Navon David
Department of Psychology, University of Haifa, Haifa, Israel.
Q J Exp Psychol (Hove). 2011 Mar;64(3):608-24. doi: 10.1080/17470218.2010.516834. Epub 2010 Nov 10.
Though figure-ground assignment has been shown to be probably affected by recognizability, it appears sensible that object recognition must follow at least the earlier process of figure-ground segregation. To examine whether or not rudimentary object recognition could, counterintuitively, start even before the completion of the stage of parsing in which figure-ground segregation is done, participants were asked to respond, in a go/no-go fashion, whenever any out of 16 alternative connected patterns (that constituted familiar stimuli in the upright orientation) appeared. The white figure of the to-be-attended stimulus-target or foil-could be segregated from the white ambient ground only by means of a frame surrounding it. Such a frame was absent until the onset of target display. Then, to manipulate organizational quality, the greyness of the frame was either gradually increased from zero (in Experiment 1) or changed abruptly to a stationary level whose greyness was varied between trials (in Experiments 2 and 3). Stimulus recognizability was manipulated by orientation angle. In all three experiments the effect of recognizability was found to be considerably larger when organizational quality was minimal due to an extremely faint frame. This result is argued to be incompatible with any version of a serial thesis suggesting that processing aimed at object recognition starts only with a good enough level of organizational quality. The experiments rather provide some support to the claim, termed here "early interaction hypothesis", positing interaction between early recognition processing and preassignment parsing processes.
尽管图形-背景分配已被证明可能受到可识别性的影响,但物体识别至少必须遵循图形-背景分离的早期过程,这似乎是合理的。为了检验直观上甚至在完成图形-背景分离的解析阶段之前,基本的物体识别是否就可以开始,要求参与者在16种备选连接模式(在直立方向上构成熟悉刺激)中的任何一种出现时,以是/否的方式做出反应。待关注的刺激目标或陪衬的白色图形只能通过围绕它的边框与白色背景区分开来。在目标显示开始之前,这样的边框是不存在的。然后,为了操纵组织质量,边框的灰度要么从零开始逐渐增加(实验1),要么突然变为一个固定水平,其灰度在不同试验中变化(实验2和3)。通过方向角来操纵刺激的可识别性。在所有三个实验中,当由于边框极其模糊而导致组织质量最低时,发现可识别性的影响要大得多。这一结果被认为与任何版本的序列论点都不相容,序列论点认为旨在物体识别的处理仅在足够好的组织质量水平上才开始。这些实验反而为这里称为“早期交互假设”的主张提供了一些支持,该主张假定早期识别处理与预分配解析过程之间存在交互作用。