面孔知觉中特征、结构和整体加工策略的眼动研究

An Eye-tracking Research on Featural, Configural and Holistic Processing Strategies in Face Perception

考察面孔知觉中特征加工、结构加工和整体加工三种不同加工方式对应的眼动模式。实验1中,将以特征信息为主的错乱面孔和以结构信息为主的模糊面孔作为线索刺激,引发对测试面孔的特征加工和结构加工,眼动分析表明:特征加工表现为对面孔各特征内更长的凝视时间,结构加工表现为对面孔各特征间高频的眼跳。实验2采用相同的研究范式,将完整面孔、轻度错乱面孔和低水平模糊面孔作为线索刺激,引发对测试面孔除特征加工和结构加工外的另一种加工方式——整体加工,表现为注视点更多地落在测试面孔中央区的鼻子部位以扩大注视范围,进而把握整张面孔信息。本研究揭示了三种不同面孔加工方式眼动模式的差异。

There were two main hypotheses attempting to explain the mechanisms of face perception. One was the holistic approach which claimed that faces were stored as relatively undifferentiated wholes, without explicitly representing the facial parts. The other was the dual-code view which claimed that faces were processed on the basis of featural and configural information and that the respective representations were stored in isolation. Indeed, neither of the two approaches could provide a detailed explanation of the mechanisms of face perception. Therefore integrated approaches were proposed. The present study in favor of these integrated systems aims at exploring featural, configural and holistic strategies during face perception. The presence of different eye movement patterns would imply how important information is extracted from facial stimuli. In the two experiments we explored the scanning behavior during face perception. In experiment 1, modified faces with primarily featural （scrambled faces） or configural （blurred faces） information were used as cue stimuli so as to manipulate the way participants processed intact faces which were presented soon afterwards. In a series of matching-to-sample tasks, participants decided whether the identity of an intact test face matched a precedent scrambled or blurred cue face. Analysis of eye movements for test faces showed more interfeatural saccades when preceded by blurred faces, and longer gaze duration within the same feature when preceded by scrambled faces. In experiment 2, a similar paradigm was used except that test faces were cued by intact faces, low-level blurred faces, or slightly scrambled faces. The results suggested that in the intact condition participants performed fewer interfeatural saccades than in low-level blurred condition and had shorter gaze duration than in slightly scrambled condition. Moreover, a few fixations in the center of the face were performed to grasp the information from the whole face. The results suggest a differentiation between featural, configural, and holistic processing strategies, which can be associated with specific patterns of eye movements. By inducing different processing strategies in experiment 1, we provided further evidence that faces can be recognized on the basis of both featural and configural information. In experiment 2 a holistic strategy characterized by the grasp of the whole face was applied, while in experiment 1 a configural strategy was adopted when the information from different locations was integrated by performing a high number of saccades between different features. In conclusion, the present study shows that three modes of face processing strategies coexist in face perception.