|
1000
|
Abstract/Summary
|
-
Recent studies have suggested a cortical representation of speech through superposition of evoked responses to acoustic edges, an idea closely related to regression-based modeling approaches for studying cortical synchronization to speech via magneto- or electroencephalography (M/EEG). However, it is still unclear to what extent speech-evoked event-related potentials (ERPs) contribute to these techniques. The present study addressed this question by re-analyzing an EEG data set obtained during a selective auditory attention task in which participants focused on one of two competing speakers. Segmenting the EEG based on acoustic edges revealed ERPs with clear P1-N1-P2 complexes and enhanced N1 components elicited by attended streams (N1 effect). Comparisons between ERPs and regression results revealed that temporal response functions were highly similar spatiotemporally to the corresponding ERPs and that stimulus reconstruction accuracies were driven by a consistent enhancement of ERPs including the N1 effect. These observations point to a direct link between ERPs to acoustic edges in speech and the linear modeling techniques. In particular, the improvement in signal-to-noise ratio produced by consistent attention-related enhancements of the N1 component was found to be critical for achieving tracking of selectively attended speech, presumably facilitating the higher-order processing of the selected stream.
|
