Attention modulates perception of transparent motion

AbstractHuman observers can extract a given motion direction from sets of random dots moving simultaneously in two or more directions in the same region of the visual field, a phenomenon referred to as motion transparency. As a necessary condition for separating transparent motion directions, low level encoding of local motion signals must generate frequency distributions of local directions with separable peaks corresponding to these directions—this process would be constrained by local stimulus attributes and the properties of local motion detectors. Furthermore, a representation of multiple directions is needed for simultaneous retrieval of several directions in a psychophysical task—this operation would be limited by higher level processes, such as attention selecting a particular direction to rise into awareness. Preliminary observations suggest that the number of directions that can be seen simultaneously is rather limited and the question arises whether this could be related to limitations of low-level encoding or higher level representations. To study specifically the effect of attention on transparent motion perception, observers were presented with sets of dots moving coherently in a variable number of directions, and were asked after the presentation whether one particular direction was present in the set. When the direction of motion was not known before stimulus onset (uncued condition), observers detected a particular motion direction among no more than 3 other directions. When direction of motion was indicated prior to stimulus onset (precued condition), however, this limit increased up to 6 directions. This attentional effect showed some inter-individual variability and appeared to benefit from spatiotemporal integration of the motion signals. A corresponding effect became apparent when observers were tested in the same paradigm whether they could separate two motion directions with variable angular difference between them. In the precued condition a typical minimum direction difference was about 60°, whereas in the uncued condition this was about 120°, suggesting that the performance in detecting one direction in a multiple direction stimulus might be limited by the ability to separate adjacent motion directions. This pattern of results suggests that attention can reliably improve transparent motion processing by affecting the separability of directional signals in low level encoding mechanisms