Neuronal population dynamics in a learned motor behavior.
We have a MSc project for a student interested in understanding the neural basis of complex motor learning through the analysis of high dimensional datasets of neuronal population activity and behavior using statistical methods and machine learning.
Songbirds are an ideal animal model for performance-driven motor learning in which the basal ganglia and motor pathways hold several parallels to those of humans, from analogies in organization and function to homologies in gene transcriptions. Variability in neural activity within the songbird basal ganglia-dorsal forebrain circuit (anterior forebrain pathway; AFP) depends on social context and its effect on the motor pathways is contingent on developmental stage. Until now, electrophysiological studies in the songbird's AFP were limited to single unit recordings, insightful yet prohibitive to understand how the observed variability percolates to behaviors that normally are driven by populations and lead to smooth motor output (as opposed to jerky bursts).
We have overcome this limitation and we can record neural activity from hundreds of sites simultaneously, covering several nuclei, in chronically implanted, freely behaving zebra finches.
The job is to stand on machine learning and statistical approaches to represent, extract features and develop models that will help understand and predict how the variability in the dynamics of these populations accounts for the variability in the behavioral performance (the song).
The candidate should be able to code tidily (ideally Python/Matlab and some ML framework) and have expertise or be willing to quickly catch up on statistics and machine learning.
ezequiel (at) ini.ethz.ch,
rich (at) ini.ethz.ch,
corinna (at) ini.ethz.ch