Abstract
How do observers decide if targets are present in briefly presented visual search displays? They could base their decision on the biggest signal in the display (max rule), on the sum of all signals in the display (sum rule), or on a subset of items in the display (limited capacity). Baldassi and Verghese (2002) argued for a max rule for an orientation search task, while Davis et. al. (in press) measured change in d' by set size functions and found that a max rule worked for some orientation tasks while others showed evidence of limited capacity. We modified Davis' method to provide a clear diagnostic for the presence of at least two decision rules. We measured sensitivity to targets in brief displays (set sizes 1, 2, 4, 8). Three tasks were tested: orientation feature search, color-orientation conjunction search, and a search for a T among Ls. By staircasing dynamic noise, we arranged for d' at set size 1 to be greater for the T vs. L task than for feature search. At set sizes 4 and 8, however, the same stimuli produced larger d' for feature search than for the T vs. L. This crossover is critical because it cannot occur if both tasks use the same decision rule. Feature search data follow a max rule while the T vs. L data suggest limited capacity of two items processed in 100 ms. We conclude that unlimited capacity models can only explain a limited set of tasks.
AFOSR grant #F49620-01-1-0071 to JMW