Visual interpolation is not scale invariant.

According to the scale-dependence hypothesis, visual interpolation depends on the retinal separation of end- points: as retinal size increases, the interpolated contour deviates from the shortest path and approaches the unoccluded angle. In the field model (Fantoni & Gerbino, 2003) as the retinal size increases good continuation increases while minimal-path decreases. We ran two experiments using the probe localization technique. Retinal size was manipulated by changing viewing distance. Observers were asked to judge if a probe, briefly superposed on the occlusion region, was inside or outside the amodally completed angle. Retinal size strongly influ- enced the penetration of interpolated trajectories in the predicted direction. We modified the field model and generated model-based trajectories that fitted empirical data closely.