Abstract
It is often assumed that direction selectivity is the machinery for extracting motion information, and that non-direction-selective mechanisms analyse spatial pattern information. This view however is challenged by the anomalous perception of moving spatial patterns presented through static apertures (e.g., Parks, 1965; Shipley & Kellman, 1997). A striking example is an advertisement display (PoleVision, AVIX inc) that shows horizontally moving characters in an array of vertical LED lines. In spite of only a fraction (one per ten lines) of the image being actually presented at any instant of time, the running catchword can be effortlessly read. Spatiotemporal interpolation phenomena, including this example, can be interpreted as implying the brain's integration of spatial pattern information over motion trajectory, which supposedly requires direction-selective spatial pattern mechanism (Burr et al., 1986). This hypothesis was tested by masking and adaptation experiments. I briefly (320 ms) presented two sets of letters (identical alphabet in each set) separately through arrays of static vertical slits located above and below the fixation point. The sets moved in opposed directions to eliminate the effects of eye tracking. For a range of inter-slit intervals, pattern movement significantly improved letter identification performance in comparison with static control conditions. In masking experiments, I filled the inter-slit areas with a random noise mask that coherently moved independently of the letter movement. It was found that the masking effect peaked when the masking noise moved in the same direction, and at around the same speed, as the letter movement. I also measured the effect of motion adaptation, and found that letter recognition was worse for the letters moving in the adapted direction than for those moving in the opposite direction. These results clearly indicate the involvement of direction selective mechanisms in spatial pattern recognition.