Abstract
Is there viewer-centered object representation in the human visual system? We used visual selective adaptation method to address this question. After adapting to an object viewed from a angle of either 15 or 30 degrees off center, the perceived viewing direction of the same object subsequently presented near the frontal views were biased away from the adapted viewing direction. Further experiments demonstrated that the strength of this aftereffect depends on the angular disparity between adapting and test viewing angles in the 3D space. The magnitude of this viewpoint aftereffect is much larger than that predicted by representational momentum. The aftereffect was obtained despite the spatial jitter added to both the adapting and test figures. This translation invariance of the aftereffect suggests that it is not caused by local feature adaptation. Viewpoint aftereffects were found within all three categories of objects (faces, cars, simple stick-figures) tested, but we failed to find cross-adaptation between two different objects. This object specificity suggests that the aftereffect is not an abstract 3D tilt (slant) aftereffect. The perceptual viewpoint aftereffect can be best explained by changes in sensitivity in object-selective neurons tuned to specific viewing angles. These results provide novel and unequivocal evidence that supports the existence of viewer-centered object representation in the human visual system.
This research is supported in part by an award from the James S McDonnell Foundation