Multiscale exploration of neocortical slow (<1 Hz) oscillations across preparations, species and conditions.

(A) Intracellular recordings from two cells (suprathreshold, top; subthreshold, bottom) from cat neocortex under anesthesia, illustrating spontaneous alternations between depolarized “Up” states and hyperpolarized “Down” states. Reproduced from Steriade et al., 1993b. (B) Example local field potential (LFP) traces recorded during natural slow-wave sleep (left) and deep anesthesia (right) from multiple sites: secondary somatosensory cortex (S1), primary visual cortex (V1), ventral posterolateral thalamus (VP), primary motor cortex (M1) and prelimbic cortex (PrL). Up/Down alternations are observed under both conditions throughout cortical and thalamic structures. Adapted from Torao-Angosto et al., 2021. (C) Left: Probability density functions of Up state (solid lines) and Down state (dashed lines) durations under deep (blue), medium (green) and light (red) isoflurane anesthesia, showing a progressive lengthening of Up/Down cycles with lighter anesthesia. Right: Scatter plot of individual Up vs. Down durations, revealing distinct clusters corresponding to each state. Reproduced from Torao-Angosto et al., 2021. (D) LFP traces recorded from deep (top), medium (middle) and light (bottom) isoflurane levels in the anesthetized mouse. The bar above the traces indicates Down (left) and Up (right) states. Reproduced from Dasilva et al., 2021. (E) Box plots of slow-oscillation frequency (Hz) measured under deep, medium and light anesthesia. Reproduced from Dasilva et al., 2021. (F) In vitro laminar recordings in acute ferret cortical slices. Top: Scheme of a 16-channel electrode array spanning supragranular (SG), granular (G) and infragranular (IG) layers and recordings of slow oscillations from four identified locations. Adapted from Rebollo et al., 2021. (Bottom) Left: Photograph of a cortical slice in the recording chamber. Right: Simultaneous intracellular (top) and extracellular (bottom) traces from layer 5. Adapted from Sanchez-Vives and McCormick, 2000. (G) LFP and MUA example of spontaneous activity in one cortical column. Correspondence between recording sites and cortical layers enabled by the superposition of the photography taken during electrophysiological recording and posterior histological processing information given by the Nissl staining. Right: Layer-specific LFP (left column) and multi-unit activity (MUA; right column) traces from layers III -VI. Note the lead of superficial layers in initiating Up transitions. Reproduced from (Covelo et al., 2025). (H) Slow oscillations in the perilesional area in humans. Left: 3D surface reconstruction (of the right hemisphere) from a subject showing one radiofrequency thermocoagulation lesion (black) and four recording bipolar contacts (white) from the same electrode. Right: Spontaneous activity recorded before RFTC (in black) and after RFTC (in grey) from each of the bipolar contact shown in the left panel. Reproduced from Russo et al., 2021.

© 1993 Society for Neuroscience. Panel A is reproduced from Steriade et al., 1993b, with permission from the Society for Neuroscience. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

Cortical travelling waves in vivo, in vitro and in silico.

(A) Slow oscillations in humans from EEG recording occurs as propagating oscillatory patterns whose centers of mass follow trajectories moving from the front to the back of the brain (dot size indicating the origin of the travelling wave). Right-top, map of wave origin, highlighting higher densities in anterior scalp regions. Right-bottom, average wavefront at different delays further showing the rostro-caudal direction of slow-wave activity across the scalp.Reproduced from Massimini et al., 2004. (B) Average spatiotemporal propagating patterns of activation waves (traveling wavefronts, thick lines coloured by time delay) in a cortical slice clustered in two different “modes of propagation”. Top, back ground example cortical slice with electrode position (circles). Reproduced from Capone et al., 2019b. (C) Schematic representation of the wave propagation modes in the mouse brain for different anesthesia levels. Right, statistics of modes velocity for an example animal are colored according to the results of k-means clustering of traveling wavefronts (i.e., rows of the time lags/delays matrix). Thick arrows, average directions and velocities of each mode. Left and right panels, “Deep” and “Light” anesthesia, respectively. (D) Average traveling wavefronts at different time delay for the modes singled out in (C), highlight the increasing complexity of slow-wave activity as anesthesia fades out (C and D reproduced from Pazienti et al., 2022). (E) (Top) Distributions of traveling wavefronteach represented as a point in the (PC1,PC2) plane under three different levels of anesthesia placed along the z-axis. (Bottom) Shannon entropy of wavefront distributions increasing from “Deep” to “Light” anesthesia levels (Wilcoxon signed-rank test with Benjamini-Hochberg corrections: * p<0.05, ** p<0.01) (reproduced from Dasilva et al., 2021). (F) Similarity matrix of traveling wavefront ordered according to the clustering provided by the dendrograms leaves for Deep (left) and Light (right) anesthesia levels in (C) of an example animal. High and low wavefront similarities, dark and light colors, respectively. Reproduced from Camassa et al., 2021. (G) Left panels, average travelling wavefronts in spiking neural network simulation clustered in different “modes of propagation" for Deep-like (left) and Light-like (right) anesthesia level of an example simulation. Center, schematic representation of the simulated network, consisting of a lattice of local excitatory-inhibitory assemblies of leaky integrate-and-fire neurons with adaptation (LIFCA). Anesthesia fading is modelled by increasing the network excitability (stronger excitatory input vext and reduced adaptation strength ga). Right, bifurcation diagram showing parameter change as a linear trajectory moving from a phase where the network prevalently stays in a low-firing asynchronous state (LAS, burst-suppression-like) and the slow-wave activity (SWA) state. Arrows, parameter combinations used in the left panels (reproduced from Pazienti et al., 2022).

