Abstract
The role of memory in repeated search tasks is contentious. Wolfe et al. (2000) contend that participants cannot learn the locations of items in repeated scenes because search time continues to increase with display size despite hundreds of repetitions. In contrast, Chun and Jiang (1998, 1999) showed that search is more efficient for repeated versus new scenes and that this learning occurs for either spatial layout or identity, but not both. We show that participants can learn that a particular item is more likely to occur in a particular location and that this learning of “bound entities” occurs rapidly. In Experiment 1 participants searched through an 8-item display for one of two targets that were present on 50% the trials. When present, each target occurred in one, high frequency (HF) location with 75% probability, and in the other target's HF location with 25% probability (low frequency, LF, location). Participants verified target presence 28-ms faster for targets in their HF versus LF locations, p[[lt]].001. On target-absent trials, a foil (different for each target) was present in the HF or the LF location. Participants accurately reported target absence 17-ms faster for foils in HF versus LF locations, p[[lt]].005. These differences were evident in the first block (62 trials); hence, bound entities can be learned quickly. In Experiment 2 we tested 4 targets; for each target, one location was more informative than another. Participants verified target presence (21 ms) faster for more informative locations, p[[lt]].01. In addition, RTs were longer in Experiment 2 than Experiment 1 indicating that, even when bound entities are learned, search slows with increased target number. Thus, increases in search time with target number (identical to display size in Wolfe's experiments) cannot be taken as evidence that learning has not occurred.
UA SBSRI grant and APA dissertation award to ES and NSF BCS 0418179 to MAP