1/64
Looks like no tags are added yet.
Sleep
A reversible state of disconnection from the environment, often with reduced mobility. Reduced movement is a good behavioural indicator of sleep, with some exceptions.
Variation in Sleep Time between Species
Varies greatly, even among mammals (e.g. dog 12-14 h vs koala 18-22 h).
Local Field Potential (LFP)
The signal produced by every ionic current across cell membranes; over large volumes and distances from the electrode, spatial averaging and synchrony dominate the signal.
Why is EEG the method of choice for measuring Sleep?
The layered organisation of cortical neurons helps interpretation: single neurons alternate between up and down states during slow oscillations, and this synchrony shows in both local and large networks.
Up and Down states
Alternating states of single neurons during slow oscillations (Up = active , down = silent); their synchrony is visible in local and large networks.
EEG + EMG in Sleep
Combined recordings (brain activity + muscle activity) used to define:
Wake (W)
NREM (N1-N3)
REM
Hypnogram
A plot of the distribution of:
Wake
NREM
REM
episodes over time.

REM Sleep
Sleep cycle stage involving high brain activity similar to wake, but muscles are atonic. Typically occurs every 90 mins and is abundant late in the night.
REM EEG Signatures
Ponto-geniculo-occipital (PGO) waves
Theta waves (about 8 Hz)
Arise in the pons and hippocampus.

NREM EEG Signatures
Slow oscillations (< 4 Hz): arises in the neocortex
Spindles (10-15 Hz); arise in the thalamocortical system.
Sharp wave-ripples (100-300 Hz): arise in the hippocampus.
Slow-wave sleep (SWS) is abundant early in the night.
Is Sleep Universal?
Yes - even the simplest life forms demonstrate rest-like periods (e.g. Cassiopea Jellyfish)
Effect of prolonged Sleep Deprivation
Lethal (where tested) - humans with fatal familial insomnia die. Death appears to be caused by metabolic dysfunction.
What hypotheses are there for why Sleep evolved?
Sleep may have evolved to optimise energy expenditure
Some physiological processes work better during rest.
Sleep is necessary, but how much is necessary remains unclear.
Cost of Wakefulness
Learning (encoding during wake) comes at a cost, and sleep may restore the brain's learning potential.
cAMP-PKA-CREB signalling
Ca²⁺ activates adenylyl cyclase
↑ cAMP
cAMP activates PKA
PKA enters the nucleus
PKA phosphorylates CREB
CREB switches on genes for plasticity proteins
New synaptic proteins and even new synaptic spines are produced
Why is cAMP-PKA-CREB signalling Important (in sleep)?
It is needed for long-lasting synaptic plasticity (LTP):
During sleep recently encoded memories are replayed causing repeated neuronal firing →
Activating cAMP-PKA-CREB signalling →
Transcription of plasticity related genes.
Effect of sleep deprivation on PDE4 (Abel et al. 2013)
SD up-regulates the cAMP-degrading phosphodiesterase PDE4 and lowers transcription of CREB-regulated genes, impairing plasticity and memory.
Rolipram
A PDE4 blocker; eliminates the effect of sleep deprivation on LTP and restores learning and memory in fear conditioning.
Vyazovskiy et al. 2011 (Nature)
Rats trained on a sugar-pellet reaching task and kept awake: single neocortical neurons entered "offline" slow-oscillation (NREM-like) states; prolonged wake puts individual neurons into a sleep-like state.
Local sleep during prolonged wake (SD1-SD4)
Sleep pressure rose across days; higher-amplitude, slower waves in motor cortex, reduced neuronal activity and poorer task performance despite the animals being behaviourally awake. Activity recovered after sleep.
Glymphatic system (Xie et al. 2013)
CSF flow in the extracellular space clears waste; CSF flow during wake is only 5% of that in sleep (per the slide).
Glymphatic link to Alzheimer's disease
β-amyloid was cleared about twice as fast in sleep; sleep is disrupted in Alzheimer's, a possible contributor to neurodegeneration.
Counter-evidence to glymphatic theory (Miao et al. 2024)
Tracer injected directly into brain parenchyma was cleared faster in wake than under anaesthesia (similar to NREM); increased tracer area under anaesthesia may reflect reduced clearance. Waste-clearance theory remains debated.
Synaptic homeostasis hypothesis (SHY)
Tononi & Cirelli: sleep is restorative; its main function is restoring learning potential through net synaptic downscaling.
SHY: why learning occurs in wake
Learning = synaptic potentiation, which must occur during wake when interacting with the environment.
Cellular costs of learning (SHY)
Energy consumption, increased noise-to-signal ratio, reduced selectivity of firing, saturation of plastic potential.
SHY: what sleep does to synaptic strength
Net reduction of synaptic strength (active downscaling).
SHY and memory consolidation (Diekelmann & Born 2010)
Compatible: downscaling removes weak associations and noise, benefiting consolidation.
