Learning faces from variability

Abstract

Research on face learning has tended to use sets of images that vary systematically on dimensions such as pose and illumination. In contrast, we have proposed that exposure to naturally varying images of a person may be a critical part of the familiarization process. Here, we present two experiments investigating face learning with “ambient images”—relatively unconstrained photos taken from internet searches. Participants learned name and face associations for unfamiliar identities presented in high or low within-person variability—that is, images of the same person returned by internet search on their name (high variability) versus different images of the same person taken from the same event (low variability). In Experiment 1 we show more accurate performance on a speeded name verification task for identities learned in high than in low variability, when the test images are completely novel photos. In Experiment 2 we show more accurate performance on a face matching task for identities previously learned in high than in low variability. The results show that exposure to a large range of within-person variability leads to enhanced learning of new identities.

Searching for faces in crowd chokepoint videos

“…In Experiment 2, we found that multiple low, but not high, variability images produced improvements in searching in comparison with single images. Although research suggests that higher variability should be expected to produce greater benefits (Menon et al, 2015;Ritchie & Burton, 2017), the current result is likely due to recency and/or context rather than variability. Although multiple images may result in benefits in terms of the information provided of the target's face, there are clearly situations in which larger variability does not necessarily produce better performance (e.g., when that variability informs regarding how someone looked several years ago).…”

“…Our results replicated this effect for searching, with increased prior familiarity with the target resulting in increased likelihood of a correct response. (Ritchie & Burton, 2017) performance, with larger variation across images of the same individual producing increased performance (Menon et al, 2015;Ritchie & Burton, 2017). Further, initial evidence suggests that viewing multiple images of a target may also prove beneficial when searching for that person in a crowd (Mileva & Burton, 2019).…”

“…Previous research suggests that providing variability information, through multiple images of an unfamiliar face, may improve both matching (White, Burton, Jenkins, & Kemp, 2014; cf. Ritchie et al, ) and learning (Ritchie & Burton, ) performance, with larger variation across images of the same individual producing increased performance (Menon et al, ; Ritchie & Burton, ). Further, initial evidence suggests that viewing multiple images of a target may also prove beneficial when searching for that person in a crowd (Mileva & Burton, ).…”

“…Recent evidence suggests that providing multiple images of an unfamiliar face may result in improvements with both matching (White, Burton, Jenkins, & Kemp, ; cf. Ritchie et al, ) and learning (Ritchie & Burton, ). Further, larger variation across images of the same individual appears to produce increased performance in both cases (Menon, White, & Kemp, ; Ritchie & Burton, ).…”

“…Ritchie et al, ) and learning (Ritchie & Burton, ). Further, larger variation across images of the same individual appears to produce increased performance in both cases (Menon, White, & Kemp, ; Ritchie & Burton, ). As such, viewing multiple images of a target may also prove beneficial when searching for that person in a crowd.…”