Learning and recognizing faces across variability in appearance: An examination of children and older adults

Abstract

Recognizing facial identity requires two skills: telling a person apart from similar looking people and recognizing them across changes in their appearance. Until recently, the vast majority of studies relied on tightly controlled images to examine face learning and recognition. Research using ambient images (i.e., images that capture within-person variability in appearance) is necessary to assess the true challenge of face learning and recognition in daily life. Only a few studies have examined face learning and recognition using ambient images in children, and, to the best of my knowledge, no studies have examined them in older adults. My dissertation was designed to address these gaps in the literature. In Study 1, children aged 6 to 11 were tested to examine two mechanisms that underlie face learning in young adults: Ensemble coding and the ability to benefit from exposure to variability in appearance in a perceptual matching task. My results revealed that both mechanisms are adultlike by the age of 6. First, children extracted the average of a set of images of an identity, regardless of whether those images were presented simultaneously or sequentially. Second, although their overall accuracy was lower than that of young adults, children showed comparable benefit from viewing multiple images of a to-be-learned identity in a perceptual face learning task. In Study 2, I examined whether younger children (4- and 5-year-olds) benefit from exposure to multiple images when learning a new face in a perceptual task. Although viewing multiple images made young children more sensitive to identity, it also led them to adopt a less conservative response bias, driven both by an increase in hits and an increase in false alarms. This increase in false alarms was not found for older children and adults in Study 1, suggesting that the ability to benefit from exposure to variability in appearance during face learning is not fully refined before the age of 6. In Study 3, I provided the first examination of face learning and recognition in older adults using a battery of tasks. On three of the five tasks, older adults showed comparable learning and recognition to young adults: 1) Older adults recognized a familiar face without error; 2) they showed ensemble coding of facial identity, regardless of whether the images were presented simultaneously or sequentially and 3) despite making more errors than young adults overall, they showed comparable benefit from viewing multiple images of a newly encountered face in the perceptual learning task. In the remaining two tasks, older adults showed a different pattern than young adults: 1) Older adults made fewer hits and more false alarms than young adults when matching images of wholly unfamiliar faces; and 2) after being exposed to low variability in appearance in a face memory task, older adults became more conservative than did younger adults, despite showing comparable benefits in sensitivity. My results reveal that the same abilities that show prolonged development during childhood are those undergo changes in aging. Collectively, my dissertation provides novel insights about learning and recognizing facial identity during childhood and aging and has important implications for understanding models of face processing.