Cerebellar granule cells, which make up the cerebellum, are the smallest and most abundant of all types of neurons in the brain. These cells are known to contribute to motor skills, attention, language and anxiety. A recent study published in Nature shows that these cells can also contribute to our expectations of whether a certain action will result in a positive reward. It’s a discovery that deviates from our previous understanding of how these types of cells function.
To investigate the function of these cerebellar granule cells, the authors used a mouse model of reward and reward anticipation. In this model, mice are trained to press a lever to get a small treat of sugar water.
When the authors looked directly at the electrical activity in the brains of these mice, they saw that some of these cerebellar granule neurons were activated during the lever-pushing task. The peak of neuronal activity coincided with the peak of physical activity for up to 20 percent of the cells. However, not all cell populations fired during the same part of the lever-pushing task, so the researchers wanted to learn more about the neuronal differences between these subpopulations of cells.
Is that good?
The authors in it Nature found that some cells fired preferentially before the reward was delivered. Other cells, meanwhile, had greater responses when the mice pushed the lever and did not receive the expected sugar water reward (called a “left-out reward”). The authors went on to study these two types of cells separately: the cells that responded to the omission of a reward and the cells that fired just before receiving a reward.
When they looked at the cell response distribution in the mouse’s cerebellum during the handle-pushing task, the researchers saw that about 5 percent of the neurons fired in response to reward, 12 percent fired with the omission of reward, and 9 percent shot in. anticipate reward. A similar distribution of neuronal responses was found when mice were trained in an entirely different reward task, suggesting that those cells were not involved in the lever-pushing activity. Instead, this division of cells is common in mice and is generally involved in reward-related tasks.
The authors then took a closer look at the firing of the neuron. They tracked the activity of individual neurons as the mice learned to perform reward tasks. After six days of training, they saw that neurons with the strongest reward anticipation responses were the same neurons that had only responded after a reward was delivered while the mice were training. This suggests that the role of those neurons is shifting from recognizing a treat to anticipating a treat.
The neurons that responded most strongly to the omission of a reward at the end of training were the same neurons that responded to the omission of a reward earlier in the training. Again, these observations suggest that the cerebellar granule neurons are specialized in terms of their responses to reward or lack of reward.
This paper contributes new information to the neuroscience understanding of how we respond to rewards. It shows that neurons are specialized to fire during different parts of a reward-related task: some neurons respond strongly in anticipation, others fire when they get the expected reward, and another group fires when the reward doesn’t appear. As neuroscientists continue to characterize the brain, the synthesis of these kinds of function-specific findings should eventually help build a general picture of how different brain regions interact and function.
Nature2017. DOI: 10.1038/nature21726 (About DOIs)