Abstract
It is well established that lowering a pattern's contrast results in a reduction of its perceived speed. This observation holds true at low temporal frequencies and the effect itself appears tuned to temporal frequency. In a similar manner, we show that a reduction in the coherence of a global motion pattern results in its perceived slowing. Our stimuli consisted of random dot kinematograms in which a proportion of the dots translated in a single direction. For the remainder of the dots, direction was random. The speed for all dots was the same and dot directions were re-determined on every frame. We initially examined the effect of reducing coherence over a range of speeds. Test and comparison dot patterns were presented in circular apertures with radii 1.41 degrees. Half of the dots were at minimum luminance, half were at maximum luminance, the remainder of the screen was set to mean luminance. The two apertures were presented side by side, separated by 1 deg with a fixation spot in the centre. Using an adaptive method of constants procedure, we matched the perceived speed of a high coherence pattern to that of lower coherence patterns. Our results show a pronounced slowing of perceived speed with reduction in coherence. The effect occurred over a large range of dot densities and lessened as speed increased. To test whether the effect was tuned to either temporal frequency or speed, we used spatial frequency tuned dot stimuli (Laplacian of Gaussian dots and Laplacian of Laplacian of Gaussian dots). A comparison of the tuning curves across our two stimulus types shows good evidence for a speed tuning of our effect. This clearly implicates mechanisms beyond primary visual cortex. Apart from their difference in tuning, the effects of contrast and coherence on perceived speed have obvious similarities. We propose that similar computational principles, operating at different levels of motion processing, underlie the effects of contrast and coherence on perceived speed.