Attentional Contributions to the Memorability Benefit in Visual Working Memory

Some visual stimuli are consistently better remembered than others across individuals, due to variations in memorability (the stimulus-intrinsic property that determines ease of encoding into visual long-term memory (VLTM)). We found that this memorability benefit also exists in visual working memory (VWM). Within VWM, memorable stimuli enjoy a dual benefit: 1) they are stored more efficiently in VWM and 2) are also more competitive at attracting VWM resources than forgettable stimuli. In this study, we examined the hypothesis that this competitive advantage of memorable stimuli reflects preferential allocation of spatial attention toward memorable stimuli. To test this, we had participants perform a face VWM task where they had to remember an array composed of both memorable and forgettable faces. Critically, on 23% of trials, letters replaced the face stimuli at differing stimulus onset asynchronies. Participants then reported the letters they remembered seeing. This enabled us to examine where spatial attention was allocated at different times during viewing. We found that participants did not report more letters superimposed on memorable faces than on forgettable faces until 450ms after the onset of face stimuli, suggesting that spatial attention is not immediately biased toward memorable stimuli. Next, we examined when the competitive benefit emerged in relation to changes in attentional allocation. Does the competitive benefit lead to differences in attentional allocation or is attention responsible for the competitive benefit? We had participants perform the same VWM task and we manipulated the stimulus duration. We found that the competitive benefit was not present until after there were differences in attentional allocation. We speculate that this may be due to memory guided attention where memorable stimuli are efficiently encoded into VWM, which then attracts attentional resources ultimately leading to the competitive benefit.