Edge detection and surface filling-in as shown by texture visual evoked potentials

Objectives: We designed a stimulation paradigm that was designed to ascertain whether specific components of the texture visual evoked potentials were attributable to edge detection or to surface 'filling-in' processes. Methods: The stimuli were textures of parallel line elements, which were either uniform (all horizontal or vertical lines) or segregated (checkerboards in which texture line elements of neighbor checks were oriented orthogonally). A sequence of 4 stimuli, two uniform followed by two segregated stimuli, was repeated cyclically. Accordingly, segregation could appear from a uniform display; it could also be maintained, with the checkerboard illusory margin location unchanged but with alteration in the orientation of all line elements. Each stimulus was presented for 487 ms and instantly replaced by the following one. Results: Segregation-related components for segregation-appearance (Sa) and for segregation-maintenance (Sm) conditions were obtained separately by subtraction. In both conditions, a negative component was obtained with a peak latency of about 140-150 ms. However, the onset of Sa was earlier than that of Sm, whereas the respective offsets were almost identical. Conclusions: We hypothesize that the segregation component in VEPs is composed of two subcomponents: an early part, which is related to the segregation of edges, and a final part, which is related to the 'filling-in' of the homogneous texture surface within the boundary defined by these edges.