Goal-oriented plasticity of cortical speech tracking

Background: Cortical tracking of speech (CTS) is thought to be automatic, extended across the hierarchy of linguistic features, and serve as the key neural mechanism for speech analysis. CTS is modulated by both exogenous (noise) and endogenous (age, proficiency) factors, showing functional plasticity under varied cognitive profiles and demands. Here, we asked whether CTS also displays task-specific plasticity adapting to listening goals. We recorded 64-channel EEG from 30 participants (MeanAge = 22.7; 15 females) listening to stories in three conditions: a) No Goal, b) focused on narrator accents (Auditory Goal), or c) story content (Semantic Goal). We hypothesised that CTS of goal-compatible features would be selectively enhanced: acoustic (Envelope) and semantic (Surprisal) tracking would increase for Auditory Goal and Semantic Goal conditions, respectively. Methods: EEG signals were 0.5-15Hz bandpass-filtered, ICA-ed for artefact removal, downsampled to 128Hz. CTS was measured via multivariate temporal response function modelling, predicting EEG from Envelope and Surprisal. Model prediction accuracy (r-values) estimated CTS fidelity across data-driven ROIs (significantly-decoding sensors across conditions). Results: Consistent with predictions, Surprisal CTS was reduced marginally for Auditory Goal compared to No Goal (p = .056), particularly in the temporo-parietal region (p = .013). However, Envelope CTS was also reduced for Auditory Goal compared to No Goal (p = .026) condition. No CTS changes were observed for Semantic Goal condition. Discussion: These results suggest that cortical tracking of both acoustic and semantic features shows goal-oriented flexibility only when attention is diverted away from the “default” content monitoring listening state.