Abstract
Purpose: Many theories of vision include the formation of a structural description of perceptual objects as an intermediate step between feature extraction and semantic analysis. We ask whether the formation of perceptual objects is affected by divided attention. Can one create multiple perceptual objects independently and in parallel during early organization of the retinal image? Or, alternatively, must one apply a serial mechanism to create perceptual objects one by one? Previously, we found that categorization of animal images (e.g., wolf vs. bear) is limited by a fixed-capacity mechanism. In that study, either a structural or a semantic analysis of objects might have imposed the bottleneck. Here, we used a physical identification task that emphasizes the structural analysis of objects and minimizes the semantic analysis.
Methods: To measure perceptual capacity, we used the extended simultaneous-sequential paradigm, which can distinguish between unlimited- and fixed-capacity processes (Scharff, Palmer, & Moore, in press). Observers searched for a specific target object among physically similar distracters. Stimuli were photographs of objects taken from several possible viewpoints. Three sets of artificial objects were used: cut-foam solids, toy-block constructions, and crumpled papers. For example, some trials required searching for a particular crumpled paper among similar crumpled papers, all seen from varying viewpoints. Thus, observers searched for a specific object, not a category or an image.
Results: Our findings were consistent with a fixed-capacity model for the physical identification of multiple objects. These findings rule out an unlimited-capacity model. We argue that there is a fixed-capacity limit to the formation of structural descriptions of objects. This is consistent with a serial mechanism, or with a parallel mechanism that acquires information from multiple objects at a fixed rate.
University of Washington Royalty Research Fund.