Exploring the neural mechanisms underlying the speech processing and phonological deficits that characterise individuals with dyslexia

Abstract

The primary aim of this thesis is to gain deeper insights into the neural mechanisms underlying dyslexia within the framework of Temporal Sampling (TS) theory, by focusing on adults. The study involved 48 adult English speakers, comprising 24 individuals with dyslexia and 24 controls without dyslexia. A comprehensive methodological approach was adopted by combining several behavioural tasks with five distinct EEG paradigms. The behavioural tasks included the Test of Word Reading Efficiency - Second Edition (TOWRE-2), two versions of Rapid Automatized Naming (RAN 1 and RAN 2), Rise Time Sensitivity, Syllable Stress Perception, Phoneme Deletion, Digit Span, Matrix Reasoning, and the Test of Written Arithmetic (TAN). TOWRE-2 was used as an inclusion measure, while Matrix Reasoning and TAN served as control measures. As predicted, significant group differences were observed in RAN 1, RAN 2, Rise Time Sensitivity, Syllable Stress Perception, Phoneme Deletion, and Digit Span, while the two groups performed comparably on Matrix Reasoning and TAN tasks. Using a Support Vector Machine approach, we found that TOWRE-2 measures were the strongest predictors followed by RAN and Digit Span tasks, while Matrix Reasoning and TAN tasks had minimal predictive value. The EEG paradigms included Resting State, Story Listening, Word Listening, and two Rhythmic Audio-Visual paradigms presented at 2 Hz and 1.5 Hz. These paradigms were designed to evaluate a broad spectrum of cognitive and sensory processing abilities. Analyses focused on the following neural measures: neural phase entrainment (delta, theta, beta, low gamma), cerebro-acoustic coherence (delta and theta), band power (delta, theta, beta, low gamma), event-related potentials (ERPs), and cross-frequency coupling, including phase-amplitude coupling (PAC) and phase-phase coupling (PPC). PAC and PPC analyses were performed across all paradigms for the delta-theta, delta-beta, delta-low gamma, theta-beta, and theta-low gamma pairs. Additionally, backward and forward Temporal Response Function (TRF) models were used to evaluate the accuracy of decoding and encoding of low-frequency speech information, providing insights into how the brain processes and represents speech signals. For the Resting State, a significant reduction in theta-low gamma PAC was observed for the dyslexic group. However, no significant differences were found in PPC, other PAC, and band-bower measures. In the Story Listening paradigm, individuals with dyslexia exhibited reduced cerebro-acoustic coherence in delta and theta bands. They demonstrated comparable speech decoding accuracy to control participants (within-participant analysis), but atypical decoding in the delta band (between-group analysis). They also showed less accurate and atypical speech decoding in the theta band. Significant differences were also observed in delta-band power during story listening in the right temporal regions and across the whole brain. However, PAC and PPC measures did not differ significantly between groups for any frequency pairs. In the Word Listening paradigm, individuals with dyslexia demonstrated significantly lower encoding accuracy in the delta and theta bands and lower cerebro-acoustic coherence in the delta band. They also showed significantly lower delta-beta PAC than control participants. No significant differences were found for other PAC, PPC measures, and band-power measures. For the 2-Hz audio-visual paradigm, individuals with dyslexia exhibited atypical phase entrainment in the beta and low gamma bands. Significant phase entrainment was observed in the delta and theta bands. Additionally, the preferred phase of entrainment in the theta band differed significantly between the two groups. Individuals with dyslexia also demonstrated atypical P2 ERP components. Significant group differences were observed in delta-band power in the right temporal region and across the whole brain. However, PAC and PPC measures did not differ significantly between groups. For the 1.5 Hz audio-visual paradigm, both groups showed significant neural phase entrainment in delta and theta bands, with no consistent beta-band entrainment. Control participants exhibited consistent phase entrainment in the low gamma band, whereas the dyslexic group did not. Band power analyses revealed no significant differences between groups in the delta, theta, beta, or low gamma bands. Similarly, PAC and PPC measures across all tested frequency pairs showed no significant differences between dyslexic and control groups. This PhD thesis makes a significant contribution to the understanding of dyslexia by uncovering critical neural differences between adults with and without dyslexia, particularly in how their brains process speech and language at different temporal scales. Through the integration of behavioural data and EEG paradigms, this work extends the scope of TS theory by providing compelling evidence of its applicability in adults with dyslexia. These findings also highlight the role of higher frequency bands (beta and low gamma) in dyslexia. Importantly, this research lays the foundation for future studies to further explore neural phase entrainment, band-power, ERP components, and cross-frequency coupling, cerebro-acoustic coherence, decoding accuracy, and encoding accuracy as biomarkers for dyslexia.