TRANSFORMING WORLD TO GOAL COORDINATES IN THE MEDIAL FRONTAL CORTEX

World models encode the relational structure of the environment (“Where am I in the world”), whereas goal models track progress toward internally-generated goals (“Where am I in relation to my goals?”). Interactions between these systems enable flexible mapping of external signals onto internal goals. We propose that neurons in the hippocampal formation (HPC) and medial frontal cortex (mFC) interact to compute a change in coordinates: distance-to-goal signals derived from hippocampal world coordinates calibrate mFC dynamics to compute goal progress. To investigate this, we trained mice on a task that allowed assessment of world models, goal models, and their interaction, enabling tests of three key predictions. First, mFC encoded both distance-to-goal before goal approach and goal progress during subsequent approach. Second, decoding errors for these signals were negatively correlated: when distance-to-goal was underestimated, goal progress was overestimated. Finally, we tested the causal contribution of hippocampal input by combining chronic mFC recordings with optogenetic inhibition of ventral hippocampus (vHPC). vHPC inhibition caused mFC neurons to fire earlier in the goal sequence, indicating overestimation of goal progress. This effect was present during early but not late learning. Together, these findings support a model where vHPC provides a distance-to-goal signal that calibrates the velocity of goal progress dynamics in mFC. This calibration is critical in novel environments but becomes dispensable with experience, potentially reflecting systems consolidation of world structure into the mFC. Our findings suggest a dynamic, computational interaction between hippocampal world models and frontal goal models supporting flexible behaviour.