Abstract
Purpose. We have found in our past study that the characteristics of gaze perception for stimuli with only two eyes differ from those with whole-face stimuli. In this study, we first compare the effects of low-pass filtering on gaze perception for whole-face and eye-only stimuli, then, we examined the effect of adding the external contour of face to the eye-only stimuli. Our past results have indicated that low-pass filtering of whole-face images has no effect on gaze perception performance.
Methods. The stimuli were real-size color digital pictures of two young persons with eyes shifted in 11 steps between 15 deg to the right and to the left from the center. The eye-only stimuli were created by cutting the eye areas from the original picture and placing them on a background of the average color/luminance of the face. For the low-pass-filtered stimuli, the two types of pictures were low-pass filtered with 3 cut-off frequencies ranging from 1.4 to 5.6 c/d. These pictures were presented on a CRT screen with a random order and subjects were asked to adjust the maker on the string running horizontally right in front of their face.
Results. For the whole-face stimuli, although there was a tendency to overestimate eye deviation for larger deviation angles, no effect of low-pass filtering was found regardless the cut-off frequencies. However, a clear effect of low-pass filtering was found for the eye-only stimuli. The overestimation increased as cut-off frequency was lowered. This effect of low-pass filtering remained even when the external contour of the face was added as a black line drawing.
The present results indicate that gaze perception relies on the measurements of the dark part of eyes relative to facial parts other than eyes themselves. Since the external contour of the face does not have much cotribution, the internal features of face such as eyebrows, nose or mouth seem to have important role in accurate gaze perception.