Abstract
Background: Previous studies have shown observers rely more on internal than external features when recognising familiar but not unfamiliar faces. Objective: We used an adaptation paradigm to examine whether this difference in internal and external feature contributions to processing is also reflected in differences in the representations of these two classes of faces in the human visual system. Methods: Twelve subjects adapted to a) whole faces, b) internal features alone, or c) external features alone for 5sec, and were then asked whether a briefly shown ambiguous whole-face most resembled the first or second person. Ambiguous faces were created by morphing between pairs of faces. One set of blocks used four pairs of celebrities, while the other used four pairs of anonymous faces. Results: We replicated the finding of face-identity aftereffects with whole face adaptors, with equivalent magnitude for both familiar and unfamiliar faces. For unfamiliar faces, adaptation to internal features alone and to external features alone also generated face aftereffects in whole-face test images, which were similar in magnitude but less than that from whole-face adaptors. However, for familiar faces, identity aftereffects were produced only by whole-face adaptors and not by internal or external features in isolation. Conclusion: Internal and external features are equivalent in perceptual representations of unfamiliar faces. Familiar faces require the whole-face context for access to their representations, which may reflect another characteristic of holistic mechanisms in face processing.
Meeting abstract presented at VSS 2012