Abstract
When two moving gratings that differ only in orientation are superimposed, the resulting plaid is perceived as a coherently moving pattern. When the same two gratings are presented separately to either eye (a dichoptic plaid), the perceived direction of motion remains that perceived monocularly, even in the presence of binocular rivalry. In recordings from binocular neurons in MT of anesthetized macaque we measured direction-tuning curves for monocularly presented gratings, and for monoptic and dichoptic plaids (component gratings separated in orientation by 120°). We calculated a pattern index to quantify the degree of pattern selectivity for each plaid type. Pattern indices in each eye were strongly correlated in individual MT neurons (r2 = 0.66, n = 67). In all cells the value of the pattern index decreased for dichoptic plaids. Of 18 cells classified as pattern selective with monoptic plaids, only 3 remained pattern selective using dichoptic plaids. We asked human subjects to identify the direction of motion of monocularly and dichoptically presented plaids, using similar stimulus conditions to the physiology. For monoptic plaids, histograms of the absolute angular separation (“delta dir”) between the perceived and the pattern direction of motion were unimodal, clustered tightly around 0°. For dichoptic plaids the histogram of “delta dir” was bimodal, with a prominent peak near 0° and a smaller one near 60°: on average, subjects perceived the pattern direction of motion on ∼70% of dichoptic trials. Thus, while the vast majority of cells in MT (including those cells that are monoptically pattern selective) show no evidence of pattern selectivity when stimulated with dichoptic plaids, the dominant motion percept evoked by dichoptic plaids is of pattern motion. We conclude that the perceptual experience of visual motion does not always correspond to the direction of motion signaled by individual neurons in MT.