Abstract
Surround gratings suppress the ability to make fine orientation discriminations on a center patch of grating when their spatial frequencies, orientations, and contrasts are similar. The interaction may be mediated by interactions among low-level analyzers, as posited in models of contrast-contrast. We measured the spatial frequency bandwidth of this effect. The patterns to be discriminated were vertical sinusoids of 4 cpd displayed in a sharp-edged 40-minute circular window, at a mean luminance of 19.8 cdm-0.5. In control conditions, no surround modulation was present. Modulated surrounds were simple vertical gratings that varied in spatial frequency ± >1 octave in 5 steps. Test and mask contrasts were held constant at 0.1. Each condition was run in a separate block of 80 trials. Differences to be discriminated were adjusted individually for each observer to yield a d′ of approximately 1.5 in control conditions. All observers were well-practiced. Difference thresholds were in the hyperacuity range and fixed for all conditions. A two-alternative signal-detection rating procedure was used to measure how discrimination performance changed as a function of center-surround similarity. The magnitude of interactions monotonically decreased with increasing differences between center and surround spatial frequency, returning to control levels at approximately ± 1 octave, with a half-amplitude, full bandwidth of approximately 0.75. Bandwidths were slightly asymmetric and narrower than either lateral-interaction bandwidths measured using apparent contrast tasks (Cannon & Fullenkamp, Vis. Res. 1991) or in estimates of spatially tuned channels in early vision (Wilson, MacFarlane & Phillips, Vis. Res. 1983) by overlaid masks. We conclude that the mechanisms underlying lateral interaction effects in orientation discriminations are different from those previously isolated or posited.
Supported by EY13953 to LAO