Purchase this article with an account.
Stéphane J. Rainville, Hugh R. Wilson; Global form perception in motion-defined radial-frequency contours. Journal of Vision 2004;4(8):36. doi: https://doi.org/10.1167/4.8.36.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: The nature of neural mechanisms that transform local motion signals into a representation of global form remains elusive. Here, we use motion-defined radial-frequency contours (RFs) as a general and systematic framework for investigating form-from-motion. Method: Stimuli consisted of 36 collinear Gabor elements arranged in a virtual circle. Each Gabor had a fixed envelope and a drifting carrier whose speed was determined by a sinusoidal function of polar angle. Randomly permuting speeds across Gabor elements produced incoherent modulations that served as ‘null’ stimuli in two-alternative forced-choice detection tasks. Thresholds were defined as sinusoidal amplitudes corresponding to 75%-correct performance. Results: Detection and discrimination data suggest that motion-RFs are optimally processed in the range of 1 to 4 radial cycles. Spatial-summation experiments (where coherent contours were replaced by incoherent contours over a variable pie-wedge section) showed that thresholds improved with coherent-contour length at a higher rate than predicted by probability summation. Results also revealed that random radial offsets in Gabor position impair spatial-RF detection but largely spare motion-RF detection if thresholds are equated via speed-to-position transfer functions measured for illusory motion-induced shifts in Gabor position [DeValois & DeValois, 1991, Vis. Res., 31, 1619–1626]. Conclusions: Mechanisms sensitive to motion-RFs are selective for contour smoothness and integrate motion structure globally. Results rule out local illusory positional shifts as a potential confound and demonstrate that motion pathways mediate shape perception for motion-RFs. Motion-RFs can be combined into arbitrary shapes via Fourier synthesis and therefore constitute a promising tool for studying the neural representations of complex motion-defined shapes.
This PDF is available to Subscribers Only