Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats

Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in... It remains unclear how specific cortical regions contribute to the brain’s overall capacity for consciousness. Clarifying this could help distinguish between theories of consciousness. Here, we investigate the association between markers of regionally specific (de)activation and the brain’s overall capacity for consciousness. We recorded electroencephalographic responses to cortical electrical ST stimulation in six rats and computed Perturbational Complexity Index state-transition (PCI ), which has been extensively validated as an index of the capacity for consciousness in humans. We also estimated the balance between activation and inhibition of specific cortical areas with the ratio between high and low frequency power from spontaneous electroencephalographic activity at each elec- trode. We repeated these measurements during wakefulness, and during two levels of ketamine anaesthesia: with the minimal dose ST needed to induce behavioural unresponsiveness and twice this dose. We found that PCI was only slightly reduced from wakefulness tolightketamineanaesthesia, butdroppedsignificantlywithdeeperanaesthesia. The high-doseeffectwasselectivelyassociatedwith ST reducedhighfrequency/lowfrequencyratiointheposteromedialcortex, whichstronglycorrelatedwithPCI . Conversely, behavioural unresponsiveness induced by light ketamine anaesthesia was associated with similar spectral changes in frontal, but not posterior ST cortical regions. Thus, activity in the posteromedial cortex correlates with the capacity for consciousness, as assessed by PCI , dur- ing different depths of ketamine anaesthesia, in rats, independently of behaviour. These results are discussed in relation to different theories of consciousness. Keywords: consciousness; ketamine anesthesia; EEG markers of consciousness; perturbational complexity index Introduction Highlights It is widely recognized that only a limited fraction of our brain activityisdirectlyinvolvedinspecifyingourconsciousexperience • We dissociate responsiveness from consciousness using (Kochetal.2016).Ideally,atheoryofconsciousnessshouldbeable a two-level ketamine protocol in rats. to precisely explain why only certain parts of the brain and types ST • We correlate activity in cortical regions with PCI , an of activity contribute to any particular experience. This reflects indicator of capacity for consciousness. two main aspects of consciousness science: understanding which • Cortical deactivation in the back, but not the front, was brain structures and activities are required for having a capacity ST associated with a significant drop in PCI . for consciousness, and which are required for particular contents Received: 15 January 2021; Revised: 9 December 2021; Accepted: 24 January 2022 © The Author(s) 2022. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 2 Arena et al. of conscious experience. The former can be identified by study- Recently, several measures aiming to objectively assess global ing how properties of brain activity change between brain states states of consciousness independently of motor or sensory where the level of consciousness is thought to change, such as functions have been developed (see for example Luppi et al. 2021), comparingwakefulnesswithdeep, generalanaesthesiaordream- with a notable convergence in evaluating the complexity of brain less sleep (e.g. Casali et al. 2013). The latter can be identified activity as indication of the capacity for consciousness (Sarasso by contrasting brain activity between conditions where particular et al. 2021). In particular, the perturbational complexity index ST stimuli are perceived or not, without otherwise altering the level [PCI; (Casali et al. 2013)], and the more general measure PCI of consciousness (e.g. van Vugt et al. 2018). [‘PCI-state transition’; (Comolatti et al. 2019)], have been shown Recently, the roles played by the frontal vs. posterior parts to reliably and consistently assess the capacity for consciousness of the neocortex in the consciousness of healthy humans have inhumansinaccordancewiththesubjects’immediateordelayed been discussed (Boly et al. 2017; Odegaard et al. 2017). In this reports of experience (Sarasso et al. 2015; Casarotto et al. 2016; ST ‘front vs. back debate’, some have argued that the ‘evidence for Rosanovaetal. 2018). Recently, PCI hasalso been shownto work a direct, content-specific involvement of the “front” of the cor- consistently in rodents undergoing propofol, sevoflurane, and ST tex, includingmostprefrontalregions, ismissingorunclear’(Boly ketamine anaesthesia (Arena et al. 2021). PCI and PCI quantify et al. 2017), while others argued that ‘the literature highlights pre- thespatiotemporalcomplexityofrepeatable,global,cortical,elec- frontalcortex’sessentialroleinenablingthesubjectiveexperience trophysiological responses to a local, direct cortical stimulation, in perception’ (Odegaard et al. 2017). Although this debate largely thus estimating how much the resulting deterministic neuronal concerned empirical data, and the issue remains unresolved, the activations are both integrated and differentiated across cortical debate illustrates that different theories of consciousness entail areas and time. PCI is inspired by the general idea that the joint differenthypothesesaboutwhichbrainregionscontributedirectly