Abstract
Controversy surrounds whether perceptual filling-in of the blind spot involves active neural completion or a passive mechanism in which the absence of information is simply ignored (Ramachandran, 1992; Dennet, 1991; Durgin, 1995). These competing accounts have proven difficult to distinguish because filling-in of the blind spot occurs so rapidly and automatically. Here we show that texture-defined stimuli fail to fill-in the blind spot and that luminance cues are necessary for active neural completion. Methods. A textured bar was presented across the visual location of the blind spot. The uniform background varied from dark to light, and included the mean luminance of the texture. Oriented texture patterns were either parallel (collinear) or orthogonal to the bar orientation. Results. Observers reported poor filling-in of the orthogonal texture when its mean luminance closely matched that of the background. In contrast, large luminance differences between texture and background restored filling-in, not only of the luminance but also of the texture. Failure of filling-in for orthogonal texture could not be explained by poor pattern visibility because strong filling-in occurred when the texture was parallel to the bar orientation, indicating collinear facilitation or grouping across the blind spot. Conclusions. These results indicate that: 1) filling-in of the blind spot involves active neural completion rather than an automatic passive mechanism, 2) this active mechanism relies on luminance-based grouping across the blind spot, and 3) pattern information alone is insufficient for filling-in but the addition of luminance allows for higher-order texture information to propagate across the blind spot. These findings demonstrate the constructive nature of conscious vision and provide constraints on how visual experiences can be actively assigned to field locations that lack direct input.