Postersession 1
Poster #: 94
Topic: Neuronal models
Wednesday, Sep 9, 2015
1st floor

High-resolution reconstruction of auditory mismatch generators using fused EEG/MEG and group inversion

Françoise Lecaignard1, Olivier Bertrand1, Sébastien Daligault2, Anne Caclin1, & Jérémie Mattout1

1DYCOG Team, CRNL, Bron, France
2MEG department, CERMEP, Bron, France

Deviant stimuli violating a regular auditory stream elicit the well-known Mismatch Negativity (MMN) whose psychophysiological underpinnings still remain debated. Recent findings in auditory deviance studies, namely earlier mismatch responses than the MMN (Escera et al, 2014) strengthen the view that deviance processing is achieved along the hierarchical auditory system. Spatially characterizing this hierarchy would benefit to the many fundamental and clinical approaches relying on mismatch responses.

Therefore, localization methods combining fine spatial and temporal resolutions are required. In SPM software, source reconstruction from scalp recordings rests on Bayesian inference (Mattout et al, 2006) and has recently been enriched with two remarkable advances: group-level inference (Litvak et al, 2008) and fused EEG-MEG inversion (Henson et al, 2009).

We combined those advances to locate early and late mismatch cortical generators elicited by frequency and intensity deviances. Bilateral sources in supratemporal cortex and inferior frontal gyrus were found for both features. Interestingly, a fine dissociation could be observed within the supratemporal plane: while intensity deviance activated the Heschl Gyrus (HG) for early and late responses, results using frequency deviance suggest activations starting within HG and progressively expanding anteriorly. Beside, using bayesian model comparison, we show that fused inversion provides an increased spatial resolution compared to unimodal ones.

Our findings provide empirical support for fused inversion using simultaneous EEG and MEG recordings. The fine spatial description of the auditory cortical hierarchy achieved here represents a crucial step prior to further hypothesis testing regarding the neurophysiological and computational mechanisms behind mismatch responses.