| dc.description.abstract | Visual items held in visual working memory (VWM) can be currently or prospectively relevant. Currently relevant items (attended) are memory items that are currently necessary to perform a certain task. Prospectively relevant (unattended) items are not currently necessary to perform a task, but could become relevant later on. Some studies suggest that these attended and unattended memory items can be stored in different brain regions, or even in different formats. However, research is inconclusive about where and how these differences exist across the cortex. In this study we aimed to get insight into where and how the brain distributes neural resources to store attended and unattended visual information. We collected 7T fMRI neural data from three participants as they performed a visual working memory task. This task was designed so that in our relevant measuring delay, a partcipant stored both an attended and an unattended visual item (a grated orientation) in visual working memory. Then, using Support Vector Machine (SVM) classification, we decoded attended and unattended visual information from the cortex. We were able to decode attended visual information from parietal, occipital and frontal regions. We were also able to decode unattended visual information in one participant from parietal and occipital regions. Furthermore, we found that attended visual information is represented more strongly in the occipital lobe than unattended visual information. We conclude that while attended and unattended visual information can be stored in similar brain regions, attended visual information is represented more strongly in the occipital lobe, where the visual cortex is located. We hypothesise that the brain could distribute more neural resources from the visual cortex to store attended visual items, as this brain region is capable of storing this information in a high-resolution format. This would thus facilitate tasks such as search, as a more high-resolution memory representation can be used to match to the visual input from our environment. Finally, we highlight benefit of using 7T fMRI and within-subject statistics with many trials per participant in decoding visual information from visual working memory. | |
