À propos
DCAP
Thesis title: Temporal dynamics of natural sounds representations in human brain: acoustic, semantic and intermediate representations
My thesis project is part of a heavily collaborative project involving Giordano and Schön (Aix-Marseille University) and Formisano (Maastricht University). The aim of this collaborative project is to gain a better understanding of the neural dynamics of the process of transforming the acoustic representation of a sound (the representation of the sound’s physical characteristics) into its semantic representation (e.g. the representation of the source of the sound: what/who, how and where; Giordano et al., (2022)). We know that auditory processing consists firstly of acoustic analyses of the sound, in particular frequency analyses at the level of the cochlea followed by spectro-temporal modulations analysis at the level of the primary auditory cortex (A1). However, the emergence of semantic representations of sound is still under debate. They appear to be already present in the superior temporal gyrus (STG) and ventrolateral prefrontal cortex (VLPFC). They then integrate a widespread neural network associated with an amodal semantic representation of sound (Binder, 2011). In order to clarify the spatiotemporal dynamics of these acoustic and semantic representations of sound, several neuroimaging techniques are being used in this collaborative project: magnetoencephalography (MEG) for its good temporal resolution (as part of my thesis project), functional magnetic resonance imaging (fMRI) for its good spatial resolution and intracranial electroencephalography (iEEG) for its good spatio-temporal resolution and the quality of the signal obtained (much higher than that obtained by indirect measurement). My thesis project therefore focuses on the temporal dynamics of acoustic and semantic representations of sounds. The general objective is to establish spatiotemporal maps of the acoustic and semantic representations of sounds in the brain in order to identify the temporal dynamics of the transformation of the acoustic representation of the sound into a semantic representation, the brain regions involved and to study the general process of object discrimination (which semantic representations are used). To this end, we are comparing acoustic and semantic representations derived from computational models with MEG brain data, over time. A representational similarity analysis (RSA) between the sounds is first performed for each type of representation (acoustic, semantic or cerebral), then different linear regressions of the cerebral response distances are performed based on the acoustic or semantic distances from the models.