September 2011
Volume 11, Issue 11
Free
Vision Sciences Society Annual Meeting Abstract  |   September 2011
Efficient integration across spatial frequencies reflected in early visual areas
Author Affiliations
  • Rachel Millin
    Neuroscience Graduate Program, University of Southern California, USA
  • Bosco S. Tjan
    Neuroscience Graduate Program, University of Southern California, USA
    Department of Psychology, University of Southern California, USA
Journal of Vision September 2011, Vol.11, 1179. doi:10.1167/11.11.1179
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Rachel Millin, Bosco S. Tjan; Efficient integration across spatial frequencies reflected in early visual areas. Journal of Vision 2011;11(11):1179. doi: 10.1167/11.11.1179.

      Download citation file:


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

      ×
  • Supplements
Abstract

Objects in a scene are composed of different spatial frequencies, and the combination of these spatial frequency components provides information about form. To perceive form, the visual system must efficiently integrate across spatial frequencies. While a number of studies have shown that integration across spatial frequencies is suboptimal for grating detection, Nandy and Tjan (2007) found it to be optimal for letter identification in both the fovea and periphery. How and at what stage of visual processing optimal integration across spatial frequencies occurs is unknown.

We used fMRI multi-voxel pattern analysis to investigate spatial frequency integration in the visual cortex. From the early visual areas (V1–V3), we measured BOLD response evoked by bandpass filtered letters X and O presented in the periphery, while subjects were engaged in a demanding task at fixation. The center frequencies for the octave-wide bandpass filtered letters were separated by two octaves, yielding a high and a low spatial frequency condition. A third composite condition was obtained by summing these two spatial frequency components. Using a linear support vector machine, we measured the accuracy with which multi-voxel activity within each visual area could be used to predict the identity of the displayed letter. The sensitivity (d′) of the decoder for discriminating X from O in the low, high, and composite conditions was determined.

We found that in V1 discrimination was no better for the composite letter than the one-octave component letters. However, in V2 sensitivity was highest for the composite, where squared d′ for the composite was approximately the sum of the squares of the components' d′s. In V3, d′ for the composite was approximately the sum of the components' ds. This result suggests that efficient integration across spatial frequencies, as seen by a linear decoder operating on BOLD signal, begins in V2.

National Institutes of Health Grants R01-EY017707, R01-EY016391. 
×
×

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.

×