Contributions of local orientation and position features to shape integration

Contour integration plays an important role in linking local elements into global shape and the binding strength among local elements depends on both orientation and position features. The very high sensitivity reported for detecting the sinusoidal deformation of circular contours may result from the presence and concordance of both orientation and position cues to shape difference. In this study, position and orientation-defined micropatch-sampled radial frequency (MSRF) patterns were employed, which permit the independent assessment of the contributions of local orientation and position features to shape integration. It was demonstrated that, while both local orientation and position features can encode shape deformation, the human visual system is more sensitive to orientation-defined shape difference than to position-defined shape difference. Furthermore, integration of the local orientation feature into shape is more than two times stronger than that of local position, and may involve a global pooling mechanism. Nevertheless, optimal shape discrimination performance requires the analysis of both local orientation and position features.