Abstract
Adaptation to configural distortions in faces has been found to show selectivity for some changes in facial identity (e.g. for different individual faces, genders, or ethnicities), while strongly transferring across some stimulus dimensions that preserve perceived identity (e.g. differences in image size or mean color). We asked how the adaptation adjusted to variations in facial expression, to assess whether a change in expression was equivalent to adaptation to the “same” or a “different” face. Stimuli were frontal view images of Dutch female or male faces with happy or fearful posed expressions from the Radboud Face Database. The images were distorted by a local horizontal expansion or contraction of the face relative to a midpoint on the nose. Observers adapted for 2 min to a single distorted face or to an alternation between two faces with opposite distortion (from the same individual with different expression or different genders with the same expression). Test faces were the same as the adapt face(s) or a face with different expression or gender, and were shown 1.5 times smaller than the adapt images to reduce low-level aftereffects. The tests alternated with 4 sec readaptation intervals, with the distortion level varied in staircases to estimate the level that appeared undistorted. In adaptation to either the single or opposing face pairs, aftereffects showed modest selectivity (equivalent to ~ 80% transfer) for the change in expression. However, the degree of transfer was roughly comparable to changes in the gender of the faces. The results thus suggest that natural forms of intra- and inter-individual variations in the face have roughly similar effects at the levels affected by the adaptation, and specifically that at least part of the adaptation precedes sites at which different expressions of the same face are represented as an equivalent identity.