Abstract
Human adults have a core computational system that allows for the rapid approximate enumeration of large set of objects (e.g. 36) in parallel. This system is shared broadly throughout the animal kingdom and may rely on representations in the parietal lobe. Humans also have the ability to attend to a small number of individual objects in parallel (approximately 4). With rare exception, studies of parallel enumeration and visual attention have continued in isolation from one another. A paradox presents itself. It has been suggested that visual attention binds the basic-level features of an object together making them into a coherent individual. But, if visual attention is limited to approximately 4 individuals, and it is visual attention that makes an object available to higher-level processes like approximate enumeration, how are we capable of enumerating upwards of 50 visual objects within a single flash of only 250 msec? Is this counting without individuals? In a series of experiments, participants were flashed arrays containing large numbers of colored dots or shapes. Between 1 and 8 different colors (Experiments 1–3) or shapes (Experiment 4) were presented. Participants were required to enumerate multiple subsets of the array in parallel, defined by color or shape, within a single masked presentation. Participants succeeded at enumerating multiple subsets (up to 4 on any given trial) when they were defined by a unique basic-level feature (e.g. color), but not when they were defined by a salient higher-order feature (e.g. shape). We demonstrate that it is the representations of preattentive vision that underlie parallel enumeration, and that humans are capable of enumerating up to 4 sets in parallel.