A new study published in Nature Communications has found that a small number of newborn neurons in the brain help convert recent experiences into lasting memories during sleep. Researchers at the University of Tsukuba in Japan found that just a few of these adult-born neurons, when reactivated during rapid eye movement (REM) sleep, appear to play a necessary role in memory consolidation. The study also suggests that the precise timing of their activity, in sync with specific brain rhythms, may be essential for this process.
Scientists have long believed that memories formed while awake become more stable during sleep, particularly during REM sleep. This is the stage of sleep associated with dreaming and heightened brain activity. A process called memory consolidation is thought to take place at this time, helping us retain important information from the day.
Previous research had shown that entire regions of the brain, such as the hippocampus, are active during REM sleep and are involved in memory. However, it remained unclear which specific neurons are responsible and how their activity contributes to storing memories. This study focused on a rare group of neurons called adult-born neurons which are created in the hippocampus throughout life. These cells are thought to be involved in learning and memory, and their loss is linked to memory disorders like Alzheimer’s disease.
Yet despite their scarcity, the removal of ABNs has a noticeable impact on memory function. This puzzling discrepancy motivated the researchers to explore how such a small population could have such a significant effect. They proposed that these neurons might play an especially targeted role during REM sleep.
“Memory researchers have long suspected that reactivating recent experiences during REM sleep helps ‘set’ memories, but we lacked direct causal evidence tying specific neurons — not just whole brain regions — to that process,” explained study author Masanori Sakaguchi, an associate professor at the International Institute for Integrative Sleep Medicine at the University of Tsukuba. “We set out to test whether tiny, well‑defined ensembles of adult‑born hippocampal neurons reactivate during REM and whether that brief replay is required for consolidation.”
The researchers used genetically modified mice to track and control the activity of adult-born neurons. These mice were designed so that the researchers could monitor neuron activity using a calcium-sensitive marker and selectively turn off specific neurons using light, a technique known as optogenetics.
To create a memory that could later be measured, the mice were placed in a setting where they received a mild foot shock. This standard “fear conditioning” procedure helps animals form a strong association between a specific environment and a negative experience. The researchers then recorded the activity of adult-born neurons during REM sleep in the hours following this learning session.
They discovered that adult-born neurons that had been active during the initial learning experience were reactivated during REM sleep. This reactivation was not random. Rather, it occurred in sync with a specific part of a repeating brain rhythm known as the theta cycle, which is commonly observed in the hippocampus during REM sleep.
Using optogenetics, the researchers were able to selectively silence only those few adult-born neurons that were reactivated during sleep. When they did this during the REM sleep period following fear conditioning, the mice showed reduced memory of the experience the next day. The mice were less likely to freeze when placed back into the setting where they had received the shock, suggesting that the memory had not been fully stored.
In contrast, silencing the same neurons later, during a retrieval test while the mice were awake, had no effect on their behavior. This suggests that the critical role of adult-born neurons lies in the consolidation phase that occurs during sleep, rather than in recall during wakefulness.
Importantly, the researchers found that only a very small number of adult-born neurons needed to be reactivated for the memory to be stored. This provides evidence that memory consolidation can rely on highly selective and sparse neural activity, rather than large-scale reactivation across the brain. Sakaguchi was surprised to find “that the ‘critical mass’ was so small — roughly three adult-born neurons per hemisphere could carry behaviorally meaningful information.”
The researchers also tested whether the precise timing of neuron activity within the theta rhythm influenced memory consolidation. To do this, they silenced adult-born neurons during specific points in the theta cycle using a feedback-controlled system that could detect the phase of the oscillation in real time.
They found that disrupting adult-born neuron activity during one particular phase of the theta cycle impaired memory formation, while silencing during other phases did not. This points to the idea that not only do specific neurons matter for memory consolidation, but also the exact moment during brain rhythms when they are active is important.
“Only one specific theta phase window in REM mattered; disrupting firing there, but not at other phases, impaired consolidation,” Sakaguchi told PsyPost. “Phase‑specific silencing impaired both contextual and trace fear memory, underscoring that when these neurons fire within the REM theta cycle is as important as whether they fire.”
This finding adds support to the broader idea that brain rhythms serve as a timing signal for coordinating the flow of information during sleep. In the case of memory, the ascending phase of the theta cycle may represent a window during which neurons are most effective at communicating with downstream brain regions involved in storing long-term memories.
“When you sleep — specifically in REM sleep — your brain replays the day’s important moments using remarkably small ‘teams’ of newborn neurons in the hippocampus; if that replay is disrupted at the wrong moment in the brain’s rhythm, the memory hardens less well,” Sakaguchi explained. “In other words, quality of REM timing, not just amount of sleep, helps determine what sticks.”
While the study provides new insights into how sleep contributes to memory, several limitations remain. The experiments were conducted in mice and focused exclusively on fear-related memories. It is not yet known whether the same mechanisms apply to other types of memories or how similar processes might operate in the human brain.
The researchers also note that they were not able to directly observe or manipulate dreams or specific memories. Instead, they tested whether the brief silencing of small groups of neurons affected the mouse’s behavior in later memory tests. This approach helps isolate the role of the neurons during sleep but does not provide access to the content of the memories themselves.
Future research may aim to identify the molecular mechanisms that allow these adult-born neurons to change their connections during REM sleep. Another important direction involves understanding how these neurons interact with broader brain circuits and how timing within brain rhythms influences their effect.
“Mechanistically, we want to identify the molecular switches that let REM‑theta timing reshape synapses on these young neurons, and to map how small adult-born neuron ensembles talk to downstream circuits during the ‘right’ phase,” Sakaguchi said. “Translationally, we’re exploring closed‑loop, non‑invasive approaches (e.g., precisely timed sensory cues during sleep) and clinical collaborations in trauma‑related conditions, where sleep‑stage dynamics may be altered; our lab has been building the AI and behavioral frameworks to test such ideas.”
“For transparency and reuse, we’ve shared code and data resources referenced in the article, and the university has released accessible summaries of the work. These provide additional context and lay explanations of the ‘~3 neurons’ and ‘theta phase’ findings.”
The study, “Transient reactivation of small ensembles of adult-born neurons during REM sleep supports memory consolidation in mice,” was authored by Sakthivel Srinivasan, Iyo Koyanagi, Pablo Vergara, Yuteng Wang, Akinobu Ohba, Toshie Naoi, Kaspar E. Vogt, Yoan Chérasse, Noriki Kutsumura, Takeshi Sakurai, Taro Tezuka, and Masanori Sakaguchi.
