We suppose that line drawings are perceived and represented in memory as a hierarchy of related parts and subparts, as dictated by Gestalt laws like common direction and spatial proximity. Therefore, a figure fragment comprising a natural part of an originally studied pattern should serve as a strong retrieval cue for redintegrating memory for the pattern, whereas an equally large fragment suggesting either no units or misleading units should lead to poorer recall. This was confirmed in an

more »

... iment in which subjects studied 33 nonsense line drawings; recall of each was tested with good, mediocre, or bad (misleading) fragments of the original patterns. Good cues had about five times more redintegrative power than bad cues. A second experiment testing multiple-choice recognition memory showed that subjects confused an originally studied pattern about four times as often with a structurally similar distractor as with a structurally dissimilar distractor (which had an equal-sized change.) Thus, memory cuing by fragments and memory confusions with slightly altered distractors indicate the significant constituents of a figure. Our primary interest here is in the way simple line drawings are represented in memory. Our second interest is in validating a particular methodology which provides evidence about such representations. Our third aim is to propose and test a theoretical algorithm that mechanically segments and assigns a psychological representation to a set of simple (nonsense) line drawings. This is important for understanding the "parsing rules" for forms shown to the human visual system. The underlying assumption is that what is remembered about a line drawing is determined by the output of a perceptual parser and by whatever associations memory adds immediately to that output (e.g., verbal labeling of what the shape resembles). Regarding the memory representation of line drawings, a number of theories differ according to the size and structure of the