Tracking the mind’s eyes: Spatiotemporal Dynamics of Domain-General and Domain-Specific Visual Mental Imagery

Visual mental imagery (VMI) enables the generation of perceptual-like experiences in the absence of sensory input. Previous work has identified a network supporting VMI that includes frontoparietal regions, domain-preferring areas in ventral temporal cortex (VTC), and a domain-general region in left fusiform gyrus, the Fusiform Imagery Node (FIN). However, it remains unclear (i) how activity unfolds over time across these regions, (ii) what representational content they encode, and (iii) how imagery dynamics relate to visual perception.To address these preregistered questions, we combined fMRI and magnetoencephalography (MEG) using model-based representational similarity analysis. Twenty-six participants performed imagery and perception tasks across three domains (Colors, Faces, Maps). First, imagery-related representations emerged after ~0.5 s and peaked around 1 s, substantially later than perception (~0.4 s), with no clear sequential ordering between domain-general and domain-preferring regions. Second, representational analyses revealed both domain-general and domain-specific information, with commonality analysis showing that the domain-general ROI (FIN) was predominantly driven by domain-general structure, while domain-specific contributions were temporally dissociable. Third, imagery and perception showed distinct temporal dynamics: perception exhibited an earlier, stronger, and more homogeneous pattern, whereas imagery was more sustained and temporally extended.Together, these findings indicate that VMI relies on distributed and interactive processes rather than a simple replay of perception. Domain-specific representations emerge relatively early, while domain-general processes, including the FIN, may contribute to sustaining and integrating imagery content over time.