Capacity Limits on Visual Imagination

Abstract

When memorizing a static set of objects with different features (e.g., red, yellow, and blue circles), or tracking a moving set of objects with identical features (e.g., purple circles), accuracy-based capacity estimates are typically 3–4 objects (Zhang & Luck, 2008). When these tasks merge, requiring a viewer to bind different features of multiple moving objects, capacity drops to 1–2 objects (Xu & Franconeri, 2015). These tasks focus the viewer on an external display, in which object features are readily available to the viewer for access and bindings can be more readily upheld. But the visual system can also represent and transform information that is generated endogenously. Might these same patterns of success and failure occur for sets of object features generated from a viewer’s own imagination, such that subjective reports of performance mirror these objective capacity measures? We showed participants sets of simple colored objects (1–4 items; either the same or different colors), followed by a blank screen, where we asked them to imagine the objects moving in a variety of ways (linear translation or rotation, as a group). They then reported the level of difficulty for each transformation (1–5 scale). Although these participants (N=30) were naïve to previously established capacity limits, their subjective reports mirrored past results. Difficulty ratings rose with increases in set size (1.1 unit increase from 1 to 4 items), and while this was true for objects of identical color (0.7 unit increase from 1 to 4 items), the rise in reported difficulty was accelerated for displays in which objects were different colors (1.5 unit increase from 1 to 4 items). We propose it is time to integrate new research themes (capacity, ensemble processing, etc) with the classic, but relatively dormant, research topic of mental imagery.