Infants' detection of shearing motion in random-dot displays.

Abstract

Random-dot kinematograms were used to estimate infants' thresholds for shearing motion in the absence of flicker and position cues. The principal advantage of these stimuli is that changes in dot position are camouflaged by the presence of numerous matching dots, thus necessitating the detection of motion before the extraction of local pattern features. Thirteen- and 20-week-old infants were tested with a forced-choice preferential looking technique. The target stimulus resembled a vertically oriented corrugated pattern that oscillated at 1 Hz, if, and only if, shearing motion was detected. Infants were tested at different velocities, ranging from 0.77s to 5.67s, and the results revealed minimum velocity thresholds of 3.57s and 1.2°/s for 13- and 20-week-old infants, respectively. Possible interpretations for these results based on position- or flicker-sensitive mechanisms are considered and are found inconsistent with the overall pattern of results. It is concluded that infants detect shearing motion in random-dot displays with a motion-sensitive mechanism. Visual perception begins with the projection of 3-D objects and surfaces onto the retina of the observer, where a 2-D image is formed. Yet, unlike a camera image, the resulting retinal image is rarely static; the eyes and head of the observer are almost always moving, and often, so are the objects that are the targets of one's visual regard. Accordingly, much of the visual information available to the observer begins as image motion on the retina. This information is used to specify many variant and invariant properties of the visual world, including surface segregation, self-motion, relative depth, and time to collision. Paradoxically, the status of image motion as a fundamental source of visual information was contested until recently (cf. Nakayama, 1985). In contrast to stereopsis or color vision, motion need not be an immediate experience but rather could be derived from changes in position over time. Consider, for example, the two hands of a clock. Whereas the second hand appears to undergo continuous movement, the minute hand changes position much more slowly, and its movement can be derived only from a comparison of past and current positions. For quite some time, it was considered conceivable that all image motion was processed in a manner analogous to the perception of the moving minute hand on a clock. Although the data accumulated over the past two decades clearly refute this position with regard to adult processing of moving stimuli (Cutting, 1987; Nakayama,