Abstract
Working memory (WM) performance can be enhanced by an informative cue presented during storage. This effect – termed a retrocue benefit – can be used to explore how participants prioritize information stored in memory for behavioral output. Previous research has shown that retrospective cues directing participants to prioritize a subset of items stored in memory (i.e., directed-remembering cues) and cues directing participants to ignore a subset of items stored in memory (i.e., directed-forgetting cues) confer similar benefits on WM performance, but it is unclear whether these cues engage similar selection mechanisms in WM. We tested this possibility by examining how directed-remembering and directed-forgetting retrocues influenced location-specific working memory representations reconstructed from human EEG activity. We recorded EEG while participants performed a retrospectively cued spatial working memory task requiring the maintenance of two locations over a short delay. During cued trials, a fully informative retrospective cue presented midway through the memory delay directed participants to prioritize (50% of valid trials) or ignore (remaining 50% of valid trials) one of the locations stored in memory. Consistent with earlier findings, directed-remembering and directed-forgetting cues both improved memory performance relative to no-cue trials, though the magnitude of this benefit was significantly greater for directed-remembering cues. Next, we examined the effects of directed-remembering and directed-forgetting cues on location-specific working memory representations reconstructed from concurrent EEG recordings. During no-cue trials, the fidelity of location-specific WM representations gradually decreased over the course of the delay period. Directed-remembering and directed-forgetting cues partially reversed this information loss, though the degree of recovery following these cues were substantially lower than the degree of recovery seen after a directed-remembering cue. Thus, directed-remembering and directed-forgetting cues appear to engage similar selection mechanisms within WM, but directed-remembering cues confer larger benefits on WM performance.