Purchase this article with an account.
Elizabeth Salvagio, Jee Hyun Kim, Mary Peterson; The mechanism for contextual influences on the configural cue of convexity. Journal of Vision 2007;7(9):904. doi: https://doi.org/10.1167/7.9.904.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Convex regions are more likely to appear as shapes (figures) than adjacent concave regions. Using displays with alternating black/white convex/concave regions, Peterson and Kim (2002, VSS) reported that (local) convexity exerts only a small influence in two-region displays (58%), but an increasingly larger influence as region number increases (85% in eight-region displays). Here we investigate the mechanism for these context effects. Experiment 1 tested whether repetition of convex regions was sufficient using displays where like regions (convex or concave) were no more similar in color than unlike regions. No context effects were observed. Experiments 2 and 3 investigated whether uniform fill (e.g., color) was necessary in convex, concave, or in both regions. Context effects were obtained with uniform concave regions (and multicolored convex regions) but not with uniform convex regions (and multicolored concave regions). We interpret these results within a competitive model of figure-ground in which (local) convexities inhibit adjacent concave regions, and inhibition is summed across non-adjacent regions filled with the same feature. This model requires inhibition of one region by a figural cue favoring the adjacent region. Experiment 4 tested whether this requirement was satisfied using displays in which multicolored straight-edged regions alternated with identical uniformly colored regions. These displays lacking figural cues revealed no context effects. Thus, context effects are obtained only when one of the alternating regions is positively cued as figure and alternating un-cued regions are filled with the same feature (e.g., color). In Experiment 5, we obtained context effects with displays in which alternating white convex/concave regions were delimited by black borders, indicating that the color of positively cued regions need not differ from that of inhibited regions. These results suggest that long-range interactions occur between inhibited regions that are uniform in color and span intermediate regions regardless of their color.
This PDF is available to Subscribers Only