Project

Back to overview

Predictive coding and cortical oscillations in speech processing

English title Predictive coding and cortical oscillations in speech processing
Applicant Giraud Mamessier Anne-Lise
Number 163040
Funding scheme Project funding (Div. I-III)
Research institution Département des neurosciences fondamentales Faculté de Médecine Université de Genève
Institution of higher education University of Geneva - GE
Main discipline Neurophysiology and Brain Research
Start/End 01.01.2016 - 31.12.2018
Approved amount 812'244.00
Show all

Keywords (3)

cortical oscillations; speech; neurocomputational modelling

Lay Summary (French)

Lead
A quoi servent les oscillations neurale montantes et descendantes dans la perception de la parole?
Lay summary

Les oscillations neurales sont souvent observées dans l’enregistrement de populations de neurones (EEG, MEG, LFP). Elles varient en fonction des stimuli sensoriels, des tâches cognitives et des états mentaux, et de nombreux rôles leur ont été attribués. Comme des phénomènes oscillatoires émergent dans les systèmes complexes, il est légitime de se demander si elles reflètent les processus dynamiques du cerveau ou si elles participent au codage de l’information neurale. Dans le domaine de la parole les oscillations neurales présenteraient l’avantage de discrétiser le flux d’information acoustique en éléments linguistiques porteurs de sens. En simulant des oscillations corticales à à l’aide d’un modèle réaliste de circuits neuronaux, nous avons montré que les oscillations lentes (theta) peuvent signaler les syllabes en s’adaptant à leur rythme, et que les oscillations rapides (gamma) encodent l’information phonémique dans les syllabes. Un troisième type d’oscillations (beta) serait important dans la prédiction de l’information de parole, lorsque celle-ci est bruitée, incomplète, etc, et fonctionnerait dans la direction opposée et en alternance avec les oscillations theta et gamma. Nous ne savons pas si ce mode d’organisation offre un réel intérêt dans le codage de l’information. Pour répondre à cette question ce projet se propose 1- de collecter des données expérimentales chez l’humain au moyen de l’EEG, la MEG et d’enregistrements invasifs réalisés chez des patients porteurs d’électrodes corticales, et 2- de poursuivre une approche de modélisation. Notre but est que comprendre les interactions entre les deux rythmes du traitement sensoriel de la parole (theta et gamma) et le rythme du traitement prédictif descendant (beta), et d’explorer leur rôle potentiel respectif dans l’encodage des prédictions et des erreurs de prédiction. Ce projet permettra en outre de comprendre les déficiences du traitement prédictif auditif dans la schizophrénie, l’autisme et la dyslexie.

 

Direct link to Lay Summary Last update: 30.10.2015

Responsible applicant and co-applicants

Employees

Collaboration

Group / person Country
Types of collaboration
CENIR/ICM Paris France (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Research Infrastructure
Neurology department HUG Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Research Infrastructure
Boston University United States of America (North America)
- in-depth/constructive exchanges on approaches, methods or results
- Publication

Scientific events

Active participation

Title Type of contribution Title of article or contribution Date Place Persons involved
Lemanic Neuroscience Annual Meeting Talk given at a conference A predictive coding model of speech perception with (and without) neural oscillations 02.09.2018 Les Diablerets, Switzerland Hovsepyan Sevada;
Symposium Window on Brain Plasticity Talk given at a conference How does brain remodeling determine cochlear implantation outcome? 07.05.2018 Brussels, Belgium Giraud Mamessier Anne-Lise;
Past, present and future of neuroimaging research Talk given at a conference The spatio-temporal geometry of speech processing 16.02.2018 Gif sur Yvette, France Giraud Mamessier Anne-Lise;
ABIM 2018 Poster A predictive coding model of speech perception with (and without) neural oscillations 07.01.2018 Champéry, Switzerland Giraud Mamessier Anne-Lise; Hovsepyan Sevada;
ABIM 2018 Poster Audio Visual Integration in the Brain: differences in neural signatures of McGurk illusions 07.01.2018 Champéry, Switzerland Delgado Jaime; Giraud Mamessier Anne-Lise; Bouton Sophie;
Salzburg Mind and Brain Annual Meeting Talk given at a conference The spatio-temporal geometry of speech processing 12.07.2017 Salzburg, Austria Giraud Mamessier Anne-Lise;
Language and Communication Poster Combining predictions and neural oscillations: A speech perception computational model 02.06.2017 Genève, Switzerland Hovsepyan Sevada;
International workshop on neural oscillations in speech and language processing. Talk given at a conference Speech processing with (and without) cortical oscillations 28.05.2017 Berlin, Germany Giraud Mamessier Anne-Lise;
Annual meeting of the German Neuroscience Society Talk given at a conference Speech processing with (and without) cortical oscillations 20.03.2017 Gottingen, Germany Giraud Mamessier Anne-Lise;
AMLAP16: Architectures and mechanisms for language processing Talk given at a conference Speech processing in auditory cortex with and without oscillations 01.09.2016 Bilbao, Spain Giraud Mamessier Anne-Lise;
Society for Neurobiology of Language Talk given at a conference Modelling neural oscillations to understand language neurodevelopmental disorders. 17.08.2016 London , Great Britain and Northern Ireland Giraud Mamessier Anne-Lise;
International Neuropsychological Symposium Talk given at a conference Modelling speech processing with cortical oscillations 21.06.2016 Baiona, Spain Giraud Mamessier Anne-Lise;


