Abstract
Resemblance theory predicts that visual search efficiency is a function of similarity along two dimensions: increased similarity between targets and nontargets decreases efficiency, while increased similarity among nontargets increases efficiency (Duncan & Humphreys, 1989). However, the theory does not specify the basis on which similarity is defined.
Tests of visual discrimination involving isolated objects have shown that symmetry relations have a direct influence on perceived similarity, with shapes that are related by symmetry over the vertical axis seen as more similar than those related by symmetry over the horizontal axis (Bornstein et al, 1978; Richards, 1978).
This study asked whether the effects of inter-object symmetry are the same for both dimensions of search efficiency in resemblance theory. Target-nontarget relations were examined by having participants search for an F among F's rotated around either a horizontal or vertical axis; nontarget-nontarget relations were examined using search for an F among distorted F's, which were symmetrical to one another over either a horizontal or vertical axis. These displays were also rotated by 90 degrees to control for any effects of shape familiarity.
The results showed that symmetry had a strong influence on search only when it involved target-nontarget relations. Symmetry among non-targets had no effect, even when a cortical scaling factor was introduced to equate the visibility of nontargets (Sally & Gurnsey, 2001).
These results imply that the two dimensions of resemblance theory do not apply uniformly to shape similarity. Further experiments are planned to determine whether this is because similarity differs for (a) target selection versus distractor exclusion, (b) explicit target comparison versus implicit distractor comparison, or (c) because shape attributes are processed differently from those that tend to group together, such as motion and color attributes.