Abstract
Purpose. We have previously shown that oriented, contrast-defined elements (2nd order), containing carriers that are both spatially localized and static, do not support reliable contour integration, even when the depicted contour is straight (Hess, et al V.R. 40, 200). We decided to revisit this issue with a different type of 2nd order stimulus, one that might be less prone to potential “self-masking” effects (interference) from either the static components of the carrier or the coarse 2nd order information at the spatial scale of the Gaussian envelope (window). Methods We used arrays of Gabor elements, each of which contained an oriented, stationary, sinusoidal waveform (1.8 c/d) that modulated either the luminance (1st order) or contrast (2nd order) of a spatially 2-d random noise carrier. The carrier was either local (confined to the Gaussian window) or global (spatially extensive) and was replaced with a new stochastic sample at various temporal rates (1 to 75 Hz). Orientation-discrimination performance for 1st and 2nd order stimuli was equated using a single element. Using a standard 2AFC task, observers chose which of two intervals contained an elongated spatial contour (path) defined solely by orientation. One interval, chosen at random on each trial, contained 158 elements of random position and orientation (background) and in the other interval (path plus background) a subset (8 elements) were arranged to lie along the invisible backbone of an elongated contour. Results Reliable linking of 2nd order spatial form is possible, particularly when the carrier is both global and dynamic, but it is much weaker than for 1st order stimuli. Also, contours composed of both 1st and 2nd order elements appear to feed into a common spatial linking mechanism suggesting a unitary association field. Conclusion 2nd order spatial cues can be linked across space but provide a relatively impoverished stimulus for driving contour integration mechanisms.
RFH is funded by NSERC grant #0GP0046528