Abstract
The existence of center-surround organization in middle temporal area (MT) neurons suggests the utilization of motion contrast detectors. We presented a psychophysical test to human observers to investigate how motion contrast stimuli are processed when concentric apertures contain different depths. The stimulus was a stereo random dot kinematogram with an inner circle (diameter 1–10 deg) and outer aperture (2–20 deg) moving independently. Observers made a 2-alternative forced choice indicating the direction of the center aperture. We tested 8 coherence levels ranging from 2–80% with surround dots (50% coherence) moving in either the same or opposite direction as center dots. Bias in the resulting psychometric functions reflected the modulatory effect of the surround on the perceived direction of center dots. This indicated a preference for reporting center motion in a direction either the same as (assimilation) or opposite to (contrast) the surround. When the inner and outer apertures had different luminances, we found a similar pattern of biases as Murakami and Shimojo (1996): assimilation bias for small apertures (<2 deg), contrast bias for larger aperture sizes (2–6 deg), and almost no bias for the largest stimuli tested (10 deg), as well as a scaling of this pattern as eccentricity increased. Center-surround MT neurons are known to be more highly activated with surround depth (Bradley, Andersen 1998), which would suggest an increase in surround modulation. However, when the outer aperture was moved to a disparity of 8 min, biases shifted toward zero for all aperture sizes and eccentricities, indicating a reduced impact of the surround. The results indicate that disparity causes a dissociation of center and surround motions that is stronger than the enhancement of motion contrast. Our results suggest that disparity provides a scene segmentation cue beyond that of luminance alone, and it prevails over surround modulation when observers perform direction discrimination tasks.
This work was supported by NIH grant R01EY007861-15 to L.M.V.