Venue : Centre Broca
Thesis defense in english
Fjola Hyseni
IMN
Teams : Computational Neuroscience (Rougier) and Network dynamics for procedural learning (Leblois / Mallet)
Thesis directed by Arthur Leblois and Nicolas Rougier
Title
Temporal dynamics in the neural representation of sensorimotor tasks: investigation of the song-timing network in birds
Abstract
Temporally precise movement patterns underlie many motor skills, yet the origin of temporal control in motor behaviors remains unclear. The zebra finch song system has shown to be an outstanding model to study temporal control and sequential neuronal activity in the order of tens to hundreds of milliseconds. Like human speech, birdsong relies on a tight muscle coordination, with its premotor nucleus, HVC, responsible for the precise control of song tempo. Current computational models of HVC rely on the synfire chain, a purely feedforward network model that can account for HVC sequential activity. Synfire chains are however not robust to noise and function for a narrow range of feedforward weights, thus requiring fine tuning during learning. On the contrary, attractor dynamics provide networks with robust functional properties that make them an alternative to feedforward models. Therefore, we propose that HVC neuronal dynamics may be modelled using a Ring Attractor with a narrow Gaussian connectivity profile, where recurrent connections allow the formation of an activity bump that remains stable across a wide range of weights. In the case of asymmetrical connectivity, the bump of activity moves across the network, generating sequential neuronal activity. We show that the width of the activity bump, and thus the duration of transient neuronal activation, can be decreased to reproduce the brief activity bursts of HVC neurons. Additionally, we reproduce a syllable duration plasticity experiment by implementing a reward covariance reinforcement learning rule in the network. Consistent with behavioral results, the change in duration is specific to the target syllable. Lastly, we investigate further with an EI network model of spiking neurons and show that with a more biologically plausible and precise model, we are able not only to reproduce HVC’s fast spiking dynamics, but also perform with specificity a behavioral learning paradigm to modify syllable duration. These findings are confronted with behavioral results of daily duration changes in birds underdoing a Conditional Auditory Feedback protocol to adaptively change syllable duration.
Key words
Timing, Oiseaux Chanteurs, HVC, Ring Attractor, AdEx
Publications
F. Hyseni, N. Rougier, A. Leblois “Attractor dynamics drive flexible timing in birdsong.” In: International Conference on Artificial Neural Networks. Springer. 2023, pp. 112–123. doi: 10.1007/978-3-031-44198-1_10
F. Hyseni, N. Rougier, A. Leblois “Comparative study of the synfire chain and ring attrac- tor model for timing in the premotor nucleus in male zebra finches.” In: ESANN 2023-European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. 2023, pp. 647–652. doi: 10.14428/esann/2023.ES2023-120
Jury
Mme. Adrienne FAIRHALL: Professeur, University of Washington – Rapporteure
M. Albert COMPTE: Professor, IDIBAPS, Barcelona – Rapporteur
Mme. Daniela VALLENTIN: Chargée de recherche, MPI, Seewiesen – Examinatrice
M. Hervé ROUAULT: Chargé de recherche, Université Aix Marseille – Examinateur
M. Alexander PITTI: Professeur, CY Cergy-Paris Université – Examinateur
M. Arthur LEBLOIS: Chargé de recherche, Université de Bordeaux – Co-directeur de these
M. Nicolas P. ROUGIER: Directeur de recherche, Université de Bordeaux – Directeur de these