Abstract
Visual discrimination is often impaired when irrelevant context items are placed near the target. This crowding effect contrasts sharply with the configural superiority effect (CSE; Pomerantz et al., 1977; Pomerantz & Portillo, 2011), in which adding identical, irrelevant contexts drastically improves the perception of targets (e.g., when discriminating positive from negative diagonals, adding identical Ls to the diagonals creates arrows and triangles that are twice as discriminable as the original diagonals). We asked Ss to locate the target in displays where three quadrants contained identical items while the fourth was odd. The items were line segments varying in orientation (e.g., a single positive diagonal in a field of verticals), and the contexts added to all four quadrants were identical (e.g., a horizontal). We compare discrimination performance against predictions based on the good news (benefits) and the bad news (costs) arising from context. We argue the benefits stem from Emergent Features (EFs) arising from pairs of line segments: Cragin’s (2010) study of two-line stimulus space identified 8 EFs arising from simple pairs of lines, including parallelism, symmetry, collinearity, and intersections. E.g., in discriminating a horizontal from verticals, adding an identical to horizontal to each results in discriminating + from =, which is much faster than the original horizontal from vertical because context creates EFs such as intersection and parallelism. In contrast to this CSE, many contexts impair performance; e.g., adding a positive diagonal to a horizontal vs. vertical discrimination impairs performance. Beyond the simple lack of EFs, we attribute these impairments to the added costs from crowding, supplemented by possible adverse effects of masking, increased processing load, and dilution of dissimilarities (Tversky, 1977). We discuss how the costs and benefits mix and assess the claim that crowding is grouping.
Meeting abstract presented at VSS 2012