Decoding Memory from Empty Space: Neural Reactivation During Gaze Reinstatement

Research on episodic memory shows that people often direct their gaze toward locations where goal-relevant information was originally encoded, even in the absence of visual input. This “looking-at-nothing” (LAN) behavior has been demonstrated to strengthen maintenance of visuospatial configurations (Olsen et al., 2014; Wynn et al., 2018), enhance the accuracy and fidelity of episodic recall (Johansson & Johansson, 2014; Johansson et al., 2022), and promote cortical reconstruction (Bone et al., 2018). It has been proposed that reinstating gaze locations serves as a retrieval cue that supports reactivation of the goal-relevant memory representations, yet direct neural evidence for content-specific reactivation during LAN has been lacking. To address this gap, we combined simultaneous EEG and eye-tracking with multivariate pattern analysis (MVPA) to decode category-specific neural activity during LAN in two independent datasets. Participants first encoded multi-element visual events composed of items from distinct categories (e.g., faces, butterflies, cars). In one dataset, they maintained the encoded event while viewing empty placeholders corresponding to the original element locations. In the other, they recalled previously encoded events while freely viewing a completely blank screen. Participants showed the expected LAN behavior, which was related to memory performance. Crucially, decoding of fixation-related EEG during LAN revealed category-specific neural activity reflecting the memory content associated with the fixated locations. Classification accuracies correlated with recall fidelity. These findings provide the first direct evidence of content-specific neural reactivation during LAN and demonstrate how gaze behavior supports episodic memory across maintenance and recollection.