Alex P. Vaz, Yina Wei, John F. Burke, Joel M. Stein, Sandhitsu R. Das, Gregory A. Worrell, Michael R. Sperling, Bradley Lega, Kathryn A. Davis, Richard Gorniak, George L. A. R. Zaghloul
5 min
Episodic memory is thought to rely on the internal replay of neural activity patterns that occurred during the initial experience of an event. While this phenomenon has been well-documented in rodent models of spatial navigation, direct evidence for the replay of neuronal spiking sequences during human memory retrieval has remained elusive. This study investigated whether human cortical spiking activity is organized into specific sequences during memory formation and if these sequences are subsequently replayed during successful recall.
The researchers recorded single-unit activity and local field potentials from the anterior temporal lobe in six neurosurgical patients using microelectrode arrays and intracranial electroencephalography (iEEG). Participants performed a paired-associates verbal memory task, which required them to encode and later retrieve associations between word pairs. The team analyzed the temporal structure of "burst events"—periods of high-frequency spiking—to determine if the firing order of neurons was consistent within trials and if these patterns were reinstated during retrieval.
The study found that cortical ripple oscillations reflect bursts of single-unit spiking that organize into trial-specific sequences during memory encoding. These sequences were not random; they were highly consistent within a single trial but differed significantly between trials. During successful memory retrieval, participants replayed these exact, trial-specific sequences. Furthermore, the researchers discovered that this replay was not uniform; it was significantly more robust when cortical spiking bursts were temporally coupled with ripple oscillations in the medial temporal lobe (MTL). This suggests that the MTL acts as a coordinator, facilitating the reinstatement of precise cortical firing patterns necessary for accurate memory recall.
These results provide the first direct evidence that human episodic memory retrieval involves the active, sequential replay of cortical neural activity. By linking human cortical spiking to MTL-mediated ripple oscillations, the study bridges the gap between rodent models of memory consolidation and human cognitive function. It suggests that memory is not merely a static storage process but a dynamic one where the brain reconstructs the temporal order of past experiences through sparse, stereotyped neural firing.
Human brain activity during memory Animal studies suggest that sequence replay of neuronal activity may underlie memory retrieval and consolidation. However, there is no direct evidence that the replay of spiking activity sequences is important for these processes in the human brain. Vaz et al. simultaneously recorded single-unit spikes, local field potential, and intracranial electroencephalography signals in the brain while participants performed a memory task. Sharp wave ripple oscillations in the temporal lobe cortex reflected bursts of neural spiking, and these bursts of spikes organized into sequences during memory formation. These sequences were replayed during successful memory retrieval. The extent of sequence replay during correct recall was related to the extent to which cortical spiking activity was coupled with ripples in the medial temporal lobe. Science , this issue p. 1131 , Human single-unit and local field potential recordings of encoding-related activity sequences during long-term memory formation and retrieval are investigated. , Episodic memory retrieval is thought to rely on the replay of past experiences, yet it remains unknown how human single-unit activity is temporally organized during episodic memory encoding and retrieval. We found that ripple oscillations in the human cortex reflect underlying bursts of single-unit spiking activity that are organized into memory-specific sequences. Spiking sequences occurred repeatedly during memory formation and were replayed during successful memory retrieval, and this replay was associated with ripples in the medial temporal lobe. Together, these data demonstrate that human episodic memory is encoded by specific sequences of neural activity and that memory recall involves reinstating this temporal order of activity.
Sam: What's the MTL doing in all this? Is the cortex running this replay autonomously?
Alex: The data argues against autonomy. When they split cortical bursts by whether they co-occurred with an MTL ripple, the coupled bursts showed substantially higher sequence fidelity than uncoupled ones. The MTL ripple seems to be gating or coordinating the cortical reinstatement — consistent with the classical view of MTL as an index that binds and retrieves distributed cortical representations, but now with direct single-unit evidence for what that coordination looks like mechanistically.
Sam: So the MTL isn't storing the memory itself — it's triggering the cortex to replay the right sequence at the right time.
Alex: That's the model the data supports. The content is in the cortical sequence; the MTL provides the retrieval signal.
Sam: Where would a careful reviewer push back?
Alex: The obvious constraint is six participants, with spatial coverage limited to anterior temporal lobe. We don't know whether this is a general cortical mechanism or something specific to this region's role in semantic and conceptual binding. The other thing worth flagging is the absence of reverse replay. Rodents show backward sequence reactivation, particularly in reward-based tasks. The authors don't find it here — they attribute that to task structure, abstract word pairs rather than spatial trajectories — but it's an open question whether the directionality of replay carries functional significance that this design simply couldn't detect.
Sam: And the sample is entirely clinical. Does that constrain generalization?
Alex: It does, though the epileptic focus wasn't in the anterior temporal lobe for these participants, so the recordings aren't obviously confounded by pathology. That said, you can't rule out that chronic epilepsy reshapes the dynamics in ways that don't generalize to healthy cortex. That's a standing limitation for all intracranial human work.
Sam: If the temporal sequence is literally the memory code, does that open anything practically — for neurodegeneration, for instance?
Alex: It's a genuine long-term implication. If you can decode the neural template for a specific memory, targeted stimulation that reinstates that sequence becomes a conceptually coherent intervention. The gap between that and a clinical tool is large, but the mechanistic framework is now more concrete than it was.
Sam: So the paper's contribution is moving from rodent models to direct human evidence — showing that episodic retrieval is the active reinstatement of a specific temporal firing pattern, coordinated by the MTL.
Alex: That's the core of it. The temporal order of spiking is the code, not just a correlate of it. It's a meaningful step toward understanding the physical substrate of human recall — and it gives the field a concrete mechanism to interrogate in larger, more spatially comprehensive studies. Thanks for listening to ResearchPod.