Session Detail

103 children 6 to 15 years judged stimuli on 13 illusion displays. Correlations examined relationships between age, PPVT, and RPM to susceptibility. Five illusions showed developmental change. The Shepard Tabletop showed positive correlations to all 3 indices; partial correlations could not tease these apart. The Helmholz square showed positive correlations to all 3 indices; partial correlations showed a correlation with RPM when controlling for age (r = .20, p = .044). The Poggendorf showed negative correlations to all 3 indices; partial correlations showed that age was correlated with susceptibility to this illusion above and beyond PPVT (r = -.36, p < .001) and RPM (r = -.36, p < .001) whereas neither PPVT nor RPM were correlated with susceptibility when age was controlled. The Jastrow showed negative correlations with both age and PPVT; partial correlations showed that age was correlated with susceptibility when PPVT was controlled (r = -.28, p = .004) but PPVT was no longer correlated with susceptibility when age was controlled. Finally, the Fick was only correlated with RPM (r = .26, p = .009). We conclude that illusions follow different developmental trajectories and cannot be regarded as a singular construct in perceptual development.

 

Collinear masking effect (CME) is the prolonged responses to a target overlapping with a collinear structure compare to non-overlapping condition. Since perceptual grouping strength would increase with display density, we assumed that the CME, which associates with grouping strength of collinearity, should also increase with display density. To measure the size of the CME and to keep search display comparable, we designed 3 search displays with fixed 9 by 9 elements but extended 11.84° x 11.84° (the highest density), 17.76° x 17.76° (the middle density), and 26.64° x 26.64° (the least density) in visual angle. The target and collinear structure could appear randomly at 3 possible locations in the display, and the overlapping probability of target and collinear structure was at chance level. A prolonged RT to overlapping trials to non-overlapping trials is defined as the CME. The results showed that the size of CME in the highest density condition (50ms) was significantly higher than that for the middle (29ms) and the least (26ms) density conditions. Our study replicated that the CME is larger when the collinear grouping strength is stronger, showing that collinear grouping is the main causes of the CME.

 

When luminance-modulated gratings move behind an opaque black occluder, the gratings appear to be continuous on the occluder. This surface completion illusion called visual phantoms disappear as the occluder’s luminance is close to the mean luminance of inducing gratings. Contrast-modulated patterns, however, induce a different type of visual phantoms even when the occluder has a mean luminance of inducing patterns. I investigated whether visual phantoms could be induced by contrast-modulation of stationary plaids, and whether the visibility of the phantoms induced by moving contrast-modulation would be higher than that of phantoms induced by stationary contrast-modulation. Participants viewed plaids oriented 45 degrees with horizontal occluder which luminance was varied, then they rated the visibilities of the visual phantoms with stationary or moving contrast-modulation. Results showed that faint illusory plaids were observed on the occluder, and their visibilities were higher when the contrast-modulation was moving than when it was stationary. These results suggest contributions of second-order mechanisms to the perception of visual phantoms.

 

Orientation gradients are thought to play a fundamental role in orientation-based texture segregation. Studies have shown that segregation can occur when there is an abrupt change in orientation across space, i.e. a texture edge, but also in their absence. Here we investigated the role edges play in the segregation process. We measured participants’ performance to segment and detect rectangular line texture figures of differing mean orientation from the background at 5 stimulus durations. The orientation change of figure from background was abrupt (Block), or varied spatially according to Cornsweet profile (Cornsweet) or a logistic curve (Blur). Performance at 3 values of orientation jitter was also measured. As a function of edge contrast (orientation contrast at the edge), the Blur profile had the lowest threshold, followed by the Block, then Cornsweet. When plotted as a function of center contrast (orientation contrast between background and center of figure), the Blur profile had the highest thresholds. We also found higher thresholds for the segmentation task compared to the detection task (especially for the Blur profile), and higher thresholds with increased orientation jitter and reduced display duration. Therefore texture properties over regions beyond the edge play a role in the segregation process.

 

We investigated the effect of the statistics of background element size distribution on the perceived size of a target. We manipulated the first, second, and third order statistics (or mean, variance, and skewness) of the background element size distribution. We used a two-interval forced-choice paradigm to measure perceived target size at different background size distributions. In each trial, the standard disk, or target, with a texture background texture was presented in one interval while a comparison disk on a blank background, the other. The task of the observers was to determine which interval contained a larger disk. We measured the point of subjective equality (PSE) for the perceived target size with a staircase procedure. The perceived target size decreased with mean background disk size. The variance and the skewness of the background element size did not affect the perceived target size. Our results showed that only the first order statistics, but not the second order statistics of the background modulates the perceived target size. We proposed a neural based model, in which the visual system extracts size information by averaging the responses of different spatial frequency channels whose response is modulated by background element size, to account for our results.

 

Textures are changes in surface reflectance that generate edge contours in images independently of the shading attributed to shape and illumination. Recent work has shown that the perception of texture contrast varies with the orientation of reflectance boundaries relative to the direction of shading gradients (Kim, Marlow and Anderson, 2014); texture gradients appear to have greater contrast when they run orthogonally to the underlying shading gradients. The perceived contrast of textures was also found to be predicted by orientation field models that compute the change in direction of gradients adjacent to texture contours. We explored the usefulness of computing the difference in local orientation field responses for analysing textures in medical images – the structure of subbasal corneal nerve fibres imaged using in vivo confocal microscopy. We computed the local difference in orientation field responses across the image and attenuated luminance as a function of these differences. The assumption here was that orientation fields should be similar along nerve contours, but different along an orthogonal axis. We found this structure enhancement technique improved texture segmentation and accounted for perceptual judgments of nerve density, suggesting that orientation fields provide diagnostic support for the automated visual analysis of biological textures in clinical scenarios.