Abstract
A tool used to differentiate the functions of cognitive “perceptual” and sensorimotor “action” visual systems has been the induced Roelofs effect, the misperception of a target's location induced by a large frame, offset from a subject's midline. The target is perceived to deviate in the direction opposite the frame's offset (a cognitive task), though observers could accurately jab the target's location (a sensorimotor task) (Bridgeman et al., VR 2000). A reinterpretation by Dassonville & Bala suggests that the Roelofs effect is caused by a frame-induced but unconscious deviation of the apparent midline. Using gaze rather than finger jab, Dassonville and Bala replicated four effects found by Bridgeman et al. (perceived Roelofs effect, delayed perceived Roelofs effect, no Roelofs effect in immediate pointing, Roelofs effect in delayed pointing) and explained them by apparent midline shifts, because gaze is calibrated to a midline shifted in the direction of the frame, compensating for the Roelofs illusion in the direction opposite the frame. After a few seconds in darkness, the apparent midline shifts back to the center, restoring the motor illusion. Thus both the two-visual-systems theory and the midline shift theory explain all four effects. Dassonville and Bridgeman developed a condition that yields different predictions in the two theories. An antijab in the direction symmetrically opposite the target's deviation from straight ahead should yield a Roelofs effect of twice the perceived illusion amplitude in the midline shift theory, but no Roelofs effect in the two-visual-systems theory. In both Bridgeman's and Dassonville's lab, antijab or antigaze tasks unambiguously yielded twice the cognitive effect, supporting the midline shift theory. Amplitudes of the perceived Roelofs effect and the frame-induced straight-ahead shift were correlated in both labs. Thus the sensorimotor system stores an egocentric calibration but not a complete map of visual objects.