Edinburgh Napier University

  Human observers are very sensitive to speeds of moving stimuli. However little is known about how such sensitivity arises independently upon the spatiotemporal properties of the stimulus. Here, we investigate the interactions between spatiotemporal frequency channels, and, in particular, how different speeds are integrated as a function of the spatiotemporal frequency content. We used naturalistic random textures ("Motion Clouds" - MCs, Simoncini et al., 2012, Nat. Neurosci. 15(11):1596-1603) to control the exact range of spatiotemporal frequencies of the stimulus. In a preliminary experiment, we combined two MCs of diverging mean spatial or temporal frequencies, and calculated the threshold up to which human observers were able to discriminate between one complex stimulus versus two interleaving stimuli. Then, we generated stimuli composed of three component MCs, which moved at one of three specific speeds and had different mean spatial or temporal frequencies. Six distinct components were arranged in four combinations in spatiotemporal space. Thus, all complex stimuli moved at the same average speed but each activated distinct channels. We measured the perceived speed of the multi-component stimuli in a psychophysical 2AFC task in which participants were asked to compare the speed of the MC stimulus with that of a random dot kinematogram. Our results suggest that systematic manipulation of speed channel interactions can affect the perceived absolute speed of the stimulus as well as the sensitivity to the perceived speed. We found that neighboring channels in spatiotemporal space interact with each other in specific directions. The results appear to partially agree with predictions made by a computational model of motion perception (Perrone, 2012, JoV 12(8):1), however they also hint at more complex, but yet unidentified rules in the weighting of different channels in the decoding of speed information.