Abstract
To generate a singular impression of depth, the visual system combines independent depth cues in a scene. However, these signals vary greatly in how accurately the underlying geometry is represented. Our goal was to determine how the Intrinsic Constraint (IC) Model of depth cue integration accounts for perceived depth magnitude in a motion parallax (MP) and binocular disparity (BD) cue combination study. We previously determined perceived depth magnitudes when MP and BD cues are combined. To test the IC model, standard deviations (σMP and σBD) for depth magnitude discriminations in singular cue conditions must be determined in order to independently estimate the signal-to-noise ratios for each individual cue. Observers viewed random-dot stimuli through a mirror stereoscope providing ocular separation for stereo stimuli, monocular viewing of the parallax stimuli, and stable convergence and accommodation. To determine σMP for perceived depth magnitude of a translating MP stimulus, MP stimuli were compared to other MP stimuli in a 2IFC, "greater depth magnitude" procedure. MP stimuli were quantified with the Pursuit/Motion Ratio whereby MP stimuli translated laterally generating pursuit (1.1 - 3.3 d/s) while dots within the stimulus window translated laterally (peak 0.09 - 0.55 d/s). To determine σBD of the perceived depth magnitude of a stationary BD stimulus, BD stimuli were compared to other BD stimuli. BD stimuli had a range of disparities (1.5 - 15 min). For each MP and BD stimulus, psychometric function parameters, including slope (β), were estimated using a cumulative normal function. The σs were calculated from β, noting that the empirical depth estimate was affected by the error in both the test and comparison stimuli. Using these σs, the IC model produced cue combination predictions that were similar to the depth magnitudes measured in the empirical cue combination paradigm, which cannot be accounted for the MWF theory of cue integration.
Meeting abstract presented at VSS 2014