This paper reflects the research and thoughts of a student at the time the paper was written for a course at Bryn Mawr College. Like other materials on Serendip, it is not intended to be "authoritative" but rather to help others further develop their own explorations. Web links were active as of the time the paper was posted but are not updated.
2003 Second Web Paper
All mammals exhibit Rapid-Eye-Movement, or REM, sleep, and yet on certain levels this type of sleep would seem to be disadvantageous. During REM sleep, which is when most dreams occur, the brain uses much more energy than during non-REM (NREM) sleep. (1) This "waste" of energy coupled with the increased vulnerability of this state on account of the body's paralysis at this time suggests that there must be a very important reason, or reasons, for the existence of REM sleep and in extension of dreams. Determining the function of dreams, however, has proved very problematic with many arguments which directly oppose each other. Some of the primary functions of dreaming have been tied to is role in development, its production of neuro-proteins, and also to how it may allow for the "rehearsal" of neurons and neuronal pathways. The influence of dreaming on learning is one of the hottest debates. Some argue that dreams aid in learning, others that dreams aid in forgetting, and yet others that dreams have no effect on learning or memory. That REM sleep seems to aid in development might argue that REM sleep may be connected to learning. It seems that most scientists believe that REM sleep aids in certain memory consolidations although some argue that it actually leads to "reverse learning.
Before discussing the role of NREM and REM in learning, it is necessary to clarify the identity of and differences between the two. This type of sleep is marked by different stages based on different the different brainwaves exhibited. REM sleep differs from NREM in that most dreams occur during REM sleep although the two activities are not synonymous. REM is also marked by an increase in brain activation, breathing , and also heart-rate while the body becomes paralyzed. (2)
Although the precise role may be arguable, REM sleep seems to play a role in development. Although newborn infants spend about half of their sixteen to eighteen hours of sleep time a day in REM sleep, adults spend only about an hour and a half in REM sleep. (1) This difference in both amount and percentage of REM sleep between infants and adults indicates the importance of REM sleep, or of dreams, in development. Several dream researchers have hypothesized that REM sleep may play an important role in infant brain development by providing an internal source of powerful stimulation which would prepare the baby for the almost infinite "world of stimulation it will soon have to face" and also by facilitating the "maturation of the nervous system." (1)
The relative amount of REM sleep in other mammals exhibits in connection with their level of development at birth also supports the idea that REM sleep must aid in development. (1) Typically, animals born relatively mature, such as dolphins, giraffes, and guinea pigs, demonstrate low-amounts of REM sleep, while animals born relatively immature, such as ferrets, armadillos, and platypuses, exhibit higher levels of REM sleep. (3) Humans fall in between the spectrum of amounts of REM sleep with platypuses having the most REM sleep and some species of dolphin and whale exhibiting none. (3)
Partially because infancy is the time when most new information must be taken in as a part of development, scientists have hypothesized the existence of a connection between REM sleep and learning. It seems that the time during infant development would be one of the times when most learning, or processing of information, occurs.
The most commonly held belief among the scientific community seems to be that REM sleep consolidates memories and aids in learning. An article in Science recently declared that "neuroscientists have long known that memory consolidation goes on during sleep." (3) A more recent discovery is that NREM sleep may also play a role, albeit a different one, in learning. Robert Stickgold from the Massachusetts Institute of Technology has found that different phases of sleep are tied to different types of learning. Learning visual skills depends on the slow-wave sleep of the first quarter of the night and also on the REM sleep of the last quarter. Learning movements relies much more on the NREM in the later part of the night. (4) In an important study in 2001, Matthew Wilson, also of MIT, found that rats dream about their activities (i.e. running through a maze) during NREM sleep in addition to in REM sleep. Unlike during REM replay, where the experience occurs approximately in real time, the memory segments that were replayed during NREM seemed to be snippets of experience. (5) Also, unlike REM sleep, slow wave sleep seemed to replay only what had happened immediately before and not something twenty-four hours ago. (5) Because of the possible time-delay REM memory reactivation, it might be representative of a more gradual reevaluation of slightly older memories. (5)
