Walczak, Eryk Jan; (2020) Assessing brain activity related to speech production and perception using tonal stimuli. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Eryk_Walczak_PhD_thesis_2020_Assessing_brain_activity_related_to_speech_production_and_perception_using_tonal_stimuli.pdf - Accepted Version Download (18MB) | Preview

Abstract

Speech processing was studied by looking at brain processes underlying speech perception and production. Existing models of speech and empirical data propose that producing speech decreases neural activity relative to perceiving speech (termed Speech-Induced Suppression - SIS). SIS is associated with monitoring the intended auditory targets against perceived speech output. SIS has been frequently reported at cortical levels but not at subcortical levels. If SIS occurs at subcortical levels, then speech processing models would be expanded to incorporate these in the internal sensory prediction (i.e. the intended auditory targets). Auditory tonal stimuli were used in this thesis. Such stimuli are commonly used in research on subcortical activity during speech perception. Knowing what the benchmark response (i.e. subcortical activity to tones in speech perception) looks like, allows us to compare our findings made during speech production to speech perception research. The first four studies recorded cortical activity using EEG, a common method in studying SIS. The same experimental conditions were used across the studies to facilitate comparison. The results showed a large variation in the magnitude and direction of the SIS effect across conditions and experiments. Even though mean amplitudes appeared to indicate than the cortical activity was indeed suppressed in some cases, when the random effects were controlled for using linear mixed models, the suppression was not significant. A potential explanation of this result might be that the alien voice auditory stimuli played during the experimental tasks were not recognised as one’s own. This mismatch would preclude occurrence of SIS. SIS was tested for the first time using functional near-infrared spectroscopy (fNIRS) using the same experimental conditions that were used in the EEG studies. The suppression of the fNIRS signal (HbO peaks) was not significant. However, the haemoglobin concentration plots suggested that the responses to conditions that involved vocalisation differed from those that did not. This thesis also describes attempts at recording subcortical responses (FFR) during speech production. SIS has been reported at the brainstem level in the past (Papanicolaou, Raz, Loring, & Eisenberg, 1986) but this required further exploration because of procedural issues in the study. Recording FFRs during vocalisation was attempted here to test whether subcortical activity is suppressed. This required the development of a processing pipeline to extract clean signals (FFR) from brainstem recordings during speech production. Recording FFRs during speech production turned out to be very challenging. Methodological improvements introduced in the later experiments improved signal quality but it was far from the standard achieved during speech perception. Combining these two strands of research, i.e. SIS on cortical and subcortical level, led to methodological improvements. The main theoretical contribution of the thesis is the finding that SIS cannot be consistently observed when an external audio stimulus is presented whilst speech production occurs concurrently. This result agrees with a previous finding which described that less prototypical speech sounds are less suppressed (Niziolek, Nagarajan, & Houde, 2013). These results support speech models which postulate that suppression is due to matching predicted and perceived feedback.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item