Santa Fe Institute Collaboration Platform

COMPLEX TIME: Adaptation, Aging, & Arrow of Time

Get Involved!
Contact: Caitlin Lorraine McShea, Program Manager,

What is Sleep?/Is sleep for remembering or forgetting?

From Complex Time

November 18, 2019
9:30 am - 10:30 am


Gina Poe (UCLA)Susan Sara (École des Neurosciences Paris Île de France)


Gina Poe – Sleep is for forgetting

It is possible that one of the essential functions of sleep is to take out the garbage, as it were, erasing and “forgetting” information built up throughout the day that would clutter the synaptic network that defines us. It may also be that this cleanup function of sleep is a general principle of neuroscience, applicable to every creature with a nervous system. I will discuss the importance of forgetting for development, memory integration and updating, and for resetting sensory-motor synapses after intense use. Sleep states and traits that could serve this unique forgetting function may be different for memory circuits within reach of the locus coeruleus (LC) vs. those formed and governed outside its noradrenergic net. Specifically, I will talk about the role of rapid eye movement (REM) and transition-to-REM (TR) sleep for hippocampal and somatosensory memories and the role of non-REM sleep for memories guided by the dorsal striatum (e.g., motor and procedural learning).

Susan Sara - Locus coeruelus in time with the making of memories during sleep

Experience -related reactivation of neuronal ensembles during sleep is a well-established phenomenon. It occurs mainly during high frequency sharp wave ripples in the hippocampus, but has been shown to occur in neocortical regions as well.  The current belief is that newly formed synapses in replay ensembles are reinforced through a potentiation process.  We have revealed an increase in firing rate of noradrenergic neurons of the locus coeruleus (LC) during nonREM sleep after learning  and a temporal relationship between LC spiking and cortical slow waves and spindles. Release of Norepinephrine by LC neurons in time with these oscillations could promote synaptic plasticity and facilitate sleep-dependent memory consolidation.

Presentation file(s)
Related files

Post-meeting Reflection

Gina Poe (UCLA) Link to the source page

I like starting our meeting “What is sleep?” with a session on function, because sleep looks different in different animals. so to figure out what sleep is we need to figure out what it does, then figure out what is necessary (conditions) to do that function then look for those conditions across ages and species.

Of course we have to start with some fundamentals and commonalities of sleep, even though we don’t know if they do - or should - define sleep or whether it is incidental-- a side effect, instead of an essential to sleep. I believe that thanks in part to the people here today, we have that starting point. We also have some basic functions of sleep identified, although it has been messy because until recently we have not known enough about those functions to adequately test whether sleep is important to those functions. But, again, we are at the point where we we have enough clues to start our reverse engineering phase.

In those early days we could not reverse engineer because we haven’t known what brain areas were responsible for what kind of learning and what neurotransmitters and electrical activity signatures were needed. But now, thanks to a horde of thousands, we know a few essential things:

Theta, gamma,



Potentiation/depotentiation, engrams, ARC, cFos, mRNA, protein synthesis,

Circuits: what is interconnected. How specifically NE and ACh targets forebrain,

Order of play: cortical registration, Hippocampcampal assembly, and eventual cortical strengthening...

Now we are ready to roll up our sleeves and figure out what sleep really is!

Why awakening-based sleep deprivation is not great:

  1. As Jerry Seigel points out, if you wait until an animal shows electrographic or behavioral sleep beforee you awaken it, then you get the arousal-response.
  2. My 2016 Sleep paper shows (replicated by Duran et al., 2017) that cortical electrographic signs of sleep misses much subcortical sleep even normally... how much more when sleep pressure is high! We also show that hippocampus is in REM when when the neocortex remains in NREM normally, but even more when the hippocampal homeostatic drive (learning-induced) is high. Thus, any sleep deprivation protocol that relies on posture, EEG, or muscle tone would and probably does miss a lot of sleep that goes on "locally". Thus any negative finding, e.g., "REM sleep is not important for X learning task" could be completely wrong if we are not measuring that particular type of sleep from the area involved in that type of learning and/or memory consolidation. Recommendation: measure the amount of sleep from the actual structure essential for the function you are testing.
  3. Some of the sleep deprivation studies use pharmaceuticals like caffeine or some other stimulant, which is good because it eliminates the awakening response discussed in point 1 above, but of course stimulants have their own effect on the function under study.

On time variances of total sleep time vs sleep bout lengths----------

One function of sleep, like the memory process of sleep, may take only a few minutes or even seconds to accomplish, while other functions of sleep, like repair and clearance, may take much longer to accomplish, and be counterproductive to the learning function, and therefore sleep may need to intersperse the learning function at intervals during the longer sleep period in order not to lose, e.g., what you consolidated.