Communication with the public

Communication Title Media Place Year
Media relations: radio, television Journée mondiale du bégaiement RFI International 2018
Media relations: radio, television Le cerveau et les maux de la parole RTBF International 2018
Media relations: radio, television Les maux de la parole RTS radio 1 Western Switzerland 2018

Associated projects

Number Title Start Funding scheme
149319 Are cortical oscillations a useful ingredient of speech processing? 01.10.2013 Project funding (Div. I-III)
182855 Predictive coding, predictive routing and cortical oscillations in speech processing 01.01.2019 Project funding (Div. I-III)
182855 Predictive coding, predictive routing and cortical oscillations in speech processing 01.01.2019 Project funding (Div. I-III)

Abstract

Background: Neural oscillations are frequently observed in neuronal population recordings as EEG, MEG, SEEG and intracortical local field potentials. They are traditionally assigned many functions, yet their causal role in cognitive operations remains difficult to establish. In the domain of speech perception, specific interactions between cortical oscillations and acoustic rhythms in continuous speech could play a role in syllable parsing (theta) and the encoding of information at phonemic timescale (gamma). The goal of our previous SNF project was to determine whether these oscillations could viably act as speech parser and phonemic encoder, a prerequisite for positing that they play a causal and critical role in speech processing. Using a biologically plausible model of oscillations we provided evidence for the capability of oscillations to encode continuous speech, arguing for their direct computational importance in cognitive operations. Aims/hypotheses: Several recent studies further indicate that different cortical oscillations, namely the gamma and low-beta oscillation are involved in bottom-up and top-down processing, respectively. It is unclear whether the asymmetry of oscillation frequency in hierarchical message passing is an epiphenomenon, a neural strategy enabling the un-mixing of bi-directional neural information transfer, or whether it serves more specific computational functions. In the framework of predictive coding, a slower oscillation for down-going than up-going information transfer could reflect the necessity to integrate more information in both space and time when making predictions than when encoding sensory stimuli. In this research project, we propose to assess whether cortical oscillations play a specific computational role for predictive mechanisms in the domain of speech perception.Methods: We will carry out experiments in humans using EEG, MEG and intracortical macro-electrode recordings to explore the relevance of distinct oscillation frequency ranges in parsing, encoding and predicting speech. In parallel, we will pursue a modelling approach with two goals 1- understanding the mechanistic and computational function of oscillations in cortical speech processing, and more specifically 2- exploring their potential causal role in conveying and coding predictions and prediction errors. We will continue biophysical modelling and expand our previously developed oscillatory network model to test the efficacy of oscillations in speech encoding and top-down decoding. In parallel and in a complementary fashion, we will develop a new approach that will assess whether oscillations can be useful in a predictive coding model of speech processing that emphasizes function over biological plausibility.Potential value of the project: The expected impact of this project reaches beyond one sensory modality or one cognitive function. Proving or disproving the usefulness of cortical oscillations in a generic organization mode of information processing, as predictive coding, is of broad interest to basic neuroscience. As cortical oscillations can be influenced by neurofeedback and brain stimulation (DBS, TMS, TDCS etc.), this project will also have an impact at a more translational level, potentially leading to targeted rehabilitation in pathologies with defective auditory to-down control, e.g., schizophrenia, autism and dyslexia.
-