A likelihood-based method for identifying replay from spike sequences

Hippocampal replay-sequential reactivation of place cells-has been implicated in memory recall, consolidation and planning, but traditional analyses rely on spatial tuning and exclude non-place cells, limiting their scope. Model-based analyses without spatial templates often depend on predefined parameters and constrained pattern definitions, making them parameter-sensitive and potentially biased. To address these limitations, we developed an analysis that estimates the likelihood of spike sequences based on pairwise firing order probabilities observed during active behavior regardless of their behavioral correlates. We validated the method using simulations, rat single-unit recordings, and mouse calcium imaging, showing strong agreement with conventional replay metrics and broad applicability. When applied to head-fixed mice passively viewing naturalistic movie stimuli, the method detected significant replay in both hippocampus and visual cortex, demonstrating its utility in uncovering structured replay across brain regions and sensory domains.