Abstract
Previous research has shown that faces are difficult to recognize when viewed in reverse contrast (Galper, 1970). In a prior study, we demonstrated that this effect can be found with a set of novel objects (Greebles; Vuong et al. 2005), and that the effect was significantly greater for both faces and novel objects with pigment as opposed to those without pigment. This suggests that surface properties are integral in the representation of both faces and objects. However, it is an open question as to whether surface properties play the same role in the recognition of common objects which are not face-like. In addition, objects and faces shown in normal contrast are on average higher in luminance than those shown in reverse contrast. Here, we normalized all images to be equiluminant to eliminate this possible confound. We tested observers in a same/different sequential-matching task using gray scale images of common objects (cars and birds) as well as faces. Replicating prior results, faces were significantly more difficult to recognize when shown in reverse contrast. Importantly, we also found a contrast effect for the common objects; this effect was the same magnitude as the face contrast effect. These results indicate that performance costs associated with contrast reversal are not due to overall luminance differences between normal and reverse contrast images. Additionally, these data suggest that surfaces properties play a similar role in the recognition of all object types, not just those that are biological or face-like.
This research was paid for by a grant from the National Science Foundation (award # BCS-0339122).