June 2007
Volume 7, Issue 9
Vision Sciences Society Annual Meeting Abstract  |   June 2007
Classification images reveal differences between spatial and spatiotemporal contour interpolation
Author Affiliations
  • Brian P. Keane
    University of California, Los Angeles
  • Philip J. Kellman
    University of California, Los Angeles
  • Cassandra M. Elwell
    University of California, Los Angeles
Journal of Vision June 2007, Vol.7, 603. doi:https://doi.org/10.1167/7.9.603
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Brian P. Keane, Philip J. Kellman, Cassandra M. Elwell; Classification images reveal differences between spatial and spatiotemporal contour interpolation. Journal of Vision 2007;7(9):603. doi: https://doi.org/10.1167/7.9.603.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Contour interpolation, the process whereby object fragments (inducers) are connected based on the geometrical relation of their contours, is relatively well-explored for static displays, but not much is known about how interpolation spans gaps in space and time. Here, we investigate characteristics of spatiotemporal interpolation with a classification image paradigm. For each condition, one naïve and one non-naive observer each judged on 4000 trials whether a vertical illusory contour of a gray rectangle appeared concave or straight. In the static condition, both inducers appeared simultaneously and motionless. In the “spatiotemporal” condition, inducers moved at 3 deg/s, and appeared in alternation from frame to frame. In both conditions, interpolation regions were embedded in a new noise field every motion frame (30 hz) and inducers were always shown noise-free at high contrast. By averaging the noise fields for each stimulus-response category, a classification image (CI) was derived for each frame of each condition. The CI effectively revealed the degree to which noise regions influenced responses. CIs from both observers showed that regions along interpolation boundaries are employed in more frames and to a greater degree in the spatiotemporal condition. To test whether the result arose solely from the difference in speed, the non-naïve observer engaged in a third “spatial-moving” condition, wherein inducers appeared simultaneously and moved at 3 deg/s. The resulting CI showed that interpolation regions were employed more than the static condition, but significantly less than the spatiotemporal condition. These results, taken together, suggest that interpolation regions are employed to a greater degree in spatiotemporal interpolation and that motion may also contribute towards interpolation strength. Finally, if interpolation strength peaks at the frame on which interpolation regions are most influential, it also appears that spatiotemporal interpolation may be faster than the alternative forms of contour interpolation.

Keane, B. P. Kellman, P. J. Elwell, C. M. (2007). Classification images reveal differences between spatial and spatiotemporal contour interpolation [Abstract]. Journal of Vision, 7(9):603, 603a, http://journalofvision.org/7/9/603/, doi:10.1167/7.9.603. [CrossRef]
 Supported by NEI EY13518 and NSF ROLE 0231826.

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.