We all have memories that we would rather forget, some embarrassing and some painful. Most of us learn to cope with these memories, but this can be difficult for those suffering from post-traumatic stress disorder (PTSD), in which reminders of traumatic events can trigger dread and fear. One way to counteract such associations is to extinguish them with repeated exposures to traumatic reminders within a safe environment. However, because this extinction process does not eliminate the fear memory, fear reactions often return, especially during stress.

An alternative approach to treating PTSD was suggested by studies1 in rodents. These studies showed that the mere retrieval of a memory triggers a reconsolidation process, during which the memory briefly becomes labile before being re-stored. Drugs that block reconsolidation can degrade the original fear memory in animals, but it has been difficult to apply such a strategy to humans because most of these drugs are toxic. On page 49 of this issue, Schiller et al.2 report that giving extinction training to humans during the reconsolidation window effectively redefines fearful memories as safeFootnote 1.

It has been known for a century that memories must undergo a consolidation process in order to be stored in the long term3. An exciting and more recent discovery was that retrieving a memory triggers a reconsolidation process that employs many of the molecular mechanisms used in the original consolidation. Reconsolidation has been observed for several types of memory across different species4, but these studies beg the question “what is the advantage of remaking a memory multiple times?” One possibility is that reconsolidation allows memories to be updated in the light of events that have occurred since the last retrieval5.

This hypothesis was recently tested6 by some of the authors of the present study. In rats conditioned to fear a tone paired with an electric shock, the researchers observed that extinction training conducted within 10 minutes of memory retrieval eliminated fear of the tone, and prevented the fear from returning under a variety of circumstances, even if a reminder shock was administered. Furthermore, rats in which fear memories had been extinguished within this critical window were resistant to re-learning the tone–shock association. So, instead of forming a new memory of safety, extinction training given during the updating window apparently converted the existing memory of fear into one of safety.

These findings6 in rodents raised questions about human memory. Does the retrieval of fear memories in humans trigger similar updating windows in which the memories could be modified by extinction training? If so, then how specific and long lasting are the memory-editing effects of reconsolidation–extinction procedures? To find out, Schiller et al.2 used a well established fear-conditioning protocol, in which human volunteers learned that the appearance of a visual cue (a blue square) predicts a shock to the wrist (Fig. 1). The authors used the participants' skin conductance as a measure of their fear — skin conductance increases with sweating, a phenomenon exploited by lie detectors.

Figure 1: Preventing fear's return.
figure 1

a, Schiller et al.2 conditioned human volunteers to fear a visual cue paired with a mild electric shock. They then extinguished the fear using extinction training, in which the cue was repeatedly presented in the absence of the shock. Fear responses extinguished in this way returned the following day when the participants were given a reminder shock. The graph shows the magnitude of a participant's fear response during the three parts of the experiment. b, Volunteers who received extinction training shortly after a retrieval trial (red line), which causes the subject to recall the fear memory, exhibited no fear response after being given a reminder shock the next day. This is because the retrieval of a memory triggers a reconsolidation window during which the memory can be updated by extinction training.

As in rodents, humans who recalled their fear memory 10 minutes before extinction training showed no fear response when tested 24 hours later. Furthermore, their fear responses did not return even if they were given a reminder shock at the start of the test day — they reacted to the blue square as if it had never been associated with a shock. By contrast, those participants whose memories were extinguished outside the critical window exhibited high fear responses when tested 24 hours later.

So are the memory-editing effects of extinction specific to the reactivated memory, or do they generalize to other memories? To address this issue, Schiller et al. conditioned volunteers to fear two stimuli, a blue square and an orange square, but only reactivated the fear memory of the blue square in a reminder trial. The participants underwent extinction training for both stimuli and were then administered a shock. This caused a return of their fear only in response to the orange square, confirming that the memory-editing effect of the reconsolidation–extinction procedure was specific to the reactivated memory.

Finally, to investigate the longevity of the effect, the authors brought a sample of their volunteers back to the lab one year later, and gave them a reminder shock. Remarkably, those who had received extinction training within 10 minutes of the reminder trial the year before continued to be immune to the shock. Taken together, Schiller and colleagues' results thus show that updating windows exist in humans, that the effects of extinction training during this window are stimulus-specific, and that the effects last for an extended period not commonly observed for other experiments in this field.

Schiller et al. studied healthy volunteers, but an exciting possibility is that their findings might be useful for the treatment of anxiety disorders such as PTSD. Current therapies involve extinction-based exposure to memory cues, but because extinction training is less effective in PTSD7, pharmacological methods are being explored to augment fear extinction8, or to block fear reconsolidation9,10. The obvious advantage of Schiller and colleagues' reconsolidation–extinction method is that no drugs are required, only a modification of the timing of standard exposure therapy.

There are, however, several issues that need to be carefully examined with regard to the potential clinical efficacy of this approach2. The aversive stimulus used in the study was a mild electric shock, which might have quite distinct effects from the kind of life-threatening events that lead to PTSD. Furthermore, it is not clear whether Schiller and colleagues' method would be effective for modifying fear memories acquired months or years before extinction training, rather than in the 24-hour period of their experiments. Finally, PTSD is a complex disorder that involves symptoms such as avoidance of traumatic reminders, emotional numbing, nightmares, flashbacks and sleep disturbances. The extent to which all these symptoms depend on aversive associations that are susceptible to editing remains to be determined. Nevertheless, Schiller and colleagues' findings are an exciting development that paves the way for mechanistic studies, in both rodents and humans, to discover how memory retrieval prepares fear circuits for updating by extinction training.