Do we understand sleep?

There is a new theory to explain changes in memory during sleep. It is not that new, Tononi introduced it a decade ago. Since then Tononi and his group have been amassing confirmations of the idea. A recent paper (see citations Tononi 2014) has discussed SHY or the synaptic homeostasis hypothesis in a general way. In essence, one reason for sleep is to promote learning (or plasticity) and that requires being disconnected from the outside world. Their logic goes: the brain must economize on energy and therefore needs to signal as little as possible (sparse signaling). Sparse signals need a very selective response (high signal to noise). The way we learn when we are awake is to strengthen the synapses that need to be enhanced, but as the day wears on, the synapses get stronger all over the brain and on individual neurons. This means that the neurons fire more frequently and also in response to noise. The neurons are stressed and use more energy. During sleep, this situation is corrected by weakening all the synapses so that the total signaling would be back at the base level. This weakening is done in a way that protects old memories, and reduces noise more than the new learning of the day. They have studied slow wave sleep as the mechanism for this weakening of synapses. The group has studied aspects of this theory from insects to humans, in many preparations and in live animals, and with computer simulations. (see citations). The theory appears to explain a number of aspects of memory and learning: consolidation of procedural and declarative memories, gist extraction, and the integration of new with old memories. There are a number of very interesting concepts in these papers and I intend to post on some in the future.


But for now I want to sound a note of caution. On a number of counts this theory should not be welcomed with open arms. First SHY really deals with a smallish part of the subject of learning and memory. In reading the papers there is very little discussion of REM sleep, the hippocampus, the amygdala, and the cerebellum – most of the work deals with slow wave sleep and the neo-cortex. Next, I find the computer simulation very informative in understanding the theory but not in convincing me that it is a theory that accurately models reality. There are just too many assumptions. In looking for studies by other groups, I found a review by Frank (see citation). Here is the abstract:


Converging lines of evidence strongly support a role for sleep in brain plasticity. An elegant idea that may explain how sleep accomplishes this role is the “synaptic homeostasis hypothesis (SHY).” According to SHY, sleep promotes net synaptic weakening which offsets net synaptic strengthening that occurs during wakefulness. SHY is intuitively appealing because it relates the homeostatic regulation of sleep to an important function (synaptic plasticity). SHY has also received important experimental support from recent studies in Drosophila melanogaster. There remain, however, a number of unanswered questions about SHY. What is the cellular mechanism governing SHY? How does it fit with what we know about plasticity mechanisms in the brain? In this review, I discuss the evidence and theory of SHY in the context of what is known about Hebbian and non-Hebbian synaptic plasticity. I conclude that while SHY remains an elegant idea, the underlying mechanisms are mysterious and its functional significance unknown. ”


Franks produces arguments that learning is characterized by a number of mechanisms and so we would expect sleep to have multiple mechanisms too, such as Hebbian long-term potentiation and depression, downscaling and upscaling. He points out that for all the evidence put forward, what is missing is a causal link between the changes during sleep and the effects on memory and learning. The only link is the computer simulation and it does not address actual realistic mechanisms.


As new experiments accumulated, their predictive power failed, and they became little theories that only explained—often imperfectly—single sleep phenomena. It is too soon to say where SHY fits in this story. SHY is a seminal theory, bold in its scope and challenging in its implications, but it seems oddly disconnected from our rapidly evolving views of synaptic plasticity. The proponents of SHY have also amassed an impressive set of supportive findings, but these have yet to be pursued in depth. These are not trivial matters. In the absence of a clearly proposed mechanism (informed by current views on synaptic plasticity), the empirical supports of SHY are hard to interpret. Therefore, the significance of SHY—and what it may one day reveal about sleep and synaptic plasticity—remains elusive. ”

Giulio Tononi, & Chiara Cirelli (2014). Sleep and the Price of Plasticity: From Synaptic and Cellular Homeostasis to Memory Consolidation and Integration Neuron, 81 (1) : 10.1016/j.neuron.2013.12.025

Hashmi A. Nere, & Tononi G (2013). Sleep-dependent synaptic down-selection (II): single-neuron level benefits for matching, selectivity, and specificity. Frontiers in Neuroscience, 4 : 10.3389/fneur.2013.00148

Nir Y, & Tononi G (2010). Dreaming and the brain: from phenomenology to neurophysiology. Trends in cognitive sciences, 14 (2), 88-100 PMID: 20079677

Frank MG (2012). Erasing synapses in sleep: is it time to be SHY? Neural plasticity, 2012 PMID: 22530156

I'm on ScienceSeeker-Microscope

2 thoughts on “Do we understand sleep?

  1. Sergio Graziosi

    (Welcome to your new blogging home, have been using wordpress myself and find it excellent all round)
    I’ve been writing about theories about sleep a little bit in the past months, and what you say here chimes with my own hunch. I don’t know if Tononi is trying to sell SHY as the unique mechanism to explain all of what happens while we sleep, it would be preposterous, but there is an obvious drive towards over-hype (not of Tononi’s making, of course) so it would be also understandable.
    Anyway, SHY looks pretty solid but can’t possibly be the only thing that happens. As for many widespread mechanisms across evolution, sleep is likely to have acquired multiple functions over time, but yes, I am convinced that SHY plays and important role. Is it the most relevant? I don’t know!
    The other thoughts that your post generated are centered on the level of explanatory theories: of course, saying the last word would need not only physiological (synaptic plasticity and thereabout) but also molecular & biophysical explanations, but we are miles from getting there. The missing explanatory link that you mention lies still at a higher level: the last time I’ve checked we had no idea of how neurons may store the kind of complex information that we “remember” every day, of course Tononi has IIT, a fantastic formal model of how this may be done (giving him a decent reason to propose a link based on computer simulations), but I suspect the best answers will come from the predictive coding* camp, if only because it is better placed to show how learning works along with memory (two ‘functions’ that have to be paired, evolving one without the other does not provide significant advantages – one could probably say that they are one single function, I suppose).
    Anyway, my own reflections on sleep are here: (on SHY and other all-round considerations)
    plus a series of (preceding) posts on dreams, mostly on Revonsuo’s threat simulation theory (TST) and

    *Predictive coding is difficult to digest! For a very good introduction (one that flowed easily even to me) see:
    Clark, A. (2012). Whatever next? Predictive brains, situated agents, and the future of cognitive science. The Behavioral and brain sciences, 1(86).

    1. JKwasniak Post author

      Thank you for your comments which are very interesting and for the links. I will be looking at them. JK


Leave a Reply

Your email address will not be published. Required fields are marked *