Abstract: Although it is acknowledged that human face is processed holistically, little is known about how global configural information interacting with local feature information during holistic face processing.
To address this issue, we measured subjects’ performance in a face learning and identification task. Sixty-three college students participated in this study which consisted of three experiments. In experiment 1, thirty-six subjects were asked to give subjective scores of face clarity modulated through variant low-pass spatial filters. Five low-pass filter levels were chosen for experiment 2 and 3. In experiment 2, fourteen subjects were instructed to identify which testing face picture was previously presented at the beginning of the task. The test pictures were modulated at two levels. At global level, low-pass filters were applied on the whole face; at local level, one of four face features was masked by a focus window containing corresponding non-filter feature. The distracting testing picture was masked at different face feature from target picture. A repeated two-factor (5 low-pass filter levels ´ 4 face features) ANOVA analysis was conducted to analyze the behavior performance data. In experiment 3, thirteen subjects performed the same task except for using inverted face instead. And same ANOVA analysis was used to analyze the performance data.
The results of experiment 2 revealed that the processing of local features within the focus window was more difficult when global configural information outside of the window filtered at lower level. The similar pattern was also observed in experiment 3. These findings support the holistic face processing and further suggest the competing relation between global configural and local feature information processing. In addition, whether face processing is based on configural or feature information depends on the extracting difficulties of feature information.
We have found the interacting pattern between global configural information and local feature information during holistic face processing. This specific pattern shows that the crucial information type for sucessful face processing is dynamic changed depending on the information extracting difficulties, which indicates the adaptability of human visual perception system.

Key words: face process, holistic, confitrual information, feature information, competition relation model