The common-sense model of memory is a filing cabinet. Something happens, it gets written down, and remembering is opening the drawer and reading. The record sits there unchanged; if it degrades, it degrades by fading.

That model is wrong in a specific way, and the way it is wrong has been oversold almost as badly as the model itself.

Retrieving a memory can make it temporarily unstable, so it has to be stored again, and what happens in that window can change what gets stored. In animals this is well established. In humans the evidence is real but fragile. The effects depend on narrow conditions, one of the most famous human studies failed a large independent replication, and nothing so far has erased a memory in a person. It is a finding about how memory updates, not a technique for rewriting your past.

Key takeaways

Reconsolidation is the re-storing a memory can need after it is recalled. A 2000 rat study made the case; the human evidence is, in one major review's words, "viable but hotly contested" (Elsey, Van Ast and Kindt, 2018).
Recall alone is not enough. In a key human study, the memory only opened up when something about the reminder was unexpected (Sevenster, Beckers and Kindt, 2013).
The best-known behavioral result did not replicate in a preregistered, independent test (Chalkia et al., 2020). Of the PTSD propranolol studies covered here, one trial was positive (Brunet et al., 2018), one found no overall difference from placebo (Roullet et al., 2021), and three small studies found no effect (Wood et al., 2015).
If a memory genuinely disturbs you, the treatments with evidence are trauma-focused therapies delivered by a trained clinician. This is not a do-it-yourself technique.

What is memory reconsolidation?

Consolidation, the idea that a new memory takes time to stabilize, is old and well established. Reconsolidation is the claim that an old, stable memory can become unstable again when it is recalled, and then has to be stabilized a second time.

If that is right, remembering is not a read operation. It is closer to a read-write cycle, with a brief opportunity for editing in the middle.

Where did the idea come from?

In 2000, Karim Nader, Glenn Schafe and Joseph LeDoux published a paper in Nature that unsettled a lot of settled thinking (Nader, Schafe and LeDoux, 2000).

They trained rats to fear a tone, waited for the memory to consolidate, then reactivated it by playing the tone. Immediately afterward they infused a drug that blocks protein synthesis into the amygdala. The fear memory was impaired. The same drug given without the reactivation did nothing.

The implication was large. The memory was not fragile because it was new; it was fragile because it had been retrieved. Reactivation had returned it to a state where it needed fresh protein synthesis to persist.

Why did researchers care so much?

Because the standard way of treating fear has a known weakness.

Exposure-based treatment works through extinction: you meet the feared thing repeatedly without the bad outcome, and the fear response drops. But extinction does not delete the original fear memory. It builds a competing safety memory alongside it (Vervliet, Craske and Hermans, 2013; Kredlow, Unger and Otto, 2016). That is why fear can come back with time, in a different setting, or after a fresh scare. Vervliet and colleagues summarized the problem as extinction being "relatively easy to 'learn' but difficult to 'remember'", and noted that relapse after exposure treatment "is not uncommon".

Reconsolidation offered a way around this. If you could change the original memory while it was open, instead of stacking a new one on top, the fear might not come back.

What has to be true for a memory to open up?

This is the part popular accounts leave out. Not every recall destabilizes a memory.

It seems to need prediction error: something about the reminder has to be different from what the memory predicts. In a human fear study, Sevenster, Beckers and Kindt found prediction error was a necessary condition for reconsolidation, and that how much was needed depended on how the memory was originally learned (Sevenster, Beckers and Kindt, 2013). Recalling a memory exactly as expected may not open it at all.

Age and strength matter too. In rats, a reactivated memory needed fresh protein synthesis to persist only when it was recent, and became less vulnerable the older it was at the time of recall (Milekic and Alberini, 2002). Human researchers work from the same assumption: the spider-fear study below was designed around the observation that "older and stronger memories are generally more resistant to change" (Soeter and Kindt, 2015). That matters for anyone hoping to apply this to something that happened years ago.

Narrow conditions, hard to hit on purpose: that is a large part of why the human results below are so uneven.

Does reconsolidation happen in humans?

You cannot infuse anything into a person's amygdala, so human studies use two other routes. Both produced striking early results, and both have run into trouble since.

The behavioral route: extinction inside the window

In 2010, Daniela Schiller and colleagues reported a striking result in Nature. People were reminded of a conditioned fear and then given extinction training ten minutes later, and the fear did not return the next day. When the same extinction came six hours after the reminder, outside the proposed window, the fear came back as usual. The authors reported the effect lasting at least a year (Schiller et al., 2010).

That result was hugely influential, and it has not held up well.

