Building spatial maps from visual and self-motion inputs

从视觉和自我运动输入构建空间地图

• 批准号: BB/W007878/1
• 负责人: Guifen Chen
• 金额: $ 62.79万
• 依托单位: Queen Mary University of London
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=BB%2FW007878%2F1
• 关键词: Building; spatial; maps; visual; self

Context of researchOur physical environment possesses many different cues that are perceived by our sensory systems. As we move through the environment, we observe a corresponding change in the sensory cues. In mammals, the hippocampus and its adjacent areas in the medial temporal lobe, have long been implicated in spatial navigation and learning. Several types of spatial neurons have been discovered in this area, including place cells and grid cells. The activity of these neurons represents an animal's current location. Despite this discovery of an internal representation (or "map") of space, it remains unclear how the brain combines environmental sensory cues (e.g. visual landmarks) with self-motion information (e.g. locomotor or optic flow cues) in order to form these maps. Hence, the focus of this project is to disentangle the effects of visual and self-motion cues on place cells and grid cells during spatial mapping.Historically, it was challenging to study the question in adult animals for three key reasons. First, separating the effects of visual and self-motion inputs is difficult to achieve in the real world. Second, place cell and grid cell networks are interconnected, hence it is difficult to study the separate networks independently. Third, spatial representations appear almost instantaneously when an animal enters a new environment. This suggests that animals learn from previous experience, possibly developing a generalized code that enables them to quickly construct new spatial representations on demand.I have recently developed a two-dimensional virtual reality (2D VR) system, providing mice with an immersive experience of navigating in a virtual world. The pioneering development places me in a unique position to answer the question which was previously challenging. The new 2D VR allows independent manipulations of visual and self-motion cues in 2D space. My preliminary data show that spatial representations in a virtual world are similar to those in the real world, but form at a much slower pace. Thus, for the first time, we have a prolonged window during which to study the formation of spatial representations, and in particular the effects of visual and self-motion cues on the formation process.Aim and objectives The project will study the distinct roles of place cells and grid cells in building spatial representations, by taking advantage of the new 2D VR system. The aim is to understand how place and grid cells interact and combine visual and self-motion cues to represent space. I will first establish the timeline of the formation of spatial representations in 2D virtual space in adult mice. Next, I will differentiate the contributions of visual and self-motion information on forming spatial representations. Finally, I will test how varying these cues affects established spatial maps. Potential applications and benefitsThe project tackles one of the key challenges outlined in the BBSRC's vision - "Understanding the rules of life", and is perfectly aligned with the BBSRC's strategic priority "Systems approaches to the bioscience". The project offers a new angle for understanding the interaction between spatial cells and their functions in spatial learning, providing the foundation for the applications in the fields of artificial intelligence, robotic navigation and ageing. First, the findings will allow computational neuroscientists to create increasingly accurate models simulating long-term memory, contributing to the development of artificial intelligence. Second, the work offers insight into teaching robots how to integrate multisensory inputs, perform complex terrain navigation. Third, the findings will help us understand the neural basis of memory processing in normal ageing, as well as neurodegenerative diseases such as dementia.

我们的物理环境拥有许多不同的线索，这些线索被我们的感官系统所感知。当我们在环境中移动时，我们观察到感官线索的相应变化。在哺乳动物中，海马体及其内侧颞叶的邻近区域长期以来一直与空间导航和学习有关。在这个区域已经发现了几种类型的空间神经元，包括位置细胞和网格细胞。这些神经元的活动代表了动物当前的位置。尽管发现了空间的内部表征（或“地图”），但仍不清楚大脑是如何将环境感觉线索（如视觉地标）与自我运动信息（如运动或光流线索）结合起来形成这些地图的。因此，这个项目的重点是解开视觉和自我运动线索在空间映射过程中对位置细胞和网格细胞的影响。从历史上看，在成年动物身上研究这个问题是具有挑战性的，原因有三个。首先，在现实世界中，分离视觉和自运动输入的效果是很难实现的。其次，位置细胞网络和网格细胞网络是相互关联的，因此很难单独研究它们。第三，当动物进入一个新环境时，空间表征几乎是瞬间出现的。这表明动物从以前的经验中学习，可能发展出一种通用的代码，使它们能够根据需要快速构建新的空间表示。我最近开发了一个二维虚拟现实（2D VR）系统，为鼠标提供在虚拟世界中导航的身临其境的体验。这种开创性的发展使我处于一个独特的位置，可以回答以前具有挑战性的问题。新的2D VR允许在2D空间中独立操作视觉和自我运动线索。我的初步数据显示，虚拟世界中的空间表示与现实世界中的空间表示相似，但形成的速度要慢得多。因此，我们第一次有了一个延长的窗口来研究空间表征的形成，特别是视觉和自我运动线索对形成过程的影响。该项目将利用新的2D VR系统，研究位置细胞和网格细胞在构建空间表征中的不同作用。目的是了解位置和网格细胞如何相互作用，并结合视觉和自我运动线索来表示空间。我将首先建立成年小鼠在二维虚拟空间中空间表征形成的时间轴。接下来，我将区分视觉和自运动信息对形成空间表征的贡献。最后，我将测试这些线索的变化如何影响已建立的空间地图。该项目解决了BBSRC愿景中概述的关键挑战之一——“理解生命的规则”，并且与BBSRC的战略重点“生物科学的系统方法”完全一致。该项目为理解空间细胞之间的相互作用及其在空间学习中的功能提供了一个新的角度，为在人工智能、机器人导航和老龄化等领域的应用提供了基础。首先，这些发现将使计算神经科学家能够创建越来越精确的模拟长期记忆的模型，为人工智能的发展做出贡献。其次，这项工作为教授机器人如何整合多感官输入、执行复杂地形导航提供了见解。第三，这些发现将帮助我们了解正常衰老过程中记忆处理的神经基础，以及痴呆等神经退行性疾病。