Abstract
The development of the perception of object unity across a spatial gap is a fundamental component of cognitive growth over the first few months of life after birth. Newborn infants have been tested for unity perception, and consistently provide evidence of perception of visible parts only, failing to link edges across a spatial gap imposed by an occluder. The earliest age at which infants have been shown to perceive object unity is 2 months, and performance becomes more robust over the next few months. Despite our knowledge of the time course of perceptual completion in infancy, little is known about mechanisms of development of this vital perceptual skill. We addressed this question by recording eye movements with a corneal reflection eye tracker in 3-month-old infants as they participated in a standard object unity task: habituation to a rod moving back and forth behind an occluder, followed by broken and complete rod test displays. We observed reliable differences in scanning patterns between “Perceivers,” those infants whose posthabituation preferences were indicative of unity perception (i.e., preference for the broken rod at test), and “Nonperceivers,” infants who appeared to perceive disjoint objects during habituation (i.e., preference for the complete rod). Perceivers scanned more reliably in the vicinity of the rod parts, and scanned more frequently across the range of rod motion relative to Nonperceivers. These scanning patterns would be expected to make available relevant visual information to support a veridical perception of a partly occluded rod. In contrast, there were not reliable differences in other kinds of scanning pattern, such as vertical scans. Perceivers were not more active in overall production of saccades. These results suggest that emerging object concepts are tied closely to the visual information available in the environment, and that developing object concepts may help guide information retrieval processes.