Purchase this article with an account.
Megumi Kobayashi, So Kanazawa, Masami Yamaguchi, Ryusuke Kakigi; Infants' face detection in natural scene. Journal of Vision 2017;17(10):454. doi: 10.1167/17.10.454.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In adult studies, it has been reported that human make use of non-facial scenic information for detecting faces (Lewis & Edmonds, 2003; Lewis & Edmonds, 2005). That is, detection of a face is significantly faster when a face appears in an intact scene than when a face appears in a scrambled scene. This suggests that we process scenes in parallel, and use the information from scene in a pre-attentive manner to detect the presence of face. In the developmental studies, previous studies have reported that infants prefer the upright face over inverted face, suggesting the early emergence of face detection. However, these previous studies presented faces in isolation from scenes, and little is known about development of infants' face detection in natural scene. Therefore, the aim of this study was to investigate whether infants utilize scenic information when they detect a face, as shown in adults. To this end, we examined 4- to 7-month-old infants' visual preference for upright image than for inverted one in two conditions: intact (a face occurs in an intact scene) and scrambled (a face occurs in a scrambled scene). In each condition, infants saw three different images. We found that 6- and 7-month-old infants preferred the upright image only in intact scene, but not in scrambled scene. However, only 7-month-olds showed significant difference between intact and scrambled scene in upright image preference. In contrast, 4- and 5-month-old infants showed upright image preference both in intact and scrambled scene. In sum, younger infants detect face regardless of whether the scene is intact, whilst older infants detect face only in intact scene. Our results suggest that infants aged over 7 months would process scenic information in pre-attentive manner in order to detect faces.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only