Most evidence for the memory consolidation hypothesis comes from indications that an increase in learning results in an increase of REM sleep, that memory processing occurs during REM sleep, and that sleep deprivation harms the ability to learn (3) In general it seems that having had enough REM sleep before and also after the learning of new information will help with the remembering of that information. The fact alone that REM sleep stimulates the learning region, the hippocampus, has been argued as an indication of REM's influence on learning. (6) Learning tasks that need high levels of concentration or the acquisition of new skills is followed by an increase in REM sleep. (1) Many studies show that learning after having reached a plateau can only take place with the help of REM sleep. A study at MIT has shown that volunteers' skill at key-tapping and speed-spotting tasks improved by 20 per cent after one night's sleep after training, and with more extra nights, it increased even more. (4) Karni and Sagi's establishing that changes in the plasticity of particular neuronal loci which underlie perceptual learning may happen during sleep, also argues for the importance of sleep to memory. (7) The increased production of proteins, which also occurs during deep sleep, may be tied to the learning process if those proteins are in fact associated with learning. (6) Finally, backing up the idea that suppression of REM affects memory consolidation is the study by Dinges and Kribbs that shows that REM deprivation impaired performance on longer tasks, while shorter tasks remained unimpaired. (5)
In apparent contradiction with the concept that REM sleep plays a part in remembering new information is the hypothesis that people actually dream to forget. Crick and Mitchison have proposed that "the function of dream sleep is to remove certain undesirable modes of interaction between cells in the cerebral cortex which could otherwise turn parasitic." (1) Their second hypothesis is that "if these hypothetical 'parasitic' modes of neuronal behavior do in fact exist, then it might be that they 'are detected and suppressed by a special mechanism." (1) Crick and Mitchison call this hypothetical process "reverse learning" or "unlearning," but explain that it "is not the same as normal forgetting." (1) They say that according to their model, "attempting to remember one's dreams should not be encouraged, because such remembering may help to retain patterns of thought which are better forgotten. These are the very patterns the organism was attempting to damp down." (1) Although dream-sleep may prevent "perpetual obsessions or spurious hallucinatory associations," Crick and Mitchison acknowledge that it would be difficult to test for the existence of the reverse learning mechanism. (1) Also it does not seem that people who routinely remember their dreams would be more prone to "hallucinations, delusions, and obsessions" than people who typically forget their dreams. (1) Doing a study to test for the existence of an increase in "parasitic" memories in individuals who take MAO inhibitors to treat depression, which block REM sleep, in comparison with people who have normal REM sleep cycles might help test out Crick and Mitchison's hypothesis.
Despite the many studies demonstrating the function of REM sleep in memory consolidation, there is debate about the validity of the claim that REM sleep aids in learning. For example, Jerome Siegel argues in Science that the evidence for the importance of the role of REM sleep in memory consolidation is contradictory and weak. (3) On the most basic level, the fact that people remember so few of their dreams may argue against their function in terms of learning (8) (3). He argues that learning does not result in an increase in REM-sleep. For example, one of his objections to research is that Smith's study, which documented the higher density of REM in college students after intensive exams, assumes that there can be a control group among humans, which is often very difficult to obtain. (3) Siegel also takes the lack of difference in the amounts of REM sleep time between students with average IQ's and those with high IQ's to prove the lack importance of REM sleep as a memory-consolidator. (3) The problem with this assertion is that it assumes that intelligence is easily quantifiable. His faith in a study involving humans also contradicts his previous objection to the Smith study. He argues against the deleterious effect of REM suppression by stating that the methods used on rats (i.e. the platform technique in which the rat wakes up whenever it starts REM sleep because it falls into water) results in an increased level of stress, which can explain the decrease in learning ability. (3)
Even if objections to the particulars of certain experiments may be valid, these objections can never prove that REM-sleep has no role in learning or memory-consolidation. Although the precise roles, which REM and NREM play, may not be understood fully, there seems to be a connection between both types of sleep with memory. REM sleep seems more conducive to the learning of extended or sequential information since "playback" periods of REM far outlast the bursts of "playback" during NREM sleep. (5) Although my feeling that REM sleep must aid in learning and memory-consolidation may be some-what instinctual, I cannot but help think about my uncle Eli Ginzberg, who wrote one-hundred-ten books (mostly on economics), always got at least nine hours of sleep a night. Although it seems fairly clear that sleep does play aid in memory-consolidation, what remains to be done is to keep testing REM sleep and the less-studied NREM sleep for evidence as to if and how it fulfills this task.
1)Lucidity Institute website, "Chapter 8: Dreaming: Function and Meaning,"
2)First Webpaper on Serendip, Great Neurological Resource
3)The REM Sleep-Memory Consolidation Hypothesis," article on Center for Sleep Research's homepage, Interesting site for sleep disorders
4)Nature website, good for scientific articles
5)MIT News website, interesting articles
6); TALK ABOUT SLEEP, Inc., basics answers about sleep
7)Harvard Undergraduate Society for Neuroscience, connected to Computer Science Program
8)UCSC Psych Website,
| Course Home Page | Course Forum | Brain and Behavior | Serendip Home |