The roles of human hippocampus and orbitofrontal cortex in cognitive map-guided social transitive inference

Cognitive maps support flexible inference by organizing relational structures across experiences. This enables humans and animals to integrate knowledge and infer indirect relationships, a process known as transitive inference (TI). While the hippocampus (HPC) and orbitofrontal cortex (OFC) are both important for cognitive mapping, their specific roles in TI remain unclear. Here, we obtained behavioral and fMRI data from 25 healthy adults to study how they learned face rankings along two social dimensions (competence and popularity) and how they inferred novel relationships between untrained face pairs. Compared with control condition, we found that during TI, the bilateral HPC, left middle frontal gyrus (MFG), insula, and superior temporal gyrus (STG) showed significantly greater activation and stronger functional connectivity (FC) within the TI network. The HPC can significantly distinguish the dimension-related from dimension-unrelated hub faces. In addition, we found that the HPC had stronger FC with the regions of the default mode network, dorsal attention network, and mid-cingulate cortex when processing dimension-related hubs compared to dimension-unrelated hubs. These findings revealed the key role of the HPC in encoding abstract relational structure and how it collaborates with large-scale brain networks to support cognitive map-guided transitive inference.