Although humans gather most of the information about the surrounding world by visual means, our knowledge about the neural basis of visual processing under natural circumstances is still very limited. This is mainly because active vision is a complex sensory-motor process, during which saccadic eye movements that serve the scanning of visual stimuli produce significant artifacts in electroencephalogram (EEG) recordings used for non-invasive analysis of human neural activity.
A novel method for analysis of fixation-related brain activity based on hierarchical linear modeling (HLM) was developed at the Brain Imaging Centre RCNS HAS. To validate the results obtained by this new approach, detailed testing of the methodology was necessary. To this end, effects of eye-movement covariates and collinearity between independent variables were investigated using data from a natural reading experiment. Findings indicate that correcting for eye-movement effects is essential for analysis of fixation-related brain activity and the model proved to be robust against the moderate collinearity present in the analyzed dataset.
The main part of this study was the preprocessing and analysis of the experimental data acquired during a natural reading experiment conducted in the Brain Imaging Centre, RCNS HAS with the participation of 24 dyslexic and 24 control subjects. Participants read sentences presented with different letter spacing size while their eye-movement and EEG activity were recorded. HLM of fixation-related EEG activity revealed significant expertise-driven configural and visual processing load effects in the first 400 ms after fixation onset both in space-time and space-time-frequency domains. These results were in agreement with previous findings. The largest limitation of the used natural reading paradigm was the relatively low spatial resolution of EEG, which could be improved with the use of additional measurement modalities. To this end, I investigated the possibility of extending the paradigm with functional magnetic resonance imaging (fMRI), and found it to be a viable improvement. However, this method would still require extensive testing and validation before practical use.