Variability and learning in birdsong


October 10, 2018 - 12:00pm
BioLabs Building, Room 1080
About the Speaker
Adrienne Fairhall
Speaker Title: 
Speaker Affiliation: 
University of Washington

Reinforcement learning is a powerful learning strategy in which one makes random perturbations in an output and reinforces changes that lead to more successful outputs. Animals learn to perform highly skilled motor tasks by trial and error, but how biological neural networks generate this variability is not known. The birdsong circuit is an ideal model system in which to test potential mechanisms for the generation of variability and how this variability can be modified by behavioral context. Birds learn their songs by trial and error, and are able to modify their songs into adulthood; the intrinsic variability of their songs is reduced in the presence of a female; and the neural architecture of the learning circuit has been mapped out. Here we will discuss potential dynamical mechanisms for the generation and control of variability, and how variability can confer robustness.