The Neural Consequences of Attentional Prioritization of Internal Representations in Visual Working Memory

Although humans can hold multiple items in mind simultaneously, the contents of working memory (WM) can be selectively prioritized to effectively guide behavior in response to rapidly changing exigencies in the environment. Neural evidence for this is seen in studies of dual serial retrocuing of two items held concurrently in visual WM, in which evidence in occipital cortex for the active neural representation of the cued item increases, and evidence for the uncued item decreases, often to levels indistinguishable from empirical baseline. Although this pattern is reminiscent of the effects of selective attention on visual perception, the extent to which more subtle principles of visual attention may also apply to visual working memory remains uncertain. In the present study we explored whether the well-characterized “same-object” benefit in visual target detection, attributed to object-based attention (e.g., Duncan, 1984; Egly, Driver, & Rafal, 1994), may also be observed for information held in visual WM. fMRI data were collected while human subjects (male and female) performed a multi-step serial retrocuing task in which they first viewed two two-dimensional sample stimuli comprised of colored moving dots. After stimulus offset, an initialrelevance cuethen indicated whether both dimensions of only the first or only the second object, or only the color or only the direction-of-motion of both objects, would be relevant for the remainder of the trial, which then proceeded with the standard dual serial retrocuing procedure. Thus, on “object-relevant” trials, the ensuingpriority cuesprompted the selection of one from among two features (“color” or “direction”) bound to the same object, whereas on “feature-relevant” trials thepriority cuesprompted the selection of one from among two features each belonging to a different object. Results of analyses with multivariate inverted encoding models (IEM) revealed a same-object benefit on object-relevant trials: Whereas, on feature-relevant trials, the firstpriority cuetriggered a strengthening of the neural representation of the cued feature and a concomitant weakening-to-baseline of the uncued feature; on object-relevant trials the cued item remained active but did not increase in strength, and the uncued item weakened, but remained significantly elevated throughout the delay period. Of additional interest, on both types of trials the secondpriority cueprompted an active recoding of the uncued item into a different neural representation, perhaps to minimize its ability to interfere with recall of the cued item. Finally, although stimulus-specific representation in parietal and frontal cortex was weak and uneven, these regions closely tracked the higher-order information of which stimulus category was relevant for behavior at all points during the trial, indicating an important role in controlling the prioritization of information in visual working memory.