Cognitive and Neural Representations of Reachable Environments

项目摘要 这项拟议的研究的广泛目标是描述人类感知和 表示可访问环境的视图。“触角空间”是我们世界不可分割的一部分 和日常体验，但这些观点是如何被视觉认知系统处理的 未知。目前的工作旨在提供对内部结构的基本见解 可达环境的表示形式。首先，建立一个庞大的空间图像数据库 将会被建造。接下来，该提案将利用最近开发的方法，该方法 采用大规模的行为实验，结合稀疏的正向嵌入模型， 它将被用来导出作为到达空间相似性结构的基础的属性 视图。最后，将测量大脑对选定的到达空间视图子集的反应，并 使用导出的尺寸进行建模。这些方法已经过测试，并已用于 物体和场景处理领域的进展；在这里，我们将它们应用于小说 触角空间的刺激区域。这项工作有可能在两国之间建立新的桥梁 通过表征高级视觉表示的视觉和视觉运动社区 并将扩大对人类如何感知环境的科学理解 视觉环境。

Project Summary The broad goal of the proposed research is to characterize the way that humans perceive and represent views of reachable environments. “Reachspaces” are an integral part of our world and day-to-day experience, but how these views are processed by the visual cognitive system is unknown. The present work aims to offer fundamental insights into the structure of internal representations of the reachable environments. First, a large database of reachspace images will be constructed. Next, this proposal will leverage a recently developed approach that employs a large-scale behavioral experiment, coupled with a sparse positive embedding model, which will be used to derive the attributes underlying the similarity structure of reachspace views. Finally, brain responses to a selected subset of reachspace views will be measured and modeled with the derived dimensions. These approaches have been tested and used to great advance in domains of object and scene processing; here, we employ them on the novel stimulus domain of reachspaces. This work has the potential to form new bridges between the vision and visuomotor communities by characterizing high-level visual representations of reachable environments and will expand scientific understand of how humans perceive the visual environment.