Abstract
Adults and infants can represent up to three ensembles in parallel (Halberda et al., 2006; Zosh et al., in revision), storing, for example, the approximate number of red and yellow dots present. Critically, in these studies ensembles were defined by a single feature (e.g., colour); leaving unknown how observers represent ensembles from arrays containing multiple orthogonal features. For example, if presented with an array containing many red items and yellow items, where items could be either circles or crosses, can observers perform two orthogonal parses (colour-defined and shape-defined) to enumerate the ensembles? In Experiment 1 we asked whether 9-month-old infants can parse an array into ensembles using only shape. Infants were habituated to arrays containing a constant number of yellow circles and yellow crosses. At test, we found that infants looked longer at novel arrays in which the number of either ensemble (circles or crosses) had doubled, relative to familiar arrays in which the numerosity of each ensemble remained unchanged. This suggests that infants represented both shape-defined ensembles in memory. In Experiments 2 and 3, we asked how infants represent an array containing orthogonal ensembles. Infants were habituated to arrays containing yellow circles, yellow crosses, red circles, and red crosses. In Experiment 2, one of the colour-defined ensembles (e.g., all of the yellow items) doubled in the novel test array, and we found that infants looked longer at these arrays. In Experiment 3, one of the shape-defined ensembles doubled; infants failed to look longer at these novel arrays. In summary, infants can represent ensembles of items based on either colour or shape. But when both features are present in a single array, infants represented colour-defined but not shape-defined ensembles. Infants appear to enumerate multiple ensembles in parallel, but their ability to track orthogonal ensembles may be limited by feature type.