Dreams and Emotional Regulation: Is REM Sleep Overnight Therapy?

Do dreams help us process emotions?

A serious line of sleep research says they might. The best-known version of the idea is that REM sleep lets the brain revisit emotional experiences under unusual chemistry, keeping the memory while softening the feeling attached to it. Matthew Walker and Els van der Helm called this possibility overnight therapy, and the question mark in their paper's title is doing honest work: it marks a hypothesis, not a finding.

This page walks through the main forms of the idea: the overnight therapy hypothesis from Walker's Berkeley lab, Rosalind Cartwright's divorce studies, Ernest Hartmann's theory of dreams that tie emotion into the rest of life, and a model that treats nightmares as this system failing. It also covers the counter-evidence, because there is real counter-evidence. Why we dream and what dreams mean has no scientific consensus, and emotional regulation is a good example of why.

Overnight therapy: sleep to forget, sleep to remember

According to Walker and van der Helm (2009), writing in Psychological Bulletin, REM sleep may decouple an emotional memory from its emotional charge: you keep what happened, but the sting fades. They called this the "sleep to forget, sleep to remember" hypothesis. The proposed mechanism is chemical: during REM, noradrenergic signaling drops away, so the brain can reactivate an emotional memory in a calmer setting than the one it was formed in.

A follow-up study from the same lab offered support. In 2011, van der Helm and colleagues reported in Current Biology that after a night of sleep, participants showed reduced amygdala reactivity and rated previously seen emotional images as less intense the next day, an effect that tracked the physiology of their REM sleep. It is one study, and later work has pushed back on it, as the counter-evidence section below sets out.

Walker's 2017 book Why We Sleep carried the idea to general readers, though some of the book's claims have since been challenged by other researchers, and it is worth noting what the 2009 review does and does not claim. It is a paper about REM sleep physiology, not about dream content: it does not claim that the dream you experience is the mechanism. That distinction turns out to be the hinge of the whole topic.

Rosalind Cartwright and the divorce studies

Rosalind Cartwright, a sleep researcher at Rush University who died in 2021, spent more than twenty years running NIMH-funded studies of people going through divorce, some of them clinically depressed. She wanted to know whether what people dreamed during a hard year had any bearing on how they came out of it.

Her 1991 paper in the journal Dreaming, "Dreams that work", reported that depressed divorcing participants who incorporated the ex-spouse into their dreams at the time of the breakup were less depressed and better adjusted at follow-up than those who did not. The people who dreamed about the problem did better than the people who dreamed around it.

In The Twenty-four Hour Mind (2010) she argued that thought and feeling run continuously from waking into sleep, and that dreaming, especially in REM, works over the emotions of the day and tempers our responses to new experience. The honest caveat is that her evidence is correlational. Dreaming about the stressor travels with better recovery. Whether it causes the recovery, or simply marks a mind already coping, no study of this design can say.

Hartmann: the dream that pictures the feeling

Ernest Hartmann, a Tufts psychiatrist who died in 2013, proposed what he called the contemporary theory of dreaming. In his account, dreaming connects new experience into wider memory networks, and emotion is the guide: the stronger the feeling, the more the dream organizes itself around it.

His signature observation was the Central Image, a single picture that carries the dream's underlying emotion. The paradigmatic case is the tidal-wave dream after trauma: the wave is not a replay of the event but a picture of being overwhelmed, and in Hartmann's dream series the intensity of such images tracked the strength of the emotion behind them. He saw this connecting work as quasi-therapeutic, something like what happens in psychotherapy.

Hartmann's evidence comes from clinical dream series and content analysis, not controlled experiments. On this site we treat his theory the way we treat Jung: a considered lens for reflection, not a demonstrated mechanism.

Nightmares: when regulation is said to fail

If dreaming regulates emotion, what is a nightmare? Levin and Nielsen (2007), in Psychological Bulletin, proposed that normal dreaming serves a fear-extinction-like function, and that nightmares are that function breaking down. Their model has two dials: affect load, the emotional pressure a person's days are putting on the system, and affect distress, a trait-like tendency to react to things with strong negative emotion.

The proposed picture in post-traumatic stress fits the same chemistry the overnight therapy hypothesis relies on, run in reverse. If noradrenergic tone stays high during REM, the normal overnight softening cannot happen, so the trauma memory keeps its charge and returns, night after night, as a nightmare. This is a coherent proposal, and it remains a proposal. The 2007 model drew a published critical commentary in the same journal issue.

A plain word of care belongs here. Nightmares are common and usually pass on their own. If yours recur, distress you, or follow a traumatic experience, talking to a professional is a reasonable step, not an overreaction.

The counter-evidence, and what it leaves open

The 2011 amygdala finding has not replicated cleanly. Baran and colleagues (2012) found that sleep preserved rather than reduced emotional reactivity, and that more time in REM sleep went with preserved rather than reduced reactivity, the opposite direction from the hypothesis. Some studies find the predicted effect and some find its reverse.

The widest view so far is a 2022 systematic review and meta-analysis by Lipinska and colleagues, covering 24 studies. It found that sleep did not reliably change physiological reactivity to negative stimuli, and that self-reported arousal to negative material was higher, not lower, after sleep. On current evidence, overnight emotional depotentiation is not a dependable laboratory result.

There is also an old puzzle the theory has to live with. Depression comes with more REM sleep, arriving earlier in the night, yet most antidepressants suppress REM without worsening mood, and Vogel reported in 1975 that three weeks of REM deprivation improved depression in severely depressed hospitalized patients about as well as a standard drug of the era. If REM were straightforwardly therapeutic, this would be hard to explain. One proposed reconciliation is that in depression, REM over-consolidates negatively toned memories.

The deepest open question is simpler than any of this: nearly every experiment in the field measures REM physiology, not the dream as experienced. What ties the regulation story to dreams themselves is the content evidence from Cartwright and Hartmann, and that evidence is correlational. Our interpreter treats emotional regulation as one honest lens among several, alongside threat simulation, continuity, and the Jungian tradition. Asking what feeling a dream carries, and what recent experience carries the same feeling, is a useful reflective question even while the science stays open.

Frequently Asked Questions

Go Deeper

Explore More Science