Targets (squares with a gap on one side) were presented 0 0 , 2 0 , or 50 from the center of the fovea. The targets were surrounded by bar masks, and the spacing between targets and masks was varied parametrically. The resulting functions were used to estimate the extent of contour interaction at different target eccentricities. Our estimates for the extent of contour interaction were approximately .1 0 for targets at the center of the fovea, .24 0 for targets located 2 0 from the center, and .8 0 for targets 50 into the periphery.Lateral masking refers to the finding that the probability of correctly identifying a target is reduced when that target is surrounded by other items. Lateral masking has been demonstrated by a number of investigators (e.g., Bouma, 1970;Estes & Wolford, 1971;Mackworth, 1965). In this paper, we examine the effect of target-mask separation at different retinal eccentricities.In a recent paper, we argued that lateral masking was a composite of at least three components. One component was sensory interaction at the level of contours or features (Flom, Weymouth, & Kahneman, 1963). We will refer to this as contour interaction. A second component was the allocation of some attention or capacity to the masks that might otherwise be available for the target (Banks, Larson, & Prinzmetal, 1979;Kahneman & Henik, 1977). This might occur because of some uncertainty as to what is target and what is mask (Wolford & Chambers, 1983), or, perhaps, because some attention is automatically allocated to the masks as well as to the target. We will refer to this as competition for attentional resources. A third component occurs when the masks are drawn from the set of possible targets. Performance is facilitated when the target and mask are assigned to the same response, but inhibited when the target and mask are assigned to different responses (Eriksen & Eriksen, 1974). We will refer to this as response competition. Wolford and Chambers (1983) argued that all three factors were involved in lateral masking. Their relative in fluences depend on a variety of factors such as the spacing between the target and the mask, the type of masks used, and the configuration of the display in terms of possibilities for perceptual grouping. They further argued that contour interaction was especially prevalent when the target and the mask were closely spaced, and demonstrated that lateral masks have qualitatively different effects at close spacing than they do at wider spacing. They concluded that, at close spacing, contour interaction was the We would like to thank Carol Fowler for her helpful comments on the manuscript. Reprint requests should be sent to George Wolford, Department of Psychology, Dartmouth College, Hanover, NH 03755. 457 dominant factor in lateral masking, but that at more distant spacing contour interaction ceased to playa role. La Heij and van der Heijden (1983) also argue that contour interaction ("feature-specific interference," in their terms) is obtained only at close spacing.Over what separation does ...