Multiple hypotheses explain the possible connections between sleep and learning in humans. Research indicates that sleep does more than allow the brain to rest. It may also aid the consolidation of long-term memories.
REM sleep and slow-wave sleep play different roles in memory consolidation. REM is associated with the consolidation of nondeclarative (implicit) memories. An example of a nondeclarative memory would be a task that we can do without consciously thinking about it, such as riding a bike. Slow-wave, or non-REM (NREM) sleep, is associated with the consolidation of declarative (explicit) memories. These are facts that need to be consciously remembered, such as dates for a history class.
Popular sayings can reflect the notion that remolded memories produce new creative associations in the morning, and that performance often improves after a time-interval that includes sleep. Current studies demonstrate that a healthy sleep produces a significant learning-dependent performance boost. The idea is that sleep helps the brain to edit its memory, looking for important patterns and extracting overarching rules which could be described as 'the gist', and integrating this with existing memory. The 'synaptic scaling' hypothesis suggests that sleep plays an important role in regulating learning that has taken place while awake, enabling more efficient and effective storage in the brain, making better use of space and energy.
Healthy sleep must include the appropriate sequence and proportion of NREM and REM phases, which play different roles in the memory consolidation-optimization process. During a normal night of sleep, a person will alternate between periods of NREM and REM sleep. Each cycle is approximately 90 minutes long, containing a 20-30 minute bout of REM sleep. NREM sleep consists of sleep stages 1–4, and is where movement can be observed. A person can still move their body when they are in NREM sleep. If someone sleeping turns, tosses, or rolls over, this indicates that they are in NREM sleep. REM sleep is characterized by the lack of muscle activity. Physiological studies have shown that aside from the occasional twitch, a person actually becomes paralyzed during REM sleep. In motor skill learning, an interval of sleep may be critical for the expression of performance gains; without sleep these gains will be delayed (Korman et al., 2003).
Procedural memories are a form of nondeclarative memory, so they would most benefit from the fast-wave REM sleep. In a study,procedural memories have been shown to benefit from sleep (Walker et al., 2002, as cited in Walker, 2009). Subjects were tested using a tapping task, where they used their fingers to tap a specific sequence of numbers on a keyboard, and their performances were measured by accuracy and speed. This finger-tapping task was used to simulate learning a motor skill. The first group was tested, retested 12 hours later while awake, and finally tested another 12 hours later with sleep in between. The other group was tested, retested 12 hours later with sleep in between, and then retested 12 hours later while awake. The results showed that in both groups, there was only a slight improvement after a 12-hour wake session, but a significant increase in performance after each group slept. This study gives evidence that REM sleep is a significant factor in consolidating motor skill procedural memories, therefore sleep deprivation can impair performance on a motor learning task. This memory decrement results specifically from the loss of stage 2, REM sleep.