Do you remember your dream last night? Was it bizarre, weird, or random? Researchers have always been fascinated by the connection between sleep and dreaming. Sleep has been divided into two broad stages: rapid eye movement sleep (REM sleep) and non-rapid eye movement sleep (NREM sleep). REM sleep has always been associated with active dreaming. The history of this linkage dates back to 1958, when Dement and Kleitman first linked “eye-movement periods with minimal ocular activity” to passive dreams, while “frequent and large eye movements” were associated with active dreams.¹ Dement termed this stage of sleep “rapid eye movement sleep,” later shortened to REM sleep.²,³ In addition to REM sleep, there is NREM sleep, where sleepers’ brain waves are usually slow and of high amplitude. Their breathing and heart rate are usually slow, with minimal body movements and less vivid dreams.⁴
Although scientists are aware of the different stages of sleep, they are still puzzled by the function of each one. Many theories have tried to explain the necessity of sleep. A few mainstream theories are the Synaptic Homeostasis Hypothesis by Tononi and Cirelli, Information Consolidation Theory by Born and Wilhelm, and Reverse Learning Theory by Crick and Mitchison.⁵,⁶,⁷ Each of these theories offers insights from a different perspective.
SYNAPTIC HOMEOSTASIS HYPOTHESIS
According to Tononi and Cirelli, learning during wakefulness is the process of changing the strength and number of synaptic connections for the long term.⁵ Their synaptic homeostasis hypothesis states that synaptic strength increases while learning during daytime, and resets during sleep at night to prepare for the next day. This hypothesis is based on a common paradigm called long-term potentiation (LTP): learning signals like glutamate first activate AMPA receptors, which then subsequently allow calcium ions to enter nearby NMDA receptors. The additional insertion of AMPA receptors into the plasma membrane ultimately makes neurons more sensitive to future signaling.
The synaptic homeostasis hypothesis is backed by substantial experimental evidence that GluA1-containing AMPA receptors in the hippocampus have higher expression levels after wakening than after sleep. The axon-spine interface (ASI), the area between the presynaptic terminal and the postsynaptic terminal of neurons, decreases during sleep for mice, and the number of dendritic spines, the protrusions from dendrites that receive signals, decreases more rapidly during sleep than waking hours, with comparable formation rates between the two states. All of the above suggest that sleep is a process of reducing synaptic strength and resetting the brain.⁸,⁹,¹⁰,¹¹
However, this theory has some pitfalls. There is no well-accepted metric for synaptic strength—spine volume, spine density, and ASI are all valid. Tononi and Cirelli did not specify why they chose ASI as the metric and why they cited ASI decrease as evidence for their theory. In addition, the theory mentions that not all synapses are affected: only smaller and weaker ones are downregulated, preserving large ones that might store important memories. The theory also contradicts another study that shows this night-resetting effect only in large synapses.⁵,¹² Hence, researchers still need to address several uncertainties and gaps in the theory.
MEMORY CONSOLIDATION THEORY
Memory consolidation theory is based on the two-stage memory model that states that memories are first stored in the hippocampus as short-term ones. Then the memories are transferred to the neocortex, where they are permanently stored as long-term memories. Born and Wilhelm raised the theory that during slow-wave sleep (SWS), the deepest stage of NREM sleep, the brain reprocesses memories and transfers them to the neocortex.⁶ Researchers have found strong evidence to support this theory. Studies show rats with spatiotemporal patterns of daytime neuronal firing reactivated during sleep, indicating that the rat’s brain is indeed reprocessing the daytime memory. When human subjects were presented with odor they experienced at daytime learning during SWS, their memories were effectively enhanced. Also, human’s odor-induced memory reactivation during SWS enhanced memories even without any REM sleep, meaning that SWS plays a direct role in memory consolidation.¹³,¹⁴,¹⁵,¹⁶,¹⁷,¹⁸
The findings are promising and exciting, but they are not conclusive. First, researchers are not certain if the odor-induced hippocampal activation during SWS detected by fMRI is due to the odor cue, some other neural conditions, or simply chance.⁶ In addition, the human odor-induced memory reactivation experiment mentioned before woke subjects up immediately after the odor cue completed. Thus, there is the possibility that SWS was not complete, weakening the conclusion that SWS is a key part for memory consolidation. It is important to acknowledge these imperfections so future researchers can refine the theory.
