Abstract
The human visual system has been assumed to integrate the information from several cues for depth perception. Landy et al. (1995) proposed a model for the process of depth cue integration, which stated that perceived depth depends on the weighted sum of the depths specified by the cues. According to the model, the weight of a particular cue depends on the estimated reliability of that cue relative to those of the other cues, and the reliabilities of cues depend on the scene contents. This model, however, has difficulty in accounting for the individual differences in the depth cue weight (Johnston et al., 1993; Sakano et al., 2002).
We hypothesized that the individual differences in depth cue weight are due to the individual differences in the past experience combining with the dependency of the depth cue reliability on the experience to particular scene contents. If this hypothesis is valid, the weight of disparity would be large for the person whose viewing distance has been biased to near distances, because the reliability of disparity would be high at near viewing distances taking the geometrical characteristics of disparity into account. To test the hypothesis, we measured the relative weights of disparity and perspective at several viewing distances and compared them with some indices of the viewing distance bias in the past experience. We used ocular refraction and normal viewing distance when performing some tasks as the indices.
The results showed that the weight of disparity decreased as the viewing distance increased for many subjects, which is consistent with the previous reports, and that the individual weight of disparity negatively correlated with the viewing distance bias in the past experience indicated by the measured indices. These results suggest that the weight of disparity in the process of depth cue integration depends on the viewing distance bias in the past experience of each person as well as the viewing distance of the moment.