Abstract
I propose a network mechanism for the determination of apparent tilt based on human psychophysics and physiology. The network model is capable of determining the apparent orientation of bars, including those in various types of tilt illusion. The model extracts local orientation information by V1-cell-like units with a Gabor-like receptive field. The activities of the units are grouped together based on co-linear synchronization observed in V1 superficial layers. The model shows good quantitative agreement with human responses in the determination of the apparent orientations of bars in a variety of bar junctions.