Oculoparalytic Illusion : Visual - Field Dependent Spatial Mislocalizations by Humans Partially Paralyzed with Curare

In darkness, observers partially paralyzed with curare make large (> 20 degrees) gazeand dosage-dependent errors in visually localizing eye-level-horizontal and median planes, in matching the location ofa sound to a light, and in pointing at a light. In illuminated, structured visualfields visual localization and pointing are accurate but errors in auditory-to-visual mqtches remain. Defects in extraretinal eye position information are responsible for all errors. The influence of extraretinal eye position information on visual localization is suppressed by a structured visualfield but is crucial both in darkness andfor intersensory localization if visual capture is 198 o n N ov em be r 24 , 2 00 6 w w w .s ci en ce m ag .o rg D ow nl oa de d fr om appear at eye-level-horizontal were now more than 0.6 m below true horizontalagain more than 200 error (5). Subsequent experiments separated the influence of the angle of the eye in the head (angle a in Fig. ig) from the angular relation of the head-and-body to gravity (angle 1), and demonstrated that the illusion was determined by a only. In those experiments the observer set a movable, peripherally viewed visual target (3) to the perceived eye-level-horizontal with a established by the vertical position of a separate foveally fixated visual target ("two-light experiment"). That only a and not 1 determined the sign and magnitude of the illusion was demonstrated as follows: (i) The error in setting the second light to the perceived eye-level-horizontal in darkness was linearly related to a (Fig. 1, a and d); this relation did not change with different settings of 13. These systematic errors were not made by the partially paralyzed observer in normal room illumination or by the unparalyzed observer either in illumination or in darkness. (ii) When the results from the one-light experiment were transformed to the coordinates of Fig. la they were indistinguishable from those of the two-light experiment at the same level of paralysis. (iii) The noillusion direction (NID)-the angle of the eye in the head for which no drop or rise of a fixated visual target was perceived when illumination was extinguishedwas also the direction of gaze for which the setting of the visual target to the perceived eye-level-horizontal was most accurate for both oneand two-light experiments. (iv) The NID was the same for all head-and-body tilts. (v) Partially paralyzed subjects pointed accurately to the proprioceptive horizontal in darkness or with eyes closed; they also pointed accurately at the horizontal direction when pointing toward a visual target located at the physical eye-level-horizontal in normal room illumination (6), but they pointed more than 200 downward (or upward) when pointing at a visually fixated target viewed in darkness which was physically at eye level but appeared to lie near the floor (or the ceiling); pointing errors were less for gaze directions at which visual errors were smaller (7). Thus, we conclude that although perception of the eye-level-horizontal must involveinformation about both angles a and 13, the paralyzing procedure only distorts EEPI about a (8). Similar results were obtained when we measured the influence of muscle weakening on visual spatial localization for horizontal changes in gaze direction in a two-light experiment. The subject set a movable, peripherally viewed visual target to the position along a horizontal arc that appeared to intersect the median plane of his body while his horizontal direction of gaze was fixed by the horizontal position of the foveally fixated target. As with the perceived eye-levelhorizontal, neither paralyzed nor unparalyzed subjects made errors in setting the visual target to the perceived median plane when fixating at different eccentricities of gaze in normal illumination.