Abstract
Perceptual studies suggest that processing facial identity emphasizes information in the upper face, whereas processing expressions of anger and happiness emphasizes the lower face. Our goal was to determine if scanning fixations showed a similar upper/lower-face difference when shifting from identity to expression, consistent with a proposed role for fixations in information processing.
We tested eight subjects on two experiments. In each experiment subjects viewed a target face and then a pair of morphed faces, and had to report which of the pair was most similar to the target. In experiment 1, the two faces differed slightly in both expression and identity simultaneously, but in one set of trials the subjects were asked to judge which morph was most similar in identity, whereas in another set the subjects were asked to judge which morph was most similar in expression. Because the same stimuli were used in both sets, differences between identity and expression processing would represent top-down effects. In experiment 2, the two faces were morphed either in expression but not identity, or in identity but not expression, and the subject was not informed which dimension differed on a given trial: hence this experiment probed for bottom-up effects. We measured the number of fixations and total duration of fixations spent on the upper versus the lower half of the morphed faces. ANOVA showed a significant interaction between experiment and dimension (expression versus identity). Identity processing showed more upper-face scanning and expression processing showed more lower-face scanning during experiment 1 (top-down effects), whereas no difference was found between identity and expression processing in experiment 2 (bottom-up effects). We conclude that fixations are related to the regional distribution of information in face perception, but that these reflect top-down guidance of information acquisition rather than bottom-up effects of the stimulus.