Characterizing the underlying population code to understand the functional organization of the hippocampus and the lateral hypothalamus

项目摘要/摘要 神经记录/成像技术和计算方法的最新进步产生了 人们对研究协调的人口活动重新产生了兴趣。了解人口密码可以帮助我们 更好地理解物质使用障碍(SUD)背后的复杂机制。一个主要观点是， 高维神经活动，例如从数百到数千个细胞的同时记录，可以 在低维流形上，使得少数几个潜在变量可以准确地描述所有 记录的神经元。外侧下丘脑是一个以其功能多样性而闻名的大脑区域。 细胞对广泛的食欲行为有很大的异质性，而促黄体生成素刺激可以唤起一种 从进食到社会互动的各种动作。本项目建议使用最新的非线性 提取代表种群活动的低维流形的降维技术 以几何方式组织不同种类的单个神经元活动的模式。然后就可以使用这些集合管 为了实现这个项目的主要目标--区分自然奖赏的黄体生成素神经种群编码 行为和不适应的寻毒行为。此外，新的计算建模方法将是 用于执行内部神经状态的无监督检测，以引导动物在这两者之间切换 追求报酬的行为。最后，最先进的细胞分辨率同时刺激和成像 显微镜将被用来通过激活来随意扰乱动物的行为和/或神经活动模式 低维流形上沿轨迹的神经元序列。重要的是，目标1和2支持 通过提供使用必要的计算和仪器技术的培训来实现这些目标。 最终，从这个项目中获得的结果将促进我们对神经机制的理解 将有害的寻毒行为和有益的自然寻赏行为分开，这样就可以 可以开发出具有更精确靶点的治疗方法，将不想要的副作用降至最低。

Project Summary/Abstract Recent advancements in neural recording/imaging technologies and computational methods have generated a renewed interest in studying coordinated population activity. Understanding the population code can help us better understand the complex mechanisms behind substance use disorders (SUD). One leading idea is that high-dimensional neural activity, such as simultaneous recordings from hundreds to thousands of cells, can lie on low-dimensional manifolds, such that a handful of latent variables can accurately describe the activity of all recorded neurons. The lateral hypothalamus (LH) is a brain area well-known for its functional diversity – individual cells respond with great heterogeneity to a wide range of appetitive behaviors, and LH stimulation can evoke a variety of actions ranging from feeding to social interaction. This project proposes to use the latest nonlinear dimensionality reduction techniques to extract the low-dimensional manifolds representing population activity patterns that geometrically organize the heterogeneous single neuron activity. These manifolds can then be used to achieve this project’s main goal – differentiating LH neural population encoding of natural reward-seeking behaviors and maladaptive drug-seeking behaviors. In addition, novel computational modeling methods will be used to perform unsupervised detection of internal neural states that guide animals switching between these two reward-seeking behaviors. Finally, state-of-the-art cellular-resolution simultaneous stimulation and imaging microscopy will be used to casually perturb animal behavior and/or neural activity patterns by activating sequences of neurons along trajectories on the low-dimensional manifolds. Importantly, Aims 1 and 2 support these goals by offering training in the use of the necessary computational and instrumentation techniques. Ultimately, results obtained from this project will advance our understanding of the neural mechanisms separating harmful drug-seeking behaviors and useful natural reward-seeking behaviors, such that SUD treatments with more precise targets can be developed that minimize unwanted side effects.