Sequential reactivation (replay) of neurons in the hippocampus has been largely observed in sharp-wave ripple events (SWRs) during quiescence in both rodents and primates, which has been associated with many mnemonic processes, including memory consolidation, navigation planning and decision making. Previous studies have shown that the decoded replay trajectories have fruitful spatiotemporal dynamics. However, the underlying mechanism of generating these different types of spatiotemporal dynamics remains largely unknown. In this study, we adopt a continuous attractor neural network (CANN) as the neural circuit model to study the rich dynamics of hippocampal place cell ensembles. The interplay between internal noise fluctuation and feedback inhibition explains various spatiotemporal dynamics during the “offline” hippocampal replay, for instance, stationary, Brownian diffusion, super-diffusion and constant propagation, etc. Analytical solutions are validated with simulation results and experimental findings.