The mystery of sleep

“What role might sleep play in supporting memory function?”

Ever wondered why we sleep? Scientists still aren’t 100% clear on the answer. But it must be really important – why else would we have evolved to spend hours each day in a vulnerable, unconscious state? There are a few theories on the role of sleep, and a key one is that it supports memory function. This essay reviews the main hypotheses about how sleep does this.

tenor (7).gif

I found this essay quite difficult. One major challenge was that I spent so long reading about this fascinating topic that I ended up with 55 A4 pages of notes. So when I tried to turn all my thoughts into 1500 words, I couldn’t possibly convey all the ideas I wanted to! I submitted the essay feeling quite dissatisfied (not to mention a little annoyed at myself).

Another challenge was that I found that reading loads of information about sleep actually made me feel really sleepy! And the studies I found seriously encouraged me to give in to the temptation to have a nap every few hours to ‘consolidate my learning’…

Despite these two major challenges, I managed to scrape a first class mark by the skin of my teeth. I celebrated with an early night.

Essay Summary

(If you fancy jumping straight into the real thing instead, just keep scrolling…)

  • Sleep is really complex. Far from just being either ‘asleep’ or ‘awake’, humans have different stages of sleep: rapid-eye-movement (REM) and 3 stages of non-rapid-eye-movement (NREM) sleep. The sleep stages are really quite different – both in terms of the electrophysiological patterns and neurochemicals found across the brain, and in terms of when each stage occurs during the night. The essay argues that if we really want to understand the role sleep plays, we need to study the role of these different characteristics.
  • Sleep plays a role in consolidation of memories. When memories are first created, they’re quite fragile, and easy to forget. Memories become stronger through a process called consolidation. There’s evidence to suggest that sleep actively plays a role in this.
  • Scientists have been trying to figure out how the different sleep stages play a role in memory consolidation. This has led to 3 key hypotheses:
    • The dual-process hypothesis says that NREM and REM consolidate different types of memories. There’s some studies suggesting this could be the case, but lots of contradictory findings too.
    • The sequential hypothesis says that NREM and REM play a complementary role, and need to occur in sequence for memories to be consolidated. There’s some evidence supporting this idea, but few studies have tested it directly.
    • The active system consolidation hypothesis says that, during sleep, new memories are re-activated and transferred from short-term memory stores to longer-term memory stores. This hypothesis is compelling because it takes into account the characteristics of specific sleep stages, but further work is needed to test it.
  • It’s not yet clear exactly how sleep supports memory function! Different studies report different findings. This, in part, might be because the methods used to study sleep aren’t perfectly accurate, and ignore the full complexity of sleep stages.
  • There’s lots more to understand about the mystery that is sleep…

If the essay looks too science-y, but the summary has whet your appetite, you might be interested in these:

  • ‘Why we sleep’ – a book by Professor Matthew Walker (a sleep scientist). Written for the general public, this describes the challenges of our modern attitude to sleep, and the benefits of getting a good night’s kip. Prof Walker argues that sleep is more important for our health than diet and exercise. Crikey. I wonder whether he also struggled to stay awake when he was writing it…
  • A clip from one of my favourite TV shows, BBC’s QI, where they briefly discuss some of the theories of sleep and what the latest view is. Sleep’s role in memory function is an example of what Sandi describes as the brain needing to do some activities ‘offline’ – “think of it as a house party… you can either entertain the guests or clean up the house, but you can’t really do both at the same time”. (Warning – they do get quite off-topic…)

If you’re up for the full essay, read on…

What role might sleep play in supporting memory function?

We all do it (hopefully!) every day. But what is sleep actually for? Evidence suggests it plays a role in memory.


Sleep is observed in a myriad of animals, from the fruit fly to the human (Tobler, 2000); characterised by reduced responsiveness to the external environment. Memory, the ability to store and recall information, is similarly a cross-species phenomenon. Thought to occur in some form in all living organisms, it underpins the ability to adapt behaviour to the changeable environment (Rasch & Born, 2013). This essay will review prominent hypotheses regarding sleep’s role in supporting human memory function. It will argue that, to fully understand sleep’s role, researchers must both overcome common methodological limitations, and adopt a more integrated approach that accounts for the complex nature of sleep.

Sleep’s complex nature

Human sleep consists of two main types: rapid eye movement (REM) sleep and non-rapid eye movement (NREM) sleep (Azerinsky & Kleitman, 1955). NREM sleep is further sub-divided into NREM1-3, with NREM3, the deepest sleep, referred to as slow-wave sleep (SWS) (Berry et al., 2012). The sleep stages differ in their temporal distribution, electrophysiology, and neurochemical environment (Peigneux, Laureys, Delbeuck, & Maquet, 2001).

