Abstract
Top-down goals, which are vital to success in everyday tasks, are thought to exert control over both visual cortex activation (Chelazzi, Miller, Duncan & Desimone, 2001) and cognitive processes such as memory (Sperling, 1960). However, it is unclear to what extent individuals can control the contents of visual short-term memory (VSTM). The current study presented two sequential 4-item displays per trial in a modified change detection task (Luck & Vogel, 1997), to examine participants’ performance when prepared in advance to encode specified stimuli and when required to exert control mid task to change or maintain VSTM contents. Perfect top-down control over VSTM contents should allow similar performance whether prepared in advance or required to exert control mid trial. Results suggest participants perform similarly whether told early in the trial that they will be tested on the second display, or whether told mid trial to forget current (first display) VSTM contents and encode upcoming (second display) contents, t(15) = -1.08, p = .297. In contrast, participants perform significantly better when instructed early in the trial to remember only the first display, than when informed mid trial to retain the previous display and ignore the upcoming, t(15) = 5.08, p <.001. Control experiments using masked stimuli to minimize individual strategy, and increased post-cue processing time, revealed the same pattern of results. These findings build on previous evidence of top-down control over the encoding of items into VSTM (e.g. Sperling, 1960). We suggest that although participants can easily update VSTM contents, by removing old items and encoding new ones, they experience difficulty inhibiting encoding of irrelevant stimuli into VSTM, whilst maintaining relevant stimuli. Knowledge of the effectiveness of top-down manipulations throughout complex stimulus presentations is an important step towards developing information presentation strategies for optimal working memory storage, and informing future models of VSTM.
Meeting abstract presented at VSS 2013