TIRCIA · General public
Can a traumatic memory be erased?
What the Inserm study on PTSD and memory really says
By Marie-Agnès Thulliez · Author · Publishing Director11-minute read
In March 2026, a headline was widely shared: researchers at Inserm, the French national institute for health and medical research, had supposedly discovered “a neuronal switch” capable of “erasing traumatic memories”. The research in question is solid and important, and what it shows is more nuanced, and just as encouraging. Here is a look at what this work really shows, and at what it changes for understanding and treating post-traumatic stress.
In brief, for everyone
The researchers showed that, in post-traumatic stress, the brain has difficulty holding back the return of intrusive memories, and that this capacity can be restored with recovery. The point is to regain the ability to hold a memory at a distance, while the memory itself remains. This is good news: the brain remains capable of change after trauma.
1. Research born of the 13 November attacks
After the Paris attacks of 2015, Inserm, the CNRS and the University of Caen launched the 13-Novembre programme, led by Francis Eustache and Denis Peschanski. It includes a brain imaging study, called REMEMBER, led by Pierre Gagnepain. It follows people exposed to the attacks, with or without post-traumatic stress, and people who were not exposed, at intervals of several years.
Two major publications have come out of it: the first in the journal Science in 2020 (Mary et al., 2020), the second in Science Advances in January 2025 (Leone et al., 2025). It was this second study that the 2026 popular-science article presented.
2. How is memory control measured?
The researchers use a laboratory task called “Think/No-Think”, developed in the early 2000s (Anderson and Green, 2001). Participants first learn associations between a word and an image. Then, when a word appears, they are asked either to think of the associated image or, on the contrary, to stop it from entering their mind.
Sometimes the image pops up anyway: this is a laboratory model of what happens during an intrusion. By observing the brain with MRI during this task, researchers can see how the prefrontal regions, which exert control, do or do not hold back the activity of the hippocampus, a key structure for memory. An important point: the images used are neutral, unrelated to the attacks.
3. First finding: a brake that works poorly
The 2020 study involved 102 people exposed to the attacks, 55 of whom had PTSD, and 73 people who were not exposed. In people with PTSD, the prefrontal regions were less successful at holding back hippocampal activity at the moment an unwanted memory surfaced. People who had been exposed but did not have PTSD retained this control (Mary et al., 2020).
The authors drew a powerful idea from this: PTSD may involve a memory engraved too deeply, and also a control system that can no longer regulate access to it. The difficulty lies less in the capacity to set a memory aside than in the way brain networks put that capacity into practice.
4. Second finding: this brake can recover
The 2025 study followed the same people with two imaging scans, in 2016-2017 and then in 2018-2019, supplemented by a symptom questionnaire in 2020-2021. It compares 34 people whose PTSD remained chronic, 19 people whose PTSD resolved between the two measurements, and 72 people who were not exposed (Leone et al., 2025). Four findings stand out:
- Control returns to normal in those who recover. In people in remission, the control mechanisms become close again to those of people who were not exposed, and this recovery precedes the decrease in traumatic memories.
- It remains impaired in chronic forms. People with chronic PTSD seem to rely more on “anticipatory” control, a form of avoidance, without managing to stop intrusions when they occur.
- The hippocampus is protected. In chronic PTSD, an area of the hippocampus (including the dentate gyrus) shrinks in volume over time. In people in remission, this shrinkage stops, and better memory control is associated with its halt.
- An early sign in some chronic patients. In some people whose PTSD remained chronic, an early improvement in control, visible at the second scan, heralded a decrease in intrusive memories reported later in the questionnaire.
The authors conclude that resilience to trauma rests on the plasticity of memory control circuits, linked to the plasticity of the hippocampus. In other words, these circuits are not fixed: they can reorganise.
5. Where the study’s conclusions stop
These results are important, but several conclusions often drawn in the media go beyond the data:
| What the study shows | What it does not show |
|---|---|
| Control of access to memories is impaired in PTSD | That a memory could be “erased” |
| This control returns to normal in those who recover | Which therapy brings about this recovery |
| It is associated with the halt in volume loss in an area of the hippocampus | That these difficulties existed before the trauma (no measurements were taken before the attacks) |
| Links over time between control and remission | A demonstrated cause-and-effect relationship |
A pre-existing vulnerability?
