Abstract
Visual surroundings can influence perception of 3D shape, as illustrated in ‘shape contrast’ biases where the perception of shape is biased in the direction opposite to the surroundings. Whereas this bias is very robust, little is known about the underlying mechanisms. In this study, we address the issue whether shape contrast biases are the result of local contrast enhancement or whether they are the result of a more global mechanism operating on an extended surround. Observers performed a sequential shape discrimination task on a hinged plane, which was defined by disparity. In the baseline condition, the central surface was presented in isolation but in the experimental conditions, surfaces were added on each side. To induce a shape contrast bias, flankers with different dihedral angle were added to the test and reference surface. These flankers were constant between surround conditions, but on each side we added four more surfaces (‘extended surround’) of which we varied the dihedral angle distribution. In a ‘homogenous’ condition, the flankers and extended surround were of constant dihedral angle, in a ‘surround average’ condition they were varied around the dihedral angle of the flankers and in a ‘central average’ condition the dihedral angle was varied around the angle of the central surface. Our results show that shape perception is influenced not only by the shape properties in the flanking surround but also by the mean of the distribution of shape properties in the extended surround. Thus, shape contrast is not locally determined and has to be understood from a global mechanism. We consider normalization of shape signals a likely candidate for such a mechanism.