Conference on Cognitive Computational Neuroscience

The brain uses cognitive maps–mental representations of relationships between items in an environment or task–to track spatial and non-spatial relationships. By representing multidimensional information in a map-like structure, the brain can leverage these relationships to make quick decisions, infer new relationships, and extrapolate to novel environments. Previous findings have suggested the involvement of the entorhinal cortex (ERC), hippocampus (HPC), and medial prefrontal cortex (mPFC) in maintaining cognitive maps. However, while cognitive maps conceptually afford flexibility it remains unclear if and how cognitive maps are representationally flexible. In an fMRI experiment, participants used a previously learned wine space to solve a new task that required cognitive map flexibility. We found univariate effects of context-dependent distance information in the mPFC, HPC, and ERC, suggesting these regions flexibly use cognitive map information. Moreover, whole brain representational similarity analysis (RSA) of context-dependent distance coding revealed activity patterns in the orbitofrontal cortex (OFC), posterior cingulate cortex (PCC), temporoparietal junction (TPJ), and dorsolateral prefrontal cortex (dlPFC). Our findings suggest that cognitive map representations are more flexible than previously thought. The brain employs multiple regions to track different cognitive map features and dynamically recruits context-relevant information to compute decision variables that meet task demands.