Abstract
Recently we reported that accuracy of open-loop manual pointing to a visual target whose elevation was perceptually mislocalized increases linearly with hand-to-body distance (Psychonomics, 1999). Manual errors equaling the perceptual errors were measured with the hand near the body; complete accuracy was approached with the arm fully extended. We now report similar distance-contingent manual errors to a visible line whose roll-tilt was perceptually mislocalized. Following monocular measurements of perceived vertical (VPV setting), the subject matched the unseen right hand's roll-tilt orientation to a visible rod set either to VPV or 5 deg cw or ccw from VPV. Errors were induced in otherwise total darkness by a variably-oriented, roll-tilted 2-line stimulus (52.8 deg-long lines). The hand was 0, 30, or 60 cm in front of the midfrontal plane. The midfrontal-plane match remained essentially unchanged near vertical although the VPV setting changed systematically with the inducer's roll-tilt. This is unsurprising since the rod's inaccurate — but constant — perception might guide the manual match although the rod-at-VPV's physical orientation varied systematically with the inducer's roll-tilt; this suggested that the rod's perception guided the manual match made by a motor system unaffected by the inducer. However, with the arm fully extended, the manual match was close to true physical orientation, becoming more clockwise as VPV did. Similar manual matching results were obtained with the rod set 5 deg ccw or 5 deg cw from VPV. Thus, the midfrontal-plane match is inaccurate but consistent with visual perception, whereas the fully-extended-arm's match is near-accurate throughout the rod's orientation range but inconsistent with perception.
Support: NIH grant EY10534.