Hippocampal mismatch signals are based on episodic memories and not schematic knowledge.

Prediction errors drive learning by signaling mismatches between expectations and reality, but the neural systems supporting these computations remain debated. The hippocampus is implicated in mismatch detection, yet it is not known whether it signals mismatches with episodic memories or generalized knowledge. Across three functional Magnetic Resonance Imaging (fMRI) experiments, we show that the hippocampus selectively responds to mismatches with episodic memories of specific events. In contrast, schematic mismatches engage Semantic Control and Multiple Demand Networks, as well as subcortical regions linked to prediction error signaling. Episodic mismatches also recruit the Default Mode Network. These findings challenge accounts that propose the hippocampus is a domain-general mismatch detector. Instead, the findings support a more specialized role for the hippocampus in learning that is underpinned by its well-established importance in processing episodic memories.