Object Recognition

Our shape segmentation experiments have led us to attempt to understand the nature of shape and object representations.  We assume that object and shape representations are distributed through the ventral pathway of the brain.  Our experiments have suggested that the representations that affect shape segmentation lie in posterior rather than anterior regions of this pathway.  Using both behavioral and brain imaging techniques, we are pursuing the hypothesis that partial overlapping representations located in posterior brain regions underlie the effects of past experience on shape segmentation (see Peterson, 2003 in Peterson and Rhodes, Eds.).  Experiments exploring these issues funded by a grant from the NSF (BCS 0425650) are in progress (See Peterson & Skow, VSS, 2005).

In addition, we have long been interested in the question of how the spatial relationships between parts of shapes are represented.  Parts and their spatial relationships can be logically separated, and, by rearranging the parts of familiar objects, we can create scrambled versions that are unrecognizable (see Segmentation and Shape & Object Perception section).  Nevertheless, it is not clear whether or not spatial relationships are represented separately from parts.  To investigate this question, we tested a brain-damaged individual who has trouble recognizing objects.  He was dramatically more impaired at detecting distractors that were constructed by rearranging the parts of shapes he knew.  In contrast, he was quite good at detecting distractors that differed from the shapes he knew by one part.  These results are consistent with the interpretation that parts and their spatial relationships are separately represented (Behrmann, Peterson, Suzuki, and Moscovitch, submitted).