Biederman and Cooper (Cognitive Psychology, 23, 393-419, 1991), using parts-deleted and features-deleted stimuli, presented evidence that object priming occurs at the level of the objects’ parts, but not features. A control condition confirmed that some priming was accrued by the (non-visual) object concept that the stimulus represented, but all visual-level priming appeared to be at the more global level of the parts of objects, rather than the local level of the individual features (edges, vertices). This outcome has long been viewed as an important piece of supporting evidence for the existence of structural descriptions (e.g., Biederman, Psychological Review, 94, 115, 1987). The original report used a naming response task, and concluded that stimuli presenting half of the parts of an object primed only those parts, whereas half-features-deleted stimuli primed both themselves and the “complementary” half, containing the features deleted from the first image, equally. The current study adapts an eye-tracking approach to enable examination of the time course of priming across an exposure to both the primed image and unprimed competitors. Parts-deleted images primed themselves quickly and exclusively, replicating the finding of Biederman and Cooper (1991). Features-deleted images showed a deviation across time, however; initially a features-deleted prime attracted looking to itself and to its complement equally, but later on, looking to the target deviated upward, demonstrating an ability to distinguish between the two versions. The outcome of the present tests provide support for the primacy of a structural parts description, while also demonstrating the existence of multiple types of representations, both global and local.

中文翻译：

---

使用眼动追踪来解析对象识别：Priming 主要激活基于部件的表示，但也激活了基于后期出现的特征的表示。

Biederman 和 Cooper ( Cognitive Psychology, 23 , 393-419, 1991) 使用部分删除和特征删除的刺激，提供了证据表明对象启动发生在对象部分的水平，而不是特征。控制条件证实，刺激所代表的（非视觉）对象概念产生了一些启动，但所有视觉级别的启动似乎都处于对象部分的全局级别，而不是对象的局部级别。单个特征（边、顶点）。长期以来，这一结果一直被视为结构描述存在的重要支持证据（例如，Biederman，Psychological Review，94, 115, 1987)。原始报告使用了命名响应任务，并得出结论，呈现对象一半部分的刺激仅启动了这些部分，而删除了一半特征的刺激启动了它们自身和“互补”的一半，包含从第一个删除的特征形象，同样。目前的研究采用眼动追踪方法，以检查暴露于已启动图像和未启动竞争者的启动时间过程。删除部分的图像快速且排他性地启动，复制了 Biederman 和 Cooper (1991) 的发现。然而，特征删除的图像随着时间的推移显示出偏差；最初，一个特征删除的素数吸引着同样地看它自己和它的补，但后来，看目标向上偏离，展示了区分两个版本的能力。本测试的结果为结构部件描述的首要地位提供了支持，同时也证明了多种类型的表示的存在，包括全局和局部。