Abstract
Performance across several visual attention tasks, such as multiple object tracking, change detection across visual memories, or rapid counting, often seems to suggest that the visual system can handle a fixed number of objects at once. These fixed capacities, often of about 3 or 4 objects, are often taken as strong constraints on possible architectures of visual attention. However, for each of these tasks, there is debate over whether capacities are truly fixed, or rather vary according to factors related to task difficulty.
In two experiments, we asked whether there is a fixed limit on the number of spatial locations that can be selected concurrently in a task. Subjects searched for a target through a cued set of search items. These items were spatially interleaved with similar looking search items that could never be the target, minimizing the potential for ‘chunking’ several cued items together into a single location. Cues either disappeared before the search, or remained throughout a trial. We determined capacity for selecting locations by finding the cued subset size where memorized cues could no longer serve visual search as efficiently as those still visible in the display.
In Experiment 1, observers could search through 5 or 6 spatial locations before response times indicated that search strayed to known distractor locations. In Experiment 2, which used denser displays with more items, subjects could only maintain about 3 or 4 locations. These results suggest that our capacity for selecting locations is not fixed. Instead, there may be a tradeoff between the number of locations that can be selected, and the precision with which their positions are encoded. These results parallel other work using multiple object tracking tasks, showing a tradeoff between the number of tracked items and the precision with which their positions are encoded (Alvarez & Franconeri, VSS 2005).