Neurons within the cerebellum respond to and learn to anticipate rewards, Stanford researchers report. The finding is a first step toward a much more exciting future for the cerebrum’s largely overlooked little brother, and one that could open up new avenues of research for neuroscientists interested in the roots of cognition.
Scientists had assumed that the cerebellum helped control muscles mostly because of what happened when it got injured. Senior author Liqun Luo, an investigator at the Howard Hughes Medical Institute, professor of biology and member of Stanford Bio-X and the Stanford Neurosciences Institute, said:
“If you have disruption of the cerebellum, the first thing you see is a motor coordination defect.”
There had been hints of a larger role for the cerebellum, but scientists had a hard time following up on those hints in part because the neurons that make up most of the cerebellum are difficult to study. Those neurons, known as granule cells, account for 80 percent of the neurons in the brain – all packed into the cerebellum – but only about 10 percent of its volume.
At that density, conventional techniques for recording cell activity don’t work well, and without an effective way of studying granule cells in real time, scientists were left with an incomplete picture of what the cerebellum was really doing.
Cerebellar Granule Cells
Mark Wagner, a postdoctoral fellow in Luo’s lab, did not set out to redeem the cerebellum. He simply wanted to study how the cerebellum controls muscles in mice using a new technique that would allow him to record granule cells in real time.
Wagner led the research with Tony Kim, a graduate student in the lab of Mark Schnitzer, an investigator at the Howard Hughes Medical Institute and an associate professor of biology and of applied physics.
Wagner had earned his PhD working with Schnitzer, who develops pioneering methods for imaging neuronal activity in fruit flies, mice and other living animals. One method, called two-photon calcium imaging, had the resolution Wagner needed to study mouse granule cells in action.
In order to study motor control, the team had to get the mice to move. In this case, mice received sugar water about a second after pushing a little lever.
While the mice pushed levers and received their rewards, Wagner recorded activity in each mouse’s granule cells, expecting to find that that activity in those cells would be related to planning and executing arm movements.
Reward Expectation Response
To some extent he was right. Some granule cells did fire when the animals moved.
But other granule cells fired when the mice were waiting for their sugary rewards. And when Wagner sneakily took away their rewards, still other granule cells fired.
“It was actually a side observation, that, wow, they actually respond to reward,” Luo said.
That discovery is something of a revelation. For 50 years, the assumption was that granule cells – and by extension the cerebellum – performed only the most basic functions. But because no one had the tools to look closely at granule cells in action, we just didn’t know, Wagner said.
Now that scientists have a better idea of what’s happening, Wagner’s hope is that it could lead to something much bigger.
“Given what a large fraction of neurons reside in the cerebellum, there’s been relatively little progress made in integrating the cerebellum into the bigger picture of how the brain is solving tasks, and a large part of that disconnect has been this assumption that the cerebellum can only be involved in motor tasks,” Wagner said.
Mark J. Wagner, Tony Hyun Kim, Joan Savall, Mark J. Schnitzer & Liqun Luo Cerebellar granule cells encode the expectation of reward Nature (2017) doi:10.1038/nature21726
Image: Cerebellum granule cells, shown as green in this image. Credit: Mark Wagner