presenceofintegrationanddifferentiationisrequiredforasystem toexperienceandwhichprovidenecessarybackgroundconditions to be conscious, which is a central part of the Integrated Informa- for consciousness. Hence, evidence showing regional differences tion Theory (IIT; Tononi 2004; Massimini et al. 2009). Later on, PCI in contribution to the capacity for, or contents of, consciousness has also been considered to be compatible with the Global Neu- may provide specific empirical support for some theories over ronal Workspace (GNW) Theory, as the measure is sensitive to others. global and sustained patterns of cortical interaction (Baars 2005; Some authors have suggested that frontal parts of the cortex Dehaene et al. 2011; Mashour et al. 2020). More importantly, PCI are crucial for consciousness (Del Cul et al. 2009) or have decoded may be considered an index of capacity for consciousness even perceptual contents from frontal regions (Levinson et al. 2021). in cases where we do not know the ground truth (Casarotto et al. Thus, the prefrontal cortex was found to be causally involved 2016; Comanduccietal. 2020). Therefore, fortheremainderofthis in determining the contents of experience (Weilnhammer et al. paper, weinterpretasignificantdropinPCIfromwhatisobserved 2021), and frontal neuronal activity correlated with visual per- in wakefulness as indicative of a relative reduction in capacity for ception (even during no-report paradigms; Kapoor et al. 2020). consciousness. However, others have pointed to evidence that the apparent While PCI is an index of the capacity for consciousness based frontal involvement in consciousness may be confounded with on evoked cortical dynamics, measures that quantify spectral task-related processes such as working memory, attention, or properties of spontaneous cortical activity have also for long preparation for motor response (Koch et al. 2016; Boly et al. 2017). been used successfully to study brain states (Loomis et al. 1937; Whilethisdoesnotimplythatfrontalcorticalregionsarenotnec- Fernandez et al. 2017; Siclari et al. 2017; Colombo et al. 2019; essary for, or causally involved in, specifying human experiences, Lendner et al. 2020). In particular, during deep stages of sleep it does remind us that brain functions associated with normal and general anaesthesia, both electroencephalography (EEG) and behaviouralresponsivenesscanbeconfoundedwiththosedirectly local field potentials are characterized by high amplitude, low involved in specifying conscious experiences (Sanders et al. 2012). frequency (LF,≤4Hz) oscillations or slow-waves (Massimini et al. The ability to respond coherently to external stimuli is used as 2004; Vyazovskiy et al. 2009; Brown et al. 2010). This slow- the main criterion for determining whether non-communicating wave activity reflects a bistable network dynamic, where neurons patients and non-human animals are conscious (Chernik et al. synchronously alternate between an up-state, with depolarized 1990; Giacino et al. 2004; Gao and Calderon 2020). However, this membrane potential and firing, and a down-state with neuronal approach is based on the assumption that unresponsive states hyperpolarizationandsilence(Steriadeetal.1993,2001;Volgushev are always unconscious, which is at odds with evidence that etal.2006;Vyazovskiyetal.2009),possiblyduetoincreasedinhibi- vivid experiences can occur in unresponsive states. For instance, tion,adaptation,andsynapticfatigue(Steriadeetal.2001;Compte dreams can occur in all stages of sleep (Nielsen 2000; Solms et al. 2003; Esser et al. 2007; Funk et al. 2017). Conversely, during 2000; Siclari et al. 2018) and during general anaesthesia (Noreika wakefulness, the EEG is mainly characterized by low amplitude, et al. 2011). Furthermore, patients can be conscious but unre- high frequency (HF,≥20Hz) oscillations, which reflect tonic neu- sponsive for decades after brain damage(Vanhaudenhuyse et al. ronaldepolarizationandfiring( Steriadeetal.1996,2001;Mukovski 2018), or they can be painfully aware during general anaesthe- et al. 2007; Vyazovskiy et al. 2009). This has inspired several mea- sia for surgery, while assumed to be unconscious (Ghoneim et al. sures for quantifying the cortical state of activation based on 2009). Conversely, quite complex behaviours can be preserved the relation between high (HF) and low (LF) frequency EEG power during conditions that are typically assumed to be unconscious, (Mukovski et al. 2007; Fernandez et al. 2017; Gao et al. 2017; Siclari e.g. sleepwalking, sleep talking (Castelnovo et al. 2018; Valomon et al. 2017; Colombo et al. 2019; Lendner et al. 2020). Thus, the et al. 2021), and presumed unconscious behaviours during certain HF/LF (power) ratio tends to drop when there is more inhibition epileptic seizures, and in unresponsive wakefulness syndrome and deactivation, whereas a higher HF/LF ratio suggests neu- (Blumenfeld 2005; Laureys et al. 2010). Thus, it is essential to ronal activation (Mukovski et al. 2007; Gao et al. 2017; Lombardi distinguish brain regions necessary for consciousness from those et al. 2017; Poulet and Crochet 2018). Coherently, the spectral that are necessary for behavioural responsiveness. relation between HF and LF powers has been associated with Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 3 the excitation/inhibition balance (Gao et al. 2017) and the ratio In addition to new data, we also included a new analysis of between LF and HF powers was found to correlate with motor datafromasetofpreviouslypublishedexperiments(Arena, Thon activity in mice (Fernandez et al. 2017). Thus, local changes in and Storm, 2019; Arena et al. 2021) with a new analysis. Specif- HF/LFratiomayindicatechangesincorticalactivation(Pouletand ically, we reanalysed data from a subset of six rats, from which Crochet 2018). werecordedepiduralEEGcontinuouslyandinresponsetoelectri- While measures of consciousness typically evaluate global calstimulationofthesecondarymotorcortexduringwakefulness brain dynamics, specific and localized changes in the HF/LF ratio andlightketamineanaesthesia(Arenaetal. 