Natalia Pallis-Hassani¹, Shruti Japee¹, Maryam Vaziri-Pashkam², Christopher Baker¹; ¹National Institutes of Health, ²University of Delaware

Our ability to extract meaning from facial expressions is critical for our social interactions. Recent work has proposed a third visual pathway specialized for motion processing, which may be vital for understanding facial expressions. While prior research on facial expression perception has typically used static facial expressions, dynamic facial motion, particularly of eyes and mouth, has recently been shown to improve recognition of facial expressions. Thus, in this study, we examined whether facial motion cues—in the absence of underlying facial features—are sufficient for recognizing facial expressions. To answer this question, we converted dynamic video stimuli of various facial expressions into random dot kinematograms (RDKs) using the underlying optic flow information in the videos. The resulting RDK videos included seven facial expressions—happy, sad, angry, disgusted, fearful, surprised, and neutral—across 22 actors. Participants were shown each of the 154 intact and RDK videos and asked to label the facial expression. Analyses of the labeling data revealed that while average accuracy rates for RDK videos were lower than those for intact videos, participants achieved above-chance accuracy for all expressions—happy: 62%, surprise: 47%, sad: 46%, disgust: 45%, anger: 28%, fear: 26%, and neutral: 98%, with chance accuracy being 14.3%. These results support the idea that motion cues in the absence of facial features are sufficient for identifying expressions. Analysis of eye-gaze data acquired while participants performed the labeling task will help identify facial motion cues that are most salient for the recognition of different expressions. In a future functional magnetic resonance imaging (fMRI) study, these RDK videos will be used to examine whether regions in the third visual pathway process facial motion cues in the absence of the underlying facial features.