Afterimages and pursuit: refining Helmholtz's theory of visual motion perception

Abstract

Why does the world appear stable during gaze shifts, yet retinally-fixed afterimages appear displaced with movements of the eyes? Helmholtz proposed these phenomena could be explained by linear summation of retinal and extraretinal motion signals. We studied the interaction between extraretinal information and the spatial patterns of afterimages during perceived afterimage motion. Six observers pursued a target which oscillated horizontally in the presence of 3 different afterimages created by fixation of large (∼50°) high contrast patterns: 1) parallel line elements oriented at −45°, 2) the same line elements oriented at +45° and 3) a grid composed of both −45° and +45° components. Subjects reported the angle of perceived afterimage motion while pursuing horizontally. When the afterimage was of oblique parallel lines, its mean perceived motion angle was determined by line orientation: −34° ± 3° SE for lines at −45° and 31° ± 4° SE for lines at 45°. In both cases the percept swings slightly toward the horizontal motion angle of the eye. When the afterimage was a grid the mean motion vector was determined by the eye movement angle: 1° ± 1° SE. In all 3 instances, Helmholtz's model predicts a perceived motion orientation of 0° since the eyes only move horizontally. Our results, however suggest that linear summation of retinal and extraretinal signals is not adequate for explaining perceived afterimage motion. Rather eye velocity must interact multiplicatively with the spatial gradient of illumination (or its afterimage). This approach predicts human perception and cortical activity correctly.