ELSC News

Article of the Month, November 2018 (Joshua's lab)

November 11, 2018

Coordinated cerebellar climbing fiber activity signals learned sensorimotor predictions

 

Author: William Heffley, Eun Young Song, Ziye Xu, Benjamin N. Taylor, Mary Anne Hughes, Andrew McKinney, Mati Joshuaand Court Hull 

 

Link:https://www.nature.com/articles/s41593-018-0228-8

 

Published in Nature Neuroscience volume 21, September 2018

 

The prevailing model of cerebellar learning states that climbing fibers signal erroneous motor output. However, this model was tested mainly in studies of behaviors that utilize hardwired neural pathways to link sensory input to motor output. We aimed to test whether this model applies to more flexible learning regimes that require arbitrary associations between sensory input and motor output.  We developed a cerebellar-dependent motor learning tasks in which mice were required to release a lever based on temporal or visual information. While mice were performing the tasks we recorded both large scale and single-dendrite-resolution calcium imaging. We found that climbing fibers were preferentially driven by correctly executed movements and other task parameters that predict reward outcome. Thus, climbing fiber activity pattern does not rely exclusively on motor errors. We suggest that climbing fibers are well-suited to drive learning by providing instructional input that signals both success and errors in behavior.