Abstract
Which image geometries count as face-like and which do not? Novel objects called Greebles have been used as stimuli in studies showing that face-specific effects can be obtained with nonface stimuli under certain circumstances, in particular with expert observers. However, this claim depends on the assumption that Greebles are not a priori treated by the putative face module as “face-like”. To examine this assumption, we worked with CK, a neuropsychological patient well-known for exhibiting severe visual object agnosia and dyslexia but intact face processing. In prior work (Moscovitch, Winocur & Behrmann, 1997), CK's preserved face recognition extended to two-tone “Mooney” faces, facial caricatures, cartoon characters and even composite faces made out of non-face objects. CK was tested with Greebles in a simultaneous-matching task and with upright and inverted Greebles in a sequential-matching task. We also assessed CK's recognition performance for common objects and faces. These results were compared to that of control subjects as well as two prosopagnosic patients tested in an earlier study. CK performed even poorer with Greebles than with common objects, yet he was normal with faces, indicating that his intact face processing abilities do not extend to Greebles. In addition, the dramatic inversion effect that he shows for faces was completely absent with Greebles, indicating that CK does not benefit from a “face-like” configuration of parts in upright Greebles. Insofar as CK is relying on face-specific visual processes, these processes do not a priori treat Greebles as faces. Taking into account both CK and prosopagnosic patients SM and CR, performance with Greebles is independent of whether particular brain-injured individuals are postulated to have a damaged or preserved “face module”. Thus, studies using Greebles that provide evidence against modularity cannot be invalidated based on the claim that Greebles are “face-like”.
Funded by the Perceptual Expertise Network (#15573–S6), a collaborative award from James S. McDonnell Foundation, and NSF award #BCS–0094491, both to IG and MJT, and an NIMH award to MB.