Facial distinctiveness and the power of caricatures

Caricatures, which increase the distinctiveness of faces, are generally recognised at least as well as undistorted images of those faces. However, caricatures seem to facilitate recognition more for some faces than others. An investigation was made into whether the effectiveness of caricaturing depends on a face's initial distinctiveness. In experiments 1-3, subjects learned names for unfamiliar faces (photographs) that varied in distinctiveness, and were tested on recognition of caricatures, anticaricatures, and undistorted images of those faces. The test images were line drawing in experiment 1 and 2 photographic images in experiment 3. Experiments 1 and 2 were identical except that subjects had more exposure to the study photographs in experiment 1. In all three experiments, distinctive faces were recognised (named) more accurately than less-distinctive faces, and caricatures were recognised at least as accurately as undistorted images and better than anticaricatures. However, distinctiveness and caricature level did not interact. Nor did a face's initial distinctiveness correlate with the degree of recognition facilitation produced by caricaturing (experiments 1-3) or with the caricature level chosen as the best likeness (experiment 4). The effectiveness of caricatures varied across faces and experimental conditions, but these differences did not relate to differences in initial distinctiveness. These results prompted a more careful analysis of the expected relationship between initial distinctiveness and the power of caricatures, which indicated that the relationship may be curvilinear rather than linear. In addition, it was found that line-drawing caricatures functioned as superportraits (recognised better than undistorted images--experiment 1) but photographic caricatures did not (experiment 3) suggesting that the forensic potential of caricatures may be limited.