REM and NREM sleep are distributed in alternating cycles throughout the night, with the first half of the night containing more NREM sleep, and the second more REM sleep (Payne, 2010). Sleep stages are marked by specific electrophysiological oscillations. For example, SWS exhibits neocortical slow oscillations, thalamo-cortical spindles and hippocampal sharp-wave ripples (Rasch & Born, 2013). Finally, four key neurochemicals vary significantly between sleep stages. Acetylcholine is almost absent during SWS but around waking levels in REM sleep, whereas serotonin and noradrenaline show the opposite trend (Hobson, Pace-Schott, & Stickgold, 2000). Cortisol follows a circadian rhythm, being lowest at sleep onset and increasing throughout the night, with cortisol peaks coinciding with REM sleep (Wagner & Born, 2008). Sleep is evidently a heterogeneous entity, necessitating an integrated research approach.

Screenshot 2018-05-30 16.32.25
The complexity of sleep. The sleep stages (REM and NREM1-3) differ in their temporal distribution throughout the night, their electrophysiology, and their neurochemical environment. Image from Rasch & Born, 2013.

Sleep plays a role in memory consolidation

Building on observations that post-learning sleep decreased forgetting (Ebbinghaus, 1885), an early hypothesis proposed a passive role of sleep whereby it protected recently encoded memories against retroactive interference, through reduced conscious awareness of the external environment (Jenkins & Dallenbach, 1924). Whilst it’s possible that sleep does play this role in memory function, the idea that it is the only role is heavily criticised (Ellenbogen, Payne, & Stickgold, 2006). Evidence against the idea includes the demonstration that the improvement effect of sleep on a word recall task was greater when participants learned an interfering word list immediately before testing (Ellenbogen, Hulbert, Jiang, & Stickgold, 2009), suggesting that sleep promotes consolidation of memories rather than only providing temporary protection from interference.

During TMS, a coil generates a magnetic field, inducing an electric current in the brain.

Many behavioural studies now support the hypothesis that sleep plays an active role in memory consolidation (Wagner, Gais, & Born, 2001; Tucker et al., 2006; Walker et al., 2003), and a recent study directly demonstrated this using a Transcranial Magnetic Simulation (TMS) protocol to induce neural plasticity similar to that seen after learning. Results showed consolidation of the plasticity increased after a nap compared to an equal period of wakefulness (Maier & Nissen, 2017).


The role of sleep stages

Much research has focused on elucidating the role of sleep stages in memory consolidation, resulting in three predominant hypotheses.

  • The dual-process hypothesis

This hypothesis asserts that NREM sleep and REM sleep promote consolidation of different memory types, specifically that SWS promotes consolidation of declarative memories (hippocampus-dependent memories that are explicitly recalled), while REM sleep promotes consolidation of non-declarative memories (hippocampus-independent memories that can be implicitly recalled) (Schacter & Tulving, 1994; Smith, 2001). A seminal study supporting this hypothesis employed the split-night paradigm (Yaroush, Sullivan, & Ekstrand, 1971) to demonstrate that recall on a declarative memory task improved significantly more after 3 hours of early-night SWS-rich sleep than 3 hours of late-night REM-rich sleep or 3 hours spent awake (Plihal & Born, 1997). Conversely, recall on a non-declarative memory task improved significantly more after the same period of REM-rich sleep than of SWS-rich sleep or wakefulness.

However, there have been many inconsistent findings. For example, one study using a non-declarative visual texture-discrimination task found performance was unaffected after late REM-rich sleep deprivation, but was affected by early SWS deprivation and additionally by total sleep deprivation (Gais, Plihal, Wagner, & Born, 2000), contradicting a previous study showing the same task was most affected REM sleep deprivation (Karni, Tanne, Rubenstein, Askenasy, & Sagi, 1994). Other studies have shown non-declarative tasks benefitting from SWS-rich sleep (e.g. Stickgold, Whidbee, Schirmer, Patel, & Hobson, 2000), and REM sleep deprivation affecting performance in complex declarative memory tasks (Empson & Clarke, 1970).

  • The sequential hypothesis

Some findings inconsistent with the dual-process hypothesis can be interpreted as supporting the sequential hypothesis (e.g. Stickgold et al., 2000; Gais et al., 2000). This hypothesis states that NREM sleep and REM sleep play complementary roles in memory consolidation, and must occur in sequence as part of a two-step process (Ambrosini & Giuditta, 2001). Proponents argue that observations from split-night studies are not due to the amount of SWS and REM sleep in each night half, but instead the different distributions of NREM/REM sleep cycles found in early and late sleep (Ficca & Salzarulo, 2004). Several behavioural studies support the hypothesis (Casey et al., 2016), for example one study demonstrated the importance an organised sleep cycle for word pair recall (Ficca, Lombardo, Rossi, & Salzarulo, 2000), while another showed a 90-minute nap containing REM and NREM sleep improved memory task performance, while a 60-minute nap containing only NREM sleep did not (Mednick, Nakayama, & Stickgold, 2003). However, few studies have directly tested the sequential hypothesis (Rasch & Born, 2013).