2021). Thesedatawill mayoccurwithinthesamebrainstateandmightaffectbehaviour be compared to those from a new set of recordings/stimulations or conscious experience. For example, local cortical sleep, which that were performed on the same six rats during a subsequent can affect task performance in rats, involves localized slowwaves period of ketamine anaesthesia with increased dosage. and neuronal silence in an otherwise awake brain state, dom- Experimental procedure inated by low amplitude and fast oscillations, with underlying tonic neuronal firing ( Vyazovskiy et al. 2011). Furthermore, pos- Epidural EEG was recorded by a grid of 16 screw electrodes terior increases in HF/LF ratio during deep stages of sleep suc- (stainlesssteel,1.2mmcalibre),whichwerechronicallyimplanted cessfully predicted whether or not humans reported dreaming through the skull, in contact with the dura. The recording elec- and were explained by a simultaneous local reduction of 1–4Hz trodes were organized symmetrically with respect to the sagittal activity and enhanced 20–50Hz activity (Siclari et al. 2017). Thus, suture and spanned most of the cortical surface of both hemi- HF/LF can vary regionally within global states, but it is not known spheres (Supplementary Fig. S1). The bilateral frontal cortex was whether local reductions in HF/LF are associated with reduced coveredbysixelectrodes, namedM2 , M2 andM2/M1thatmedi- R C consciousness levels. More generally, it is still unknown whether ally overlaid the rostral and caudal part of the secondary motor any particular localized spectral properties are related to the cortex and part of the primary motor cortex, respectively. Other brain’s global capacity to sustain consciousness. six electrodes covered left and right parieto-occipital associa- In this study, we aimed at investigating this relation, asking tive cortices: PA electrodes covered lateral parietal cortex, RS/PA whether specific regional changes of cortical activation, assessed medially covered retrosplenial and parietal cortex, while RS/V2 by the HF/LF ratio, are associated with changes in capacity for electrodes covered the posteromedial cortex, over the caudal part ST consciousness, assessed by PCI , irrespectively on the specific ofretrosplenialcortexandmedialpartofsecondaryvisualcortex. content of experience. We also aimed to dissociate levels of The last four electrodes, bilateral S1 and V1, overlaid the primary behavioural (un)responsiveness (assessed by responses to pain somatosensory cortex and the primary visual cortex respectively. stimuli) from the capacity for consciousness by comparing wake- Event-related potentials (ERPs) were recorded in response to elec- fulnesswithtwodistinctlevelsofketamineanaesthesia.Indeed,it tricalstimulationoftherightsecondarymotorcortexbyabipolar has been shownthat the unresponsivestate induced by ketamine tungsten electrode (see Supplementary Fig. S1 for detailed elec- can subtend vivid conscious experiences, with wakefulness-like, trodelocationswithrespecttobregma;PaxinosandWatson2007). high PCI (Sarasso et al. 2015; Arena et al. 2021). While it was The standard surgical procedure under a regime of controlled also previously reported that when ketamine plasma level is par- general anaesthesia/analgesia was adopted for implantation of ticularly high, such as soon after bolus injection (Akeju et al. chronic electrodes, and after 3days of recovery, rats were habit- 2016), the EEG alternates between low and highly complex activ- uated to head and body restriction in at least 3 subsequent days, ity patterns (Li and Mashour 2019). In this condition, slow-wave as previously described (Arena et al. 2021). The electrophysiologi- oscillations also occur, interrupting an enhanced HF activation, cal recording/stimulation began only when rats did not show any resultinginthegamma-burstactivitypattern(Akejuetal. 2016; Li signofdistressandwerecalmwithintherecordingsetup,withthe and Mashour 2019), and suggesting transient moments of uncon- head connected to a fixed head-bar by two chronically implanted sciousness at high ketamine doses. Here, we carefully control clamps and with the body inserted in a transparent acrylic tube, ketamine dosage by adopting intravenous infusion at two differ- with a natural posture. The tail was left outside the tube to test ST entconstantratesandinferconsciousexperiencerelyingonPCI reflex motor responses to pain stimulation. level, in comparison with wakefulness condition. Then, we inves- The six rats were subjected to electrophysiological record- tigatewhetherregionalHF/LF ratioreliablycovarieswithchanges ing/stimulation sessions during wakefulness and subsequent ST in PCI , within and across conditions. The results may be used ketamine anaesthesia. Ketamine (Vetoquinol, Ittigen, Switzer- to validate predictions and