Reaction Time Measures of Feature Saliency in a Perceptual Integration Task

Abstract

Research on cue saliency in facial recognition has relied heavily on two basic paradigms which can broadly be described as (i) recognition and (ii) recall. The recognition paradigm involves the identification of a familiar face from one or more of its individual features or examining the effects of alteration of particular features on identification. The recall paradigm utilises the frequency with which fragments or features appear in subjects descriptions of a previously presented stimulus. Highly salient features which subjects attended to would be expected to constitute a substantial part of their description. Recognition and recall studies have established that particular aspects of the face are more salient than others (c.f. Shepherd, Davies and Ellis 1981). When Caucasians view caucasian faces the hair and the eyes emèrge as probably the most important cues in both paradigms. The nose features prominently in verbal descriptions of faces (Ellis et al. 1980, Shepherd et al., 1977) but does not appear to be very affective as an isolated cue for identification (Goldstein and Mackenberg 1966, Seamon et al., 1978). It is also apparent that subjects select the hair and forehead region first, then the eyes followed by the nose and mouth when they construct Photofit faces and the accuracy of feature selection follows the same order (Ellis et al., 1977). Laughery et al. (1977) and Ellis et al.(1977) report that the amount of time devoted to the hair/forehead and eye region in Photofit and Identikit reconstruction and when instructions are given to sketch artists is greater than that spent on lower facial features.