High-performing face recognizers use eye-eye distance and eye-nose distance more than low-performing face recognizers

摘要

It is widely understood that faces are recognized either holistically or configurally, in terms information that is spatially distributed across much of the face. Individuals with developmental prosopagnosia demonstrate profound difficulty when recognizing faces whereas "super recognizers" are individuals with exceptional ability to remember faces with very little effort (Russell & Nakayama, 2006). The current study explores face recognition in super recognizers in more detail, focusing independently on face discrimination ability and face recognition decision strategy using a two-part experiment. Face recognition ability was assessed using the Cambridge Face Memory Test (Duchaine & Nakayama, 2006) and participants who performed in the top 10% and bottom 10% were operationally defined as high- and low-performing face recognizers, respectively. In Part 1 of the experiment, we measured face discrimination thresholds for seven internal face dimensions (i.e. eye shape, nose-mouth distance, etc) using a 2-interval same-different task. Face stimuli were displayed in grayscale, and an adaptive thresholding procedure (QUEST) was used to measure the 79%-correct threshold for face discrimination. In Part 2 of the experiment, we measured face recognition strategy using a 3-alternative forced error (3-AFE) design, a modified 3-AFC, delayed match-to-sample task in which there is no correct answer (Schwartz & Chang, 2008). Distractors were calibrated using thresholds measured in Part 1 and each distractor varied in only one dimension from the target so participant response indicated the dimension with the least utility. Results show that high-performing face recognizers rely more on eye-eye distance and eye-nose distance compared to low-performing face recognizers. High performers also deemphasize eye shape and nose-mouth distance compared to low performers.