Texture brightness filling-in

The masking procedure by Paradiso and Nakayama (1991) (Vision Research, 31, 1221-1236) was used to investigate brightness filling-in within textures made of line elements: a texture stimulus was masked by a second stimulus containing a square contour. When a uniform texture was presented, the texture region inside the masking square appeared darkened and a small number of texture elements were perceived with a degenerated shape, appearing as dim dots or shorter line elements; it is as if the line element expanded from a bright point to fill the entire region defined by its contour. If the texture stimulus was a texture patch segregating from the surrounding texture by an orientation gradient and this patch was inside the square mask, darkening was not as strong as in the previous condition, and masked line elements preserved their elongated shape. Brightness spreading was measured in two experiments using dichoptic presentations. Experiment 1 used an adjustment task and showed that the brightness of texture line elements spread from equiluminant borders between segregating textures. Experiment 2 used a matching task and demonstrated that spreading was blocked by segregation borders dependent on the orientation gradient between texture line elements. The selectivity for line orientation began 40-80 msec after texture onset and maximal spreading occurred at approximately 120 msec. These findings may indicate that two processes subserve filling-in within textures: the first spreads isotropically the mean stimulus luminance at an initial processing stage of image analysis; at a later stage, the second spreads a texture flow (both brightness and shape of line elements) directed along the orientation of texture line elements. The texture flow mechanism fills in with a texture surface the region bounded by segregation contours.