Explicit memory for rejected distractors during visual search

Abstract

Although memory for the identities of examined items is not used to guide visual search, identity memory may be acquired during visual search. In all experiments reported here, search was occasionally terminated and a memory test was presented for the identity of a previously examined item. Participants demonstrated memory for the locations of the examined items by avoiding revisits to these items and memory performance for the items’ identities was above chance but lower than expected based on performance in intentional memory tests. Memory performance improved when the foil was not from the search set, suggesting that explicit identity memory is not bound to memory for location. Providing context information during test improved memory for the most recently examined item. Memory for the identities of previously examined items was best when the most recently examined item was tested, contextual information was provided, and location memory was not required.

Acknowledgments

This work was supported by grants from the National Institute of Health to MSP (R01 MH64505) and the Office of Naval Research to AFK and WRB. Parts of this research were presented at the 2005 Vision Sciences Society in Sarasota Florida.

Notes

¹Particularly high miss rates in this paradigm may be attributed to the dynamic nature of the search display or the low probability of a target occurring. Because the item that was just examined would often be removed from the screen after the eyes moved to a new item, participants were often unable to reexamine an item just after it was examined. Therefore, if participants fixated the target without responding, they rarely had a second chance to view the target. This is unlike traditional search paradigms in which participants might fixate the target, move their eyes without recognizing it, and move their eyes back to the target location later during the trial. As noted by Peterson et al. (2001), refixations during a static search display were often directed toward the target after the eyes had fixated too briefly the first time. Another potential explanation might be the low incidence of target occurrence in this paradigm (only 25% of the nonprobe trials). This may have caused participants to adopt a stringent criterion for detecting the target.

²Williams et al. (2005) used a different type of search task than was used in the experiments reported here. Instead of asking participants to search for the presence of a target, they were asked to report how many targets were present (0, 1, 2, or 3). The search target was different than that given in the experiments reported here; participants were told to search for a specific object of a specific category (e.g., “yellow phone”). Furthermore, distractors could be from the same specific category as the target (e.g., green phones). Any of these differences could have lead to better performance on the LTM test than would have been found using a search task like what was used in the experiments reported here.

³The number of objects’ worth of information stored in memory during visual search was calculated similarly to how it was calculated in Irwin and Zelinsky (2002). We corrected for guessing (Busey & Loftus, 1994) by applying the formula p = (x – g)/(1 – g), where x is the percentage correct averaged across all lags, g is the probability of guessing (g = 1/n, were n equals the number of objects examined prior to the memory test, six in Experiment 1 and five in Experiments 2a, b, and c), and p is the percentage correct after correcting for guessing. P was then multiplied by the total number of objects examined (n) for an estimate of the number of objects remembered during visual search (Sperling, 1960).