Abstract
Most studies of face recognition use tightly controlled images but in the real world extrinsic (e.g., camera, lighting) and intrinsic (e.g., make-up, viewpoint) factors influence the appearance of a face. Two pictures of the same person can look very different and pictures of two different people can look very similar. Adults easily recognize familiar faces despite within-person variability in appearance, but frequently perceive two photos of an unfamiliar identity as belonging to different people (Jenkins et al., 2011), an error that is more common in children (Laurence & Mondloch, 2015). We examined the process by which a newly encountered face becomes familiar. In Experiment 1 adults (n=72) watched a 10-minute video in which a model read a storybook as filmed on a single day (low variability, LV) or across three days to incorporate variability in intrinsic and extrinsic factors (high variability, HV). Participants then were given a pile of novel photographs and asked to find all of the pictures of the model (n=9). Both training groups recognized more photos (Ms=7.0 [HV] and 6.8 [LV]) than the no-training control group (M=5.4), ps< .01, with no effect of condition on false alarms (misidentifying a similar looking distractor as the model). In Experiment 2 we assessed the effects of variability and training duration; MTurk workers (n=288) watched a 1- or 10-minute video in the HV or LV condition. Accuracy was not affected by either variable, ps>.20. We conclude that dynamic stimuli (which more closely resemble how we learn faces in daily life) incorporate sufficient variability to support face learning and that adults learn faces efficiently. An ongoing study with children (aged 6 to 12 years; n = 36 to date) suggests they learn less efficiently than adults.
Meeting abstract presented at VSS 2016