Proceedings of The Physiological Society

University of Cambridge (2004) J Physiol 555P, C162


Does the short wavelength system contribute to human stereoscopic depth perception?

J. S. Lauritzen * and E. Negro †

* University of Ulster, Coleraine BT52 1SA, UK and † ESIL, Marseille, France

The short wavelength branch of the visual system (SWS) has poor spatiotemporal resolution, and S-cones are more fragile, so that conditions like glaucoma selectively damage this branch of the visual system first. Psychophysical studies of the SWS in isolation can be carried out by adapting the L and M-cones with a yellow background, and presenting stimuli at the preferred wavelength of the S-cones.

We investigated whether people use information carried by the SWS to extract stereoscopic depth cues. Stimuli were displayed on a computer monitor with the red and green guns disabled. The entire apparatus was contained in a lightproof case with a viewing port at one end. A yellow adapting background could be interposed using a beam-splitter. Pictures of objects were viewed binocularly through LCD shuttered glasses, and could be displayed as page-flipped stereograms. This allowed 4 viewing conditions: non-adapted, no stereo display; yellow-adapted, no stereo display; non-adapted, with stereo display; and yellow-adapted, with stereo display. Three two-alternative forced-choice experiments, with feedback given after each trial, were carried out under the 4 viewing conditions. An experiment consisted of 20 presentations each of 10 different stimuli, presented in pseudorandom order:In experiment 1, subjects (n = 7) judged whether a wire frame box was taller or shorter than a perfect cube. This task requires a judgment based on 2D shape only. In experiment 2, subjects (n = 7) judged whether a wire-frame box was deeper or flatter than a perfect cube. This task uses a combination of 2D shape and stereo-cues. Experiments 1 & 2 used ten boxes deviating from a cube by ± 2-10 %. Lastly, subjects (n = 5) judged which of two solid spheres appeared closer to them. The spheres were offset by 5 to 25 voxels. The 2D outline of the two spheres was virtually identical for all trials, so that the task relied primarily on stereo-cues.

We found no significant differences between the viewing conditions in experiments 1 & 2 (thresholds ± S.E.M.: 2.2 ± 2.8 % deviation and 5.9 ± 1.2 % deviation; ANOVA, P = 0.99; P = 1.0), with neither the stereo-cues nor yellow-adaptation affecting performance. In experiment 3, there was a large difference between the stereo and non-stereo conditions, with subjects performing at chance for all levels of offset without the stereo-cues; there was good discrimination when stereo-cues were allowed (P = 4.2x10-4), and with significantly better performance when not adapted (threshold offset: yellow adapted: 13.1 ± 0.65 voxels, non-adapted: 7.6 ± 4.4 voxels; P = 0.013).

This suggests that the visual system does make use of SWS information for stereoscopic depth perception, although the lower spatial resolution of this system means that discrimination based on it is less accurate.

Where applicable, experiments conform with Society ethical requirements