Gilestro et al. 2009
In flies: synaptic proteins needed for memory are down-regulated during sleep (net decrease of synaptic proteins in sleep).
Vyazovskiy et al. 2008
In mammals: GluR1-containing AMPA receptors and GluR1 phosphorylation are lower after sleep, consistent with LTD-like weakening.
How does net downscaling occur in NREM?
By removal of AMPA receptors from spines (Tononi & Cirelli 2018).
Renormalisation: which synapses are spared?
Neurons and synapses most active during sleep; spines active during learning are protected.
Role of UP states in synaptic fate
UP states of the slow oscillation can weaken or preserve spines by spike-timing-dependent plasticity (STDP); co-active postsynaptic neurons are important for preservation.
GSK3β phosphorylation
Candidate tag for preservation of synapses (blocks GSK3β's role in LTD).
Arc and Homer1a
Candidate tags that localise to weak spines and mark them for downscaling.
Stickgold 2005
Sleep helps all types of memory. Prior sleep improves acquisition in the next wake period; subsequent sleep improves consolidation (stabilisation and integration) of newly encoded memories.
Declarative learning and sleep
Enhanced by sleep (e.g. word-list recall); even a 1-2 hour nap, or an ultra-short 6-minute nap, after training improves retest performance (Lahl et al. 2008).
Procedural learning and sleep
E.g. serial reaction time task: a full 8-hour night gives the greatest improvement; prolonged training without intervening sleep does not improve performance.
Active systems consolidation (model)
Reactivation of the circuits involved in encoding allows offline consolidation of memories during sleep.
Evidence base for active systems consolidation
Mostly episodic, hippocampus-dependent memory; the process can last days, months or years.
Reorganisation of memory in systems consolidation
Redistribution of memory representations from hippocampus to neocortex for long-term storage and integration with existing memories; some synapses strengthen while others weaken.
Why reactivate memories during sleep rather than wake?
Ensures representations are not "hallucinations" (not mistaken for current input).
Acetylcholine (ACh) across states
Low in SWS; high in wake.
High ACh effect
Enhances hippocampal input and encoding.
Low ACh effect
Favours replay (hippocampal-to-cortical communication).
Replay
Temporal firing sequences from wake (e.g. place cells on a circular track in mice) reappear during SWS in hippocampus and neocortex, as a sharp wave followed by a ripple.
Replay speed and location
5-20x "fast forward"; also seen in subcortical networks (dopaminergic VTA, amygdala) and can occur during quiet wake.
Girardeau et al. 2009 (causal test of replay)
Disrupting ripples after the sharp wave in rats doing a Y-maze task impaired performance with no change in sleep architecture.
Neocortical slow oscillation (SO) role
Generates up and down states that set the temporal frame for neocortex-hippocampus communication and redistribution of memories.
Thalamocortical spindles role
Coincide with the rising phase of the slow wave and prime the neocortex for plasticity at the same time as hippocampal replay.
Hippocampal sharp wave-ripples role
Accompany reactivation of hippocampal neuronal ensembles.
Origin of spindles
Thalamocortical and intrathalamic network activity.
Spindle density after learning
Increases after intense learning; spindles are thought to be a key step in consolidation.
Niethard et al. 2018 (PNAS)
SO-spindle interaction mechanism: in cortical pyramidal neurons (Ca2+ imaging), somatic inhibition with dendritic disinhibition creates conditions for dendritic spine plasticity.
Latchoumane et al. (Neuron)
Causal test: experimental manipulation of spindle activity changes memory.
Temporal coordination of oscillations
The SO synchronises spindles and sharp wave-ripples, enabling hippocampal memory reactivation and neocortical reorganisation; up states and spindles allow cortical plasticity, which continues in REM (insertion of new AMPA receptors at synapses tagged by replay).
SHY vs active systems consolidation: nature
SHY = restorative; active systems consolidation = active.
SHY vs active systems consolidation: core process
SHY = net downscaling of synapses (AMPA receptor removal); active systems consolidation = replay and transfer with selective strengthening.
SHY vs active systems consolidation: gain
SHY = restores capacity, improves signal-to-noise, saves energy; active systems consolidation = stabilises and integrates memories in neocortex.
SHY vs active systems consolidation: key oscillations
SHY = slow waves; active systems consolidation = SO-spindle-ripple coupling.
Can SHY and active systems consolidation coexist?
Yes: active systems consolidation can be embedded within global synaptic downscaling; they may be different, coexisting processes (global renormalisation plus selective strengthening of replayed memories).
Summary: sleep across species
Present in virtually all life forms; time asleep differs between species (ecology, metabolism).
Summary: sleep, learning and plasticity
Positive correlation between sleep and learning/memory; sleep may be required to renormalise the potential for plasticity and new learning in wake.
Summary: replay
An essential step for reorganising memories for long-term storage.
Summary: network coordination
Spatially distant cortical and subcortical networks show coordinated, temporally regulated activity.