Abstract
Theoretical models emphasize the importance of top-down signals in the mechanisms that support working memory (WM; Curtis & D’Esposito, 2003; Sreenivasan et al., 2014). Yet, the nature and mechanisms by which top-down signals support WM functions remain unknown. Previously, we demonstrated that WM representations of an item’s spatial position (Rahmati, Saber, & Curtis, 2018) and a non-spatial feature (i.e., orientation; Rahmati, et al., 2018) could be reconstructed from the pattern of activity in visual cortex, even when those representations are remapped to a part of the visual field that was not retinally stimulated. Although these results implicate top-down signals in remapping WM, we could not distinguish among mechanisms by which the top-down signals operated. Here, we test how top-down signals facilitate remapping of WM content to distinguish between models that posit that top-down signals carry precise feature information about WM representations, or they simply enhance the gain of neurons with receptive fields matching prioritized attentional space. Using fMRI and an inverted encoding model (Sprague et al., 2013), we reconstructed the stimulus orientation of a gabor maintained in WM during a retention interval from visual field maps defined in visual and parietal cortex. In order to reveal what information was carried by the top-down signals, participants compared the rotated 90-degree version of the orientation of a sample gabor presented in one quadrant, e.g., upper right, to thethe gabor tested in the quadrant diagonal from the sample, e.g., lower left. In parietal cortex, the delay period activity across voxels with receptive fields that mapped the diagonal probe quadrant contained information about the rotated orientation. In visual cortex, voxels mapping the probe quadrant contained information about both the original and rotated orientation. Therefore, WM representations in parietal and visual cortex are supported by top-down signals carrying spatial and non-spatial features.
Acknowledgement: NYU Global Seed grant for collaborative research (to CC & KS), NIH F32-EY028438 (to TCS) and NVidia Hardware Grant (to TCS)