Abstract
The lack of a percept of motion transparency with locally-balanced dots moving in opposite directions indicates the existence of a local-motion-pooling (LMP) stage prior to global-motion (GM) pooling. We investigated the tuning properties of LMP units to luminance polarity, speed (temporal frequency) and multiple directions over time. It has previously been shown that luminance-polarity information (On and Off pathways) is kept independent at the local-motion stage, but is pooled at the GM stage. Whether polarity information is still independent at the LMP stage was investigated. Earlier studies have indicated that, while speed tuning occurs at the GM stage, it doesn't occur at the LMP stage. However, the speed combinations used in those studies would have driven common speed-tuned GM units. If speed tuning does occur at the LMP stage it would likely reflect the observed GM tuning. Consequently, LMP tuning was tested with speeds that drive independent GM units. Using locally-paired, orthogonally-moving dots, we showed that luminance polarity information is combined at the LMP stage and that independent, speed-tuned (TF) LMP units exist. Finally, it has been shown that adapting to a transparent-motion stimulus, in which the signals differ only in their direction of motion, results in an unidirectional MAE, opposite to the vector-average direction of the two adapting signals. It has been proposed that this MAE results from the integration of the different directions in each LMP unit over the course of adaptation, such that the net adaptation in each LMP unit is to the vector-average direction. This theory was tested with a multi-aperture stimulus in which a single motion direction was presented in each aperture. It was possible to generate a transparent MAE with this stimulus, supporting the theory.