Purchase this article with an account.
Xiaoye Wang, Aaron Fath, Winona Snapp-Childs, Mats Lind, Geoffrey Bingham; Inhomogeneity of Perceived Slants With Different Motion-Based Visual Information. Journal of Vision 2016;16(12):654. doi: https://doi.org/10.1167/16.12.654.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Todd and colleagues (Todd et al., 2007; Todd, Christensen, & Guckes, 2010) reported a strong bias in perceived slant presented with static texture gradients with restricted viewing angle. People tend to underestimate the slant using texture when the viewing angle is small. The current study explored the similar distortion of perceived slant with different motion-based visual information. Methods We used random dot stereograms to display planar surfaces with 24 different slants (from 27° to 73° with 2° increment and 0° tilt) rotating around a vertical axis through the surface center, viewed under stereomotion, monocular structure-from-motion and texture, and combined visual information. Participants were instructed to adjust a line to match the slant of the surface. Results Judged slant error means and SDs were computed for each visual information condition and slant, and regressed onto actual slants. For all information conditions, error means were best fitted using concave downward quadratic trends, with maxima at 55° and intercepts with the x-axis at 40° and 70°. Judgments of slants between 27° and 40° overestimated and between 40° and 70° underestimated. SDs were also best fit by quadratics, but the peak occurred at the 40° intercept for the means with minima as slants approached 0° and 90°. Conclusion The current study extended the previous findings of bias in slant perception to different motion-related visual information. The results indicate that people tend to identify reference angles, where perceived slants were accurate and more precise. Mid-range slants were accurate but least precise. Bottom half of the range was underestimated and the top half overestimated.
Meeting abstract presented at VSS 2016
This PDF is available to Subscribers Only