On the relative contributions of motion energy and transparency to the perception of moving plaids

Stoner, Albright and Ramachandran [(1990) Nature, 344, 153-155] found that moving rectangular-wave plaid patterns that admitted a transparency interpretation appeared to segment that "slide" past one another as the plaids were translated, while the components of plaids that did not admit a transparency interpretation appeared to unify and move rigidly in the direction of translation of the plaid. In experiment I, we show that the magnitude of the effect reported by Stoner et al. is due largely to their repeated-trials experimental protocol, in which plaids moving in a particular direction, upward or downward, are repeatedly presented. This protocol leads to a direction-of-motion-specific adaptation that diminishes the effectiveness of processes that are presumably involved in the unification of the various sensory signals evoked by a moving plaid. In the second experiment, we measured frequencies of nonrigidity for a larger class of moving plaid-like patterns that moved either upwards or downwards on a pseudorandom schedule identical to that employed by Stoner et al. Some of the patterns admitted a transparency interpretation, while others did not. The overall pattern of results could not be accounted for within the context of Kim and Wilson's [(1993) Vision Research, 33, 2479-2489] model of motion integration that considers only the oriented motion energy present in a moving plaid stimulus. The results indicate that additional factors, distinct from though perhaps related to the visual analysis of transparency, must also be incorporated into models of perceived plaid motion.