Le cerveau multidimensionnel

Accueil du site Neuro-PSI > L’Institut > Conférences > Neurospin

Conférence Neurospin


Lundi 6 janvier 2020 - 11h00, salle 183, Amphi NeuroSpin
Centre d’études de Saclay

« Time-locked cortical processing of continuous speech : from sound to words, and effects of selective attention »

Christian BRODBECK, University of Maryland, USA

Abstract :

In most realistic contexts, speech is experienced as a continuous acoustic signal. MEG and EEG have the temporal resolution to track neural processes with millisecond accuracy, but the inherent temporal dynamics of natural, continuous speech pose a challenge for traditional data analysis techniques. In this talk I will introduce an approach modeling MEG responses to speech as a continuous, linear response to multiple concurrent, continuous predictor variables. This approach allows studying brain responses to naturalistic speech stimuli, such as an audiobook. By generating predictor variables based on cognitive models of neural processes, we can distinguish different cognitive processes involved in speech processing.

I will first talk about isolating neural responses related to lexical processing of clean speech. A central step in speech perception is integrating the information contained in the sequence of phonemes to detect words. Behavioral research has resulted in the well established cohort model of lexical processing. Based on this model, we identified brain responses time-locked to phonemes, and associated with different aspects of how these phonemes are used to detect words. These responses provide evidence for a mechanism involving early lexical segmentation, predictive coding of phonemes and lexical competition.

These neural markers of lexical processing are then used to address long-standing questions about the processing of background speech during cocktail-party listening, while employing more naturalistic stimuli than possible in behavioral studies. Participants listened to an equal loudness mixture of a male and a female talker and were instructed to attend to one of them and ignore the other. Previous psychophysical research has yielded conflicting results on whether in such a situation the ignored speech is processed lexically. MEG responses indicate that, in our continuous listening paradigm, time-locked lexical processing is restricted to the attended speech. However, in more recent research we identified a neural representation of acoustic features of the ignored speech, which likely plays a role in the segregation of the two speech sources, and possibly also in the behavioral effects of ignored speech.

Together, these results illustrate the potential of an approach that models processing of natural, continuous speech, and its relevance for understanding real world speech comprehension.





Ajouter un événement iCal

  Webmaster Plan du site Planning Crédits
  Suivez-nous !
  Tablets & Smartphones
ICN Neuro-Cog Neuro-Dev Neuro-Syst