August 2012
Volume 12, Issue 9
Free
Vision Sciences Society Annual Meeting Abstract  |   August 2012
Psychophysical reverse correlation of motion perception
Author Affiliations
  • Jacob Yates
    Center for Perceptual Systems, University of Texas at Austin\nInstitute for Neuroscience, University of Texas at Austin
  • Alexander Huk
    Center for Perceptual Systems, University of Texas at Austin\nInstitute for Neuroscience, University of Texas at Austin
  • Lawrence Cormack
    Center for Perceptual Systems, University of Texas at Austin\nInstitute for Neuroscience, University of Texas at Austin
  • Jonathan Pillow
    Center for Perceptual Systems, University of Texas at Austin\nInstitute for Neuroscience, University of Texas at Austin
Journal of Vision August 2012, Vol.12, 746. doi:10.1167/12.9.746
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Jacob Yates, Alexander Huk, Lawrence Cormack, Jonathan Pillow; Psychophysical reverse correlation of motion perception. Journal of Vision 2012;12(9):746. doi: 10.1167/12.9.746.

      Download citation file:


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

      ×
  • Supplements
Abstract

We used psychophysical reverse correlation to probe the mechanisms of motion processing. Specifically, we investigated whether short-range stroboscopic (apparent) motion processing can be explained by canonical motion processing mechanisms, or by a template-matching mechanism tuned to the apparent-motion stimulus. Observers were asked to detect a 3-step apparent-motion bar embedded in spatiotemporal Gaussian white noise. Half the trials contained the signal bar plus noise; the rest contained only noise. The ideal observer for this task is a linear template matched to the expected space-time locations of the bar, followed by a threshold. We estimated a linear and second-order kernel for each observer, using the maximum a posterior estimate under a generalized linear observer model (GLM), with a prior encouraging smooth and sparse filters. (These estimates were substantially better at predicting responses on held-out data than filters fit by "classical" psychophysical reverse correlation). The observed second-order kernels exhibited compelling direction selectivity (i.e., spatiotemporal orientation) and reveal integration across a larger range of spatiotemporal locations than the actual signal. Surprisingly, the second-order kernel alone provided a more accurate description of observers’ data than the linear GLM, despite the fact that the target signal contained white bars only. To further analyze nonlinear contributions to motion perception, we reverse correlated observers’ choices with the energy at each possible velocity in the noise from each trial. This analysis, using a radon transform to sum over all orientations in the Fourier domain, revealed perceptual filters tuned for the implied velocity of the apparent motion signal. Our results indicate that sign-invariant motion-energy calculations (Adelson & Bergen, 1985) underlie the detection of apparent motion, and demonstrates the feasibility of reverse correlation methods for analyzing the nonlinear mechanisms underlying motion processing in both psychophysics and physiology.

Meeting abstract presented at VSS 2012

×
×

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.

×