Representation, Retention, and Recognition of Information in Visual
Working Memory

David E. Fencsik, Travis L. Seymour, Shane T. Mueller, David
E. Kieras, and David E. Meyer

Poster presented at the 43rd Annual Meeting of the Psychonomic
Society, Kansas City, MO November 21-24, 2002

Previous experiments on visual working memory (WM) have inspired the
hypothesis that information about briefly displayed objects is stored
in separate object files whose contents may endure for several seconds
(e.g., Vogel, Woodman, & Luck, 2001). Supposedly, visual WM can
contain about four such files. Yet data from these experiments seem
problematic to interpret and were perhaps contaminated by some
artifacts. To help resolve such problems, we have conducted new
studies that required same-different judgments and cued recall under
conditions where successive object displays changed in various
ways. Our results suggest that at least two distinct types of
extra-iconic visual storage contributed to performance there. One type
involves unintegrated location-independent visual features. The other
involves integrated location-dependent features, consistent with the
object-file hypothesis. However, it appears from our results that
visual WM has a capacity of only about three rather than four object
files.

Sponsored by the Office of Naval Research