One sometimes reads that these control difficulties might exist before the trauma and make some people more vulnerable. This is the interpretation the researchers themselves favour. For Giovanni Leone, first author of the study, these alterations in control “are much more likely to be the cause than the consequence of PTSD” (Inserm press release, 2025). Their reasoning rests on the dynamics observed: changes in control accompany changes in symptoms and, in some patients, precede them.
Some caution is nonetheless in order. The study began after the attacks: no measurements were taken before the trauma. This hypothesis therefore remains a sound inference rather than a direct demonstration. It is consistent with other, older work suggesting that a smaller hippocampus could be a pre-existing risk factor (Gilbertson et al., 2002). For clinical practice, the study’s main message remains the same: these mechanisms are not fixed and can change.
Limitations to bear in mind
The groups are modest in size, particularly the remission group (19 people). The study observes associations over time, without being able to prove a cause-and-effect relationship. Finally, the treatments received did not differ clearly between groups: the study therefore cannot tell us which form of care promotes this recovery.
6. The “neuronal switch” image
A popular-science article published on 19 March 2026 announced in its very title “this neuronal ‘switch’ discovered at Inserm that makes it possible to erase traumatic memories”. The image of a switch, turning a memory on or off, is vivid, but it greatly simplifies what the data show.
The brain does not work on an all-or-nothing basis. Access to a memory can be more or less regulated, more or less intrusive, and it changes over time. What changes is the memory’s conditions of access, its emotional charge and the way it is experienced, while the memory itself remains.
This nuance has practical consequences. Talk of “erasure” can raise in patients the hope that a trauma could be deleted, or the disappointment of failing to achieve this. Neurobiology and therapy both point instead towards a memory that loses its hold.
The article itself does in fact qualify its claim: it states further on that the prefrontal cortex “does not destroy the memory file”. But its title and conclusion do speak of erasure, and several statements go beyond what Inserm published. Comparing the article with Inserm’s press release of 8 January 2025:
| What Inserm says | What can be read in the popular-science article |
|---|---|
| The prefrontal cortex regulates the activity of the hippocampus | The “switch” supposedly also targets the amygdala |
| In PTSD, this control fails to hold back the hippocampus | Attempts at suppression supposedly “paradoxically” worsen hippocampal activity |
| In those who recover, control of access to memories returns to normal | Memories or their emotional charge could be “erased” |
| No therapy was tested in the study | CBT and EMDR supposedly receive a “resounding neurobiological validation”; eye movements supposedly unlock the switch |
| Future avenues: stimulating memory control, studying the GABA alpha 5 receptor | Transcranial magnetic stimulation and neurofeedback are presented as avenues being explored |
| No statement of this kind | “Inserm scientists are keen to reassure” the public about the erasure of memories |
7. What this changes for clinical practice
This work is consistent with what therapists observe: recovering from trauma means being able to remember without reliving, rather than forgetting. The memory moves from the status of an intrusive memory, which imposes itself as if the event were still happening, to that of an integrated memory, which can be recalled as a past event.
It also invites us to distinguish between two things that look alike: avoiding a memory, by trying never to think about it, and regulating access to it, that is, being able to set it aside when it surfaces. People with chronic PTSD seem to rely more on avoidance, without controlling intrusions any better. Trauma-focused therapies, such as EMDR, prolonged exposure or cognitive processing therapy, in fact do the opposite of avoidance: they lead the patient to approach the memory, within a structured framework (see our article EMDR and exposure therapy).
Finally, the researchers mention avenues for the future. Pierre Gagnepain envisages complementary therapies that would stimulate memory control mechanisms and encourage their plasticity, with one possible advantage: acting on these networks “without making the patient relive the traumatic emotions”. The team is also studying the role of a brain receptor, the GABA alpha 5 receptor, located mainly in the hippocampus, which may be involved in forgetting and in silencing memories.
