What targeted memory reactivation actually does
Targeted memory reactivation, or TMR, is a way to nudge the sleeping brain to replay a specific memory. Before sleep, a person learns something paired with a sensory cue—say, a sound or a word. Then, during sleep, that cue is quietly played again. The idea is that the cue reactivates the associated memory, and the sleep stage during which this happens may shape what the brain does with it [1].
In the study that directly tested emotional arousal, participants' subjective arousal responses were measured across a night after TMR was applied in either REM or slow-wave sleep. Only the REM-timed cues significantly decreased arousal, and the effect was strongest for the more negative stimuli [1]. That is a specific, measurable change in how emotionally charged the memories felt—not just how well they were remembered.
Why REM sleep might be the stage that takes the edge off
One prominent hypothesis—sleep to forget, sleep to remember—proposes that reactivating emotional memories during REM sleep underlies sleep-dependent habituation, meaning the gradual blunting of emotional responses [1]. The finding that REM-timed TMR reduced arousal, but slow-wave-timed TMR did not, is consistent with that proposal [1]. In plain terms, REM may be a state where the brain revisits an emotional memory without the full emotional response, allowing the charge to fade.
It is important to separate this proposed function from what was actually observed. The study measured subjective arousal, not a neural mechanism, and the authors describe the result as supporting one aspect of the hypothesis rather than proving it [1]. So the most accurate statement is that REM reactivation was followed by reduced arousal in this experiment—not that REM sleep is established as the cause of emotional blunting in general.
Reducing arousal is not the same as weakening the memory
A separate line of work shows that reactivating emotional memories during sleep can affect memory strength in ways that differ by sleep stage. In a daytime-nap study, reactivation during slow-wave sleep enhanced memory, while reactivation during REM impaired it [3]. That study focused on memory performance and did not measure a reduction in emotional arousal responses, so it does not directly answer the arousal question [3].
These two results are not contradictory: one is about how vividly or strongly a memory is recalled, the other about how emotionally arousing it feels. The nap study also found that emotional memories benefited from TMR more than neutral ones, and that the emotional valence of cued items modulated sleep spindles and delta/theta power [3]. That suggests sleep is doing several things at once—preserving some memories, letting others fade, and possibly reshaping their emotional tone—and the stage and timing matter for which outcome dominates.
Other sleep stages and other kinds of learning
TMR during REM has also been linked to benefits outside emotion. In a whole-body procedural learning task, learning improved most when TMR was applied in REM sleep compared with a control sleep group, and performance was greater when task-related dream imagery also occurred in REM [5]. That study did not examine emotional arousal, so it broadens the picture of what REM reactivation can do without speaking to the arousal question [5].
Meanwhile, work on emotional memory modulation during slow-wave sleep has focused on enhancing therapy-like memory change, but it has not compared REM with slow-wave sleep, so it cannot tell us whether REM is uniquely suited to reducing arousal [4]. And a study of oscillatory correlates during NREM versus REM examined memory consolidation and brain rhythms but did not assess changes in emotional arousal responses [2]. Taken together, the direct evidence for the REM-specific arousal reduction rests on one experiment, while the surrounding studies show that TMR effects are real but depend heavily on what is measured and when.
Sources used in this answer
Targeted memory reactivation in REM but not SWS selectively reduces arousal responses
TMR during REM sleep, but not slow-wave sleep, significantly decreased subjective arousal, driven by more negative stimuli, supporting one aspect of the sleep-to-forget, sleep-to-remember hypothesis.
Emotional arousal modulates oscillatory correlates of targeted memory reactivation during NREM, but not REM sleep
TMR during REM and NREM sleep was studied with oscillatory correlates of memory consolidation, but emotional arousal responses were not assessed.
Emotional memories are enhanced when reactivated in slow-wave sleep but impaired in REM
During a daytime nap, TMR enhanced emotional memory in slow-wave sleep but impaired it in REM, and emotional memories benefited from TMR more than neutral ones; arousal responses were not measured.
Improving emotional memory modulation with targeted memory reactivation during slow wave sleep
A planned study of TMR during slow-wave sleep after imagery rescripting aims to measure emotional memory characteristics including arousal, but does not compare REM with slow-wave sleep.
Whole-body procedural learning benefits from targeted memory reactivation in REM sleep and task-related dreaming
TMR in REM sleep benefited whole-body procedural learning, especially when accompanied by task-related dream imagery, but emotional arousal responses were not examined.