Sleep, Memory & Brain Rhythms Brendon O. Watson & György Buzsáki Abstract: Sleep occupies roughly one-third of our lives, yet the scienti½c community is still not entirely clear on its purpose or function. Existing data point most strongly to its role in memory and homeostasis: that sleep helps maintain basic brain functioning via a homeostatic mechanism that loosens connections between overworked synapses, and that sleep helps consolidate and re-form important memories. In this review, we will summarize these theories, but also focus on substantial new information regarding the rela- Downloaded from http://direct.mit.edu/daed/article-pdf/144/1/67/1830530/daed_a_00318.pdf by guest on 30 September 2021 tion of electrical brain rhythms to sleep. In particular, while REM sleep may contribute to the homeostatic weakening of overactive synapses, a prominent and transient oscillatory rhythm called “sharp-wave ripple” seems to allow for consolidation of behaviorally relevant memories across many structures of the brain. We propose that a theory of sleep involving the division of labor between two states of sleep–REM and non- REM, the latter of which has an abundance of ripple electrical activity–might allow for a fusion of the two main sleep theories. This theory then postulates that sleep performs a combination of consolidation and homeostasis that promotes optimal knowledge retention as well as optimal waking brain function. Sleep is clearly a basic human drive, yet we do not fully understand its purpose or function. One could argue that quiet but conscious rest could be just as ef½cient as sleep for recuperating certain parts of the body and would be less dangerous, since the brain would not be closed to outside inputs. From the evo - lutionary point of view, then, unconscious sleep must offer an unseen advantage to the brain. In attempting to understand the neural implica- tions of sleep and neural activity during sleep, the BRENDON O. WATSON is a clinical ½eld has focused on the view–well supported by data psychiatrist and a research fellow –that sleep bene½ts memory and general neural func - at Weill Cornell Medical College at tion. In more recent years this claim has been split Cor nell University and is doing into two subdomains: 1) a hypothesis centered on post doctoral research work at the homeostasis, wherein sleep reverses the overelabo- Buzsáki Lab at the New York Uni- ration and exhaustion of neural networks brought versity School of Medicine. about by prolonged waking states; and 2) a hypoth- GYÖRGY BUZSÁKI is the Biggs Pro - esis that sleep consolidates important memories for fessor of Neural Sciences at the New long-term storage. In sleep theory, as in neurosci - York University School of Medicine. ence, much attention has recently been focused on (*See endnotes for complete contributor synaptic connections, which carry information be - biographies.) tween neurons. Yet at the level of the syn apse, these © 2015 by the American Academy of Arts & Sciences doi:10.1162/DAED_a_00318 67 Sleep, two theories seem to conflict: while the haps resultantly, volleys of action poten- Memory homeostatic theory states that synapses, tials are often generated by coordinated & Brain Rhythms in general, are weakened, the con sol i - populations of neurons in a cohesive dation theory states that selected synap- manner. tic connections should be strengthened All of this complexity must be harnessed during sleep as a way to consolidate and organized somehow. This is partly ac - mem ory. complished through the spatial segrega- We seek here to summarize the major tion of neurons into subdivisions of the con cepts in the neuroscience of sleep (and brain (often referred to as nuclei or simply refer the interested reader to a more com- regions) such as the hippocampus, the thal - prehensive review of the relationship be - amus, or the neocortex. Each region and its tween sleep and memory).1 We propose interactions are thought to handle spe - Downloaded from http://direct.mit.edu/daed/article-pdf/144/1/67/1830530/daed_a_00318.pdf by guest on 30 September 2021 that electrical brain rhythms are key phys - ci½c neural tasks: controlling breathing iological features that allow the brain to rhythms, enabling visual perception, han- carry out all aspects of the tasks of sleep dling emotions, or navigating places and and that offer important insight into those memories. To orchestrate these spatially tasks. We also seek to determine whether distinct and seemingly task-speci½c re - these two apparently opposing views on gions, the brain employs a temporal orga- sleep might be reconciled. nizational scheme using periodic electri- cal oscillations (regular fluctuations in elec - Before proceeding to examine the rela- trical potential occurring simulta neously tionship between sleep and brain rhythms, in many neurons, which are measurable it is worth reviewing some aspects of brain even from outside the brain) to achieve structure and function that are pertinent two fundamental operations. to the topic. Our current understanding The ½rst is perpetual