explanations from theories of con- land) was infused at a constant rate via a 26GA catheter in sciousnessbyconstrainingwhichcorticalregionsmainlyunderlie the tail vein. Subcutaneous injection of glycopyrrolate 0.01mg/kg consciousness as opposed to behavioural responsiveness. was also performed to reduce the increased salivation. Since rats keep eyes open during general anaesthesia, eye ointment was applied to maintain eyes humid and body temperature was kept at 36.5–37.5 C by a heating pad, as previously described in Materials and methods detail. During the recording session, the stimulating electrode Animal model and experimental data was connected to an isolated current stimulator (Isolator HG203, Six adult, male, Sprague–Dawley rats (n=6; body weight ∼370g) High Medical, London, UK) triggered by a voltage pulse genera- were used in this study. All the experiments and animal care tor (2100, A-M System, Washington DC, USA), while the epidural procedures were conducted at the University of Oslo and were EEGelectrodeswereconnectedtoa16-channelunipolaramplifier approved by the Norwegian Authority, Mattilsynet (FOTS: 11812) referenced to ground (RHD2132, Intan Technologies, Los Angeles, in agreement with the Norwegian law of animal handling. Efforts CA,USA),andcontrolledbyOpenEphyssystem(Siegleetal.2017), were made to avoid or minimize animals’ pain and distress. Rats which acquired and digitized the electrophysiological signal at 10 were caged in enriched environments, with ad libitum access to or 30kHz, 16-bit resolution. food and water and were exposed to a 12:12 hour light–dark cycle The EEG activity was continuously recorded from all 16 at 23 C constant room temperature. channels, in a dark environment, in which all rats received 4 Arena et al. ∼100 electrical monophasic current pulses of 50µA, 1ms, deliv- baseline window, α=0.01) was used to conserve only the ITPC ered at 0.1Hz, at first during wakefulness (W), and during incrementsthatdifferedsignificantlyfromtherespectivebaseline subsequent ketamine anaesthesia. The anaesthesia was induced (from −500 to −200ms). The time point of the last significant ITPC by an intravenous (iv) bolus injection of ketamine 30mg/kg value within 800ms after stimulation and in a broad frequency and then maintained at the initial constant rate of ketamine range (5-80Hz) was detected for each channel and considered to 1.75mg/kg/min iv, by a syringe pump. After 10min from induc- be the duration of the phase-locked EEG response, which quan- tion,thetailwaspinchedthreetimesbyforcepstocheckthepres- tified the temporal extension of the deterministic effect of the enceofbehaviouralreactionto painstimulation. Thebehavioural electrical stimulation [ITPC drop time; (David et al. 2006; Pigorini ST responsetypicallyconsistedofalateralandwidemovementofthe et al. 2015; Rosanova et al. 2018; Arena et al. 2021)]. PCI was used tail and/or in the alteration of the respiratory rhythm, with the to estimate the capacity for consciousness (Comolatti et al. 2019; occurrence of a deeper breath, accompanied by a sudden chest Arena et al. 2021) and was assessed in the full ERP window, from movement. The infusion rate was stepwise increased by adding 0 to 600ms, and across time, in short, moving windows of 100ms, ST 4%oftheinitialdosageuntilthebehaviouralresponsewasabsent with 50ms of overlap. PCI was computed using the available (3min between each increment). The resulting minimal dosage code at [github.com/renzocom/PCIst] with the same parameters of ketamine that abolished behavioural reaction to pain stim- previously described (Comolatti et al. 2019; Arena et al. 2021). The ulation was 1.78±0.02mg/kg/min, iv (mean±SEM across rats; functional connectivity across cortical areas in response to stim- 1.8mg/kg/minforsimplicity)anddefinedthecondition‘ketamine ulation was also used to estimate the level of integration and 1’ or K1. In order to evaluate possible dose dependencies, the differentiation of the cortical network, as previously described in same recording/stimulation was also repeated afterwards, during detail (Arena et al. 2021). The inter-site phase clustering (ISPC) deeperketamineanaesthesia (‘ketamine 2’ or K2), at the constant was assessed as the consistency across trials of the phase differ- rate of 3.5mg/kg/min iv, which corresponded to two times the ence across channels for all time-frequency points (Cohen 2014; initial constant rate. Arena et al. 2021). ISPC was calculated for each channel pair, the respective mean ISPC of the baseline (from −500 to −200ms) was subtracted and bootstrap statistic (500 permutations; posi- Analysis of electrophysiological signal tiveandnegativethresholdsbasedontheobtaineddistributionof The acquired electrophysiological data were analysed in MAT- the maximum and minimum ISPC values in the baseline window, LAB2016a (Math Works, Natick, Massachusetts, USA) and Origin α=0.05) was adopted to conserve only the significant variations 9.1 (OriginLab, Northampton, Massachusetts, USA) and prepro- from baseline. The ISPC values that could be determined by vol- cessed as described before (Arena et al. 2021). Raw epidural EEG ume conduction (clustering around 0 or pi) were set to 0 and wasbandpassfiltered(0.5–80Hz, Butterworth, thirdorder), down- the remaining ISPC values were averaged in the frequency range sampled to 500Hz and ERP epochs of 10s were extracted for each 5–14Hz and between 180 and 400ms. The proportion of the num- channel, centred