REVERSE LEARNING THEORY
Crick and Mitchison think the purpose of sleep, specifically REM sleep, is to actively detect and suppress certain memories to modulate storage space in the brain. This is based on the assumption that mammal brain cortices are like computer networks: like a computer, the brain likely generates “parasitic,” or useless memories that would impede future storage—therefore, it needs to be cleaned periodically.⁷ This cleaning process, termed “reverse learning,” happens during sleep because the brain needs to temporarily stop receiving and sending signals. The mechanism proposed is that the brain needs to randomly activate memories to look for parasitic ones. For this purpose, random memory-related dreams are generated during REM sleep. The strongest evidence comes from John Hopfield and another group of researchers, who independently came up with the reverse learning model and showed it indeed improved neural networks.¹⁹
There are many implications and pitfalls to this theory. Some scientists propose that schizophrenia may be related to defects in reverse learning, since failed REM sleep can lead to hallucinations, delusions, and obsessions—all of which are symptoms of schizophrenia. This theory also has implications for A.I., since this learning-and-forgetting method can resolve memory overloading in neural networks. A major pitfall, however, is that this theory is extremely hard to test, and so far the only empirical evidence comes from computers, making scientists question whether or not the reverse learning theory fully explains the mechanism of human dreams.
Although all the theories mentioned above have their drawbacks, they are bold steps on the road to understanding more brain functions. As research on sleep and dreaming continues to get more attention, we will have both the technology and the data to unravel the secrets of sleep in the near future, and potentially link sleep deficits to illnesses like depression, autism, and schizophrenia.
Dement W, Wolpert EA. The relation of eye movements, body motility, and external stimuli to dream content. J Exp Psychol. 55(6), 543-553 (1958).
Dement WC. Recent studies on the biological role of rapid eye movement sleep. Am J Psychiatry. 122, 404-408 (1965).
Cohen HB, Duncan RF, II, Dement WC. Sleep: the effect of electroconvulsive shocks in cats deprived of REM sleep. Science. 156, 1746-8 (1967).
Martin JM. et al. Structural differences between REM and non-REM dream reports assessed by graph analysis. PLoS One. (2020).
Cirelli C. Sleep, synaptic homeostasis and neuronal firing rates. Curr Opin Neurobiol. 44:72-79 (2017).
Born J, Wilhelm I. System consolidation of memory during sleep. Psychol Res. 76(2):192-203 (2012).
Crick F, Mitchison G. The function of dream sleep. Nature. 304(5922):111-114 (1983).
Vyazovskiy V, Cirelli C, Pfister-Genskow M, Faraguna U, Tononi G. Molecular and electrophysiological evidence for net synaptic potentiation in wake and depression in sleep. Nat. Neurosci. 11:200-208 (2008).
de Vivo L. et al. Ultrastructural evidence for synaptic scaling across the wake/sleep cycle. Science. 355:507-510 (2017).
Maret S, Faraguna U, Nelson AB, Cirelli C, Tononi G. Sleep and waking modulate spine turnover in the adolescent mouse cortex. Nat. Neurosci. 14:1418-1420 (2011).
Yang G, Gan WB. Sleep contributes to dendritic spine formation and elimination in the developing mouse somatosensory cortex. Dev. Neurobiol. 72:1391-1398 (2012).
Diering GH. et al. Homer1a drives homeostatic scaling-down of excitatory synapses during sleep. Science 355:511-515 (2017).
Pavlides, C., Winson, J. Influences of hippocampal place cell firing in the awake state on the activity of these cells during subsequent sleep episodes. The Journal of Neuroscience. 9, 2907-2918 (1989).
Wilson, M. A., McNaughton, B. L. Reactivation of hippocampal ensemble memories during sleep. Science, 265, 676–679 (1994).
Sutherland, G. R., McNaughton, B. Memory trace reactivation in hippocampal and neocortical neuronal ensembles. Curr.Opin.Neurobiol. 10, 180–186 (2000).
Ribeiro, S. et al. Long-lasting novelty-induced neuronal reverberation during slow-wave sleep in multiple forebrain areas. PLoS Biology, 2, E24. (2004).
O’Neill, J., Pleydell-Bouverie, B., Dupret, D., Csicsvari, J. Play it again: reactivation of waking experience and memory. Trends in Neurosciences, 33, 220-229 (2010).
Diekelmann, S., Bu ̈chel, C., Born, J., Rasch, B. Labile or stable: opposing consequences for memory when reactivated during waking and sleep. Nature Neuroscience, 14, 381-386 (2011).
Hopfield, J.J., Feinstein, D.I., Palmer, R.G. ‘Unlearning’ has a stabilizing effect in collective memories. Nature 304,158-159(1983).
WRITTEN BY ANQI (ANGELA) WANG
Sixth College, B.S. Neurobiology, UC San Diego 2023, Glasgow Lab, Department of Neurobiology
FROM SALTMAN QUARTERLY VOL. 18
To read the original version, please click here. To read the full version on our website, please click here. To read more individual articles, please click here.