Abstract
Visual working memory is a capacity-limited system that requires processes to choose what information enters memory and to select what information guides a response. These processes are referred to as working memory gating mechanisms. Input gating regulates what information is allowed into working memory, and output gating regulates what information is used to guide behavior. Our previous work investigated which brain regions are recruited for input compared to output gating, finding that both processes recruit a frontostriatal network. (Chatham et al., 2014). However, this gating task did not require subjects to maintain a high-fidelity visual code. The purpose of the current study was to adapt this gating paradigm to include visual features, allowing us to test whether the same areas are recruited when subjects are required to maintain a visual code in working memory. We adapted the original task into a delayed estimation task, requiring subjects to remember one or two Gabor patches that they responded to by turning an orientation wheel. Additionally, subjects were presented with a higher-order context-cue that indicated which of the Gabor patches was relevant. This context-cue could either come first, putting task demands on input gating, or last, putting task demands on output gating. Subjects performed this modified gating task in the fMRI scanner across multiple sessions. We collected population receptive field mapping data from these subjects in order to identify regions of interest. We found that input gating conditions, where the context-cue was presented ahead of the items, and output gating conditions, where the context-cue was presented after the items, both activated a common set of frontoparietal areas, including anterior premotor cortex and inferior frontal sulcus. These results suggest that working memory gating engages a similar network of frontoparietal regions, regardless of whether the remembered stimuli require a visual or non-visual code.