REM sleep replays emotional memories, but replay alone may not strengthen them
During sleep, the brain spontaneously reactivates memory traces, including emotional ones. One line of evidence indicates that emotional memory traces are replayed during sleep to enhance later retention [4]. In a study using auditory cues to reactivate emotional memories during REM sleep, researchers observed enhanced cortical responses to those memories during subsequent retrieval [4]. This suggests that reactivation during REM can make emotional memories more accessible, at least in terms of brain response.
However, a different study using targeted memory reactivation found that reactivating emotional stimuli during REM sleep actually impaired memory [3]. In that daytime-nap experiment, cueing emotional items during REM led to worse recall, and the percentage of time spent in REM was correlated with worse recall in a group that napped without any cueing [3]. The same study found that emotional memories benefited from reactivation during slow-wave sleep, especially when participants had a high product of slow-wave and REM sleep time [3]. Taken together, these findings suggest that REM sleep is not simply a strengthening stage for emotional memories; its effect may depend on what happened in other sleep stages first.
REM sleep can also suppress fear memories through theta rhythms
A biophysically based model of REM sleep examined how rhythmic interactions between the medial prefrontal cortex and limbic structures process fear memories [1]. The model focused on the infralimbic cortex, a part of the medial prefrontal cortex that is thought to play a critical role in suppressing fear memories [1]. When theta-frequency inputs (4–12 Hz) reached a cell assembly in the infralimbic cortex representing an emotional memory, connections from the infralimbic cortex to the amygdala strengthened while connections from the amygdala to the infralimbic cortex weakened [1]. This pattern suppressed the activity of fear expression cells associated with that memory [1].
The model also found that lower-frequency theta inputs (4 Hz) produced these synaptic changes over a wider range of input strengths, while inputs at other frequencies were ineffective [1]. Under simulated PTSD REM sleep conditions, rhythmic activity dissipated and 4 Hz theta inputs became ineffective, but higher-frequency 10 Hz theta inputs induced changes similar to those seen with 4 Hz inputs under normal conditions [1]. These results suggest a possible mechanism by which REM sleep could weaken fear memories rather than strengthen them, and they may help explain why PTSD patients repeatedly experience the same emotionally charged dreams [1]. The model is a simulation, not a direct measurement in human brains, so the proposed mechanism remains a hypothesis.
Reward circuits in the ventromedial prefrontal cortex are involved in REM-related stabilization
A study of visual perceptual learning investigated how reward-processing circuits contribute to sleep-related stabilization [5]. Participants were trained on texture discrimination tasks before and after sleep, and researchers measured the balance between excitatory and inhibitory neurotransmitter concentrations in the ventromedial prefrontal cortex, a part of reward-processing circuits [5]. During REM sleep, this excitatory/inhibitory balance decreased relative to baseline, and the amount of decrease was correlated with the degree to which pre-sleep learning was stabilized [5]. No such change occurred during non-REM sleep [5]. This suggests that the ventromedial prefrontal cortex, as part of reward-processing circuits, is involved in stabilizing learning during REM sleep, although the study examined visual perceptual learning rather than emotionally salient memories [5].
The same study notes that reward provided during training prolongs subsequent REM sleep and strengthens visual perceptual learning after sleep [5]. This is consistent with the idea that reward-related signals can influence how REM sleep processes recently acquired information, but it does not directly show that emotional or reward circuit activations strengthen high-value emotional memories. The link between reward circuits and emotional memory consolidation during REM remains suggestive rather than established.
The hippocampus coordinates with other regions during REM, but the emotional strengthening story is incomplete
A review of the hippocampus in emotional memory consolidation during sleep emphasizes coordination between dorsal and ventral hippocampal regions during both REM and non-REM sleep [2]. It notes that remapping toward rewards and neurons directly encoding reward increase from REM sleep, and that studies in humans point toward a role for REM sleep in emotional memory consolidation [2]. This fits with the broader picture that REM sleep engages reward-related and emotional circuits, but the review does not specify a single mechanism by which high-value memories are strengthened.
Across these studies, the evidence supports a nuanced answer: REM sleep can reactivate emotional and reward-related circuits, and this reactivation is associated with changes in cortical responses and learning stabilization [4][5]. But reactivation during REM alone may impair rather than enhance emotional memory, unless it follows slow-wave sleep reactivation [3]. And in the case of fear memories, REM-like theta activity may suppress rather than strengthen them [1]. The practical implication is that the strengthening of high-value memories during REM sleep is not a simple amplification; it depends on the type of memory, the prior sleep stages, and the specific circuits involved.
Sources used in this answer
Emotional Memory Processing during REM Sleep with Implications for Post-Traumatic Stress Disorder
In a biophysically based model, theta-frequency inputs to the infralimbic cortex during REM sleep strengthened infralimbic-to-amygdala connections and weakened amygdala-to-infralimbic connections, suppressing fear expression cells; under simulated PTSD conditions, 4 Hz inputs were ineffective but 10 Hz inputs produced similar suppression.
The role of the hippocampus in the consolidation of emotional memories during sleep
A review emphasizes coordination between dorsal and ventral hippocampal regions during REM and NREM sleep, notes that remapping toward rewards and reward-encoding neurons increase from REM sleep, and points toward a role for REM sleep in emotional memory consolidation in humans.
Emotional memories are enhanced when reactivated in slow-wave sleep but impaired in REM
In a daytime-nap study using targeted memory reactivation, reactivating emotional stimuli during REM sleep impaired memory, whereas reactivation during slow-wave sleep enhanced memory, especially in participants with high combined slow-wave and REM sleep time.
Sleep, cognition and emotion in adults and adolescents
Using fMRI and intracranial EEG, reactivating an emotional memory trace during REM sleep with auditory cues enhanced cortical responses to those memories during subsequent retrieval, and specific oscillations triggered reactivation particularly in the orbitofrontal cortex.
The stabilization of visual perceptual learning during REM sleep involves reward-processing circuits
During REM sleep, the excitatory/inhibitory balance in the ventromedial prefrontal cortex decreased relative to baseline, and the amount of decrease correlated with the degree of stabilization of pre-sleep visual perceptual learning, suggesting involvement of reward-processing circuits.