at the stimulus onset (from −5 to 5s). All epochs ber of mean positive ISPC values for each channel was defined as were offset corrected by subtracting the average voltage of their the connectivity degree of the electrode (Cohen 2014; Arena et al. respective baseline (from −1 to 0s), and trials with high voltage 2021). artefacts in their baseline were removed. The first 90 trials of pre- The spontaneous cortical activity associated with the differ- processed signals were used for analysis, maintaining the same ent experimental conditions was quantified from 90 epochs of temporal sequence across animals and conditions. ERPs were not 5s of epidural EEG signal that preceded the electrical stimula- normalized and the electrical noise was similar along with the tion (from −5 to 0 s). A Morlet wavelet convolution (6 cycles, whole duration of recordings. 80 wavelets, linearly spanning from 1 to 80Hz) was performed The cortical excitation in response to electrical stimulation on each epoch for all channels. Spectral powers were extracted was quantified by the root mean squared (rms) amplitude of and averaged across samples, trials and channels, obtaining a the first 50ms of the mean ERP for all electrodes, and then global estimation of the power of each frequency for each ani- averaged across electrodes. Spectral powers and phases of ERPs mal and condition. The resulting periodogram was linearly fitted were obtained from Morlet wavelet convolution (three cycles in Log-Log coordinates, in the frequency range 20–40Hz. The wavelets, linearly spanning from 1 to 80Hz, with 1Hz resolution), slope of the obtained linear function was the spectral exponent which was performed for each trial and channel. The spectral of the 1/f function and was used to quantify the distribution powers were normalized over the mean power of the baseline of frequency powers in the spontaneous EEG activity (Gao et al. (−500 to −200ms) across trials, for each respective frequency and 2017; Colombo et al. 2019; Lendner et al. 2020; Arena et al. channel. The mean relative powers across trials were then dB 2021). Instantaneous powers were normalized by 1 and con- converted and bootstrap statistic (500 permutations; positive and verted in dB, and also averaged in high and low frequency ranges negativethresholdsbasedontheobtaineddistributionofthemax- (HF 20–80Hz, LF 1–4Hz, respectively), across and for each sin- imum and minimum dB values in the baseline window, α=0.05) gle channel, before dB conversion. The ratio between HF power was applied for each frequency and channel to conserve only the and LF power (HF/LF ratio) was also computed to estimate the significant dB variations from respective baseline. The resulting level of activation for the cortical area underlying each channel relative spectral power was averaged across trials in the HF range (Fernandez et al. 2017; Poulet and Crochet 2018) and its contribu- (20–80Hz)toestimatethetemporaldynamicoftheneuronalacti- tion to cortical complexity and capacity for consciousness (Siclari vation underlying the EEG signal (Mukovski et al. 2007; Pigorini et al. 2017). et al. 2015; Rosanova et al. 2018; Arena et al. 2021). Late incre- ments of HF power (> 0dB) were detected in a time window from Statistics 80 to 800ms (Arena et al. 2021). The inter-trial phase clustering [ITPC; (Cohen 2014)] was computed for each frequency and chan- Allresultsareexpressedasmean±SEManderrorbarsandshades nel and bootstrap statistic (500 permutations; threshold based represent SEM in the figures. Shades in linear regressions rep- on the obtained distribution of the maximum ITPC values in the resent the 95% confidence band. The topographical plots in the Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 5 figures report the Laplace interpolation of a variable over the dor- of ketamine. We also repeated the electrophysiological experi- sal surface of a skull, anchored to the true electrode locations. ment with an increased administration rate, to test for a possible The function of the colour maps is only for better visualization dosage dependency. As reported previously (Arena et al. 2021), since all the analyses were performed at the level of single we adjusted the first, low ketamine infusion rate (K1) for every channels. Parametric statistics were adopted after assessing the single rat, to the minimal dose that induced behavioural unre- normality of distribution of the measured variables, by apply- sponsiveness (i.e. no motor response to pain stimulation). This ing the Shapiro–Wilk test, in a population of 12–14 rats during first dosage, K1, was approximately 1.8mg/kg/min, iv, while the wakefulness. All the variables were tested in a repeated mea- secondsubsequentketaminedose, K2, wassettoabouttwotimes sure design. Thus, principal and interaction effects were tested the first one, at 3.5mg/kg/min, iv. withone-wayortwo-wayrepeatedmeasuresANOVA(rANOVA),in During wakefulness, the spontaneous EEG activity was char- which Greenhouse–Geisser correction was applied when spheric- acterized by low amplitude, fast oscillations, typical of cortical ity could not be assumed, while group comparisons were tested activation (Fig. 1a). Interestingly, despite the loss of behavioural with Student’s paired-samples t-test. Because one channel was responsiveness, the fast oscillations persisted in both K1 and K2 removed from the analysis in two rats, two-way ANOVA was used conditions (Fig. 1a), and a