The neural language of song: Brain mechanisms for sensorimotor syntax control

Description du projet

Les substrats neuronaux de la syntaxe hiérarchique

Les règles syntaxiques hiérarchiques définissent la manière dont les unités simples sont organisées pour former des unités plus complexes. Les séquences motrices obéissent à des règles syntaxiques hiérarchiques, qu’il s’agisse de lacer ses chaussures ou de chanter une chanson. Financé par le Conseil européen de la recherche, le projet NeuralSyntax vise à étudier le chant du canari et à déchiffrer les règles syntaxiques hiérarchiques par l’étude des séquences comportementales et neuronales. S’appuyant sur leurs études antérieures qui ont permis d’identifier des états neuronaux cachés qui gardent en mémoire les syllabes chantées pendant plusieurs secondes et prédisent les transitions à venir, l’équipe tentera de disséquer les mécanismes de transition. Elle tentera de révéler l’activité neuronale qui sous-tend les transitions entre les syllabes, la sélection de la syllabe suivante, l’adaptation des règles syntaxiques aux nouvelles chansons et le remodelage saisonnier des règles syntaxiques.

Objectif

Our motor sequences obey syntax rules. From tying shoes to giving a speech, choosing the next action requires our brain to remember past actions, bridge across many seconds, and apply rules that create hierarchical syntax. Whereas past studies revealed how primary motor cortex drives actions, we poorly understand the transition mechanisms by which the brain strings actions into hierarchical variable sequences. The key obstacle is identifying deep hierarchical syntax rules in behavior and in neural sequences. This obstacle is met by studying birdsong that naturally segments to sequences of syllables. In the zebra finch, juveniles learn once from a tutor and fix a single sequence driven by a robust song-locked activity. Here, to study hierarchical syntax rules I will use canaries: virtuosos that learn and adapt such rules along and across seasons. Canaries produce a huge repertoire of songs that enable separate analysis of neural states and syllable acoustics when studying their rich and dynamic syntax rules. My previous work in canaries identified hidden neural states that keep a memory of sung syllables over several seconds and predict upcoming transitions. Here, I leverage this model to dissect transition mechanisms and elucidate: 1) neural activity in the premotor nucleus HVC that underlies flexible transitions between syllables; 2) integration of auditory and thalamic inputs for applying hierarchical syntax rules to select the next syllable; 3) neural state dynamics when adapting syntax rules for new songs; and 4) seasonal remodeling of HVC via neurogenesis. We use imaging, electrophysiology, and neural network simulations that integrate data across behavior, neural activity, and circuit levels of analysis. Our work will map the process by which basal ganglia and thalamocortical circuits generate and adapt hierarchical syntax rules over time, sensory, and social contexts, and might offer insights to dysfunction of motor sequence control in maladaptive conditions.