Recent work (Shimojo, Silverman & Nakayama, 1989; Vision Research, 29, 619-626) suggests that the visual system must discriminate between extrinsic boundaries (boundaries created by front occluding surfaces) and intrinsic boundaries (real object boundaries) in order to recognize objects and that this would importantly affect the way it solves the so-called "aperture problem" in motion. With the aid of a series of demonstrations (plus two formal experiments) we (1) propose a new explanation for the fact that edge line terminators in a "barber pole" display are perceived as intrinsic; (2) show that inner line terminators in a plaid pattern (i.e. those resulting from the intersection of the two sets of bars) specify coherent or incoherent motion depending on the availability of occlusion-related depth information; and (3) suggest a unitary scheme integrating the two current alternative solutions to the aperture problem, the one based on line terminators and the one based on a velocity space combination of orthogonal components.