research :: standardized facial expression stimuli
Cognitive research on face processing is constrained by several factors, but perhaps most importantly is the quality of the stimuli. Differences would be expected in subjects’ performance and the cognitive processes underlying this performance if the face stimuli are not realistic, or otherwise impoverished, such as using line drawings of faces or other schematics. Even presenting naturalistic photographs of face images is not without its difficulties. The subjects being photographed are often behind glass or bars, such as in a zoo setting, have their backs turned towards the camera, are moving so as not to produce high quality images. This is further confounded if the stimuli to be gathered are facial expressions. Researchers must wait for expressions to occur naturally, unprovoked, the individuals must not be moving too quickly, move behind structures, etc. Moreover, it is additionally difficult to capture facial expressions at their peak intensity, as these signals have rapid onset and offset points. Human studies overcome many of these issues by asking subjects to pose facial expressions according to strict criteria so that only prototypical and peak intensity movements are captured. Obviously doing this with nonhuman primates is impossible.
Therefore, in order create a standardized set of facial expression images, we have turned to a 3D computer animation software (www.efrontiers.com) called Poser to model peak intensity prototypical movements in the chimpanzee (and soon the macaque). The resulting stimuli are standardized across individual and show peak intensity movements as identified using the ChimpFACS system (see Parr et al., 2007). The figure 1 below shows an example of these movements next to a prototypical example taken from a naturally occurring context. These animations allow for the precise control of individual facial movements, the intensity of these movements and their overall configuration.
Figure 1
These images are then presented to subjects for discrimination using the Matching-to-Sample format, shown below in figure 2. In a) subjects see an example of a Poser facial expression (the bared-teeth display) and they must match this to one of two comparison images (b) showing the same or another category of facial expression.
Figure 2
Additionally, because Poser enables the construction of expressions movement by movement, we can deconstruct these facial expression images to show only individual movements. Below in figure 3, several individual movements are shown, corresponding to individual action units (AUs) identified by ChimpFACS. From left to right these are Action Unit 10: the upper lip raiser, AU12: lip corner puller, AU16: lower lip depressor, AU22: lip funneler, and AU27: mouth stretch.
Figure 3
