Introduction

Retinal images are first segmented into regions corresponding to different objects. The ownership of the borders between adjacent regions has to be resolved before 3D object can be extracted from its background. The questions here are: how does the visual system resolve the border ownership, how does the neuron do the border assignment.

Previous studies of our group showed that neurons in area V2 encode border ownership. Those neurons fired more strongly to a contrast border that belongs to an object on one side of their receptive field. When the border belongs to an object on the other side of their receptive field, those neurons responded much weaker.

Fig. 2 shows one of those neurons. The raster plots show that this neuron responded more strongly to a border which belongs to a square on lower left side of its receptive field. When the border belongs to a square on the upper right side, the response was weaker. The two stimuli are locally identical but with different global configurations. Reversed the luminous contrast, the cell still preferred a border which belongs to a square on lower left side of its receptive field.