Purchase this article with an account.
James L. Dannemiller, Melanie A. Lunsford; Element grouping with parabolic contours. Journal of Vision 2006;6(6):329. doi: https://doi.org/10.1167/6.6.329.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Humans are remarkably good at detecting contours even when there are contour breaks and uncertainty regarding the global orientation of the contour (Geisler et al., 2001). We used a parabolic contour that comprised 13 oriented segments, each 0.2 deg in length separated by a center-to-center distance of 0.4 deg to examine the accuracy of contour detection when the contour shape was constant, but its orientation was uncertain on each trial. Orientation jitter was added to the contour elements in the range from +/-25 deg, and these elements were embedded in a 7.7 deg field of 364 distractor elements with precisely the same orientations on each trial as those that comprised the orientation-jittered contour. Thus, we ensured that the distribution of distractor orientations was exactly uniform and identical on each trial to the distribution of the noise-jittered contour elements. The contour appeared randomly in one of four orientations for 167 ms with its apex offset from the center of the field. Across three observers (5120, 5120 and 2560 trials), we observed percentages of correct judgments of 92%, 86% and 87%. Based on the stimulus parameters in Geisler et al. most closely matching our stimulus (e.g., orientation jitter, contour smoothness and curvature), our high accuracy rates suggest that observers may be able to use the known shape of a contour to facilitate its detection in noise. This could imply additionally that grouping of elements in noise can be influenced by top-down information.
This PDF is available to Subscribers Only