general increase of EEG amplitude to compare the spatial distribution of variables between left and occurred with ketamine infusion, as shown by the averaged peri- right hemispheres. Linear fitting was performed with the least- odogram across channels, which scaled up from wakefulness to square method. To evaluate correlations and goodness of fit, the K1 and further to the K2 condition (Fig. 1b). In line with this, coefficient of determination, R , was computed and a t-test was the mean HF power (20–80Hz) across channels increased from performed to test the null hypothesis of slope=0, establishing wakefulness to K1 and further to K2 (Fig. 1c; W: 42.42±0.54dB, a P-value. When multiple hypotheses were tested across con- K1: 47.52±0.83dB, K2: 49.15±0.64dB; one-way rANOVA, princi- ditions or along cortical areas, the Bonferroni–Holm correction pal effect of condition, P=9.1942*10-8; paired-samples t-test, W −5 was adopted (number of conditions=3, corrected α=0.01666; vs. K1, P=0.0005, W vs. K2, P=6.8217*10 , K1 vs. K2, P=0.0032). numberofcorticalareas=8,corrected α=0.00625foreachhemi- Likewise, the mean LF (1–4Hz) power increased from wakeful- sphere). Gaussian v-test was used to test volume conduction in ness to K1 and further to K2 (Fig. 1d; W: 73.57±1.26dB, K1: connectivity analysis (Cohen 2014; Arena et al. 2021). All statistics 77.75±0.72dB, K2: 79.75±0.54dB; one-way rANOVA, principal are two-tailed. The statistical significance in the figures are repre- effect of condition, P=0.0021; paired-samples t-test, W vs. K1, sented as follows: P<0.05 *, P<0.01 **, P<0.001 ***, P≥0.05 ns (not P=0.0440, W vs. K2, P=0.0271, K1 vs. K2, P=0.0271). The significant). spectral exponent of the mean periodograms was similar across conditions, indicating a general scaling across all frequencies (Fig.1e;W:−1.82±0.17,K1:−1.41±0.12,K2:−1.57±0.12;one-way rANOVA, principal effect of condition, P=0.1959; paired-samples Results t-test, W vs. K1, P=0.1719, W vs. K2, P=0.3077, K1 vs. K2, Perturbational complexity was independent of P=0.2838). behavioural responsiveness but was reduced by Across conditions, the same electrical stimulation induced a increasing ketamine dosage similar initial cortical excitation, quantified by the RMS ampli- Inpreviousstudiesinhumans,ketaminehasbeenfoundtoinduce tude of the early ERP deflections ( Fig. 2a and b; W: 52.63±5.13µV, unresponsiveness combined with vivid, dream-like experience, K1: 47.65±3.91µV, K2: 44.84±4.31µV; one-way rANOVA, prin- and high cortical complexity (Sarasso et al. 2015). Thus, to dis- cipal effect of condition, P=0.1943; paired-samples t-test, W sociate cortical complexity from behavioural responsiveness, we vs. K1: P=0.2942, W vs. K2: P=0.1613, K1 vs. K2: P=0.1132). recorded spontaneous and evoked EEG activity in six rats, during Consistently, the electrical pulse evoked an early, broad- wakefulness (W) and subsequent constant intravenous infusion band increment in power that was similar across conditions. Figure 1. The averaged power spectrum of spontaneous EEG was scaled up from wakefulness to ketamine anaesthesia and by increasing ketamine dosage. (a) Example of spontaneous EEG (5s) from the retrosplenial/parietal cortex (RS/PA) of one rat during wakefulness (W), light ketamine anaesthesia (ketamine 1, K1; administration rate: 1.8 mg/kg/min i.v.) and deep ketamine anaesthesia (ketamine 2, K2; administration rate: 3.5 mg/kg/min i.v.). (b) Mean periodograms from 16 channels and 6 rats exposed to the same conditions of A (shades represent SEM across rats). (c) Variations of mean high frequency power (HF, 20-80 Hz) and (d) low frequency power (LF, 1-4 Hz) are shown for each rat across conditions and increased from W to K1 and from K1 to K2. (e) The spectral exponents of the averaged periodograms across channels are also reported for each rat and condition 6 Arena et al. Figure 2. The spatiotemporal dynamics of ERPs revealed a drop in complexity from low to high dosage of ketamine. (a) Top, superimposition of mean ERPs from all 16 electrodes in response to single pulse stimulation (1 ms, 50 µA; dashed line) of the right secondary motor cortex (M2), from the same rat during wakefulness (W, left) and light and deep ketamine anaesthesia (K1, middle and K2, right respectively). One averaged ERP from the same channel over the right primary somatosensory cortex (S1) is in bold for clarity. Middle, power spectrogram (dB) and, bottom, phase-locking across trials (ITPC) from the channel shown in bold above (right S1). The black arrows indicate the moment of relaxation of HF power (dB ≤ 0) that follows the first response to stimulation. The ITPC drop time is indicated by vertical continuous lines. (b) The rms amplitude of the early ERP (first 50 ms from stimulus onset) averaged across channels is shown for all rats and conditions. (c) The ITPC drop time (in frequency range 5–80 Hz) averaged across channels is plotted for each rat and condition. (d) The ITPC drop time and the onset of later increased HF power was averaged across channels from all rats and shown for each condition. (e) Left, time courses of mean PCIST (moving windows of 100 ms, 50 ms overlap) and standard errors (shaded) are plotted for all conditions (horizontal lines indicate periods of statistically significant difference between conditions, P<0.05). Right, PCIST in range 0-600 ms is shown for each rat and condition This activation was quickly followed by a period of relaxation, principal