Closed-loop control of neural dynamics and action selection in C elegans

项目摘要/摘要 为了对新的感官刺激做出反应，动物必须决定下一步该做什么。逃避行为 是研究行为的神经生物学基础的优秀模型，因为它们是 基本的和可解释的。在这里，我们建议研究影响概率的神经因素 线虫逃避厌恶感觉神经元灰尘刺激的行为。具体来说，我们 将决定行为的差异在多大程度上是由正在进行的全球神经动力学造成的，即 时变的活动模式，编码主要行为，如向前、向后、背部和 腹部爬行。在动力学过程中，给定的神经元受到不同程度的兴奋和抑制。 来自它的突触伙伴，这可能有助于行为看起来是概率的。 在具有细胞分辨率的全脑成像期间访问整个大脑的活动 在动物的同时，我们会在光遗传刺激神经元的信号活动期间 与爬行行为相关的闭环钙成像定量和光发生 刺激平台，以扰动动力学。我们将描述刺激强度和力度如何 相互作用，激活感觉神经元和下游神经元。我们还将研究变量Forward或 当蠕虫从没有明显动态的状态受到干扰时出现的反转行为， 睡吧。我们将描述神经元如何对不同强度的刺激做出反应并适应生理学 与激活功能和动作选择相关的模型的数据。 总之，这些实验使用了一个简单的动物模型来阐明如何唤起 神经元的活动受到整个网络中持续模式的调节。这些实验 强调感觉和运动表征的分布式视图，并探讨其对 高度相互联系的神经系统中的动力学。这份提案中的研究和培训计划 绝对是实验和计算的混合体，每个都必须由两个领导者指导 以合作著称的加州大学旧金山分校的纪律。

Project Summary/Abstract In response to new sensory stimuli, an animal must decide what to do next. Escape behaviors are excellent models with which to study the neurobiological basis of behavior because they are essential and interpretable. Here we propose to study neural factors which influence a probabilistic behavior where C. elegans escapes stimulation of an aversive sensory neuron ASH. Specifically we will determine how much variance in behavior is accounted for by ongoing global neural dynamics i.e. time-varying patterns of activity which encode major behaviors such as forward, backward, dorsal, and ventral crawling. During dynamics, a given neuron receives different levels of excitation and inhibition from its synaptic partners, which may contribute to behaviors seeming probabilistic. During whole-brain imaging with cellular resolution to access the activity of the entire brain of the animal simultaneously, we will optogenetically stimulate neurons during signatures of activity correlated with crawling behaviors using a closed-loop calcium imaging quantification and optogenetic stimulation platform to perturb dynamics. We will characterize how stimulus intensity and dynamics interact to activate sensory and downstream neurons. We will also examine a variable forward or reversal behavior which emerges when worms are disturbed from a state with no obvious dynamics, sleep. We will characterize how neurons respond to different intensities of stimulation and fit physiology data to a model relating activation function and action selection. Taken together these experiments use a simple animal model to shed light on how evoked neuronal activity is modulated by ongoing patterns throughout the network. These experiments emphasize a distributed view of sensory and motor representations and explore its implications on dynamics in a highly interconnected nervous system. This research and training plan in this proposal is decidedly hybrid experimental and computational, necessarily guided by two leaders in each discipline at the famously collaborative University of California, San Francisco.