When I was in middle school, I had insomnia for three months. I struggled to fall asleep, and when I finally did, I had very vivid dreams. I would open my eyes and find myself floating in an ocean in my room. I could smell the saltiness of the water, the current gently lifting me, the waves rolling next to my body, and the cold wind blowing over me. I couldn’t move, yet I could sense and think. This frequent vivid dreaming lasted a few months, and each time I felt trapped. When I finally awoke, I assured myself that at least I got some sleep, despite how intense the experience was. Eventually my insomnia subsided, but even today the dreams visit me, especially when I am short on good quality sleep.
I know now that the cause of these dreams is in fact a rebound of a phase of sleep called rapid eye movement (REM) sleep. With poor sleep or sleep deprivation, REM sleep is often truncated, as it requires several stages of non-rapid eye movement (NREM) sleep to occur first. This loss of REM sleep leads to a rebound when sound sleep finally occurs, maintaining an apparent homeostasis of this sleep phase.
But why is REM sleep needed in the first place, and why is it so important that we get enough of it? And when it causes dreaming as lifelike yet surreal as my own, what purpose could it possibly serve to the survival of our species? The answer may lie in four key features of REM sleep in humans: eyes rapidly moving, dreams occurring, muscles unable to move, and NREM sleep always occurring first.
Why does REM involve rapid eye movement?
The REM phase was initially named due to the rapid eye movement behavior observed in sleeping humans, which does not occur during phases of NREM or wakefulness. One explanation is that eye movement reflects our changing scenes in the visual world of our dreams. This fact could indicate humans’ reliance on visual systems, as those who do not depend on the visual system might not display similar behavior. Indeed, congenitally blind humans with no visual exposure have a reduction in their rapid eye movements during REM sleep, and blind mole rats that evolved to have regressed visual systems also do not display eye movements during REM [1–2]. Likewise, whales that rely much more on sound for navigation and find it hard to move their eyes due to physiological limitations also barely move them during REM sleep [3]. Thus, it seems that this behavior is a trait of REM sleep that possibly pertains to the specific sensory systems we rely on in life, and which create detailed dreams in our sleep.
What is the purpose of dreaming during REM?
There are many proposed explanations for the purpose of dreams. First, dreams in REM sleep are thought to reduce the negative emotional components of memories, allowing one to remember their adverse experiences while not feeling as intensely upset by them. While there hasn’t been any direct causal evidence to support this claim, there is significant indirect evidence, including (1) evidence suggesting emotional dreams may predict better recovery from disorders like depression; (2) the observation that the one emotion-processing center, the amygdala, displays higher activity during REM sleep after viewing horror images than when those images are revisited; and (3) evidence that changes in REM sleep might be indicative of mood disorders [4]. Furthermore, according to this explanation, REM sleep mainly regulates negative emotions, which are proposed to play a more vital role in evolution and survival than positive emotions. Indeed, it has been shown that disrupting REM sleep can worsen depression and anxiety associated with negative experiences in humans [5]. By dissociating one’s emotion from their experience, the emotional component itself can be modulated and reduced independently [6].
An alternative hypothesis is that REM sleep drives dreaming to help consolidate memories of the previous day’s experiences. The primary evidence for this argument is derived from observations of impaired memory performance in rodents after depriving or disrupting REM sleep and its associated dreams [7]. However, suppressing REM sleep and thus dreaming in humans seems not to impair memory, and patients with lesions in a brain region called the pontine tegmentum, which leads to reduced REM sleep, do not experience memory impairment. Furthermore, dreams in the REM phase are often seen as random and seemingly irrelevant to reality. But this perception is biased towards our limited knowledge of dreams, as we only know about the subset that we consciously remember, which may be a small portion of our total dreaming. Dreams we don’t remember may in fact be less random or more relevant to our previous experiences than we think.
