Dynamic composite faces are processed holistically

Abstract

Holistic processing is considered one of the hallmarks of face recognition. Recent studies using the composite task claim to show a lack of holistic processing for dynamic faces, however they only presented moving faces in the learning phase and tested with static composite images. So while previous research has addressed the question of whether moving faces influence the processing of subsequently viewed static faces, the question of whether moving faces are processed holistically remains unanswered. We address that question here. In our study participants learned faces in motion and were tested on moving composite faces, or learned static faces and were tested on static composite faces. We found a clear composite effect for both upright static and dynamic faces, with no significant difference in the magnitude of those effects. Further, there was no evidence of composite or motion effects in inverted conditions, ruling out low level or other motion signal properties as explanations of performance in upright faces. Together, these results show that upright moving faces are processed holistically, in a similar manner to static faces, extending decades of research with static faces and confirming the importance of holistic processing to familiar face recognition.

Holistic processing of static and moving faces.

“…Favelle et al, 2015;Xiao et al, 2012). Note that there are many methodological differences between Xiao et al (2012) and the present study (e.g., angular differences between successive face views, 26° vs 5°, ranges of face views, 180° vs. 60°, whole faces with hairstyle vs.…”

“…Rigid facial motion (i.e., head rotation and nodding) has been shown to eliminate the composite face effect (Xiao et al, 2012), but non-rigid facial motion (e.g., smiling) often does not (Cook et al, 2015;Favelle et al, 2015;Steede & Hole, 2006;Xiao et al, 2013). Such discrepant results occur not only between studies of rigid and non-rigid facial motion but also between different studies investigating non-rigid facial motion.…”

“…Such discrepant results occur not only between studies of rigid and non-rigid facial motion but also between different studies investigating non-rigid facial motion. For instance, while one study showed that non-rigid facial motion (i.e., simultaneous eye-blinking and mouth chewing) reduced the composite face effect in comparison to static faces (Xiao et al, 2013), other studies found that faces expressing natural facial movements (e.g., happy expression) are processed as holistically as static faces (Cook et al, 2015;Favelle et al, 2015;Steede & Hole, 2006). Thus, how facial motion affects holistic face processing not only remains to be characterized across different theoretical hypotheses, but also remains to be synthesized across different empirical studies.…”

“…For instance, Favelle et al (2015) asked participants to learn the names of top-half faces showing happy expression, and then to perform a naming task for the top face halves embedded in composite faces. The composite faces were made of learned face-tops and novel face-bottoms, with the two halves were either aligned or misaligned.…”

“…This finding also holds true for dynamic faces animated by natural, non-rigid facial motion. Faces displaying dynamic happy expression are processed as holistically as static happy faces (Favelle et al, 2015). However, holistic processing seems to be less prevalent for faces moving non-rigidly but unnaturally than for static faces (e.g., faces exhibiting eye-blinking and mouth-chewing simultaneously, Xiao et al, 2013).…”