High and low: the resolution of representations in visual working memory

Abstract

Visual working memory (VWM) has long been considered to be limited in capacity, but the way in which it is limited remains unclear. Despite differences in predictions of the number of objects that can be stored, both the slot and resource models agree that resolution in VWM declines as the number of objects increases. Here we investigate the relationship between the resolution of items and the number of items in VWM by separating different types of resolution measures. In this study, we adapted the paradigm of Awh, Barton, and Vogel (2007) to provide separate measures for low-resolution (i.e., categorical judgment) and high-resolution (i.e., within-category fine discrimination) representations of an item in memory. Participants were asked to remember a mixture of objects from two categories, varying in set size and display time. After a 1s retention interval, the location of one item was highlighted and participants were first tested on the category of this item-to-report. The nature of the second response was contingent on the first: depending on which category response was made, participants either adjusted a color wheel or selected a cube from an array (Experiment 1), or reported the color or orientation of gabor patches (Experiment 2). In both experiments, precision of high-resolution representations declined monotonically until the set size reached four items, fitting to the predictions from the standard mixture model of Zhang and Luck (2008). In contrast, we observed that the accuracy for low-resolution representations remained very high and did not decrease with set size, which is not consistent with either model. We propose that the inverse relationship between the number and resolution of representations in VWM is perhaps not the only possible relationship: different types of resolution representation exist in VWM, and people can maintain both high- and low-resolution representations of an object.