“CRCNS: Computational principles of memory based decision making in Drosophila”

点击翻译按钮获取中文摘要

Decision making largely depends on the integration of prior sensory experiences that are stored in the brain as memories. One of the best-understood model systems of a two-choice decision process is associative olfactory memory in Drosophila. In the fly, learning occurs at the synapse between Kenyon cells (KCs) and mushroom body output neurons (MBONs). By pairing olfactory stimuli with a reward or a punishment flies can reliably learn to distinguish two initially neutral odors. As in mammals, dopaminergic neurons (DANs) represents this contextual valence and modulate the strength or connectivity of KC>MBON synapses within distinct compartments of the MB, consequently altering behavior. To understand how this decision making is encoded in memory we thus must first understand the computation performed within individual DAN-KC-MBON modules. In preliminary work, we built a realistic computational model of the MBON-3 neuron including the precise synaptic connectivity all 948 innervating KCs based on the complete synaptic connectome of the MB. Our model incorporates precise membrane properties of the MBON-3 neuron that we obtained by performing patch-clamp recording in vivo. We demonstrate that MBON-3 is electrotonically compact and show that activation of complex KC input patterns reflecting physiological activation by individual odors in vivo are sufficient to robustly drive MBON spiking. Here, we will combine experimental analysis in vivo with computational modelling to determine the mechanisms controlling MBON activation under baseline conditions and in response to memory-induced plasticity. We will identify the minimal number of KCs and KC synapses required for robustMBON activation for different subsets of approach and avoidance MBONs in vivo. These data will guide computational approaches to identify general and specific features of the KC-MBON interactions. The results will then serve to identify compartment-specific plasticity mechanisms in vivo, and corresponding computational mechanisms in silico. Finally, we extend these approaches to simultaneous analyses at multiple KC-MBON modules to provide a computational basis for decision-making.