Decoding the details of visual working memory representations from ERP and alpha oscillation activities

Working memory (WM) can temporarily store and process perceived information to guide upcoming actions. However, even if the information is encoded in WM with sufficient attention, "attribute amnesia" occurs, where part of the information is not reselected by WM, resulting in the information being forgotten. In this study, we investigated whether different types of feature information of objects have different priorities to be reselected by WM and the changes in the underlying EEG signals, focusing on whether EEG signals contain detailed feature information. We collected EEG signals from participants while they performed a change detection task and analyzed the data using a combination of univariate analysis and multivariate pattern analysis (MVPA). The behavioral results show that the color on the same object is reported more quickly and accurately than shape. ERP results indicate that while both color and shape can evoke the P300 component at the posterior scalp, the P300 amplitude is larger for color. Additionally, there is no significant difference in the intensity of posterior alpha oscillatory activity evoked by color and shape. Notably, MVPA results reveal that specific color and shape feature values can be decoded using EEG data. These results show that different features have different priorities in WM reselection and that the details of these features are stored in WM.