Abstract
This study examined the difference between the processes for faces and non-face objects at the level of visual recognition and at the level of working memory. In Exp. 1, the ability of spatio-temporal integration in face and non-face object recognition was examined. Stimuli were pictures of faces, non-face objects and Japanese character Kanji, and each was divided into 16 elements (4 × 4 matrix). The picture elements were presented sequentially in a random order with various SOA (18 ms – 212 ms). Subjects answered who or what was shown in the picture. It was expected that the holistic (configural) information, which is thought to be important for face recognition, would be degraded more for longer SOA. The results showed that for longer SOA, performance for face was worse than non-face object, but the performance for Kanji was worst of the three conditions. Thus, it is suggested that the recognition of face needed holistic information compared to non-face object, but this conclusion may be ambiguous. In Exp. 2, the visuo-spatial short-term memory for face, non-face object and Kanji was examined using a working memory paradigm. Subjects observed a series of three (whole) pictures presented at different locations on the screen divided into 3 × 3 matrix, followed by a delay (4 s), and then a test picture appeared at a location of the matrix. Subjects judged whether the test picture or location was the same as one of the three pictures/locations. Subjects' performance of the location task was not different between faces, non-face objects and Kanji. However, the performance of the picture identity task was qualitatively different between faces, non-face objects and Kanji. For faces, the third picture was most easily identified (recency effect), and for non-face objects, the first and the third pictures were identified easily in comparison to the second picture (primacy and recency effects), but for Kanji, there were no significant difference. This suggests a possible specific short-term store for faces. These findings suggest that the processes for faces, non-face objects and Kanji are different at the level of visual working memory.