Subitizing sets and set-based selection: Early visual features determine what counts as an individual for visual processing

Abstract

Human adults can select a set of items that share an early visual feature (e.g. the red dots) from a scene that contains many spatially overlapping sets (e.g. different colored dots completely spatially intermixed) and encode set-based summary representations for this selected set, e.g. approximate number of items (Halberda et al, 2006). This selection appears to take place in parallel for all items in the set across the visual field. While it is well known which features support pop-out for a single item among distractors, virtually nothing is known about which features will support the selection of an entire set of items in a visual scene. Adults viewed brief presentations of spatially overlapping sets that differed in a single feature (e.g. tilted lines among verticals) and estimated the approximate number of targets. A wide variety of features were tested separately including all those considered to support item “pop-out” and key ones that do not (e.g. complex shape). Results indicate that, just as early visual features support parallel search for a single item, they support the selection of an entire set of items in parallel (e.g. 13 tilted lines intermixed with 15 verticals) (Experiment 1). This set-based selection also shows classic asymmetry effects (e.g. tilted easier than vertical) (Experiment 2). And, as further evidence that sets function as individuals for attention, adults showed classic “subitizing” signatures for precisely enumerating up to 3 sets accurately and in parallel (Experiment 3). These results extend what counts as an “individual” for visual processing to include entire sets of items that share an early visual feature.