Either the undoing of one sleep state's function by another or the processes or side effect outcomes of one state requiring another sleep state to clean up after it may be why healthy sleep proceeds in an orderly fashion and unhealthy sleep, even if it is adequate in total length is disordered: REM sleep occurring too soon in depression, REM sleep occurring at sleep onset in narcolepsy, etc. If REM sleep serves to clean up synapses (strengthens those spuriously weakened but tagged for keeping, and weakens synapses spuriously strengthened by SWS or those ready to be erased now that slow-oscillation coupled spindles have consolidated them to their final place) then that would explain why the length of REM sleep is related more to the length of the prior NREM state than to the length of waking.

It may be that the cycles of sleep are timed as they and the states follow each other in order as they do because should our planned long sleep cycle be interrupted, we will at least get some of all of the different kinds of necessary work done. even though having only one cycle leaves it incomplete, it would be better than nothing. Interrupting a sleep cycle midway (e.g. being awakened out of SWS) may be like interrupting the wash cycle midway, leaving clothes soaked and soapy rather than rinsed and clean. A power nap with only stage 2 sleep may be like a quick rinse and spin-dry which is better than the dry but filthy state that the party of wakefulness has left us. But a sleep disturbed from slow waves may, like the middle of the wash phase with soap and soaking wet, leave our brains in a worse state than the previous dry filthy one.

Susan Sara (École des Neurosciences Paris Île de France) Link to the source page

impact of the talks and discussion on my perspective:

The quest for a universal definition of sleep and an overarching function may be misdirected. There may be sleep functions that are species specific. Moreover, functions dependent upon sleep in one given species may be sleep-independent in another species (elephants sleep 2 hours/day and have good memories). Therefore, we need a more phylogenetic or ecological approach.

From an ontogenetic point of view, the function of sleep, especially REM sleep may change with development.

Reference Material

Here is an excellent paper clearly spelling out all the hypotheses related to sleep for learning and forgetting (besides Susan and my J Neurosci 2019 articles!) - Jesse J. Langille, Remembering to forget: A dual role for sleep oscillations in memory consolidation and forgetting. Frontiers in Cellular Neuroscience 13:1, 2019. dob: 10.3389/fncel.2019.00071

Title Author name Source name Year Citation count From Scopus. Refreshed every 5 days. Page views Related file
Remembering to forget: A dual role for sleep oscillations in memory consolidation and forgetting Jesse J. Langille Frontiers in Cellular Neuroscience 2019 0 1
Sleep is for forgetting Gina R. Poe Journal of Neuroscience 2017 0 4
The function of dream sleep Francis Crick, Graeme Mitchison Nature 1983 0 1
Experience-dependent phase-reversal of hippocampal neuron firing during REM sleep Gina R. Poe, Douglas A. Nitz, Bruce L. McNaughton, Carol A. Barnes Brain Research 2000 0 0
Input source and strength influences overall firing phase of model hippocampal CA1 pyramidal cells during theta: Relevance to REM sleep reactivation and memory consolidation Victoria Booth, Gina R. Poe Hippocampus 2006 0 0
REM restriction persistently alters strategy used to solve a spatial task Theresa E. Bjorness, Brett T. Riley, Michael K. Tysor, Gina R. Poe Learning and Memory 2005 0 0
The human emotional brain without sleep - a prefrontal amygdala disconnect Seung Schik Yoo, Ninad Gujar, Peter Hu, Ferenc A. Jolesz, Matthew P. Walker Current Biology 2007 0 1
Cognitive neuroscience of sleep Gina R. Poe, Christine M. Walsh, Theresa E. Bjorness Progress in Brain Research 2010 0 1
Antidepressant suppression of non-REM sleep spindles and REM sleep impairs hippocampus-dependent learning while augmenting striatum-dependent learning Alain Watts, Howard J. Gritton, Jamie Sweigart, Gina R. Poe Journal of Neuroscience 2012 0 0
Abnormal Locus Coeruleus Sleep Activity Alters Sleep Signatures of Memory Consolidation and Impairs Place Cell Stability and Spatial Memory Kevin M. Swift, Brooks A. Gross, Michelle A. Frazer, David S. Bauer, Kyle J.D. Clark, Elena M. Vazey, Gary Aston-Jones, Yong Li, Anthony E. Pickering, Susan J. Sara, Gina R. Poe Current Biology 2018 0 0
Sleep contributes to dendritic spine formation and elimination in the developing mouse somatosensory cortex Guang Yang, Wen Biao Gan Developmental Neurobiology 2012 0 0
Homer1a drives homeostatic scaling-down of excitatory synapses during sleep Graham H. Diering, Raja S. Nirujogi, Richard H. Roth, Paul F. Worley, Akhilesh Pandey, Richard L. Huganir Science 2017 0 1
REM sleep selectively prunes and maintains new synapses in development and learning Wei Li, Lei Ma, Guang Yang, Wen Biao Gan Nature Neuroscience 2017 0 2
Title Author name Source name Year Citation count From Scopus. Refreshed every 5 days. Page views Related file
Sleep to remember Susan J. Sara Journal of Neuroscience 2017 0 3
Reactivation, retrieval, replay and reconsolidation in and out of sleep: Connecting the dots Susan J. Sara Frontiers in Behavioral Neuroscience 2010 0 4