Abstract
The respective roles of the occipital, parietal, and frontal cortices in visual working memory (VWM) maintenance have long been under debate. Considering the previous mixed findings on whether the multi-voxel response patterns in the parietal and frontal regions convey mnemonic information (e.g., Ester et al., 2015; Riggall & Postle, 2012), one possibility is that attentional salience based on temporal order can modulate the mnemonic signals in high-level frontoparietal regions. To test this hypothesis, we examined whether temporal recency could lead to changes in the mnemonic representations. On each trial, participants viewed two gratings with different orientations in succession, and were cued to remember one of them. After a long delay of 10.4 s, participants rotated a test grating until its orientation matched the remembered orientation. Using fMRI and a forward encoding technique (Brouwer & Heeger, 2009; 2011), we reconstructed population-level, orientation tuning responses in the occipital (V1-V3), parietal (IPS), and frontal (FEF) areas during VWM maintenance. Unlike the early visual cortex where robust tuning responses were observed regardless of whether the remembered target was the first or second in the sequence, the parietal and frontal cortices showed better tuning responses when participants remembered the second grating, indicating the effect of recency. To exclude the possibilities that this effect was caused by the residual perceptual or motor preparative signals elicited by the second grating, we conducted a control experiment where participants performed a change detection task on serially presented color patches that were masked by a colored pattern. The results again demonstrated superior representation for the color of the second patch in IPS, but at later time points during retention. These results suggest that attentional state, such as attentional salience brought by recency, has a strong influence on the mnemonic representations in the parietal cortex.
Meeting abstract presented at VSS 2017