effect of condition, P=0.001; paired-samples t-test, W with HF activity similar to or below the baseline (dB≤0), which vs. K1: P=0.7068, W vs. K2: P=0.0014, K1 vs. K2: P=0.01). lasted 150.52±12.04ms, averaged across rats and conditions The long-lasting response in wakefulness and K1 also showed a (Fig. 2a, see also Supplementary Fig. S2 for spectrograms and later increase of HF power in most of the channels (W: 100% of phase-lockingplotsinthebroadfrequencyrange,from1to80Hz). channels; K1: 92.36±4.33% of channels). Such HF activation was However, after this first dynamic, the ERP developed differently largely associated with the deterministic response, since its onset throughtimeindifferentconditions.Weevaluatedthephasecon- preceded the ITPC drop, as shown by comparing the times of sistency of the ERPs across trials, by computing the ITPC at each these events from the electrodes with such late increase in HF timepoint(Cohen2014;Arenaetal.2021)andquantifiedthedura- power (Fig. 2a and d; paired-samples t-test, ITPC drop time vs. −45 tion of the deterministic effect of the stimulus as ITPC drop time LateHFpoweronset;inW:n=94channels,P=7.8199*10 ;inK1: −13 (David et al. 2006; Pigorini et al. 2015; Rosanova et al. 2018; Arena n=87channels, P=2.6807*10 ). Incontrast, withhighketamine et al. 2021). During wakefulness and with low ketamine dose, dose, a late HF power activation was still detected in some elec- the ERPs showed long-lasting waveforms that were phase-locked trodes(46.6±14.93%ofchannels), butwasnotphase-locked,asit across trials, thus still deterministically caused by the stimula- occurred after the ITPC drop (Fig. 2a and d; paired-samples t-test, tion.Conversely,withthehighketaminedosage,thephase-locked ITPC drop time vs. Late HF power onset, in K2: n=44 channels, −5 response quickly died out, indicated by an earlier ITPC drop, P=2.1199*10 ). These results were also associated with an over- averaged across channels (Fig. 2a and c; W: 347.35±18.85ms, all reduction of the functional connectivity across channels in K1: 369.76±59.86ms, K2: 152.32±24.17ms; one-way rANOVA, the K2 condition, which also corresponded to reduced diversity Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 7 of cortical connectivity (Supplementary Fig. S3). Coherently, K2 The strong reduction in perturbational ST changed the time course of PCI , which was initially high and complexity was associated with a selective quicklydecayedsoonafterthestimulationinallconditions. How- deactivation of bilateral posteromedial cortex ST ever, in wakefulness and with low ketamine dose, PCI built up We found that high-dose ketamine caused a non-linear, dose- ST ST again, reaching similar values after 200ms. Then PCI dropped dependent drop of PCI (Fig. 2), but this drop was not associated again, faster in K1 than W, until fading. Conversely, with high with a clear overall transition of spontaneous EEG towards slow ST ketamine dose, PCI never recovered after the early decay and waves and reduced HF power (Fig. 1). Thus, we investigated remainedcloseto0(Fig.2aande).Thetimecoursewasinlinewith the spectral features of spontaneous activity from single cortical the perturbational complexity calculated for the entire response areas with higher detail, by computing the local, instantaneous ST window (0–600ms), thus PCI showed slightly higher values in HF/LF ratio (Fernandez et al. 2017; Siclari et al. 2017; Poulet and wakefulness than in K1 condition, but the difference was not Crochet 2018; see Fig. 3a). By averaging across time (from −5 to statistically significant for this sample [although a significant dif- 0 s) and trials for each electrode, we obtained a topographical ference was found for a larger data set (Arena et al. 2021), see distribution of the HF/LF ratio, which was similar between left ST Discussion]. With a high ketamine dose, however, PCI was sig- and right hemispheres, and this symmetry persisted in all condi- nificantlylowerthantheotherconditions( Fig.2e; W:78.47±8.44, tions (Fig. 3b; two-way ANOVA, principal effect of lateralization, K1: 59.57±5.77, K2: 23.75±6.55; one-way rANOVA, principal in W: P=0.6934, in K1: P=0.9902, in K2: P=0.9789). By com- −5 effect of condition, P=4.3026*10 ; paired-samples t-test, W vs. paring the mean HF/LF ratio across conditions, for each cortical K1, P=0.0535, W vs. K2, P=0.0028, K1 vs. K2, P=0.0005). area, we found that it significantly decreased from wakefulness Figure 3. The HF/LF ratio of spontaneous EEG in bilateral posteromedial cortex was selectively reduced by increasing ketamine dosage and correlated with the level of PCIST. (a) Up, Example of spontaneous EEG (5s) from the posteromedial cortex (RS/V2) of one rat during wakefulness (W), light ketamine anaesthesia (ketamine 1, K1) and deep ketamine anaesthesia (ketamine 2, K2) and below, the relative spectrogram and the ratio between high frequency (HF, 20–80 Hz) and low frequency (LF, 1–4 Hz) powers (HF/LF ratio) in time. (b) The colour maps show the topographical distributions (R-C: rostral-caudal) of the 16 EEG electrodes (small green circles) and, above, the spatial interpolations of the HF/LF ratio, averaged across time, trials and rats, for each condition. Below, the colour maps report the spatial interpolation of the t-scores (paired-samples t-test) from comparing the HF/LF ratio across conditions for each channel (left, wakefulness vs. ketamine 1; middle, wakefulness vs. ketamine 2; right, ketamine 1 