© 2004 Society for Neuroscience. Panel A is reproduced from Massimini et al., 2004, with permission from the Society for Neuroscience. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

© 2019 Oxford University Press. Panel B is reproduced from Capone et al., 2019b, with permission from Oxford University Press. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

© 2022 Pazienti et al.. Panel C and D are reproduced from Pazienti et al., 2022 (published under a CC-BY-NC-ND license). Further reproductions must adhere to the terms of this license.

© 2021 Institute of Electrical and Electronics Engineers. Panel F is reproduced from Camassa et al., 2021, with permission from the Institute of Electrical and Electronics Engineers. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

Computational models of slow oscillations.

Computational models of Slow Oscillations and Waves.

(A) (Reproduced from Mattia and Sanchez-Vives, 2012) SOs as relaxation oscillators. a. Experimental MUA (black) (equivalent to firing rate v(t) in model) and reconstructed adaptation variable a(t). b. Phase plane (a, v) with nullclines and fixed points. Arrows and circles trace the orbit of the relaxation oscillator. c. Excitation-adaptation bifurcation diagram (external excitatory input Iext vs. adaptation strength ga) showing four regimes: LAS, HAS (low/high-firing asynchronous states), and SO (slow oscillations, i.e., limit cycle) and bistable AS (coexistence of LAS and HAS). Vertical dashed line: trajectory modeling the awakening from sleep/anesthesia with sample activity traces in (A-d). d. Example firing rates from simulations modeling the awaking across burst-suppression, slow-oscillations and the desynchronized state (from dark to light traces, parameters as in A-c). (B) (Adapted from Di Volo et al. 2019) Asynchronous and synchronous dynamics in AdEx networks. a. Network architecture of excitatory RS (blue) and inhibitory FS (red) AdEx neurons. b-c. Mean firing rates in asynchronous and synchronous regimes, differing only in injected SFA. d-e. Time courses of population firing rates and adaptation current We. (C) Mean-field (MF) networks with spatial structure. a. 2D multimodular arrangement with varying local/global connectivity; darkest circles indicate highest connectivity. b. Nullclines and fixed points in the (v, c) plane across connectivity levels. c. Energy landscapes around fixed points (top), and zoomed phase-plane (bottom), both color-coded as in a. d. Time-lapse snapshots of Up wavefront propagation across the cortical slice. White dashed line: most excitable region (MCS); color: simulated MUA. Reproduced from Capone et al., 2019b. (D) Hodgkin-Huxley (HH) neuron model in 2D cortical network. a. Schematic of spatial connectivity for pyramidal (blue) and inhibitory (red) neurons, with example connectivity distributions. b. 50*50 neuron grid with 50% local excitatory (7*7) and 90% local inhibitory (5*5) connections. c. Simulated spontaneous MUA activity. d. Intracellular voltage (Vm) and sodium concentration in three pyramidal cells. Used with permission from Compte et al., 2003. Also reproduced from Barbero-Castillo et al., 2021a. (E) Digital reconstruction of neocortical microcircuit. a. Virtual slice with seven unitary microcircuits showing spontaneous activity. b. Rastergrams from random neurons under in vitro- and in vivo-like conditions. Reprinted from Cell, 163(2), Markram et al., 2015. (F) Interactive simulation interface. A. UI for selecting parameters: external current, adaptation, oscillatory modulation amplitude. B. Time courses offiring rates from three sample populations and full-network activity (color-coded). C. Parameter space plot of external current and adaptation, including preset regimes. D. Sequential frames of slow-wave propagation in data and simulations.