local-global com- of the brain is that the basic currency of munication, where by the results of local computation is a collection of electrical computations are broadcast to wide- signals transferred from one cell to another. spread brain areas so that multiple struc- This occurs via action potentials (electrical tures are simultaneously informed about signals within neurons that are triggered any given local effect.2 Re latedly, in the after neurons have received suf½cient ex - reverse direction, global brain activity is citatory input) and highly adaptable chem - made available to individual local circuits ical synaptic contacts (specialized junctions by electrical oscillations; this is often between neurons that allow information to referred to as “top-down” con trol.3 The pass between them). The action po tential second fundamental feature of the brain signals are generated by individual neu- is its persistent activity; that is, the ability rons at rates ranging from one per minute of an input to induce and main tain a long- to tens or even hundreds per second. They lasting activity trace long after the input are large enough in amplitude to be mea - has already vanished, even during sleep.4 sured from outside the neuron, and extra- Electrical oscillations appear to facilitate cellular recordings are often used by neu- these functions via their capacity to coor- roscientists as measures of information dinate groups of neurons and to divide transmission by a given neuron or popula- information into transmittable chunks. tion of neurons. The synaptic connections among neurons are relatively sparse and The collective electrical activity of the are often structured rather than random, neurons of the brain is such that signals creating functional “circuits”; and per- from large populations of neurons can be 68 Dædalus, the Journal ofthe American Academy of Arts & Sciences recorded, either with high ½delity from tion in neural networks. Second, an even Brendon O. electrodes inside the tissue of the brain more oscillation-centric interpretation is Watson & lfp György (local ½eld potential recording, or ), that synchronization of various brain nu - Buzsáki or in an attenuated form from outside the clei is the embodiment of perception, and head (through electroencephalography, that since oscillations would be essential to or eeg). Both during sleep and in waking synchronization, they are the key to per- states, the lfp and eeg show perpetually ception.6 changing activity (see Figure 1A). Some- Most forms of brain rhythms result from times large-amplitude slow oscillations are rhythmic inhibitory synaptic transmission predominant, while at other times small- (inputs from neurons with net inhibitory amplitude fast oscillations are present, but effects on the other neurons around them) most often many rhythms coexist simul- onto bulk neuronal populations includ- Downloaded from http://direct.mit.edu/daed/article-pdf/144/1/67/1830530/daed_a_00318.pdf by guest on 30 September 2021 taneously. ing the information-carrying excitatory Neuronal oscillations have been found neurons.7 The rhythmic inhibitory volleys to exist in the brains of all animals. In from these cell populations provide win- mammals, electrical signals over a broad dows of alternating reduced (inhibited) range of frequencies from as low as one and enhanced excitability and offer natu- wave every forty seconds (0.025 Hz) to as ral temporal frames for grouping or tem- high as six hundred waves per second (600 porally “chunking” neuronal activity into Hz) have been recorded. In addition, per- what appear to be functionally related haps the best documented but least em - groups of action potentials. The neurons phasized fact about brain dynamics is that generating these grouped action potentials spectral features of the eeg and lfp are are called cell assemblies and constitute the similar in all mammals, independent of basic units of information processing. This brain size. Every known eeg pattern of the stop-start parsing function of neuronal os - human brain is present in all other mam- cillators (and their hierarchical cross-fre- mals investigated to date. Furthermore, quency coupling organization, de tailed be - the correlations of various families of fre- low) can support “syntactical” rules for quencies with aspects of overt behavior neu ral communication that are known to and cognition both within and across both send er and receiver, making com- spe cies have led to the idea of frequency munication more straightforward than if bands: groups of oscillation frequencies areas of the brain had to interpret long in the brain that act as single functional uninterrupted messages or stochastic entities (for example, all frequencies from patterns of action potentials. 5–8 Hz may act similarly; Figure 1B). Sci- In addition to instantaneously organiz- entists have classi½ed at least ten mutually ing neurons and inducing them to ½re ac - interacting oscillation bands that are main- tion potentials together, oscillations may ly de½ned by their behavioral correlations.