Abstract
When a television show or movie is paused, people who are frozen in mid-action typically look much less flattering than they do in natural motion. What causes this frozen face effect? Here we conducted two experiments to quantify the effect and determine its cause. In the first experiment, we presented 40, two-second video clips of people speaking in naturalistic settings (e.g., news programs, talk show interviews, etc). We also presented all of the static frames that comprised each video, interleaved in a random order within the same session. Using a 7-point Likert scale, subjects rated how flattering each stimulus was. Flattery ratings of the videos were significantly higher than average flattery ratings of static images derived from the videos. In the second experiment, we used an ABX discrimination task to measure recognition of the static faces that comprised each video. In each trial, a video was presented, followed immediately by two static images: a target image from the video, and a lure from another very similar video. Subjects were required to judge which static image was a member of the preceding video. Overall, there was a strong recency effect; subjects were more accurate at recognizing static images near the end of the video, despite the random order of presentation. More interestingly, subjects were more accurate at recognizing target images that had flattery ratings close to the source video's rating (Experiment 1). The results suggest that when viewing dynamic faces, the visual system selectively filters or suppresses outlying, unflattering faces.