A test by Merel Kindt and Marieke Soeter, measuring fear three different ways, found that extinction inside the window did not prevent fear from returning (Kindt and Soeter, 2013).
A meta-analysis of studies up to mid-2014 found a small-to-moderate average benefit in humans (g = 0.40) over standard extinction (Kredlow, Unger and Otto, 2016). That is real, but it is an average over mixed results, and it predates the next study.
In 2020, a highly powered, preregistered, independent replication of Schiller's first experiment found no benefit of reminder-then-extinction over ordinary extinction (Chalkia et al., 2020). A companion report by the same team reanalyzed Schiller's original data. It found that the exclusion criteria and the number of excluded participants had been misreported, that qualitative decisions by the experimenters had determined who was included, and that the reported differences between the groups depended on those decisions (Chalkia, Van Oudenhove and Beckers, 2020). The original authors' own 2018 addendum had already said inclusion rested partly on a qualitative judgment of participants' skin-conductance responses (Schiller et al., 2018).

The honest summary is that the behavioral version may work under some conditions nobody has pinned down, and does not work reliably under the conditions first published.

The drug route: propranolol

Kindt, Soeter and Bram Vervliet gave people propranolol, a common beta-blocker, before reactivating a conditioned fear. The next day, the startle response that measured their fear was gone, and it did not come back. The effect was specific to the emotional expression of the memory (Kindt, Soeter and Vervliet, 2009), which is how the authors summarized their finding in later work (Kindt and Soeter, 2013). In plain terms: people still knew what had happened in the experiment; their bodies stopped reacting as if it was happening again.

A follow-up with people afraid of spiders went further. After a two-minute exposure to a tarantula, a single 40 mg dose of propranolol turned avoidance into approach, an effect that lasted at least a year, while the same dose without the tarantula exposure did not. The groups were small, 15 people each, and participants' own reports of their fear only caught up with their behavior months later (Soeter and Kindt, 2015).

The clinical trials in PTSD are where it gets mixed:

In a trial of 60 adults with long-standing PTSD, people took propranolol 90 minutes before a brief weekly session of recalling the trauma, for six weeks. Their symptoms fell more than with placebo. The authors called for replication (Brunet et al., 2018).
A trial of 66 adults using the same schedule found symptoms fell in both groups, by about 39% with propranolol and 34% with placebo, with no overall difference between them. In a subgroup with severe symptoms, scores kept falling for three months after propranolol and rose after placebo, a secondary finding (Roullet et al., 2021).
Three smaller studies giving propranolol, mifepristone, or mifepristone plus d-cycloserine at the time of trauma recall found no group differences in physiological reactions or PTSD symptoms (Wood et al., 2015).

Propranolol for reconsolidation is an experimental idea with inconsistent results. It is not an established or guideline-recommended treatment, and it is not a reason to buy beta-blockers.

The human studies at a glance

Study	Design and n	Intervention	Result
Schiller et al., 2010	Lab fear conditioning in volunteers	Extinction 10 minutes vs 6 hours after a reminder	Fear did not return in the 10-minute group; reported lasting at least a year
Kindt and Soeter, 2013	Lab fear conditioning, fear measured three ways	Reminder trial before extinction, inside the window	Fear still returned on all three measures
Kredlow et al., 2016	Meta-analysis, 63 comparisons in animals and humans, studies to mid-2014	Extinction after a reminder vs standard extinction	Small-to-moderate benefit in humans (g = 0.40)
Chalkia et al., 2020	Preregistered, highly powered direct replication of Schiller's first experiment	Same as Schiller	No benefit over standard extinction
Kindt et al., 2009	Lab fear conditioning	Oral propranolol before a reminder	Startle fear response gone 24 hours later and did not return
Soeter and Kindt, 2015	Spider-fearful adults, 3 groups of 15	40 mg propranolol after 2 minutes with a tarantula, vs placebo, vs propranolol without the tarantula	Avoidance turned to approach for at least a year, only with the tarantula exposure
Brunet et al., 2018	Randomized trial, 60 adults with long-standing PTSD	Propranolol or placebo 90 minutes before brief trauma recall, weekly for 6 weeks	Larger symptom reduction with propranolol
Roullet et al., 2021	Randomized trial, 66 adults with long-standing PTSD	Same schedule as Brunet	Both groups improved; no overall difference
Wood et al., 2015	Three small studies in PTSD, 8 to 16 per group	Propranolol, mifepristone, or mifepristone plus d-cycloserine at trauma recall	No group differences in physiology or symptoms

What does this explain about everyday memory?

Why your most confident memories are not your most accurate

The day after the September 11 attacks, 54 Duke students wrote down how they first heard the news, along with an ordinary recent memory. Retested weeks or months later, both kinds of memory had become less consistent at the same rate. But the students' sense of vividness and their belief in the 9/11 memory's accuracy stayed high, while for the everyday memory they declined (Talarico and Rubin, 2003). Flashbulb memories were special in how accurate they felt, not in how accurate they were.

Reconsolidation is one proposed reason memories drift: if recall can reopen a memory, new information can slip in. There is direct evidence for that in ordinary memory. Students who were reminded of a list of objects just before learning a second list later mixed items from the second list into their recall of the first. The reminder was what made the difference (Hupbach et al., 2007).

Vividness and confidence are not evidence of accuracy. The memories you have told most often are the ones that have had the most chances to change.

