AbstractCognitive flexibility, the adaptation of mental processing to changes in task demands, is thought to depend on biological neural networks’ ability to rapidly modulate the dynamics governing how they process information. While extensive work has elucidated how network dynamics can be reshaped by slowly occurring structural changes, e.g. the gradual modification of recurrent synaptic patterns, much less is known about how dynamics might be reconfigured over faster timescales of seconds. One compelling example of rapid and selective modulation of network dynamics potentially involved in cognitive flexibility is observed in rodent hippocampus, where short bouts of exploratory behavior cause new activity sequences to preferentially “replay” during subsequent awake rest periods without continued sensory input. Fast mechanisms for selectively biasing sequential activity through networks, however, remain unknown. Using a spiking neural network model, we asked whether a simplified version of sequence replay could arise from three biophysically plausible components: recurrent, spatially organized connectivity; homogeneous, stochastic “gating” inputs; and rapid, activity-dependent scaling of gating input strengths, based on a phenomenon known as long-term potentiation of intrinsic excitability (LTP-IE). Indeed, these enabled both forward and reverse replay of flexible sequences reflecting recent behavior, despite unchanged recurrent weights. Specifically, activation-triggered LTP-IE “tags” neurons in the recurrent network by increasing their spiking probability when gating input is applied, and the sequential ordering of spikes is reconstructed by the existing recurrent connectivity. In a proof-of-concept demonstration, we also show how LTP-IE-based sequences can implement temporary stimulus-response mappings in a straightforward manner. These results elucidate a simple yet previously unexplored combination of biological mechanisms that converge in hippocampus and suffice for fast and flexible reconfiguration of sequential network dynamics, suggesting their potential role in cognitive flexibility over rapid timescales.