Another possible reason for dreaming in REM sleep has been attributed to the overfitted brain hypothesis [4] - this hypothesis asserts that repeated learning of a task reduces our ability to generalize that task, which is then mitigated by dreaming. In other words, we are more likely to dream about things that we repeatedly learn or think about during wakefulness, and this mechanism helps integrate new feelings and thoughts into our previous knowledge, expanding our ability to apply that knowledge to new scenarios. Besides human observations, this hypothesis is supported by neural network training, in which introducing noise (like dreams in humans) helps these networks avoid overfitting to training datasets and strengthens their ability to acquire and adapt to new inputs.
Finally, dreaming in REM might reduce noisy connections between neurons to help generalize learned information [8]. This hypothesis proposes that unnecessary neuronal synapses are pruned and eliminated during dreams [9–10]. But this proposal may not match our intuitions, as it often appears that dreams only introduce rather than filter out noise in our thoughts and perceptions. One might explain that the information needs to be activated to be eliminated, but more evidence is needed to substantiate this idea.
Why are muscles paralyzed during REM?
Losing strength in muscles, called muscle atonia, is another unique trait of REM sleep. A dominant explanation is that muscle atonia prevents animals from acting out their dreams. Another hypothesis posits that muscle atonia allows rejuvenation of spinal intervertebral discs, which constantly work against gravity during wakefulness and NREM sleep [11]. Supporting evidence for this claim includes the fact that REM sleep is dramatically shortened for astronauts in space, who of course don’t experience the gravitational forces we do on Earth [12]. Additionally, it appears that animals allow themselves to enter the REM phase when they feel safe. Completely relaxing muscles might be an eventual period of rest for those muscles to finally take a break. However, this relaxation may still pose a number of survival risks, which might explain why some animals like birds evolved to sleep with one leg standing [13].
This feature of REM sleep is so unique that when muscle atonia fails, such as when sleeping individuals act out during REM sleep, it is labeled as a pathology called REM Sleep Behavior Disorder (RBD). RBD is a core feature of neuronal degenerative diseases, likely linked to early damage or death of muscle-controlling neurons [14].
Lastly, why can REM only come after NREM?
Sleep begins with NREM sleep, then transitions to REM sleep in normal conditions. From REM, rodents do not transit back to NREM, while humans often do. However, without NREM sleep first, strangely no known species can enter REM sleep. If REM is such a necessary sleep stage, why can't it be directly entered to receive its maximal benefits?
The first reason might be that the neuronal circuit controlling the transition from NREM to REM represents a one-directional process determined by brain physiology. Indeed, studies have identified neurons that promote the transition from NREM to REM in rodents [15]. But why might this sequence have arisen in the first place?
It might be that NREM is the major resting period among brain states, while REM is a third state between NREM and wakefulness, concluding the rest and preparing for wakefulness [16]. Alternatively, transitioning directly from wake to REM may be dangerous, so the brain has evolved to set a gate or a switch before REM sleep which only NREM sleep can cross over. This switch is now known to be orexin, a neuropeptide signal in brain regions such as the tuberomammillary nucleus and the locus coeruleus, which is essential for maintaining wakefulness and suppressing REM sleep [15].
************
In sum, there are several features that make REM interesting and unique not only among sleep stages but among behavioral states overall. Just as diverse as animal behaviors during wakefulness, REM sleep differs greatly depending on what species we are discussing, let alone the fact that REM sleep can differ significantly under stress or pathological conditions. Ultimately, REM is likely a survival-oriented function that prepares or improves an animal’s response to uncertain challenges during wakefulness, regardless of which hypothesized mechanism it uses to achieve this goal.
With the development of new methods, it is exciting for scientists to identify and explain even more fascinating features of REM sleep, such as a recent discovery that brain waves compete between hemispheres during this phase [17], providing an even better understanding of the mysterious other world that we inhabit each and every night.
Edited by Nick Bulthuis
References:
J. A. E. Christensen, S. Aubin, T. Nielsen, M. Ptito, R. Kupers, and P. Jennum, "Rapid eye movements are reduced in blind individuals," J Sleep Res, vol. 28, no. 6, p. e12866, Dec 2019, doi: 10.1111/jsr.12866.
