Abstract
Purpose:When pattern elements have random orientation, human ability to detect changes in orientation of a single element deteriorates dramatically with the total number of elements (change blindness set-size effect). However, when elements have orientations that are tangent to a circle, there is little, if any, set-size effect (Kempgens, Loffler & Orbach, VSS 2006). What are the factors producing this surprising ability to detect change in multi-element configurations?
Methods:Stimuli were composed of multiple D6 (6 cpd) elements, which were positioned as if on circular contours. We studied the effects of set-size (3–40 elements), inter-element spacing (0.7° – 4.7°), closure (a circle vs. a circular arc), circle radius (1.2°–6.2°), element orientation (tangential, radial, or jittered by ±2° to ±45° with respect to tangential), element position (radial jitter of up to ±50% of the 2.7° base radius), and the number of objects (three 120° arcs cut from a circle and everted) formed by the pattern elements. Subjects indicated in which of the sequentially presented stimuli one element had changed its orientation.
Results:Orientation change detection (threshold approximately 10°) was largely independent of set-size, inter-element spacing, closure, radius and small orientation (up to 6°) and radial position (up to 10%) jitter. However, jittering position or orientation above these levels drastically impaired performance, as did increasing the number of objects formed by the pattern elements.
Conclusions:Our results are in agreement with global mechanisms, which are relatively broadly tuned for orientation and position, sampling shape information within circular annuli. The relation between thresholds for single-circle patterns and patterns consisting of three objects (a triplet of non-concentric arcs) is consistent with probability summation between these units.