Summary: A new study has solved a fundamental biological mystery: how the brain remains flexible enough to learn new information while remaining stable enough to safeguard past knowledge. The study reveals that the brain reuses a dedicated core of cells to process multiple distinct memories without mixing them up or erasing old data.
By tracking hundreds of individual neurons simultaneously in moving mouse models, investigators unmasked a “memory switchboard” within the hippocampus that separates incoming and outgoing signals through divergent firing patterns, protecting long-term memories from being overwritten.
Key Facts
- The Plasticity Stability Enigma: Neuroscientists have long struggled to explain how mammalian brains continuously absorb new experiences without causing retroactive interference or erasing existing memory maps.
- The CA1 Core Hub: The research team identified that approximately one in four memory cells within the cornus ammonis 1 (CA1) region of the hippocampus acts as a shared physical “hub” connecting incoming and outgoing signals.
- Divergent Firing Channels: This cellular switchboard routes signals using a precise structural mechanism: a minority of CA1 neurons collect fast-changing incoming data from the cornus ammonis 3 (CA3) region, but when transmitting that data onward to the navigation-heavy retrosplenial cortex, the exact same cells fire in a completely different pattern to keep the channels separate.
- Simultaneous Multi-Region Recording: To capture this real-time transformation, the team trained six mice on a rewarded track, using high-density electrodes to monitor hundreds of individual neurons across multiple deep-brain and cortical zones simultaneously while the animals moved around naturally.
- The Nighttime Replay Loop: These critical CA1 hub neurons do not rest after waking behavior; they remain highly active during nighttime sleep, replaying waking patterns inside brain events known as sharp-wave ripples to solidify memories and keep the pathway from the hippocampus to the cortex open.
- Clinical and AI Implications: Co-senior authors Dr. Zhe S. Chen and Dr. György Buzsáki note that this switchboard blueprint provides vital clues regarding circuit failures in Alzheimer’s disease, while offering a biological template to fix “catastrophic forgetting” in next-generation artificial intelligence networks so they can update continuously without overwriting past data.
Source: NYU Langone
The brain may reuse some cells to store many different memories without mixing them up with or erasing older memories, a new study in mice suggests.
Led by NYU Langone Health researchers, the study revealed that about one in four memory cells in a brain area called the hippocampus acts as a shared “hub” that links incoming and outgoing signals.
A report on the findings was published online May 13 in the journal Nature.
Scientists have long wondered how the brain can be flexible enough to learn new information while also being stable enough not to forget past knowledge.
To shed light on this mystery, the investigators focused on a chain of connected areas linking the hippocampus, which sits deep inside the brain and helps organize new experiences into memories, and the neocortex, the brain’s outer layer, which stores long-term information. These included the cornus ammonis 3 (CA3), a hippocampal region that sends in fast-changing information; cornus ammonis 1 (CA1), a hippocampal region that acts as a central hub; and the retrosplenial cortex, which plays a key role in navigation and scene reconstruction.
The team found that a minority of hippocampal CA1 cells (neurons) carry most of the incoming messages that were sent by CA3. Then, when CA1 sends signals to the retrosplenial cortex, those same cells fire in a different pattern, creating a separate outgoing channel.
In this way, messages coming in and going out stay separate even though they reuse many of the same neurons, much like how an electronic switchboard can manage many calls without crossing the lines. This setup may help the retrosplenial cortex maintain its map’s stability while the other two regions continue learning from experience.
“Our findings help explain how memory can be both moldable and enduring,” said study co-lead author Joaquín Gonzalez, PhD, a postdoctoral fellow in the Department of Psychiatry at NYU Grossman School of Medicine. “By changing how the same cells fire together instead of turning on new cells, the brain can keep information organized and protect older memories.”
Additional findings showed that the key CA1 neurons that handle daytime communication remain active at night during sleep, in brain events known as sharp-wave ripples.
Because the same core of cells handles both daytime processing and nighttime replay, the pathway from hippocampus to cortex can remain open and help solidify memories.
“Our study shows how learning and memory consolidation can coexist in the same network,” said study co-lead author Mihály Vöröslakos, MD, PhD, a postdoctoral fellow in NYU Grossman School of Medicine’s neuroscience department. “Our discovery was made possible because for the first time, we were able to record hundreds of individual neurons across all the key regions simultaneously in animals that were moving around naturally.”
“Our discovery of a ‘memory switchboard’ deep in the hippocampus may provide clues as to how memory circuits fail in Alzheimer’s disease and other conditions that affect the brain’s ability to recall events and find places,” said study co-senior author Zhe S. Chen, PhD, a professor in the Departments of Psychiatry and Neuroscience at NYU Grossman School of Medicine.
