Abstract
A new technique for determining red-green isoluminance for stereopsis has been suggested (Champion & Simmons, ECVP 2001). This technique is an adaptation of the motion-nulling technique described by Lu et al (1999, PNAS 96 8289–8294) but in this case chro matic and luminance images are presented to different eyes rather than in different frames of a motion sequence. If stereoscopic performance with this dichoptically mixed stimulus is equivalent to that obtained with chromatic stimuli in both eyes, then it follows that there could be a luminance component, or possibly a luminance artifact, contained in the chromatic stimulus which supports this performance in the absence of a truly chromatic stereopsis mechanism.
Experiments were performed with 0.5 cpd Ga bor stimuli in each eye. One eye was presented with a high contrast chromatic stimulus and the other with a variable contrast luminance stimulus. It was found that, for a range of R/R+G ratios, stereoscopic performance in a front/back depth identificatio n task with this dichoptically mixed stimulus could not match that obtained with the binocular chromatic stimulus. However, there was normally a small range of luminance contrasts for which performance was significantly above chance. This above-chance per formance might represent intrusion of a non-linear luminance artifact (i.e. one which does not share the exact spatio-temporal characteristics of the luminance stimulus). Further experiments with luminance Gaussian stimuli replacing the luminance Gabor co nfirmed this suspicion. Care was also taken to measure performance at a R/R+G ratio where dichoptic masking between chromatic and luminance stimuli was negligible. The reduced performance found could not therefore be explained by interocular suppressive e ffects.
In conclusion, despite the presence of relatively strong luminance artifacts in these nominally isoluminant stimuli, there still seems to be enough evidence to support the existence of a chromatic-contrast-sensitive stereopsis mechanism.