ReDAS - Reconstructive Dynamics of Visual Working Memory under Attentional Shifts

This collaborative project investigates how attention shapes the representational format of information in visual working memory (VWM).

Recent findings show that when multiple items are maintained in VWM, semantic information (conceptual meaning)can be accessed faster and more accurately than perceptual details, suggesting retrieval dynamics similar to those observed in episodic long-term memory. However, when only a single item is maintained under full attention, this semantic prioritization disappears (Kerrén et al., 2022).

The project tests whether the balance between semantic and perceptual information can be dynamically modulated by attention and encoding conditions. Specifically, we examine whether retrocues can preserve high-fidelity perceptual representations and whether simultaneous vs. sequential encoding influences the format in which information is later retrieved.

To address these questions, the project combines behavioral measures, EEG decoding of neural representations, and drift diffusion modeling (DDM) to track how working memory representations evolve under different attentional states.