An intriguing set of studies carried out by Roger Shepard and his associates has provided additional information about the nature of mental images, their distinctiveness from verbal information, and the role they play in cognition. In an early study, Shepard had subjects think about such questions as the number of windows in their house. He noted that the time required to produce an answer increased with the number of windows counted, consistent with the idea that individuals actually were mentally manipulating some sort of image. Further, subjects described themselves as taking a "mental tour" of their house in order to respond to this question. At least subjectively, there was a strong impression of mentally picturing - looking at or scanning - images.
In a later series of studies, Shepard and his co‑workers showed that mental images generated by persons underlie a number of cognitive operations. In one set of studies, for example, persons were asked to judge whether three‑dimensional objects presented in different orientations were identical; see Figure 3.1. The fascinating result was that the time required to make the judgments increased linearly with the extent of rotation required. That is, it appeared that persons were mentally rotating the objects in order to make the comparison; the greater the rotation, the longer it took to make a judgment.
More recently, Stephen Kosslyn and his colleagues have demonstrated other interesting effects. For example, in one study, persons were asked to memorize a map of an island on which such objects as a tree, rock, or hut were depicted at varying locations; see Figure 3.2. After the map was committed to memory, they were asked to focus on a named object on the map. They then were given the name of a second object and told to locate it by imagining a black speck moving in a straight line from the first object to the second. Objects were, of course, varying distances from one another on the map. If the mental image is being scanned, as Kosslyn hypothesized, then time required to move from one object to the next should vary directly with the distance on the image. In fact, this was what Kosslyn and his associates found. "Distant" objects took longer to reach than "near" objects, demonstrating that images, like pictures, contain information about the spatial relations among objects.
Figure 3.1: Pairs of patterns with different orientation.
These pairs of figures are similar to those used by Shepard and XletzIer (1971) in their study of the mental rotation of three‑dimensional objects.
Figure 3.2: An island map.
Using persons' ability to form images of different sizes (e.g., a large rabbit versus a very tiny rabbit) and at different locations (e.g., nearby versus far away), Kosslyn also has shown that when persons are asked to verify certain features of mental images (e.g., "Do rabbits have whiskers?"), details of "small" images (e.g., a small rabbit) take longer to verify than those of "large" images (a large rabbit). According to Kosslyn, such evidence points to the conclusion that images have a "grain" or resolution. Thus, portions of images visualized as subjectively smaller actually make details harder to discern.