Abstract
Purpose. We wanted to know if the site of orientation integration (Dakin, JOSA 2001) was at an early or late stage in the visual pathway relative to the site of binocular integration.
Methods. We used a task in which subjects had to judge the mean orientation of an array of oriented Gabors. The Gabor orientations were samples from a Gaussian orientation distribution of variable bandwidth and mean that was left or right of vertical. The internal noise and number of samples were estimated from fitting a standard summation-variance model to the data. These orientation samples were either presented to one eye or to both eyes under dichoptic viewing. When presented to both eyes they could be in the same disparity plane or in different disparity planes. In some cases, signals of random orientation (termed noise) were added to the signal orientations in one or other of the above conditions.
Results. Performance on this task depended on whether the signal and noise were presented in different disparity planes. Furthermore similar results were obtained for dichoptic and monoptic viewing conditions. Interestingly, noise significantly (p<0.05) disrupts performance when it is presented to the dominant eye, leading to higher thresholds, higher internal noise, and decreased sampling efficiency.
Conclusions. Our results suggest that the site of orientation integration is not only after the site of binocular integration but also after the site where disparity is encoded. The finding that the effectiveness of noise depends on the eye to which it is presented, even though this information is not known to the subject, suggests that there are also monocular pathways through which this type of integration can occur although their sensitivity must be reduced compared with their binocular counterpart.