Perception of Photographic-Quality Caricatures of Emotional Facial Expressions

Abstract

G Rhodes, S E Brennan, S Carey (1987 Cognitive Psychology19 473 – 497) and P J Benson and D I Perrett (1991 European Journal of Cognitive Psychology3 105 – 135) have shown that computer-enhanced (caricatured) representations of familiar faces are named faster and rated as better likenesses than veridical (undistorted) representations. Here we have applied Benson and Perrett's graphic technique to examine subjects' perception of enhanced representations of photographic-quality facial expressions of basic emotions. To enhance a facial expression the target face is compared to a norm or prototype face, and, by exaggerating the differences between the two, a caricatured image is produced; reducing the differences results in an anticaricatured image. In experiment 1 we examined the effect of degree of caricature and types of norm on subjects' ratings for ‘intensity of expression’. Three facial expressions (fear, anger, and sadness) were caricatured at seven levels (−50%, −30%, −15%, 0%, +15%, +30%, and +50%) relative to three different norms; (1) an average norm prepared by blending pictures of six different emotional expressions; (2) a neutral expression norm; and (3) a different expression norm (eg anger caricatured relative to a happy expression). Irrespective of norm, the caricatured expressions were rated as significantly more intense than the veridical images. Furthermore, for the average and neutral norm sets, the anticaricatures were rated as significantly less intense. We also examined subjects' reaction times to recognise caricatured (−50%, 0%, and +50%) representations of six emotional facial expressions. The results showed that the caricatured images were identified fastest, followed by the veridical, and then anticaricatured images. Hence the perception of facial expression and identity is facilitated by caricaturing; this has important implications for the mental representation of facial expressions.