Local versus global minima in visual pattern completion

A critical discussion of Kellman and Shipley's (1991) theory of occlusion phenomena

“…They therefore predict a preference for an occlusion interpretation with the square in front of the other figure. However, Figure 3 C has been presented to subjects in several studies (Dinnerstein and Wertheimer, 1956;Buffart et al, 1981;Boselie, 1988), and in all cases the preferred interpretation is a nonocclusion interpretation, namely "a cross next to a square".…”

“…Yet, as Kellman and Shipley themselves note, occluded triangles are reported as triangles. In fact, the phenomenon that edges that are oriented at acute angles connect phenomenally behind an occluding object, forming an acute angle behind the occluder, has been reported time and again in the literature (Boselie, 1988;Boselie and Wouterlood, 1989;Gerbino and Salmaso, 1987;Kanizsa and Gerbino, 1982, Wouterlood and Boselie, in …”

“…A, B these displays are preferably interpreted as instances of occlusion, edges running into each other at an acute angle, producing a kite in A, a trapezoid in B. Note that these solutions are preferred over occlusion interpretations with obtuse angles, for instance, a regular pentagon in A, a regular hexagon in B, which are simpler and more regular than the forms with acute corners (Boselie, 1988) C a cross next to a square. D, E though these displays do not differ in the relatability of the black edges leading into the gray region, the shape of the black parts is easily seen in D (crosses), but not in E. F depth ordering in the absence of interposition and surface quality differences; the four three-line junctions are of the arrow type.…”

Competing global and local completions in visual occlusion.

“…During past decades several studies on visual occlusion and amodal completion have been done by various researchers (cf. Boselie, 1988;Boselie & Wouterlood, 1989;Buffart, Leeuwenberg, & Restle, 1981;Chapanis & McCleary, 1953;Dinnerstein & Wertheimer, 1957;Gerbino & Salmaso, 1987;Kanizsa & Gerbino, 1982;Kellman & Shipley, 1991;Sekuler & Palmer, 1992;van Lier, van der Helm, & Leeuwenberg, 1994;. In the literature two types of amodal completion are frequently described: According to a global completion, the completed shape is as simple as possible, and according to a local completion, the completion itself is as simple as possible.…”

“…However, the global minimum principle has been criticized too by several researchers who demonstrated that the preferred interpretation does not always coincide with the interpretation with the simplest shape (cf. Boselie, 1988;Kanizsa, 1985;Rock, 1983). The pattern given in Figure 3 was brought in by Kanizsa (1985).…”