Effects of Head Orientation on the Variations in the Discrimination of Shading Orientation

We have previously shown that errors in judging the direction from which a smooth surface is illuminated (ie shading orientation) are not random (Moshonkina et al, 1993 Perception 22 Supplement, 100), but vary with this direction in a regular manner. In the experiments described here we investigate whether the same regularities occur when the head is tilted—specifically, whether the gravitational or the head frame of reference is predominantly used in estimation of shading orientation.

Stimuli were computer images of spheres (23 deg diameter) presented briefly (for 100 ms, excluding effects of eye movements). The spheres were apparently illuminated from one of 24 directions spaced at equal (15°) intervals around the sphere in the plane of the display. One of these was randomly chosen for each trial. The task was to define the direction of the apparent illumination on a clock scale with half-hour accuracy. The subject's head was either upright or tilted to the left by 30°. Viewing was left monocular.

The errors in the estimation of illumination direction showed the same systematic dependence on ‘true’ direction regardless of head orientation. These regularities can be described as follows. (1) Estimates were most accurate when the shading gradient was along the horizontal axis (ie with the illuminated pole of the sphere on the right or the left), with greater errors when the gradient was along the oblique axis or vertical. (2) With oblique shading, the estimates showed a systematic tendency to bring the illumination direction closer to the horizontal axis than in ‘reality’. (3) The largest errors occurred when the illuminated pole was oriented downwards. This suggests that the gravitational frame is important in the discrimination of shading orientation.

A significant effect of head orientation was observed when the illuminated pole was oriented downwards. The most probable estimation error was −30° with head upright but −60° with head tilted. This difference is opposite to that expected from application of the head frame of reference.