between RSA brain response and RSA semantic memory model. In addition, the oscillatory data signals for a different response modulated by theta and alpha powers depending on the cluster presented.