Abstract
In this talk I present experiments demonstrating that processing an ensemble group of many items requires a single “slot” in visual attention and visual working memory; an ensemble group functions as a single item. Experiments 1–3 used subitizing as an assay for studying the deployment of attention to both ensemble groups and individual objects. When subjects were asked to enumerate individual objects, the classic subitizing elbow was seen with rapid and accurate enumeration for 1, 2, and 3 objects and slower more error-prone enumeration above 3. When similar displays appeared containing ensemble groups of many items, this same result obtained with speed and accuracy depending on the number of groups present, not on the number of objects inside each group. In Experiments 2 and 3, known factors that interfere with subitizing (e.g., presenting concentric rings rather than spatially separated rings) were found to affect both objects (rings) and ensembles (groups of dots) to the same extent, further suggesting that each ensemble group behaves like a single item for visual attention. Experiments 4–6 used change detection and partial report measures of visual working memory capacity and found that each ensemble group requires a single “slot” in VWM and multiple ensemble features can be stored for each group inside one slot (e.g., approximate number, average orientation, average size). Experiment 6 explored the time-course of consolidation of ensemble features in VWM and suggests that an ensemble is first selected as an item and then feature information is consolidated in parallel, similar to the consolidation seen for individual objects. These results inspire a reconsideration of the importance of selection prior to computing ensemble statistics and expand what can count as a single item for attention and memory.