For now, these are research avenues, rather than available treatments. They leave trauma-focused therapies fully in place, as the effectiveness of those therapies rests on other studies.
Clinical vignette
A fictional case, composed from common clinical situations.
Julie, 36, is in therapy for post-traumatic stress following an assault. One morning she arrives with a newspaper article: “They’ve found a switch to erase traumatic memories. Why don’t we do that, instead of going back over what happened?”
The therapist reads the article with her. She explains that the study erased nothing: it observed that, in people who recover, the brain becomes better and better at holding back the intrusion of memories. These are research findings, rather than a treatment. And trauma-focused therapies help people remember without reliving, rather than making them forget.
Julie then says what she was really afraid of: that working on the memory would plunge her back into fear. Together they go over how the sessions unfold and what she can do if the emotion rises. Julie decides to continue. A few weeks later, she notices that the image comes back less often, and that she can let it pass.
Frequently asked questions
Can a traumatic memory be erased?
No, and that is not what the study shows. Effective therapies change the way the memory is experienced and the place it takes up, while the memory itself remains.
Should I learn to stop thinking about the trauma?
Not in the sense of avoidance, which keeps the disorder going. What helps is being able to recall the memory without being overwhelmed, which is what trauma-focused therapies work on.
Can the brain really recover?
Yes, that is the study’s main message: in people who recover, memory control circuits return to normal and the volume loss in an area of the hippocampus stops.
Do these findings apply to all kinds of trauma?
The study concerns people exposed to the 2015 attacks. The mechanisms described are probably general, but this remains to be confirmed for other types of trauma, in particular repeated childhood trauma.
What is established, and what is not yet
- Established: in PTSD, control of access to unwanted memories is impaired, and it returns to normal in those who recover.
- Probable: this recovery contributes to remission and to the protection of the hippocampus.
- Not established: the existence of a “switch” that would erase memories, and the pre-existing nature of these difficulties, which the authors consider probable without being able to demonstrate it.
In one sentence
Recovering from trauma means regaining the ability to hold a memory at a distance, rather than erasing it, and the brain remains capable of this.
The Inserm research describes a gradual reorganisation, rather than a switch.
References
- Agence nationale de la recherche (2025). Réapprendre à oublier : plasticité cérébrale et résilience au trouble du stress post-traumatique [Relearning to forget: brain plasticity and resilience to post-traumatic stress disorder]. Link
- Anderson, M. C., & Green, C. (2001). Suppressing unwanted memories by executive control. Nature, 410(6826), 366-369. Link
- Gilbertson, M. W., Shenton, M. E., Ciszewski, A., et al. (2002). Smaller hippocampal volume predicts pathologic vulnerability to psychological trauma. Nature Neuroscience, 5(11), 1242-1247. Link
- Inserm (8 January 2025). Stress post-traumatique : la plasticité cérébrale, un mécanisme clé de la résilience au trauma [Post-traumatic stress: brain plasticity, a key mechanism of resilience to trauma]. Communiqué de presse [Press release]. Link
- Leone, G., Casanave, H., Postel, C., Fraisse, F., Vallée, T., de La Sayette, V., Dayan, J., Peschanski, D., Eustache, F., & Gagnepain, P. (2025). Plasticity of human resilience mechanisms. Science Advances, 11(2), eadq8336. Link
- Mary, A., Dayan, J., Leone, G., Postel, C., Fraisse, F., Malle, C., Vallée, T., Klein-Peschanski, C., Viader, F., de la Sayette, V., Peschanski, D., Eustache, F., & Gagnepain, P. (2020). Resilience after trauma: The role of memory suppression. Science, 367(6479), eaay8477. Link
- Singer, A. (19 March 2026). Mémoire : ce « switch » neuronal découvert à l’Inserm qui permet d’effacer les souvenirs traumatiques [Memory: this neuronal “switch” discovered at Inserm that makes it possible to erase traumatic memories]. Amphisciences / Ouest-France. Popular-science article analysed here. Link
This article offers general guidance, up to date at the date of publication. It is no substitute for the advice of a doctor, who alone can make a diagnosis, or for support from a psychologist or psychotherapist.
