A learning-evoked slow-oscillatory architecture paces population activity for offline reactivation across the human medial temporal lobe
Memory processing requires the coordinated engagement of neuronal populations across distributed brain networks and across time. How such coordination is organized in the human medial temporal lobe (MTL) remains unclear. Here, we show that MTL population activity is dynamically structured by a transient slow-oscillatory architecture that emerges during online learning to promote offline consolidation and later recall. Using intracranial recordings that combine single-neuron spiking activity and local field potentials in human participants, we find that mnemonic engagement evokes on-demand slow-oscillatory bursts in the hippocampus. These hippocampal bursts synchronize gamma-band patterns across MTL regions, defining discrete coordination events that pace cross-regional coactivity motifs during learning. These learning-time population motifs are then selectively reactivated during hippocampal ripples in post-learning rest, and the strength of their reactivation predicts subsequent recall accuracy. Together, these findings identify a multi-scale coordination mechanism that links distributed population activity across learning, consolidation, and recall in humans.