Abstract
The location of a target presented within a static frame that is offset from the subject's midline has been reported to be perceptually mislocalized in the direction opposite that of the frame (Roelofs 1935). However, when asked to point to the target, the effects of frame offset are smaller, or even nonexistent in some subjects (Bridgeman et al. 1997). This dissociation of perceptual and sensorimotor localization appears to support the hypothesis that the brain contains separate maps of visual space for perception and action. However, an alternative explanation is equally viable. Previous measurements of the target's perceived location required subjects to compare the location of the target with a remembered array of targets seen earlier. Perhaps the subjects correctly perceive the current target but the remembered location of the comparison array is distorted by Roelofs' illusion. To test this alternative hypothesis, subjects were required to sit in complete darkness and indicate the remembered locations of 5 targets (0.5 degrees diameter) positioned in a horizontal array −4, −2, 0, 2 and 4 degrees from the midline. A large frame (21 × 8.5 degrees, 1-s duration) presented either centered or shifted 5 degrees to the left or right was accompanied by a verbal command to indicate the location of one of the remembered targets with a saccadic eye movement. Accuracy was found to be systematically biased, with the reported position of the targets skewed in the same direction as the frame offset. A second experiment demonstrated that subjects' report of perceived straight-ahead was similarly shifted in the direction of the frame. Thus, the presence of the offset frame distorts remembered visual space and the perceived straight ahead in a way that precisely explains the previous findings with Roelofs' illusion, and argues against the idea of separate maps of space for perception and action.
Supported by NSF grant BCS — 9996264