Fig 1.
Theoretical outcomes of a delayed discrimination experiment for disparity memory.
The test disparity where participants choose nearer and farther with equal probability (horizontal dashed line) in the test phase of the trial is the point of subjective equality (PSE). If disparity is retained accurately in memory, the PSE is expected to correspond to the reference disparity to be memorized (blue). A shift of the memory trace either towards nearer or farther disparities results in corresponding shifts of the PSE (shown in green and orange, respectively). ‘Relative test disparity’ on the abscissa refers to the difference from the reference disparity, which is different from what is usually meant by ‘relative disparity’ in the literature (see Fig 2 and Materials and methods for details).
Fig 2.
A. Illustration of reference (indicated in bold) and corresponding test disparities in relation to the monitor and the observer expressed as absolute disparities. The vertical lines represent the perceived order of the various disparities relative to the monitor plane. For crossed disparities (-), greater values represent stimuli closer to the observer, whereas higher uncrossed disparities (+) seem further away from the monitor plane. The top row indicates the small test disparity range (reference value: 17.5’), and the bottom row represents the higher disparity range (reference value: 28’) employed in this paradigm. B. Illustration of reference (indicated in bold) and corresponding test disparities in relation to the monitor and the observer expressed as relative disparities. Zero (0’) disparity represents the actual reference stimulus, and test disparities are expressed as the difference between the reference and the test disparities. C. Stereo pair for free fusion for illustrating one of the stereograms used as reference or test stimulus in the study. When viewed by crossed (convergent) fusion, a horizontal bar is seen in front of the image plane. Throughout the experiments all stimuli were presented on a 3D monitor and viewed through polarized goggles.
Fig 3.
Schematic representation of the delayed disparity discrimination task.
A. No-mask experiment (Experiment 1), B. Masking experiment (Experiment 2). Temporal sequence of the stimuli is indicated by the horizontal arrow. At the beginning of each trial, a reference stimulus with a predefined disparity appeared for 1 s. After a given Interstimulus Interval (ISI = 0, 1, 2 or 4 s), a test stimulus was shown, and observers were requested to indicate whether the test appeared closer or farther to the observer than the reference stimulus. Mask conditions were presented in separate blocks, whereby a 667 ms long mask stimulus of variable disparity appeared in the middle of the 2 s long ISI.
Fig 4.
Example proportion “farther” curves (responses) as a function of relative test disparity.
Data of the 9 participants (grey lines) for +17.5’ reference disparity and 0.8 contrast with the fitted logistic functions (blue). The blue dot indicates where the fitted function crosses 0.5 on the ordinate; its abscissa is the point of subjective equality (PSE). The difference of the PSE from the reference disparity (vertical blue lines) suggests a shift of remembered disparity in short-term memory. Interstimulus intervals (ISI) are shown at the top of each panel.
Fig 5.
Mean accuracy (proportion correct) of responses as a function of interstimulus interval (ISI).
Data for four different reference disparities (shown in different colors), at various relative test disparities (RTD, columns) and two different stimulus contrasts (rows). Test stimulus disparities are expressed relative to the reference disparity with negative values meaning nearer, and positive values meaning farther perceived depths than the reference depth, respectively. Each data point shows the mean ± SD of 9 participants.
Fig 6.
Proportion of “farther” judgements as a function of relative test disparity (PF curves) for Experiment 1.
Data are shown for four different reference disparities (in different line colors), four different interstimulus intervals (ISI, increasing from left to right) and two different stimulus contrasts (different rows). Each data point shows the mean ±SD of 9 participants. Test disparity is shown relative to the reference disparity. Horizontal bold grey lines indicate 50% “farther” responses, vertical lines represent reference disparity.
Fig 7.
Change in PSE as a function of ISI.
Data points show the mean and SD of bootstrapped data sets. Contrast levels 0.2 and 0.8 are indicated by light grey and black data points, respectively. The horizontal blue lines represent the disparity of the reference stimulus, which is also indicated by blue numbers to the right of the respective data series. Linear regression on the data points is shown by solid lines. Dashed horizontal lines indicate the PSE values at the calculated crossing of the linear trends for pairs of high and low reference disparity (from top to bottom, 22.21’, 21.21’ -19.67 and -21.07’). Separate lines are shown for each contrast level.
Fig 8.
Proportion of “farther” judgements as a function of relative test disparity (PF curves) for Experiment 2.
Data are shown for four different reference disparities (in different line colors), four relative mask disparities (RMD, different columns), two stimulus contrasts (different rows). Relative mask disparity equals |mask disparity| − |reference disparity|, therefore negative values mean masks closer to zero disparity than the reference stimulus. Each data point shows the mean ±SD of 9 participants. ISI was 2 s throughout. Horizontal grey lines indicate 50% “farther” responses, vertical lines represent reference disparity.
Fig 9.
The point of subjective equality as a function of relative mask disparity.
Reference disparity is indicated by the horizontal blue lines. The two panels on the left show data for uncrossed disparities, whereas those on the right show data for crossed disparities. Contrast levels 0.2 and 0.8 are indicated by light grey and black colors respectively, of the data points. Each data point shows the mean and SD of bootstrapped data sets (see Methods for details), and trend lines are linear fits.