Visual enumeration of spatially overlapping subsets
Observers are able to extract summary statistical properties, such as numerosity or the average, from spatially overlapping subsets of visuals objects. However, this ability is limited to about two subsets at a time, which may be primarily caused by the limited capacity of parallel representation of those subsets. In our study, we addressed several issues regarding subset representation. In four experiments, we presented observers with arrays of dots of one to six colors and instructed them to judge the number of colors. We measured both speed and accuracy of those judgments. Following standard criteria used for the interpretation of object enumeration data, we recognized two modes of subset representation: a) parallel, effortless and strategy-independent representation of no more than two subsets, and b) serial representation modulated by different attentional strategies and a working memory template. We also found an advantage of large sets over small ones, demonstrating that subset representation can be formed based on some statistical accumulation of information from individual objects.