Abstract
Less than a handful of studies have explored contour detection during early infancy. The theoretical relevance of investigating contour detection in young human infants lies in the finding that the development of long-range projections characteristic of the V1 begin to emerge around 4 months. These projections are believed to be the physiological implementation of the ’good continuation’ or colinearity contour grouping rule. One prediction is that young infants' ability to spatially integrate orientation elements into contours should be limited, and this limitation cannot be explained solely by improvements in visual acuity. The present study assessed contour detection in 4- and 8-month-old infants using an eye tracker. Infants observed images consisting of pseudo-randomly oriented and positioned Gabor elements. A circle, defined by the alignment of 12 elements, was present in half of these images. Noise density also varied across images. Observers could use proximity cues in addition to good continuation to detect the circle at the lower noise levels. Only good continuation, however, would enable observers to detect the circle at the higher noise levels. The dependent measure was a difference score: fixation time on the circle minus fixation time on the corresponding region in the noise (i.e., circle-absent) stimuli. Four-month-olds fixated more on the circle only at lowest noise density. Eight-month-olds showed more fixations on the circle even at the higher density levels. Overall, the findings support the notion that grouping by good continuation is present, but limited, in young human infants.
Supported by NICHD grant no. HD-38315-R01 to P.G.