Abstract presented at the OSA/UCI "Vision & Color" Meeting 2001

Abstract published in Optics Express 9(8).

PDF Hand-Out | Abstract | Related Publications

((W.H.A. Beaudot, K.T. Mullen))

McGill Vision Research, Department of Ophthalmology, McGill University, Montreal, Canada.

Purpose. We investigated the limitations of color vision compared to luminance vision in two tasks requiring the spatially long-range integration of local orientations: i) a contour integration task associated with early cortical stages, and ii) a global shape discrimination task associated with higher cortical stages.

Methods. The contour detection task requires the linking of orientation across space to detect a path consisting of 10 aligned Gabor elements embedded in an array of randomly-oriented elements (14 x 14), measured using a temporal 2AFC method. The global shape discrimination task consists in the comparison between circular and non-circular (radially modulated D4) stimuli, measured as a radial modulation threshold. In both sets of experiments, stimuli were band-pass with a low peak spatial frequency and isolated the red-green, blue-yellow, and achromatic post-receptoral mechanisms. Contrast was matched in multiples of stimulus detection threshold.

Results. First, we found that all three mechanisms perform similarly on contour integration, show the same dependence on path curvature, and have similar internal orientation noise and relative efficiency. Second, although the blue-yellow mechanism performs worse than the red-green mechanism for the global shape discrimination, with the achromatic mechanism performing best, chromatic global shape perception can still reach hyper-acuity performance levels comparable to luminance vision.

Conclusions. These experiments demonstrate that color vision is only slightly worse than luminance vision when pre-cortical factors are taken into account such as contrast sensitivity and spatio-temporal resolutions, and suggest that color vision has sufficient orientation selectivity and processing capabilities to perform basic 2D shape perception.

Research supported by CIHR Grants to K.T. Mullen (MOP-10819)

Related publications:

© 2001 McGill Vision Research Centre