The tree in the bathroom: the role of the incongruous in understanding the gist of a scene

我们可以对真实世界的场景一目了然(卧室、田野等)。在部署注意力和个体对象识别之前形成场景的初始表示。这一场景《主旨》是基于全局信息，在整个场景中集成的。然而，自然场景中可能包含不协调的物体，这可能会与主旨相矛盾(‘嵌合’场景)。例如，一个场景的主旨可能是‘客厅’，它将包含沙发、椅子和地毯等相同的对象。一个嵌合的场景可能包含不协调的物体，如树和湖，通过窗户看到。不一致的信息会延迟或干扰场景主旨的快速计算。或者，这种全球过程可能足够复杂，足以打消不一致的信息，甚至支持两位科学家(在我们的例子中，客厅和森林湖)。我的研究将检验这一假设，即人类的主旨处理对不协调是稳健的，并将检验这种稳健性的限度。这项工作将阐明人类场景感知的基本构件；日常视觉的基本组成部分，以及从驾驶车辆到医疗诊断的许多专门任务的要求。

We can categorize a real world scene in a single glance (bedroom, field, etc). An initial representation of a scene is formed before the deployment of attention and individual object identification. This scene “gist” is based on global information, integrated across the entire scene. However, natural scenes may contain incongruent objects in them that could contradict the gist (‘chimeric’ scenes). For example, a scene whose gist might be ‘living room’ would contain congruent objects like sofas, chairs, and carpet. A chimeric scene might contain incongruent objects like trees and a lake, seen through a window. Incongruent information could delay or disrupt the rapid calculation of scene gist. Alternatively, this global process may be sophisticated enough to discount incongruent information or even to support two gists (living room & forest lake in our example). My research will test the hypothesis that human gist processing is robust against incongruity and will examine the limits of that robustness. This work will elucidate basic building blocks of human scene perception; a fundamental component of everyday vision and a requirement for many specialized tasks from driving vehicles to medical diagnosis.