Building surfaces from borders: A neural basis for edge-induced brightness in the Cornsweet illusion

The perception of surface brightness is determined by both surface luminance and by surrounding border (or edge) contrasts. The Cornsweet illusion is one which isolates the border contribution. In the Cornsweet illusion, two equiluminant surfaces appear different in brightness due to the contrast edge between them. I have used the Cornsweet illusion to probe the encoding of edge-induced brightness in the cerebral cortex.; Using single-unit recording and intrinsic optical imaging methods, I have studied cortical responses to edge-induced brightness contrast (Cornsweet) and perceptually equivalent Real luminance contrast. These studies have demonstrated the existence of cells in cat visual cortex (primary and second visual areas, Areas 17 and 18) whose responses are correlated with the perceived (illusory) brightness change. Both single-unit and optical imaging results suggest that responses to the Cornsweet parallel those to Real luminance change, in that the responses are greater in Area 18 than Area 17. Consistent with the percept, these responses extend as far as 10–20 degrees from the contrast border. Though responses to the Cornsweet are weak, they are significant, as shown using bootstrap statistical methods. Thus, single cells in Area 18 that encode actual luminance may also encode illusory brightness.; To examine how border contrast information is propagated to the surface, I have studied the interactions between ‘border’ and ‘surface’ cells. Two primary hypotheses were tested: a 17-17 lateral ‘filling-in’ hypothesis of propagation and a 17-18 feedforward propagation. Pairs of single-units in Areas 17 and 18 (both 17-17 and 17-18 border-surface cell pairs) were recorded and their connectivity analyzed by evaluating spike timing interactions. Interestingly, spontaneous (no stimulation) interactions revealed an intrinsic surface-to-border bias. Real and Cornsweet stimulation shifted the interaction in a border-to-surface direction. Analyses of latency and strength of interaction suggest that a filling-in model of propagation is unlikely. Rather, these data suggest that a 17-to-18 border-to-surface circuit underlies edge-to-surface propagation during both Real and Cornsweet brightness processing.; In summary, these findings suggest that the second visual area plays a significant role in the perception of edge-induced (illusory) brightness. In addition, they indicate that a feedforward 17-18 circuit underlies this percept.