Beyond the magic number four: Remapping high-capacity, pre-attentive, fragile working memory.

Abstract

Visual short term memory allows us to access visual information after termination of its retinal input. Generally, a distinction is made between a robust, capacity-limited form (working memory, WM) and high-capacity, pre-attentive, maskable forms (sensory memory, e.g. fragile memory, FM). Eye movements pose a challenge for spatial memory because retinotopically coded coordinates need to be updated with every saccade (spatial remapping). This computationally intense process has been considered strictly limited to WM. To answer the question whether fragile, pre-attentive memory items are also remapped we directly compared trans-saccadic WM (tWM) and FM (tWM) capacity. Participants memorized oriented bars and reported on a change in probe displays after making a saccade. A predictive retro-cue indicated future targets and protected FM from being masked by the memory probe, which is known to increase capacity estimates due to the inclusion of FM items (retro-cue benefit). If only stable, attended memory items can be remapped, then capacity after a saccade should be that of tWM, even if capacity was high before the saccade (FM). We found that tFM capacity was considerably higher than tWM capacity. This demonstrates that in addition to attended items in WM, non-attended items in FM were also remapped. In a second experiment, we retroactively masked items to disrupt the retro-cue benefit. The results show that masks at the retinotopic location as well as at the spatiotopic location reduced the retro-cue benefit. This confirms that FM items have been remapped to task-relevant, world-centered coordinates. We provide conclusive evidence for the existence of high-capacity, trans-saccadic, fragile visual memory and challenge the strongly held belief that trans-saccadic memory is identical to robust WM. This has important implications for our understanding of spatial remapping, which is thought to be intimately linked to spatial attention. Our results demonstrate that un-attended items can also be remapped. Meeting abstract presented at VSS 2017