Variable and intricate dynamics are core to cortical communication in the human brain. Electrical stimulations in intracranial EEG (iEEG) can help probe effective connectivity over time and brain states through the measurement of cortico-cortical evoked potentials (CCEPs).In 15 epilepsy patients undergoing iEEG for diagnostic reasons, we probed effective connectivity with single-pulse electrical stimulations (1ms, 3mA). For any given pair of iEEG contacts at shorter (<20mm) or longer (≥20mm) range, effective connectivity was expressed as a probability of observing a single-trial CCEP in one or the other direction and tested for significance over time against surrogate recordings. Significant modulations of CCEP magnitude by brain states (asleep vs. awake) were expressed as rank-biserial correlation r and tested with a two-way ANOVA accounting for brain regions.Across participants, 41’152 effective connections showed a probability and directionality that depended on the identity and distance between connected brain regions (p<0.01). Among these, 9’784 connections (24%) showed significant shifts in connectivity during NREM in a region-specific manner (p<0.01). Notably, connections from hippocampus and amygdala to orbitofrontal showed a decrease (median [IQR] r: -0.57 [-0.69, -0.44]) and an increase (0.41 [0.18, 0.57]) in connectivity, respectively.Using longitudinal cortical stimulations directly in humans, we unraveled shared dynamics in regions-specific effective connectivity. We found that sleep redistributes effective connectivity by imposing a bidirectional modulation of the strength of certain connections. This work furthers our understanding of non-linear dynamics in the human cortex.