CRII: RI: Navigational Circuitry of Brain: Novel Neural Codes with Diversity for Robust and Adaptive Location Processing

该项目的主要目标是开发一种新的编码框架的导航电路的大脑，并在此模型的背景下，研究神经多样性和神经合理的信息处理机制如何有助于一个强大的导航系统。基于神经编码原理的鲁棒定位技术可能在机器人技术中有应用。此外，这里开发的框架将导致更好地理解大脑的导航电路和相关的大脑区域，这与阿尔茨海默病和其他大脑疾病有关。大脑的导航电路包括细胞，其活动取决于动物在其环境中的位置。海马体中的位置细胞通常在环境中的某个区域活跃，但在其他区域保持沉默。另一方面，内嗅皮层中的网格细胞在多个区域做出反应，每个细胞的活动都遵循一个惊人的周期性模式，形成一个六边形的空间。地点和网格种群都表现出多样性（例如，神经元的激发场通常沿着背腹轴增加）和适应性（例如，细胞响应于环境的变化而改变它们的活性水平）。关于这些细胞之间的关系，有一种理论认为，网格细胞形成了位置细胞，而位置细胞又负责执行导航计算，但这种观点与最近的神经生理学数据相矛盾。相反，研究者的中心假设是，大脑有一个由位置细胞和网格细胞组成的神经代码，当用神经上合理的、低复杂度的算法处理时，这个代码的码字形成了一个鲁棒的、自适应的导航处理机制。该项目的重点是编码模型，包括位置细胞，提供边信息解码的网格细胞活动，并单独读出人口执行计算与位置和网格细胞迭代。利用这些模型，研究人员建议研究神经多样性在实现鲁棒性方面的作用，分析代码的模块化处理能力，通过降低码率获得更广泛的估计来理解神经自适应性，并将这些发现与实验结果进行比较观察。

The main objectives of this project are to develop a novel coding framework for the navigational circuitry of the brain, and in the context of this model, to investigate how neural diversity and neurally plausible information processing mechanisms contribute to a robust navigation system. Robust localization techniques based on neural coding principles may have applications in robotics. In addition, the framework developed here will lead to a better understanding of brain's navigational circuitry and associated brain areas, which are implicated in Alzheimer's disease and other brain disorders.The brain's navigational circuitry includes cells whose activities are modulated depending on where an animal is in its environment. Place cells in the hippocampus are typically active over a certain region within an environment, but remain mostly silent otherwise. Grid cells in the entorhinal cortex, on the other hand, respond at multiple regions, where the activity of each cell follows a striking periodic pattern, forming a hexagonal tiling of the space. Both place and grid populations exhibit diversity (e.g., firing fields of neurons typically increase along the dorsoventral axis) and adaptivity (e.g., cells change their activity levels in response to changes in environment). One theory of how these cells are related is that grid cells form place cells, which in turn are responsible for performing navigational computations, but this view contradicts recent neurophysiological data. The central hypothesis of the investigator, in contrast, is that the brain has a neural code consisting of both place cells and grid cells, and, when processed with neurally plausible, low-complexity algorithms, codewords of this code form a robust and adaptive navigational processing mechanism. This project focuses on coding models consisting of place cells that provide side information in decoding of the grid cell activity, and separate read-out populations that perform computations with place and grid cells iteratively. Utilizing these models, the investigator proposes to study the role of neural diversity in achieving robustness, to analyze the modular processing capacity of the code, to understand neural adaptivity through obtaining broader estimate by lowering the code rate, and to compare these findings with experimental observations.