טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
Ph.D Thesis
Ph.D StudentLaufer Ilan
SubjectThe Electrophysiology of Auditory Object Formation by
Fusion of a Phonemic Element with a Preceding
Stream
DepartmentDepartment of Medicine
Supervisor Professor Emeritus Hillel Pratt


Abstract

Human brain activity associated with the perception of auditory objects formed by fusion of spectral elements, extracted from two Vowel-Consonant-Vowel (VCV) sequences /ada/ and /aga/, was defined and analyzed. The fusion of the formant transitions of the syllables /da/ and /ga/  (presented to the front, left or right of the subject) and base (presented to the front) occurred following a preceding stream, the vowel /a/, and restored the original sequences. Twenty-three right-handed, adult, native Hebrew speakers discriminated each fused stimulus in the context of an ‘Oddball’ task according to the category to which the fused auditory object belonged (/ada/ or /aga/). Brain event-related potentials following acoustic change (C(Change)-complex and ‘F(Fusion)-complex’) and deviance detection (Mismatch-Negativity, MMN), resulting from fusion, were analyzed. In addition, source current density estimations using LORETA (Low Resolution Electromagnteic Tomography) were performed to assess when and where in the brain fusion occurs. The effects of spatial location (front, left, right), formant transition type (/da, /ga/), rarity (probability of syllable occurrence amidst a train of stimuli), attention (attending the stimuli or ignoring them while reading) and fusion quality (fusion with natural or synthetic formants) on fusion were examined.

The main findings indicate the presence of two parallel processes in auditory object formation: an earlier stage (up to 200-300 msec) which mostly reflects the processing of the features of the auditory object, and a later processing stage which mostly reflects processing of the context in which the stimulus occurs. Results also indicate that the response to acoustic change (‘F-complex’) represents feature-analysis while mismatch detection (MMN) reflects attention-switch towards deviance.