Cross-dimensional interaction and texture segregation

Abstract

revision accepted for publication October 2, 1987. How do different dimensions (e.g., color and form) interact when the visual system locates boundaries in surfaces? Seven experiments are reported in which subjects located a target "singleton" element that was unique in a given dimension. Minimal interference occurred when another dimension, known to be irrelevant, was varied randomly, even if the specific identity of the target feature was unknown (Experiments 1-4). These results argue against the hypothesis of a single difference-detection process that operates on multidimensional representations. However, performance was also little affected by whether the subjects knew in advance which dimension contained the singleton (Experiment 5), so an attentional set to suppress the irrelevant dimension was not implicated. On the other hand, singletons in the (prespecified) irrelevant dimension did produce major interference when the target was unspecified (Experiments 6 and 7), which indicates that texture boundaries of similar types cannot be accessed completely independently for separable dimensions. Together, the results suggest that boundaries may be registered by detectors that respond to specific types and locations of boundaries specified by variation within a particular dimension; selection by boundary type is effective, but selection or suppression by dimension is relatively ineffective. Parallels with some recent neurophysiological findings are noted. © 1988 Psychonomic Society, Inc.