Abstract
Previous studies have shown that information held in visual working memory is represented in a broad network of brain regions, including the occipital, parietal, and frontal cortices (Ester et al., 2015). However, less is known about whether and how the mnemonic information in parietal and frontal regions can be modulated by task demand. In our previous work (Yu & Shim, VSS 2015), we demonstrated task-modulated working memory representations in the occipital cortex, and potentially in the parietal and frontal cortices. In the current study, we further examined how the working memory representation of each feature in multi-feature objects was modulated by task and spatial location. On each trial, two colored gratings were presented simultaneously with one in each hemifield. Participants were cued to remember either color or orientation of one of the gratings for 10 s. Using fMRI and a forward encoding model (Brouwer & Heeger, 2009; 2011), we reconstructed population-level, feature-selective tuning responses in occipital, parietal and frontal cortices during memory delay. We found that not only orientation but also color information can be maintained over the delay in early visual cortex as well as higher-order parietal and frontal cortices (e.g., IPS and FEF) when it was cued to be remembered. Furthermore, regardless of whether color or orientation was cued, the remembered, task-relevant feature was represented not only in the contralateral but in the ipsilateral hemisphere as well, suggesting a global spread of the representation of the remembered, task-relevant feature. Conversely, neither the task-irrelevant feature of the remembered object, nor any feature information of the not-remembered object was represented. Taken together, these results indicate that the information of remembered features in occipital, parietal, and frontal cortices can be flexibly modulated by task demand, suggesting a highly selective mechanism of visual working memory that encodes and maintains task-relevant features only.
Meeting abstract presented at VSS 2016