Predicting visual mental imagery: structural and transcriptomic signatures in the human brain

Visual mental imagery (VMI) vividness varies substantially across individuals, yet the structural and molecular bases of this variability remain elusive. A central debate concerns whether VMI ability depends primarily on early visual cortex (V1) or arises from distributed higher-order cortical systems. Here, in a cohort of 229 adults, we apply parcel-wise cortical morphometric modeling with 20-fold cross-validation with random splits to predict individual differences in VMI ability. We show that the morphometric signature spans ventral temporal, parietal, and prefrontal cortices, with no dominant contribution from early visual regions. This distributed profile is selectively associated with individual performance in tasks requiring high versus low imagery demand. Transcriptomic alignment using the Allen Human Brain Atlas reveals gene expression enrichment in neuronal system development, synaptic signaling, and inorganic ion transmembrane transport. Together, these findings support a distributed cortical architecture underlying VMI vividness and provide a molecular context for its neurobiological processes.