Why talking about things helps, sometimes

Therapy often involves recalling something and then having a different experience in the presence of it: someone responds without disgust, the feared consequence does not happen, you notice something you had missed. If that recall opened the memory, the new experience could go into what is stored.

One influential proposal is that therapies which look very different on paper share this as an active ingredient: reactivate, introduce a mismatch, let the memory be stored again under new conditions (Lane et al., 2015). It is a hypothesis, not a settled account, and it has the same evidence problems as everything above.

Why going over something a hundred times doesn't help

Here is an inference, not a tested result. If a memory only opens up when something is unexpected, then retrieval that keeps the same content, frame and conclusion should change very little. That is a fair description of rumination.

What is tested is that repetitive thinking is not all the same. A major review found it can help recovery from upsetting and traumatic events or harm mood, depending partly on whether the thinking is concrete and specific or abstract and evaluative (Watkins, 2008). "Why does this always happen to me?" is the abstract kind. "What exactly did he say, and what did I do next?" is the concrete kind. We cover the rest of that in the science of rumination.

What gets overclaimed?

This concept has been adopted enthusiastically by people selling things, and the popular version has run well ahead of the evidence.

"Reconsolidation therapy can erase traumatic memories." No. The best human results show a reduced fear response, not deleted content. People remembered what happened; they stopped reacting to it as strongly. And the replications are mixed.

"Just recall it and it becomes changeable." No. Recall without prediction error appears not to destabilize anything.

"This proves therapy X is uniquely effective." Several therapies claim reconsolidation as their mechanism. At best it is a plausible ingredient shared by many of them, and it belongs to none.

"You can rewrite your own memories at home." The window is short, the conditions are narrow, and deliberately stirring up a traumatic memory without support is a good way to have a bad week rather than a breakthrough. This is not a self-help technique, and that is the most important sentence in this article.

What can you actually use?

Not much on your own, and a few things worth knowing.

Retelling somewhere the response is different may matter. Not repeating the same account to the same audience in the same frame, which is rumination, but telling it where the reaction is not what you expect. That fits the mismatch idea, though nobody has shown it works through reconsolidation. It is also the line between processing a difficult emotion and suppressing it.

Distrust your most confident memories, especially the ones you have told many times.

Check whether you are rehearsing or processing. If you have gone over something many times and nothing has shifted, ask whether anything ever comes out different. If it never does, more repetition is unlikely to help.

If a memory floods you in the moment, grounding can get you through the next ten minutes. It interrupts attention; it does not treat anything, which is the whole argument of grounding is not a treatment.

If a memory genuinely disturbs you, the useful move is a clinician trained in trauma work, not an article. The UK's NICE guideline recommends trauma-focused CBT for adults with PTSD, which includes cognitive processing therapy, cognitive therapy for PTSD, narrative exposure therapy and prolonged exposure. It also recommends EMDR for adults whose PTSD follows a non-combat trauma more than three months earlier (NICE NG116). The US National Center for PTSD names cognitive processing therapy, EMDR and prolonged exposure as the talk therapies that work best, usually over about 8 to 16 sessions (VA National Center for PTSD). These are built to handle exactly the stirring-up this post describes.

Getting started. If you have never been to therapy, what your therapist wishes you knew covers the first session, and our guide to therapy covers finding one.

Frequently asked questions

What is memory reconsolidation? The proposed process by which a recalled memory becomes temporarily unstable and has to be stored again. During that window, what happens can change what is stored. It is well established in animals and contested in humans.

Can memories be erased? Not in humans, on current evidence. The strongest human studies show a reduced emotional or physiological response to a memory while the person still remembers what happened, and even those results have replicated inconsistently.

How long is the reconsolidation window? In Schiller's 2010 study, extinction ten minutes after the reminder made a difference and six hours after did not. That timing comes from one study that later failed a large replication, so treat it as a rough indication, not a figure.

Is this how EMDR works? Possibly in part; it is not established. Reconsolidation is one proposed mechanism among several. EMDR is recommended for PTSD by NICE and the US National Center for PTSD, while how it works remains debated.

Can propranolol erase trauma? No. One trial found it helped when taken before recalling the trauma; another with the same schedule showed no overall difference from placebo. It is experimental for this use, and any medication question belongs with a doctor.

Can I use this on myself? Not reliably. The conditions are narrow, and deliberately reactivating a traumatic memory without support is more likely to distress you than to help.

Sources and further reading

Curated authoritative references. More pieces written the same way, with the limits left in, are in what the research actually says. ILTY is a mental-health support tool, not a substitute for professional care. If you are in crisis, call or text 988 in the US, 9-8-8 in Canada, or find your country's line at findahelpline.com.

ILTY is not a trauma therapy and does not do reconsolidation work. If a memory is traumatic, that work belongs with a trained therapist, and AI therapy vs real therapy explains where an app stops. For the everyday version, the argument you keep replaying or the decision you keep relitigating, Mr. Relentless will ask the concrete question you have been avoiding, so the retelling has a chance to come out different.

Download ILTY

Memory Reconsolidation: Why Remembering Something Can Change It