Medial Temporal Default Mode Network Selectively Encodes Autobiographical Visual Imagery
Andrew J. Anderson, Adam Turnbull, Feng V. Lin preprint
The medial temporal subsystem of the default mode network (MT-DMN) encodes visual details of imagined autobiographical scenes. Fifty participants imagined personal experiences from generic prompts while undergoing fMRI. Using Stable Diffusion to create personalized images from participants' own descriptions, researchers found that MT-DMN activity patterns matched the representational structure of those images, even after accounting for semantic content. This effect was not seen in other brain networks or during reading without imagination, identifying the MT-DMN as a key area for reconstructing visual experiences from memory.