Abstract
Previous visual working memory studies have shown that viewers can store the visual properties (typically, colors) of approximately 3-4 items at a time (Brady, Konkle, & Alvarez, 2011; Zhang & Luck, 2008). If viewers are asked to remember the locations of each color (by testing detection of color swaps instead of replacements), this capacity drops to 2-3 items (Alvarez & Thompson, 2009). If the items move, this capacity drops as low as a single stored color-location pairing (Horowitz, Klieger, Fencsik, Yang, Alvarez, & Wolfe, 2007; Saiki, 2003; Saiki & Miyatsuji, 2009). These previous tasks allowed the viewer access to an external display, in which the object and its features were readily available. But how would viewers fare when these objects need to be represented and manipulated endogenously with visual mental imagery? Would the same factors that limit capacity in visual working memory also limit visual mental imagery?
To test the capacity limits of observers without online access to an object and its features, participants were asked to store colored circles (1-4 items), and imagine these circles translating or rotating (10, 60, 90, or 110 degrees). We then measured subjective task difficulty and objective change detection performance. Both measures suggest a capacity-limited system. Subjective responses indicated that updating color locations across transformations was more difficult with greater set sizes, for rotation transformations, and across greater distances, and objective performance worsened with an increase in set size. The capacity limiting factors found here in both the subjective difficulty ratings and objective change detection parallel those found in working memory, suggesting that visual working memory and visual mental imagery may share similar capacity constraints.