Item

Abstract

Using computer animated laserscans of human faces we have recently shown that complex non-rigid facial motion patterns applied to previously unfamiliar faces can bias the perception of identity (Knappmeyer, Thornton, Bülthoff, 2001). Here, we further explore this effect by varying the amount of form (shape and texture) information available at training. Our intuitive hypothesis was that the form cue might dominate over the motion cue. Thus enhancing the form cue at training should weaken any biasing effects of facial motion. This was true when individual texture was added to the training faces. However, the motion effect was still observable and further tests with an animated average head showed that observers had nevertheless learned the individual motion patterns under these conditions. Simply enhancing the shape information of the inner facial features did not affect the motion bias at all. Finally, facial motion still biased observers’ identity decisions when the faces were presented upside down. In sum, the motion bias seems to be more robust than initially expected suggesting that both changeable and invariant aspects of faces can be useful cues to identity. Using similar techniques, we are currently investigating whether facial motion also affects other aspects of face processing such as facial attractiveness.