Becoming a “Greeble” Expert: Exploring Mechanisms for Face Recognition

Abstract

Sensitivity to configural changes in face processing has been cited as evidence for face-exclusive mechanisms. Alternatively, general mechanisms could be fine-tuned by experience with homogeneous stimuli. We tested sensitivity to configural transformations for novices and experts with nonface stimuli ("Greebles"). Parts of transformed Greebles were identified via forced-choice recognition. Regardless of expertise level, the recognition of parts in the Studied configuration was better than in isolation, suggesting an object advantage. For experts, recognizing Greeble parts in a Transformed configuration was slower than in the Studied configuration, but only at upright. Thus, expertise with visually similar objects, not faces per se, may produce configural sensitivity.

Perceptual learning and inversion effects: Recognition of prototype-defined familiar checkerboards.

“…The results, however, showed that features of upright faces were better recognized in their original configuration than in the other two conditions. Gauthier and Tarr (1997) showed that this advantage for a configuration (better feature identification in the original than in the distorted configuration) could also be obtained with Greebles (an artificial category) once the participants became experts with these stimuli. In 5 addition Gauthier and Tarr (1997) as well as Tanaka et al (1997) showed that experts could display a whole/part advantage with stimuli such as Greebles, cars or cells.…”

“…Gauthier and Tarr (1997) showed that this advantage for a configuration (better feature identification in the original than in the distorted configuration) could also be obtained with Greebles (an artificial category) once the participants became experts with these stimuli. In 5 addition Gauthier and Tarr (1997) as well as Tanaka et al (1997) showed that experts could display a whole/part advantage with stimuli such as Greebles, cars or cells. Tanaka and Gauthier (1997) suggested that, although novices may sometimes rely on first-order relational information (for a facial stimulus an example of this would be the nose in relation to the mouth) only experts seem to rely on second-order relational information (the variations in first-order relations relative to the prototype for that stimulus set).…”

“…In 5 addition Gauthier and Tarr (1997) as well as Tanaka et al (1997) showed that experts could display a whole/part advantage with stimuli such as Greebles, cars or cells. Tanaka and Gauthier (1997) suggested that, although novices may sometimes rely on first-order relational information (for a facial stimulus an example of this would be the nose in relation to the mouth) only experts seem to rely on second-order relational information (the variations in first-order relations relative to the prototype for that stimulus set).…”

Image-based object recognition in man, monkey and machine

“…The degree of experience an individual has had with a class may help to determine the default level of access for recognition, how sensitive recognition is to image transformations, e.g. brightness reversal, and to changes in configural information (Gauthier and Tarr, 1997a;Tanaka and Sengco, 1997;Gauthier et al, 1998). Models of recognition must be sufficiently plastic to adapt as experience with an object class accumulates.…”

“…Second, extensive experience with the same features in a consistent configuration will give rise to more complex features. These simple statistical learning mechanisms offer an explanation for the configural sensitivity found in cases of perceptual expertise, including face recognition (Gauthier and Tarr, 1997a;Tanaka and Sengco, 1997). Consider that the acquisition of expertise is marked by extensive practice differentiating similar instances from within a class.…”