In their Correspondence article (Active and effective replay: systems consolidation reconsidered again. Nat. Rev. Neurosci. https://doi.org/10.1038/s41583-019-0191-8 (2019))1, Antony and Schapiro agree that the concept of contextual binding (CB) described in our Opinion article (A contextual binding theory of episodic memory: systems consolidation reconsidered. Nat. Rev. Neurosci. 20, 364–375 (2019))2 is crucial for explaining episodic memory, and that many aspects of standard systems consolidation theory are probably incorrect. They also point to some recent sleep and replay studies that might be taken as evidence for a form of systems consolidation that maintains that the hippocampus rapidly trains the cortex during offline periods of sleep or rest. Although we agree that these findings are compatible with a modified form of systems consolidation, we contend that the studies summarized by Antony and Schapiro do not necessitate such an account.

For example, demonstrations that replay correlates with subsequent memory, or that sharp wave–ripple (SWR) activity in the hippocampus correlates (in some instances) with activity in the cortex3, are consistent with other theoretical accounts such as CB. That is, because episodic memory involves hippocampal binding of item information and context information that is in the cortex, residual or potentiated encoding activity across the hippocampus and cortex should be observed after the nominal encoding event is over, even if the hippocampus is not actively training the cortex at the time. Furthermore, although experimental manipulations of SWRs in the hippocampus and/or neocortex can affect subsequent memory4, the studies performed to date do not show whether those manipulations prevented the hypothesized transfer of information from the hippocampus to the cortex, or whether they affected the hippocampal or cortical representations themselves. For instance, it would be reasonable to infer that interfering with hippocampal function during an SWR could disrupt a hippocampal (or cortico-hippocampal) memory trace, or that reactivating hippocampal memory traces during a delay could affect subsequent memory performance, even if the hippocampus is not active in training the cortex.

Although the present evidence does not compel a systems consolidation account, future studies could provide more definitive evidence. For example, studies show that as little as 60 minutes of sleep leads to considerable reductions in forgetting. If these rapid effects of sleep reflect systems consolidation, hippocampal lesions should produce accelerated rates of forgetting over delays that include sleep. Studies reviewed in our paper do not support this prediction5, but perhaps future studies will show otherwise.

Another important question for future research will be to determine whether replay observed in rodents is related to putative replay reported in humans. Rodent research on replay and its potential role in systems consolidation has focused largely, if not entirely, on reactivation of place cell sequences during sleep and quiet wakefulness, and have focused on examining activity observed in well-learned spatial environments. It would be important to know whether one can observe this type of replay in tests of memory in humans that are more analogous to those described in theories of systems consolidation. Regardless of whether future studies of post-encoding activity are ultimately found to provide stronger support for context models or systems consolidation models, we agree with Antony and Schapiro that such studies will continue to have an important impact on the field.