Deep in the midbrain, one type of neuron has two crucial jobs when it comes to acting while anticipating a reward—a state also known as Pavlovian conditioning. Called dopaminergic (DA) neurons, they can link a signal, such as a sound or smell, to the reward. They can also tell a trained animal to perform an action, such as sticking out its tongue for a treat.
It’s been difficult to untangle those two seemingly distinct neural activities—i.e. initiating movement versus learning to make that movement. The authors of a recent Nature Neuroscience study sought to do so by deactivating DA neurons at different times as mice responded to a signal and reaped the reward. “We asked the question, not whether it’s playing a role in these behaviors, but when,” says senior author Sotiris Masmanidis, a neuroscientist at the University of California, Los Angeles. The results suggest that DA neurons have the biggest impact after the reward arrives, when they maintain the brain’s association between signal and behavior.
First, the researchers trained mice to associate a whiff of banana smell with the arrival, 2 seconds later, of a droplet of sweetened milk. Soon enough, the mice stuck out their tongues as soon as they smelled banana, even before the treat arrived—tantamount to Pavlov’s dogs drooling in anticipation of a snack.
Then the team used optogenetics, delivering a pulse of light to the mouse’s brain in order to quiet DA neurons at two different times. For one time point, they turned off the neurons in that 2-second window between signal and reward, just as the animals stuck out their tongues for the expected milk. If DA neurons were most crucial to generate this action of licking—as Masmanidis predicted, based on previous literature—then silencing them should reduce licking. It did, though the effect was small, suggesting DA neurons weren’t heavily involved at this stage.
For the other time point, the researchers waited to shut down DA neuron signaling until after the reward arrived. Unexpectedly, this had a bigger effect. Mice that experienced this treatment licked less in future trials. In other words, without that signal from the DA neurons, the association the mice learned between banana smell and the reward weakened. In fact, the researchers observed a similar response when they reduced the size of the rewards. “It suggests the main role [of DA neurons] is indeed in learning,” says Masmanidis.
Masmanidis cautions, though, that his group only examined one highly controlled behavior. DA neurons might demonstrate an expanded role in the case of other actions. For example, DA neurons disappear from the brains of people with Parkinson’s disease, causing movement difficulties. Masmanidis speculates that a different task, such as walking, might require activity from these neurons both for learning and for the action itself.
The study addresses a longstanding puzzle in the study of DA neurons and learning, says Luke Coddington, a neuroscientist at Howard Hughes Medical Institute Janelia Research Campus in Ashburn, VA, who was not involved in the current paper. For many other neural systems, researchers understand whether neurons fire to report a signal, or to create an active response. DA neurons, because they’re able to do both, have been trickier to parse. Based on Masmanidis’ results, “You can make a strong conclusion that it’s a learning process, rather than just a one-off regulation of behavior,” says Coddington, who also works on DA neurons and learning. “It helps us constrain our thinking.”
Plenty of questions remain. “Why is it that dopamine has such a profound effect on learning?” asks Masmanidis. “What is the disease relevance?” To find out, he now plans to investigate the other parts of the brain that DA neurons communicate with as these critical cells reinforce or create behaviors.