A Control Theoretic Approach to Addressing Hippocampal Function

项目摘要 海马结构与学习和记忆密切相关。神经变性疾病如 阿尔茨海默氏病严重影响这一领域，导致严重和渐进的记忆丧失。的 海马体似乎是外部世界的非自我中心的认知地图的所在地。这张地图至关重要 不仅对空间认知，而且对过去经验的有意识回忆。海马体是 被认为将个体项目和经验事件绑定在一个连贯的时空框架内， 允许经验被存储和检索作为一个有意识的记忆。几十年的调查 海马位置细胞和最近发现的网格细胞揭示了这种认知地图来自于 外部感觉输入与内源性产生的神经动力学的相互作用（潜在的 导航策略称为“路径集成”）。行为动物的经典神经生理学研究 充分说明了环境地标对这些射击地点的强大影响， 空间细胞将这种方法扩展到定量研究路径整合的内部过程， 技术上的挑战。虚拟现实技术与系统理论相结合， 来解决这些问题。我们设计并制造了一种新的装置，使我们能够操纵 视觉输入（地标和光流）可用于大鼠导航真实的圆形轨迹，作为其 运动，同时保持正常的走动和前庭经验。将记录在此 装置复制已知的标准现象学。在初步实验中，我们诱导了一个持续的， 增加了地标信息和路径整合之间的冲突。结果表明， 系统在受到这种持续冲突的挑战时重新校准路径集成器。此外，我们还开发了 一种新的方法，用于分离光流和其他自运动线索对神经元更新的贡献， 位置的表示，不受地标的竞争影响。具体来说，我们开发了一种 在线人口解码器，并使用解码输出来控制行为过程中的这种认知表示 通过光流的实时反馈操作。这种方法将构成小说的基础 研究计划旨在全面分析认知的外部与内部决定因素 地图此外，该程序有望揭示与一般神经计算相关的重要原理。 大脑如何整合外部感官输入与内部认知表征的问题，最终 产生对精神分裂症特征的思维障碍和幻觉的见解， 其他精神障碍。 1

PROJECT SUMMARY The hippocampal formation is critically involved in learning and memory. Neurodegenerative disorders such as Alzheimer’s Disease dramatically impact this area, leading to severe and progressive memory loss. The hippocampus appears to be the locus of an allocentric, cognitive map of the external world. This map is critical not only for spatial cognition, but also for the conscious recollection of past experience. The hippocampus is thought to bind the individual items and events of experience within a coherent spatiotemporal framework, allowing the experience to be stored and retrieved as a conscious memory. Decades of investigation of hippocampal place cells and the recent discovery of grid cells have revealed that this cognitive map arises from the interaction of external sensory inputs with endogenously generated neural dynamics (underlying the navigational strategy known as “path integration”). Classical neurophysiological studies with behaving animals have amply characterized the powerful influence of environmental landmarks on the firing locations of these spatial cells. Extending this approach to quantitatively investigate the internal processes of path integration has proven technically challenging. Virtual reality technology, in combination with systems theory, offers opportunities to solve these problems. We have designed and constructed a novel apparatus that allows us to manipulate the visual inputs (both landmarks and optic flow) available to a rat navigating a real circular track as a function of its movements, while preserving normal ambulatory and vestibular experience. Place cells recorded in this apparatus replicate known standard phenomenology. In preliminary experiments, we induced a sustained, increasing conflict between landmark information and path integration. Results demonstrate the capacity of the system to recalibrate the path integrator when challenged with this sustained conflict. Further, we have developed a novel approach for isolating the contribution of optic flow and other self-motion cues to the update of the neural representation of position, free of the competing influence of landmarks. Specifically, we have developed an online population decoder, and used the decoded output to control this cognitive representation during behavior through real-time feedback manipulations of the optic flow. This approach will form the foundation of a novel research program aimed at a comprehensive analysis of the external vs. internal determinants of the cognitive map. Furthermore, this program promises to reveal important principles of neural computation relevant to general problems of how the brain integrates external sensory input with internal, cognitive representations, ultimately generating insights into the disordered thinking and hallucinations that are characteristic of schizophrenia and other mental disorders. 1