A novel closed-loop paradigm for investigating roles of Drosophila neurons in specific behavioral contexts using optogenetic neural manipulations

Abstract

Recent developments in optogenetics, the use of genetically encoded ion channels that open in response to exposure to light, have enabled investigations of roles of neural subsets in producing natural behaviors in freely moving animals. Because many neural circuits have been shown to have functions that are specific to behavioral contexts, we sought to enable optogenetic activations that are time-locked to onsets of specific behavioral elements. We have developed a novel paradigm for closed-loop optogenetic control of neural activity with high temporal specificity to patterns of acoustic production during fruit fly courtship. We have implemented real-time detectors of courtship song modes with convolutional neural networks to achieve speed and accuracy. The song-mode detectors perform with true positive rates as high as 92% and false positive rates as low as 0.5%, and the framework responds to detected song modes with optogenetic stimuli with delays as short as 12 ms. Using our framework, we show evidence that one subset of dopaminergic neurons can change song patterning over short timescales but is likely not involved in song patterning over the longer timescale of a courtship-bout duration. Our novel closed-loop paradigm can be applied to investigations of many other neuronal subsets and circuits, and their roles in courtship behavior.