Order-specific and non-specific motion responses in the human visual system

A series of experiments measured direction discrimination in two-frame random block kinematograms. Blocks were presented against a uniform grey background, and were filled either with uniform grey (darker or brighter than the background; first-order blocks) or with random microtexture (isoluminant with the background; second-order blocks). Experiment 1 found that when blocks maintained their order from frame to frame, performance declined from near-perfect to chance levels as block displacement increased. When blocks switched order between frames, performance was generally worse (65-75% correct at best), but still above chance levels. Results from control experiments established that it is important to remove intensity cues in second-order patterns using a psychophysical technique, and that above-chance responses with order-switching patterns persisted, even when such intensity cues were removed or randomised. The last experiment measured the effects of block density manipulation. First-order and second-order patterns showed the same decline in Dmax performance as pattern density increased, and results from patterns containing a mixture of first- and second-order blocks could be predicted from performance obtained with each set of blocks presented separately, except at very low densities. It is concluded that both order-specific and non-specific responses are available during motion analysis, but order-specific responses tend to predominate.