For the study, the research team trained six mice to run back and forth on a straight track with water rewards at each end. While the animals explored, the scientists used high-density electrodes to record activity from hundreds of neurons at once. They also tracked the rodents’ positions so they could match each spike of brain activity with the mouse’s behavior at that moment.
The team then looked for shared patterns of activity between regions to see how signals from CA3 were transformed by CA1 before reaching the retroplenial cortex. In additional sessions, the researchers recorded the mice while they slept and found that the waking patterns were replayed many times but differently within the hippocampus and across the hippocampus and neocortex.
According to the authors, these findings may help address a major challenge faced by artificial intelligence tools, which tend to ‘forget’ what they have learned when trained on new tasks.
“By showing how the mammalian brain can safeguard memories during learning, our research may offer a biological blueprint for designing next-generation AI technology that can update itself continuously without overwriting what it has already acquired,” said study co-senior author György Buzsáki, MD, PhD, the Biggs Professor of Neuroscience at NYU Grossman School of Medicine and a member of NYU Langone’s Institute for Translational Neuroscience.
Dr. Buzsáki, who is also a member of NYU Grossman School of Medicine’s Department of Neurology, said that the researchers’ next plan is to examine whether similar switchboardlike channels appear in other memory circuits.
Because the study was conducted in mice in a controlled environment, the researchers cannot draw firm conclusions about what happens in more natural environments or in the human brain, cautioned Dr. Buzsáki.
Funding: Funding for the study was provided by National Institutes of Health grants RF1DA056394, P50MH132642, R01MH122391, and U19NS107616.
Along with Drs. Gonzalez, Vöröslakos, Chen, and Buzsáki, NYU Langone researchers involved in the study were Deren Aykan; Nina Soto, PhD; Noam Nitzan, PhD; Rachel Swanson, PhD; and Mursel Karadas, PhD.
Key Questions Answered:
A: By changing their rhythm instead of changing the cells themselves. The NYU Langone team discovered that the hippocampus acts like an electronic switchboard; the exact same core neurons process both incoming and outgoing messages, but they fire in completely different patterns for each channel, keeping the data lines separate and past knowledge protected.
A: To find out how memories are permanently locked into long-term storage. By tracking the mice during sleep, the researchers discovered that the same hub cells used during the day stay active at night, replaying the day’s events during sharp-wave ripples to transfer information from the temporary hippocampus to the long-term neocortex.
A: By solving the major AI flaw known as catastrophic forgetting. Current AI tools tend to completely wipe out their older training when learning a new task, but using this newly discovered mammalian blueprint allows developers to design next-generation AI networks that can update themselves continuously without overwriting previously acquired data.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this memory research news
Author: Shira Polan
Source: NYU Langone Health
Contact: Shira Polan – NYU Langone Health
Image: The image is credited to Neuroscience News
Original Research: Closed access.
“Subspace communication in the hippocampal–retrosplenial axis” by Joaquin Gonzalez, Mihály Vöröslakos, Deren Aykan, Nina Soto, Noam Nitzan, Rachel Swanson, Mursel Karadas, Zhe Sage Chen & György Buzsáki. Nature
DOI:10.1038/s41586-026-10481-z
Abstract
Subspace communication in the hippocampal–retrosplenial axis
The capacity of hippocampal circuits to transform inputs into downstream outputs is fundamental to navigation and memory, yet the circuit-level mechanisms that enable this flexibility in adapting to experience remain unclear.
Here we approach this problem by performing large-scale (up to 1,024 channel) recordings across the hippocampal–retrosplenial cortex (RSC) circuit in behaving mice, enabling simultaneous access to spiking activity in dentate gyrus (DG), CA3, CA2, CA1 and RSC. On the basis of a linear dimensionality-reduction technique known as partial canonical correlation analysis, we identify low-dimensional communication subspaces between two regions while accounting for influences from a third area.
These subspaces captured distinct input–output transformations in the CA1 region, linking upstream hippocampal activity (DG, CA3 and CA2) to downstream cortical targets (RSC). Intrinsic firing properties and anatomical location constrained subspace memberships—members were mapped to deep sublayers of the CA3–CA1–RSC axis during both spatial and non-spatial tasks.
These subspaces could recombine overlapping neuronal pools to support distinct interareal interactions across changing experiences and brain states. Reactivation patterns of CA1–CA3 subspaces, but not those of CA1–RSC, during post-experience sleep correlated with replay, reflecting a plasticity–stability balance in the input–output transformation along the hippocampal–retrosplenial axis.
Our findings suggest a model in which hippocampal–neocortical communication reconfigures predetermined circuit motifs to flexibly encode experiences.
