Purchase this article with an account.
Francesca Pei, Yoram Bonneh, Vanitha Sampath, Chuan Hou, Anthony M. Norcia; Texture detection in infants. Journal of Vision 2002;2(10):88. doi: https://doi.org/10.1167/2.10.88.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We investigated whether infants can differentiate between orientation-organized textures and random images. We defined textures as patterns that have regularities in spatial scale and orientation. Texture-detection requires the registration of the statistical distribution of local orientation and scale measurements. We recorded Visual Evoked Potentials (VEP) to textures based on Gabor patches. In experiment 1, we measured responses to an organized texture (all patches vertically oriented) alternated at 1Hz with a random image (each element randomly oriented). We compared this test condition with a control condition consisting of the alternation between two independent random configurations. We collected data from 10 adults and 2 groups of infants (8.4–35.4 wks of age). In experiment 2, the global orientation of the organized pattern changed randomly, rather than always being vertical (10 adults and 2 groups of infants between 8.8–30.7wks of age). The carrier of each Gabor was 2.45cpd. Two full cycles were visible. The center-to-center separation between each Gabor was 3l.
VEP vector-amplitude was measured for the first to the sixth harmonics. In both experiments, infants and adults showed significant odd harmonics in the test condition that were not present in the control condition which indicates that the response to the transition from organized texture to random stimuli was not the same as that from random to organized.
The stimulus of experiment 1 had a local cue — every other update of the image contained a vertical patch. Local mechanisms, such as orientation-specific adaptation could possibly have contributed to this response.
Discrimination between the two configurations in experiment 2 could only have been accomplished by comparing relative-orientation of two or more patches, as no local information at a single patch distinguished the two configurations. Therefore, we conclude that texture discrimination can be accomplished on the basis of orientation information alone no later than 2–3 mths of age.
This PDF is available to Subscribers Only