Internship

Magnetoencephalography data acquisition and analysis for voice perception

From :   to

The internship consists of acquiring and analyze Magnetoencephalography data to study voice perception in humans.

Description

In the context of the comparative study of voice perception in primates (ERC project COVOPRIM led by Prof. Pascal Belin), this project aims to investigate the cerebral representation of identity (e.g. gender, age, speaker identity) in voice perception. We already acquired functional magnetic resonance imaging (fMRI) data and obtained highly activated voice areas while passively listening to 13h of human voice. We use Deep Learning to explore the cerebral representation of identity (e.g., gender, age) in voice perception. In particular, we seek to perform decoding to find a linear mapping between the brain activity and the stimulus (Naselaris et al., 2011); such a mapping allows to reconstruct voice (Fig. 1).
Benefiting from the temporal resolution of the magnetoencephalography (MEG) neuroimaging technique, the current project aims to investigate the temporal flow of cerebral representation of identity while listening to a human voice as well as developing a Deep Learning model to reconstruct voice from the temporal brain activity.
The beginning of the project consists of preparing the stimuli for the coming MEG data acquisition. Based on the previously acquired fMRI data, an optimal subset of the stimuli will be selected, while preserving the balanced categories.
In the second place, the experimental protocol will be designed and the experiment set up. Next, 2 participants will be scanned while listening to the stimuli.
Once the acquisition phase is done, the preprocessing of the MEG data will be realized using MNE-Python software (Gramfort et al., 2013) as well as sanity checks and validation controls to demonstrate that the acquired data are proper.
At last, more advanced analyses may be done to assess our scientific questions, such as MEG forecasting (Chehab et al., 2021), temporal representational dissimilarity (BL Giordano et al., 2021; TC Kietzmann et al., 2019).

Desired profile

Strong interest in Cognitive Neuroscience, neural bases of communication; willingness to do a PhD thesis in this or a related topic; good knowledge of programming a plus.

Host institution

The team BaNCo investigate the cerebral bases of communication via neuroimaging techniques.

Your opinion is important to us

Description de la soumission d'un avis

Your vote :
Your opinion is important to us