Abstract
Mirror-image symmetry between target and distractor can result in more difficult search than symmetry about an oblique axis (e.g., Wolfe and Friedman-Hill, 1992). Search for a deviant target stimulus (e.g., a tilted line) among prototypical distractors (e.g., vertical lines) is easier than vice versa (Treisman and Gormican, 1988). A corollary is that search should be more difficult if both the target and distractors are deviant stimuli than if only the target is deviant. Finally, apprehending spatial relations of objects' parts is not easy (e.g., Logan, 1994; Palmer, 1994; Davis and Peterson, 1998). In sum, feature search sometimes is difficult. Why? To determine why, we decomposed visual processing into component parts using psychophysical techniques (Davis and Peterson, 1998). We used three different types of stimuli: (a) letters, (b) pacmen, and (c) lines. For each stimulus type there was a mirror-image condition, in which the homogeneous distractors were mirror-images of the target (e.g., T:/ & D:\), and a standard condition in which they were not (e.g., T:/ & D:|). Relevant set size was either 2 or 4. All subjects were first tested in a discrimination experiment of set size 2, to equate target-distractor discriminability across conditions and stimuli, before participating in the visual search experiment. Our results indicated that mirror-image symmetry about the vertical axis affected the discriminability of target from distractors, but may not adversely affect search performance once discriminability was balanced across mirror image and standard conditions. Moreover, we found no evidence that searching for a deviant target among deviant stimuli was more difficult than for a deviant target among prototypical stimuli. Finally, when search involved apprehending spatial relations of parts, performance was worse than predicted by noise-limited processing (Shaw's 1980 Boundary Condition test), suggesting capacity limitations occurred.