Rebecca Yang, Duke University
While Drosophila can be classically conditioned to associate cues with reward, it was unclear whether they can learn more complex reward rules through trials and errors. Here we show that Drosophila readily learn a reward rule that requires them to perform a specific action at a specific location. We first identified specific neurons whose optogenetic activation was profoundly attractive to flies. We next developed a high-throughput closed-loop system that automatically delivers rewards optically to flies if their actions and positions conform to the rule we specified. Flies quickly learned the place-action reward rule and adjusted their behavior when the rule changed. Lastly, we showed that this form of learning depended on a specific subset of dopaminergic neurons that are different from the ones that mediate olfaction classical conditioning. Our results suggest that Drosophila are able to solve complex reward rules, enabled by a powerful and previously unknown reward system.