© 2019 Oxford University Press. Panel C is reproduced from Capone et al., 2019b, with permission from Oxford University Press. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

© 2003 American Physiological Society. Panel D is reproduced from Compte et al., 2003, with permission from the American Physiological Society. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

© 2003 Elsevier. Panel E is reproduced from Markram et al., 2015, with permission from Elsevier. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

Emerging from slow waves to wakefulness:

(A) Micro-scale: single cell membrane potential; Up states and Down states from in vivo intracellular recording of a cat cortical neuron during slow wave sleep, awakening and wakefulness. Reproduced from Mukovski et al., 2007. (B) In vivo patch clamp recording from a rat cortical neuron during sevoflurane anesthesia and awakening (from C. Hönigsperger, A. Arena & J.F. Storm [2019, unpublished]). Also shown is simultaneous recording of the local field potential (LFP) from the cat. (C) Meso scale: MUA, LFP; On and Off periods. (Top) decreasing incidence of slow waves in (bottom) LFP recordings and unit activity from rats during early and late sleep, and towards period of wakefulness (from (Vyazovskiy et al., 2009). (D) LFP recordings during decreasing depth ofgeneral anesthesia with ketamine + medetomidine (from Tort-Colet et al., 2021). The Off periods (neuronal silence) become less frequent and duration of On periods (firing of action potentials) increases from early to late sleep stages. A similar effect is seen by decreasing the level of general anesthesia. (E) Macroscale: ECoG and EEG. At the ECoG/EEG level from rats, the emergence from anesthesia to wakefulness is characterized by a reduced signal integration (less globally synchronized) and more segregation, therefore increasing in complexity (from (Nilsen et al., 2024)). (F) (Top) Intracranial EEG recordings from a rat during general anesthesia (sevoflurane 2.6%) and wakefulness, with relative periodogram (top right) illustrate how the activity changed from low to high frequency during the transition from general anesthesia to wakefulness (from (Arena et al., 2021), This redistribution of frequency powers from slow waves to desynchronized and high frequency activity is characterized by a higher slope (or spectral exponent) of the periodogram in range 20-40Hz. This feature is consistent across species (rats vs humans) and across conditions (wakefulness vs general anesthesia with various drugs, and sleep (from Arena et al., 2021; Colombo et al., 2019; Lendner et al., 2020). (G) A similar redistribution of power from low to high frequency was also observed in cortical slices in vitro, after bath-application of norepinephrine and carbachol (a cholinergic agonist) with physiological Ca2+ concentration and temperature to mimic wakefulness. This caused a switch from a bistable dynamic (Up and Down states) to a more depolarized state with asynchronous firing (from (Barbero-Castillo et al., 2021a). (H) A cortical network model showed similar changes of the slope of the periodogram (in a similar high frequency range) associated with a shift in the ratio between excitatory/inhibitory synaptic currents. Reproduced from Gao et al., 2017.

© 2017 Oxford University Press. Panel A is reproduced from Mukovski et al., 2007, with permission from Oxford University Press. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

© 2009 Vyazovskiy et al.. Panel C is reproduced from Vyazovskiy et al., 2009 (published under a CC-BY-NC-ND license). Further reproductions must adhere to the terms of this license.

© 2021 Tort-Colet et al.. Panel D is reproduced from Tort-Colet et al., 2021 (published under a CC-BY-NC-ND license). Further reproductions must adhere to the terms of this license.

© 2019 Colombo et al.. Panel F is reproduced from Colombo et al., 2019 (published under a CC-BY-NC-ND license). Further reproductions must adhere to the terms of this license.

© 2017 Elsevier. Panel H is reproduced from Gao et al., 2017, with permission from Elsevier. It is not covered by the CC-BY 4.0 licence and further reproduction of this panel would need permission from the copyright holder.

Whole-brain simulations of synchronous and asynchronous states.

Whole-brain simulations of synchronous and asynchronous states. (A) Mouse brain models consisting of 98 nodes each described by the AdEx mean-field model and connected according to the mouse connectome information from the Allen Mouse Brain Atlas. The activity of individual nodes is superimposed, for asynchronous states (left) and synchronized slow-wave oscillations (right). (B) Similar paradigm simulated with the macaque brain (82 nodes), where the connectivity information was obtained from the Cocomac database. (C) Asynchronous and synchronized slow-wave states obtained for the human brain (68 nodes). All simulations were done in TVB and are available in EBRAINS. (Reproduced from Sacha et al., 2024).

From spontaneous activity to perturbation.

(A) Time course of one channel of scalp EEG during N3 sleep showing spontaneous activity and single trial responses to TMS (vertical dashed lines). (B) Time course of average Delta activity across all EEG channels during periods of spontaneous activity and TMS (green horizontal bars). (C) Spontaneous activity from four EEG channels in the perilesional area of a stroke patient. (D) Responses to TMS when stimulating the perilesional area of the same patient shown in panel c. Highlighted channels correspond to the four channels showing the largest responses and correspond to those shown in panel c. (E) Time-frequency representation of the responses to TMS of the four channels highlighted in panel D. F, G, H. Same as panels C, D and E, respectively, but from a contralesional area. I, J, K, L, M. Responses to TMS and time-frequency representations of the four highlighted channels for a UWS patient, and a neurotypical subject during N3 sleep and wakefulness.