Abstract
The edges of an object in a natural image can be defined by various contrasts such as luminance, color, and texture. Edges created by texture boundaries can be as perceptually salient as those created by simple luminance contrast. Early visual cortex is responsive to first-order (luminance) and second-order (orientation) contrasts (Larsson, Landy, Heeger 2006). Here, we measure BOLD responses to region transitions defined by luminance and luminance variance, referred to as first- and second-order contrasts.
To compare early visual responses defined by first and second-order contrasts, we used either a luminance increment or a luminance variance increment to define a foveal disk subtending 4°, against an 8° background of dynamic white noise (mean gray, 25% contrast) The dynamic noise reduced the effects of neural and retinal adaptation. We used the method of adjustment to find the point of subjective equality (PSE) for the salience of luminance increments relative to a 50% Michelson contrast increment. We then measured BOLD fMRI responses in the occipital lobe using a block design. For each stimulus type (first- or second-order contrast), the mean luminance and variance of the central disk were constant throughout the scan. The background mean luminance or variance was modulated to match the disk interior during “off” blocks, and lowered during the “on” blocks (to define the disk). We used a block localizer consisting of checkerboard annuli to select regions of interest inside and outside the disk for analysis.
Response to the image regions outside the disk was significantly modulated in the second-order contrast, but not the luminance case. Inside the disk, we observed a biphasic positive BOLD modulation: a positive transient response when the disk appeared and when the disk disappeared.
This work was funded in part by R01 EY015261 and T32 HD00715, MIND Foundation BTRR P41 RR008079 and P30 NS057091.