Abstract
In any moment, the quality and the quantity of information represented in visual working memory (VWM) is severely limited. Additionally, individuals tend to overestimate the accuracy of representations held in VWM, leading to confident-but-inaccurate responses (e.g., Adam & Vogel, 2017). Since it is difficult to improve the capacity of VWM, we instead aimed to improve the accuracy of metacognitive assessments of VWM representations. In the first experiment, participants remembered an array of simple objects (e.g., coloured squares or oriented lines) over a one second retention interval. Using a whole-report method (e.g., Adam & Vogel, 2016), participants then reported the colour or orientation of each object along with the confidence of their report. Importantly, performance feedback was given for each response to promote confident-and-accurate responses while discouraging confident-but-inaccurate responses. More precisely, confident-and-accurate responses led to larger reward points than not-so-confident-but-accurate responses, while confident-but-inaccurate responses resulted in larger penalty points than not-so-confident-and-inaccurate responses. Results demonstrated that in both younger and older adults, our feedback improved metacognitive accuracy for VWM representations, leading to a significant reduction of confident-but-inaccurate responses. Furthermore, this improvement was observed even after the performance feedback was removed. Interestingly, this training effect was more immediate and larger in magnitude when compared to a control experiment without feedback (Experiment 2). In Experiment 3, the training effect was found to generalize to an untrained stimulus. That is, when participants were given feedback training on one type of stimulus (e.g., colour) and then tested on the untrained stimulus (e.g., oriented lines), their metacognitive accuracy for the untrained stimulus was higher than before training. Therefore, we demonstrate that not only does feedback training improve metacognitive accuracy, but this training effect also generalizes to an untrained stimulus.