You probably think of your worst memory as a file. It sits somewhere in the dark, unchanged, and you take it out and look at it. Therapy, in that picture, is the unpleasant act of opening the drawer.

The neuroscience suggests something stranger. A memory is not retrieved so much as rebuilt, and when it is rebuilt it may be briefly open to revision. The drawer is more like wet clay than a filing cabinet. That is both the reason talking about the past can change how it feels, and the reason it sometimes goes badly.

The idea that overturned the filing cabinet

For most of the twentieth century, the dominant view was that memories are stabilized once, in a process called consolidation, and then they stay put. Recalling one just reads it out.

In 2000, Karim Nader, Glenn Schafe and Joseph LeDoux published a now-famous result in Nature. They trained rats to fear a tone, then reminded the rats of the tone and injected a drug that blocks protein synthesis into the amygdala. The fear memory, which should have been stable, was disrupted. It seemed that reactivating a memory made it fragile again. It had to be re-stored, and the re-storing, called reconsolidation, could be interfered with.

The finding fit with older observations. Elizabeth Loftus spent decades showing that human recollection is reconstructive and can absorb new information, sometimes false information, at the moment of recall. Your brain doesn't play a recording. It assembles a scene from fragments, and what is going on while you assemble it leaves traces.

What this has to do with a therapist's office

Here is the part worth taking slowly. In 2010, Daniela Schiller, Elizabeth Phelps and colleagues reported an experiment in humans. People learned to fear a colored square paired with a mild shock. A day later, some were briefly reminded of the fear and then, within a limited window, went through extinction training, where the square appears repeatedly without a shock. Those who got the training inside the window showed no return of fear a year later. Those who got it outside the window did.

The interpretation was that when an old memory is reactivated and then meets new, contradicting experience while it is still malleable, the updated version can replace the old one rather than merely compete with it.

Some researchers, such as Bruce Ecker, who developed Coherence Therapy, have proposed that this is part of how psychotherapy works. A painful memory is brought to life, and something in the room doesn't match what the memory predicted. The feared rejection doesn't come. The shame is met with steadiness. The memory is stored again with that mismatch folded in.

I want to be careful here. This is a promising hypothesis, not a settled account of therapy. Reconsolidation findings in humans have been harder to reproduce than the early headlines suggested, effects are often modest, and nobody can point to a specific moment in a session and say a memory was rewritten. Treat it as a useful lens, not a guarantee.

Why the lens is still useful

Even held loosely, the idea explains several things people find baffling about therapy.

Why remembering can make things feel worse first. To be changed, a memory has to be brought back. Reactivation comes before revision. The raw feeling is not a sign that the work is failing. It may be the point at which the work is possible.

Why simply replaying a memory isn't the same as healing it. Rumination reactivates a memory over and over, but nothing new ever meets it. You rehearse the same pain, and each rehearsal can strengthen the path. The ingredient that seems to matter is the mismatch: some experience that doesn't fit what the old memory said would happen.

Why your story of an event shifts over time. If you notice that you tell the same memory differently now than you did a year ago, you are not lying or losing your grip. You are seeing reconstruction at work. The facts may be intact, while the meaning, the emphasis and the emotional charge have moved.

Why an insight can feel true in the room and thin by Thursday. The updated version has to compete with the old one, which has been rehearsed for years. New learning usually doesn't delete old learning so much as gain ground against it.

The risk nobody mentions

If recall is reconstruction, then memory is also vulnerable to suggestion, to the mood you're in, and to the story you're hungry for. Loftus's work is a standing warning: confidence and vividness are not proof of accuracy.

This matters in therapy for a practical reason. You can walk out with a coherent, emotionally satisfying account of your childhood that is partly assembled from what felt right in the moment. That doesn't make therapy worthless. Emotional truth and factual precision are different things, and healing mostly needs the first. But hold your reconstructions with some humility, especially about specific events you can't check.

It also means that reopening a memory is not something to do carelessly on your own at 2 a.m. Reactivating without a way to meet it with something safe and new can simply deepen the groove. For trauma in particular, pacing belongs with a trained professional.

Your next moves

  • Date your versions. Next time a memory comes up in session, write two or three sentences about it the same day, with the date. Do it again when it comes up weeks later. Compare. Don't judge which is right; look at what moved: the emphasis, the villain, the feeling.
  • Write down the mismatch. After a session, answer one question: what happened in the room that the old memory wouldn't have predicted? Maybe you said the shameful thing and your therapist didn't flinch. That single line is the most important data you have.
  • Name how it felt before and after recall. Rate the charge of the memory from 0 to 10 at the start and end of the same session. A drop within a session is worth noticing. So is no drop. Either is information to bring back.
  • Stop the solo replay. If you catch yourself running the same memory on a loop, interrupt it by saying aloud what is different today: your age, who is in the room, what you know now that you didn't then. That gives the memory something new to meet.
  • Ask your therapist about pacing. Try this question: "When we go back to this memory, how will we know when to stop and what to do afterward?" A good therapist will have an answer, and asking it makes you a collaborator, not a patient being opened up.

Where a record helps

If memories are rebuilt every time, then what you were able to hold onto from the session itself becomes unexpectedly valuable: what you felt, what shifted, what you said, in the words you used that day, before the next reconstruction reshaped it. That is the part Sesh is built for. It's a private place to capture what happened in a session while it's still fresh, so that over months you can see how your relationship to a memory changed, not just how you remember it changing. If you'd like that kind of record, you can find it at sesh.lumenlabs.works.

The past is not fixed the way you were taught to fear. It also isn't infinitely soft. It is somewhere in between: a thing you rebuild each time you visit, and so a thing you can, with care and company, learn to rebuild a little more kindly.