Individual working memory capacity affects quantitatively but not qualitatively the deployment of attentional resources

Working memory capacity (WMc) has been traditionally thoughts as involving a fixed number of objects that can be stored at a time (“slot” models). More recent models argued instead that WMc may be based on fixed resources, which can be divided according to the complexity of to-be-stored objects (“resources” models). Here we report evidence showing that both accounts can be valid. We presented participants with everyday pictures for a short period (4 secs). After a retention period of 8 s, we asked participants to verbally report as many objects/details as possible of the previous scenes. We then computed how many times the objects located at either the peak of maximal or minimal saliency in the scene (as indexed by a saliency-map) were recollected by participants. We also measured WMc (K score) using an independent visuo-spatial WM test, namely, the change location task. The results showed that high WMc (i.e., high K score) predicted the amount of objects successfully recollected, in line with a fixed-slot account of WMc. However, the current level of individual WMc failed to predict the probability to recollect maximal- vs. minimal-saliency objects in the scene. This indicates that the deployment of attention resources, within the individual number of fixed-slots, follows bottom-up capture by low-level sensory features (i.e., visual saliency). These findings are in line with the idea that memory representation in complex, real-life situations, can be driven by bottom-up, saliency-related factors that control the allocation of WM “resources” within “fixed” constraints operating at the individual level.