M. S. Blumberg, J. A. Lesku, P. A. Libourel, M. H. Schmidt, and N. C. Rattenborg, "What Is REM Sleep?," Curr Biol, vol. 30, no. 1, pp. R38-R49, Jan 6 2020, doi: 10.1016/j.cub.2019.11.045.
O. I. Lyamin and J. M. Siegel, "Sleep in Aquatic Mammals," Handb Behav Neurosci, vol. 30, pp. 375-393, 2019, doi: 10.1016/b978-0-12-813743-7.00025-6.
E. Hoel, "The overfitted brain: Dreams evolved to assist generalization," Patterns (N Y), vol. 2, no. 5, p. 100244, May 14 2021, doi: 10.1016/j.patter.2021.100244.
P. McNamara, S. Auerbach, P. Johnson, E. Harris, and G. Doros, "Impact of REM sleep on distortions of self-concept, mood and memory in depressed/anxious participants," J Affect Disord, vol. 122, no. 3, pp. 198-207, May 2010, doi: 10.1016/j.jad.2009.06.030.
M. Vandekerckhove and Y. L. Wang, "Emotion, emotion regulation and sleep: An intimate relationship," AIMS Neurosci, vol. 5, no. 1, pp. 1-17, 2018, doi: 10.3934/Neuroscience.2018.1.1.
J. Born, B. Rasch, and S. Gais, "Sleep to remember," Neuroscientist, vol. 12, no. 5, pp. 410-24, Oct 2006, doi: 10.1177/1073858406292647.
S. Izawa et al., "REM sleep-active MCH neurons are involved in forgetting hippocampus-dependent memories," Science, vol. 365, no. 6459, pp. 1308-1313, Sep 20 2019, doi: 10.1126/science.aax9238.
G. R. Poe, "Sleep Is for Forgetting," J Neurosci, vol. 37, no. 3, pp. 464-473, Jan 18 2017, doi: 10.1523/JNEUROSCI.0820-16.2017.
W. Li, L. Ma, G. Yang, and W. B. Gan, "REM sleep selectively prunes and maintains new synapses in development and learning," Nat Neurosci, vol. 20, no. 3, pp. 427-437, Mar 2017, doi: 10.1038/nn.4479.
J. C. Fryer, "Is a purpose of REM sleep atonia to help regenerate intervertebral disc volumetric loss?," J Circadian Rhythms, vol. 7, p. 1, Jan 5 2009, doi: 10.1186/1740-3391-7-1.
B. Wu, Y. Wang, X. Wu, D. Liu, D. Xu, and F. Wang, "On-orbit sleep problems of astronauts and countermeasures," Mil Med Res, vol. 5, no. 1, p. 17, May 30 2018, doi: 10.1186/s40779-018-0165-6.
N. C. Rattenborg, J. van der Meij, G. J. L. Beckers, and J. A. Lesku, "Local Aspects of Avian Non-REM and REM Sleep," Front Neurosci, vol. 13, p. 567, 2019, doi: 10.3389/fnins.2019.00567.
D. A. Barone and C. Henchcliffe, "Rapid eye movement sleep behavior disorder and the link to alpha-synucleinopathies," Clin Neurophysiol, vol. 129, no. 8, pp. 1551-1564, Aug 2018, doi: 10.1016/j.clinph.2018.05.003.
T. E. Scammell, E. Arrigoni, and J. O. Lipton, "Neural Circuitry of Wakefulness and Sleep," Neuron, vol. 93, no. 4, pp. 747-765, Feb 22 2017, doi: 10.1016/j.neuron.2017.01.014.
W. R. Klemm, "Why does rem sleep occur? A wake-up hypothesis," Front Syst Neurosci, vol. 5, p. 73, 2011, doi: 10.3389/fnsys.2011.00073.
L. A. Fenk, J. L. Riquelme, and G. Laurent, "Interhemispheric competition during sleep," Nature, vol. 616, no. 7956, pp. 312-318, Apr 2023, doi: 10.1038/s41586-023-05827-w.