Direction specific masking and the analysis of motion in two dimensions

We measured the effects of moving two-component cosine grating masks on the detectability of a moving spatially localized test pattern with a 1.0 octave spatial frequency bandwidth. Masking was used to distinguish between two-component patterns with fluid motion (blobs) and those with rigid motion (plaids). The two gratings which made up the two-dimensional masking patterns were always of the same spatial frequency and contrast, but moved in different directions. We find that plaid masks consistently produced threshold elevations that are 2.0-4.0 times greater than are produced by a single component mask at twice the contrast. Furthermore, this effect is nearly independent of the angle between the two mask components. For fluid motion, however, masking is determined by the mask component whose direction of motion is closest to that of the test. The results obtained with moving two-dimensional patterns demonstrate that, for blobs, the motion of the pattern as a whole has no effect on the degree of masking, whereas, for plaids, the signals arising from the two components interact in a nonlinear manner, thus producing a substantial enhancement of masking, which is clearly related to the coherent motion of the entire pattern. These data shed light on the properties of higher order motion units (possibly in MT cortex) that respond to the direction of two-dimensional pattern motion, suggesting that they combine, in a nonlinear manner, the outputs of units which respond independently to the direction of each mask component.