Almost everyone has a memory they would happily get rid of: an awkward moment, a painful scene, something that keeps coming back at the worst possible moment, often late at night. Johannes Gräff, a neuroscientist at EPFL in Lausanne, studies whether such memories can really be deleted. His answer is yes, in principle, but with a catch that matters a lot once you move from mice to people — and at the end you can check how you deal with your own difficult memories.
Before asking whether a memory can be erased, it helps to agree on what “erased” actually means. According to the CORDIS article published on 14 August 2026, there are two broad schools of thought on the question.
The first goes back to Ivan Pavlov, the Russian Nobel Prize winner best known for his experiments with dogs. In the 1920s, Pavlov suggested that a memory fades away through an active process of suppression. In other words, the brain does not delete anything; it simply learns to hold the old memory down.
The second, more recent view turns this idea around. Some scientists have proposed that erasing memories, and traumatic ones in particular, could involve going back to them rather than pushing them away.
Why would revisiting a bad memory help? The explanation is surprisingly simple. Each time a person remembers something, the memory becomes malleable for a while and can be changed.
That flexibility is not just a laboratory curiosity. It is the very phenomenon that exposure therapy makes use of. In this kind of therapy, a person deliberately returns to a difficult experience in a safe setting, so that the memory can be reshaped instead of simply buried.
Johannes Gräff works at the Brain Mind Institute of the Federal Institute of Technology Lausanne (EPFL) in Switzerland. He points to one idea that has changed the field in recent years.
“A key breakthrough in the past few years has been the concept of how memories are stored in so-called engram cells,” says Gräff. “The idea is that each time we learn or experience something new, cells in our brain are activated, leaving a memory trace.”
So a memory is not floating somewhere in the brain in an abstract form. It is tied to a particular group of cells that were active when the experience happened. That makes it, at least in theory, something researchers can locate and study.
Gräff and his team tested this in the Remote memory traces project, funded by the European Research Council. Working with mice, they were able to identify and tag the neurons that are activated by one specific traumatic memory.
The experiments produced several key findings:
Gräff is careful about the wording, though. “Follow-up tests suggested that mice were able to successfully unlearn, or, if you like, erase the memory of a traumatic incident,” he says. “Though it is more accurate to say that the association to a particular traumatic event has been erased.”
That distinction matters. What disappeared in the mice was the link between an event and the fear attached to it, not necessarily every trace of the event itself.
The obvious next question is whether this knowledge could be used to target and erase one particular memory on purpose. “In theory this is possible,” Gräff remarks. “In mice, we were able to insert a molecular switch in engram cells. Turning this ‘off’ effectively got rid of the memory.”
A switch that removes one memory and leaves the rest untouched sounds like science fiction, yet in mice it worked.
Here comes the catch. Erasing a memory in the human brain would be orders of magnitude more complex than doing it in mice.
Gräff puts the dilemma plainly. “Hypothetically, could we use gene therapy to switch specific neurons on or off?” he asks. “I would say yes, but with the caveat that at the moment we would be going in blindly, not knowing for sure that we would be erasing one memory and not the other. So right now, this approach is just not feasible.”
In short, the problem is not the idea itself but precision. Without knowing exactly which cells hold which memory, an attempt to delete one experience could just as easily damage another.
Instead of a memory-deleting procedure, Gräff is looking at something more practical. He wants to build on the findings about “unlearning” traumatic events and see, for example, whether safe pharmacological solutions could be used to enhance the benefits of exposure therapy.
If that works, it could bring major benefits for people with post-traumatic stress disorder (PTSD). The goal would not be to wipe their past clean, but to help them confront traumatic memories and move on.
Memory erasure is no longer pure fantasy, at least in the lab. Research on engram cells shows that memories are stored in specific groups of neurons and that “unlearning” happens in those very cells. In mice, a molecular switch was even enough to remove a memory. In humans, however, the same approach would be far more complex and, for now, not feasible. The most promising path is to support existing treatments such as exposure therapy, which could help people with PTSD face difficult memories rather than forget them.
5 questions · one minute
EN 
CS
DE
ES
FR
HU
IT
PL
PT
РУ
SK
TR
УК
AR
中文