Purchase this article with an account.
Cheng Qiu, Daniel Kersten, Cheryl Olman; Segmentation effects on the tilt illusion: contrast and depth. Journal of Vision 2012;12(9):1291. doi: https://doi.org/10.1167/12.9.1291.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In the tilt illusion, when the orientation of the center and surround gratings differ by a small angle, the center grating appears to tilt away from the surround orientation (repulsion); however, for a large difference in angle, the center appears to tilt towards the surround orientation (attraction). Schwartz et al. (2009) showed that a segmentation model based on the Gaussian Scale Mixture model of natural image statistics in terms of orientation features could account for both attraction and repulsion. We measured the effect of two other sources of segmentation information, contrast and stereo disparity differences between the center and surround, on the strength of the tilt illusion in human observers. By plotting the degree of the tilt bias as a function of the difference in orientation between the center and surround, we observed: 1) when the center contrast was high (70%), both low-contrast surround and stereo depth segmentation cues reduced the amplitude of both the repulsion and attraction effect, relative to a high-contrast, 2D surround; 2) when the center and surround were both low contrast (10%), the repulsion was stronger than in the condition in which they were both 70% contrast; 3) stereo depth cues in this low-contrast condition reduced the repulsion effect but had little effect on attraction; 4) a higher surround contrast relative to the center decreased the repulsion effect but increased both the range and magnitude of the attraction effect. We modified the segmentation model from Schwartz et al. in order to include contrast and depth segmentation features. The probability of perceptual grouping of the center and surround predicted from the psychophysics results by the modified model is consistent with natural scene statistics in terms of contrast features. We also show that the predicted neural population responses are consistent with existing electrophysiology and brain imaging data.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only