Abstract
Previous research suggests that learning to categorize faces along a novel identity dimension (created by morphing between two unfamiliar faces) changes the perceptual representation of the category-relevant dimension, increasing its discriminability, its separability from other dimensions, and the information used to identify faces varying along the dimension. An open question is whether categorization training modifies the encoding of face identities at the extremes of the category-relevant dimension (i.e., the parent faces used to create such dimension). This encoding has been proposed to be norm-based, a hypothesis supported by studies showing that recognition of a face identity is facilitated by adaptation to an "anti-face". Here, we trained a group of participants to categorize faces that varied along two morphing dimensions, one of them relevant to the categorization task and the other irrelevant to the task. A control group did not receive such categorization training. During test, thresholds for the identification of one of the extremes of the category-relevant dimension were obtained using the method of constant stimuli, with stimuli varying from an average face to the target identity. For each participant, thresholds were obtained in three conditions: no adaptor, anti-face adaptation, and category adaptation (exposure to the other extreme in the category-relevant dimension). Surprisingly, categorization training had very little effect on identification thresholds in the absence of an adaptor, suggesting that increments in discriminability commonly found after categorization are specific to the trained morphed dimension, and do not transfer to a task involving detection of a single dimension extreme. Secondly, anti-face adaptation produced a reduction in thresholds that was comparable in both groups, suggesting that categorization training does not influence the norm-based encoding of face identity. Finally, category adaptation produced a reduction in thresholds that was slightly stronger after categorization training.
Meeting abstract presented at VSS 2018