Abstract
Working memory (WM) is one’s ability to maintain and manipulate a limited amount of information temporarily. Previous studies have shown that subjects can introspect the quality of a single cued WM representation (Rademaker et al., 2012; Li et al., 2021), or report which of several WM representations was maintained most precisely (Fougnie et al., 2012; Suchow et al., 2017). However, it remains unknown whether participants can similarly introspect the relative quality of their spatial WM representations. Here, we employed a memory-guided saccade task to test whether participants can correctly intuit the best of several WM representations on a trial-by-trial basis. In one condition, human subjects remembered a single position over a delay period and reported its location with a saccade. The other two conditions both contained two targets, each presented in a different color. In one of those two conditions, the subjects were cued to recall the location of one of the targets, randomly chosen on each trial. For the other condition, the subjects were asked to recall the location of the target that they believed they remembered the best. A repeated-measures one-way ANOVA revealed a main effect of the condition, (p < 0.05). We observed an effect of set size such that errors were greater for two targets compared to one target. Most importantly, when subjects could choose the best WM representation to report, errors were lower compared to trials when subjects were cued which WM representation to report. These results indicate that participants can successfully monitor trial-by-trial fluctuations in the relative quality of spatial WM representations to guide behavior.