Generating visual images: units and relations

Three major conclusions were drawn from the results of the experiments reported in this article: First, mental images may be constructed by amalgamating images of individual parts, and an increment of time is required to add each additional part to an image. This was true when "parts" were defined by the Gestalt laws of proximity, continuity, or similarity, when parts of objects were presented on separate pages initially and the subject mentally "glued" them together into a single image, and when the number of parts was varied by altering the way an ambiguous geometric form was described. Second, descriptive information can be used in constructing images. Subjects were able to image scenes in accordance with descriptions that specified the relative distances between component objects of the scene. More time was required to form images of scenes containing more objects, and more time was later required to scan between two imaged objects if they were mentally pictured at greater distances. Third, the ease of imaging a unit depends on how much material is included in each unit and on how difficult it is to locate where the unit should be placed relative to the existing portions of an image. This conclusion was supported by the finding that subjects require less time to image arrays composed of units containing fewer letters and require less time if arrays are composed of relatively discriminable letters than if arrays are composed of relatively indiscriminable letters. Finally, in two of the experiments nonimagery control groups were tested to demonstrate that generating an image is not the same as simply retrieving memorized verbal information or reviewing information in some more abstract format.