Abstract
Large changes in manual accuracy that are linear with distance of the hand from the body have been measured for three different sorts of visually-guided (open-loop) behavior under visual induction that also generates perceptual errors in visually perceived eye level (VPEL; up to 18°) and visually perceived vertical (VPV; up to 12°) (Welch & Post, 1996; Li & Matin, 1999, 2004, in press; Post, et al., 2004). The three manual behaviors are: (1) pointing/reaching to a visual target, (2) height matching to a visual target, and (3) roll-tilting the hand to feel vertical. In each case, accuracy of the unseen hand increased linearly with distance of the hand from the body, approaching complete accuracy at full extension; with the hand close to the body, the manual errors equalled the magnitude of the perceptual error. The large errors are generated by visual induction by a single eccentric, pitched-from-vertical line, or by two roll-tilted lines in darkness; the inducers also induced separately-measured distance-dependent errors in the felt elevation or roll-tilt orientation of the arm. Both the manual errors and the perceptual errors are quantitatively well-accounted for by a new development (Proximal/Distal Model) of an earlier treatment that explained changes in VPEL and VPV as a consequence of a weighted average of inputs from vision and from the body-referenced mechanism (extraretinal inputs); relative weights change systematically with distance of the hand from the body. The large errors in felt elevation of the arm at full extension brought pointing/reaching to nearly complete accuracy but errors in felt roll-tilt of the arm only brought the manual roll-tilt to about half accuracy. These results show that accuracy at full arm extension is not a consequence of motoric immunity from the distorting influences of visual stimulation; perception and action are not dissociated as suggested by a recent version of the two visual system hypothesis.
Support: NIH grant EY10534.