EEG decoding reveals sensory storage for both active and passive states in visual working memory
Poster Presentation: Tuesday, May 19, 2026, 8:30 am – 12:30 pm, Banyan Breezeway
Session: Visual Working Memory: Models, neural
Schedule of Events | Search Abstracts | Symposia | Talk Sessions | Poster Sessions
Xinran Chen1, Mowei Shen1, Hui Chen1, Yingtao Fu1; 1Zhejiang University, Hangzhou, China
Previous research has shown that contents of visual working memory (VWM) can be decoded from neural activity in early visual cortex, supporting the sensory recruitment hypothesis. However, it remains debated about whether different representational states of VWM—namely active and passive states—rely on sensory storage. Here, we used electroencephalography (EEG) to demonstrate that memory load of both states could be decoded from multivariate pattern of posterior neural activity. In a VWM task, participants were required to memorize either two or four colors, and the representational states were manipulated by a retrospective cue: cued items were maintained in the active state, whereas uncued items were in passive state. The memory load of both states was independently manipulated as either low or high. We found that activity from posterior channels during the delay period reliably classify high versus low load for both active and passive state. Besides, decoding for passive state showed less robust result compared to the active state. Together, these findings suggest that both active and passive state of VWM recruit sensory storage, although passive representations may be maintained with weaker memory strength.