Abstract
In visual-search tasks, adding context can dramatically hamper or facilitate the latencies for detecting a target-item. An appealing demonstration of the latter was provided by Pomerantz, Sager, and Stoever (1977). Participants were presented with a display containing four diagonal lines and were then asked to indicate the location of the odd line - which differed from the other three by its orientation. In a second condition, a fixed context (carrying no task-relevant information) was added to each of the lines: an L shape that created a closed figure of a triangle when combined with the odd line, but not with the other diagonals. The localization of the odd item in the ‘with-context’ condition was much faster than in the original condition (Configural Superiority Effect, CSE). Presumably, novel properties such as closure may emerge when we combine distinct features (lines) into a unified configuration (a triangle). Recently, we have applied Townsend and Nozawa's Systems Factorial Technology to study the processing of these stimuli, using a slightly different search task. Topologically dissimilar alternatives, although not concurrently presented, led to faster processing. Most visual-search tasks with timed responses involve the presentation of super-threshold stimuli that result in error-free performance. In the current study we test if the advantage of configural figures exists for accuracy as well. To impair performance, we added random Gaussian noise to each display. Our preliminary data show CSE for speeded detection, but a reversed accuracy effect for noisy displays. ‘Ideal observer’ analysis was conducted for the no- and with-context conditions.
Supported by NIH-NIMH grant MH0057717