Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

Oscillatory Bursts as Neural Markers of Predictive Speech Processes

Poster Session F - Tuesday, March 10, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballroom

Lindsey Power¹ (), Sylvain Baillet¹; ¹McGill University

Predictive speech processing relies on oscillatory dynamics in the auditory and higher order language regions, with theta (4-8 Hz) oscillations tracking syllabic rhythms and supporting bottom-up auditory signalling, and beta (15-30 Hz) oscillations mediating top-down inhibitory control and prediction. These signals often manifest as transient bursts with precise temporal characteristics offering insight into neural timing. However, the functional roles of bursts in continuous, natural speech processing remain unresolved. Our work investigates beta and theta bursts in naturalistic speech processing using MEG data from 11 healthy participants listening to TED talks. An artificial neural network predicted upcoming words, and outputs were used to derive proxy measures of contextual uncertainty and lexical surprise for each word. We mapped cortical sources of MEG data and detected beta and theta bursts in 200 regions of interest. Throughout natural speech, beta and theta bursts changed with auditory input, with beta bursting increasing in bilateral auditory and language regions during pauses and theta increasing in similar left-lateralized regions during word presentation. Furthermore, uncertainty and surprise of presented words modulated bursting responses: predictable words elicited early beta burst suppression and reduced theta bursting, while surprising words elicited delayed but prolonged beta burst suppression and increased theta bursting. Spatial analysis revealed both bottom-up and top-down beta and theta burst propagation relative to the auditory cortex depending on contextual uncertainty. Our results shed light on the temporal dynamics of speech processing, demonstrating complementary roles of beta and theta bursting in signalling predictive speech features.

Topic Area: LANGUAGE: Lexicon