Purchase this article with an account.
Ethan Meyers, Yuri Ostrovsky, Pawan Sinha; Visual de-fragmentation via high spatial frequencies. Journal of Vision 2005;5(8):353. doi: 10.1167/5.8.353.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In the study of visual development, researchers have observed two key progressions. First, infants' percepts of the visual world apparently start out fragmented, and over several months, come to cohere into assemblies that correspond to objects. Second, along a similar timeline, children's acuity gradually improves from being rather poor at birth to near adult levels. While the underlying reasons for the second progression are reasonably well understood, the causes for the first one are still largely unknown. Here, we suggest that these two progressions might not be independent, but rather may be causally linked, with improving acuity facilitating the extraction of integrative cues. Low-resolution information permits image segmentation into regions of homogenous color and luminance attributes. But, since objects can include regions of different photometric attributes, a purely region-based parsing will lead to over-fragmentation. High-resolution information, as it becomes available, acts as ‘perceptual glue’ to link together regions based on cues such as good-continuation, contour alignment across occluders and junction structure. To test this theory, we conducted experiments, as part of Project Prakash, with a 28 year old individual who has an unusual condition of congenital aphakia (lack of lenses) in both eyes, preventing the normal progression of acuity. We find that his analysis of images exhibits a marked tendency to over-segment them, and an inability to use cues of contour relatability or junction geometry. These results provide tentative support to our theory of the developmental role of high spatial frequencies in ‘de-fragmenting’ the visual image, and thereby, facilitating the extraction of object-concepts for recognition. Additionally, the findings suggest that extended visual experience and neuronal maturation, in the absence of high spatial frequency information, are inadequate for explaining the development of visual integration abilities.
This PDF is available to Subscribers Only