  • The active system consolidation hypothesis

This hypothesis seeks to explain the neural mechanisms of sleep’s role in memory consolidation, and thus incorporates sleep’s neurochemical and electrophysiological characteristics (Rasch & Born, 2013). It suggests that, during sleep, memories encoded in a temporary store are reactivated and redistributed to different brain regions for long-term storage (Frankland & Bontempi, 2005). This active system consolidation is thought to depend on the specific neurochemical environment, particularly during SWS (Plihal & Born 1999; Gais & Born 2004). The proposed mechanism for declarative memories involves hippocampal sharp-wave ripples and thalamo-cortical sleep spindles, most associated with SWS, driving the redistribution of memories from the hippocampus to the neocortex, coordinated by neocortical slow oscillations (Corkin, 2002; Takashima et al., 2006). Redistributed memories are proposed to be stabilised by synaptic consolidation during REM sleep (Rasch & Born, 2013).

Screenshot 2018-05-30 18.34.46
A model of active system consolidation during sleep. From Rasch & Born, 2013.

The plethora of studies showing reactivation of newly-encoded memory traces during post-learning SWS support this hypothesis. First observed in rat hippocampal place cells (Wilson & McNaughton, 1994), this reactivation has been observed in humans and associated with enhanced performance in related memory tasks (Peigneux et al., 2004). While reactivation studies do not directly demonstrate consolidation, a recent rodent study showed that post-learning sleep supported formation of new branch-specific synapses (Yang et al., 2014). Other studies focus on providing support for the proposed mechanisms by manipulating electrophysiological (Marshall, Helgadottir, Molle, & Born, 2006) or neurochemical (Gais & Born 2004) features of sleep. Further work is required to test the hypothesis, particularly for non-declarative memories.

Methodological Limitations

The literature on sleep’s role in memory consolidation is littered with discrepant findings, which impede validation of current hypotheses. This may be partly due to methodological limitations. For example, the finding that SWS benefits non-declarative memory tasks could be interpreted as evidence against the dual-process hypothesis, or as reflecting declarative elements of these tasks. Indeed, some argue that no task is process-pure (Jacoby, 1991). To overcome this, and other limitations of behavioural data, protocols such as TMS and transcranial direct current stimulation (tDCS) can be employed, to directly manipulate electrophysiology in the brain (e.g. Marshall, Mölle, Hallschmid, & Born, 2004).

Such protocols could also overcome limitations of many sleep paradigms. For example, the split-night paradigm has been criticised for ignoring possible roles of NREM2 sleep, which is distributed nearly equally across the night (Tucker et al., 2006) and is implicated in motor memory (Smith & MacNeill, 1994). Other paradigms subject participants to sleep deprivation or repeated waking, which can introduce stress as a confounder to task performance (Peigneux et al., 2001).

Many paradigms insufficiently account for the complex nature of sleep, by focusing on sleep stages, rather than the neurochemical and electrophysiological characteristics of sleep. For example, acetylcholine, serotonin, noradrenaline and cortisol have all been shown to play a role in memory function, and are not perfectly correlated to the sleep stages (Payne, 2010). Similarly, neocortical slow oscillations, thalamo-cortical spindles and hippocampal sharp-wave ripples are all associated with declarative memory consolidation (Payne, 2010). Based on the observation that there are huge variations in the sleep of healthy adult humans (Appleman, Albouy, Doyon, Cronin-Golomb, & King, 2016), a focus on isolating the roles of the neurochemical and electrophysiological markers of sleep, rather than sleep stages, offers a promising future direction.


Sleep’s role in supporting memory function remains slightly mysterious. A growing literature gives inconsistent support to prominent theories, but strong support for sleep’s role in memory consolidation. In order to elucidate the underlying mechanisms, the field must embrace new protocols to overcome traditional methodological limitations of sleep and memory research, and adopt a more integrated approach that incorporates the neurochemical and electrophysiological complexity of sleep.



One thought on “The mystery of sleep

  1. I certainly agree with the writers conclusion, a more integrated approach may remove some of the mystery/conflict in the historic research. On a personal note, I certainly find a Power nap before playing bridge in the evening really helps my concentration and memory recall!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s