vs. ketamine 2). The horizontal black line in the colour bar indicates the threshold for statistical significance (t5 =4.5258, Bonferroni–Holm corrected). White arrowheads indicate the channels with statistically significant differences across conditions. (c) Above, the colour map shows the spatial distribution of the coefficient of determination R2 from the correlation between PCIST and HF/LF ratio, across rats and conditions, for each channel. Below, the colour map reports the spatial interpolation of t-scores, assessing the statistical significance of the correlation for each channel. The horizontal black line in the colour bar indicates the threshold for statistical significance (t16 =3.1458, Bonferroni–Holm corrected). White arrowheads indicate the channels showing statistically significant correlations. The correlations of right M2C and left RS/V2 are reported in (d) with respective R2 and P-values 8 Arena et al. −3 to low ketamine dosage only in right M2/M1 cortex (HF/LF*10 , the variation of behavioural responsiveness could not be sepa- paired-samples t-test; W: 4.28±0.60, K1: 1.13±0.25, P=0.0372) rated from changes in perturbational complexity, hence a signif- −3 ST and S1 cortex (HF/LF*10 , paired-samples t-test; W: 6.48±0.69, icant correlation between PCI and the HF/LF ratio was detected K1: 3.48±0.51, P=0.0372), although weak, non-significant reduc- in both frontal cortex and posteromedial cortex (Supplementary tions were also seen elsewhere (Fig. 3b). In contrast, high doses of Fig. S5). ketamineselectivelyreducedthemeanHF/LFratiointhebilateral Inprinciple,thereductionoftheHF/LFratiocanbedetermined posteromedial cortex (left and right RS/V2 channels), indicating by an increase of LF powers, by a decrease of HF powers, or by a specific deactivation of this part of the cortex, with respect to a combination of the two events. Thus, in order to explain the the K1 condition (Fig. 3a and b, Supplementary Fig. S4; left RS/V2, reduction seen in the posteromedial cortex, we assessed the LF −3 HF/LF*10 , paired-samples t-test; K1: 1.82±0.19, K2: 1.09±0.17, andHFpowersatthelevelofeachcorticalarea, acrossconditions −3 P=0.0088; right RS/V2, HF/LF*10 , paired-samples t-test; K1: (Fig. 5a). At first, we tested for possible lateralization, without 1.50±0.20, K2: 0.91±0.14, P=0.0431). In line with these results, finding any clear difference between left and right hemispheres the HF/LF ratios at both bilateral RS/V2 and right M2/M1 cortex in each experimental condition, for both LF powers (Fig. 5a; two- were also reduced from wakefulness to the K2 condition (Fig. 3b). way ANOVA, principal effect of lateralization, in W: P=0.7581, To test whether the level of activation of any cortical area was in K1: P=0.9459, in K2: P=0.9589;), as well as for HF power effectivelyabletopredictthecomplexityofglobalcorticaldynam- (Fig. 5a; two-way ANOVA, principal effect of lateralization, in W: ics,weassessedpossibleregionalcorrelationsbetweenHF/LFratio P=0.8284,inK1:P=0.9664,inK2:P=0.6748).However,LFpowers ST andPCI (Fig.3c).WefoundthattheHF/LFratiofromthesponta- weredifferentiallydistributedacrosscorticalareas,andanoverall neous activity of bilateral posteromedial cortex (left, right RS/V2) increase in power could be detected in relation to the increment ST was highly, linearly correlated with the PCI value across con- of ketamine dosage (Fig. 5b; two-way rANOVA, principal effect ditions (Fig. 3c and d; left RS/V2: linear fit, R =0.590, P=0.0016; of cortical areas: P=0.0009, principal effect of ketamine dosage: right RS/V2: linear fit, R =0.568, P=0.0024). A weaker but sig- P=0.0132). Similar effects were also found for HF powers (Fig. 5b; nificant correlation was also identified only at the level of the two-way rANOVA, principal effect of cortical areas: P=0.0109, right secondary motor cortex (Fig. 3c and d; right M2 : linear fit, principal effect of ketamine dosage: P=0.0041), and were in line R =0.417, P=0.0264). with the scaling up of the mean periodograms, seen by averag- Next, we assessed the same correlations between local HF/LF ing across electrodes (Fig. 1). Nevertheless, by comparing powers ST ratio and global PCI , this time distinguishing between brain between K1 and K2 conditions for every single channel, a signifi- states and state transitions. We first evaluated the correla- cant increase of LF power was only found at the level of bilateral tion in wakefulness condition alone, where both behavioural posteromedial cortex (left and right RS/V2), while the increase of responsiveness and high perturbational complexity were present HF power was more spatially sparse, without a clear clusteriza- (Fig. 4a and b; data from three different recordings, performed in tion (Fig. 5b). To more directly compare the relative increment three different days on the same rats were considered only in of powers that occurred by increasing ketamine dosage, we com- this condition, to increase the number of observations). Then, putedtheratiobetweenK2andK1conditions(K2/K1), forbothHF we repeated the estimation of the correlations by considering the and LF powers, at the level of each cortical area (Fig. 5c). Overall, conditions of wakefulness and low dosage of ketamine, when per- the power increments were differentially distributed across corti- turbationalcomplexityisstillhigh,butbehaviourtransitionsfrom cal areas, and an overall difference between HF and LF could not responsiveness to unresponsiveness (Fig. 4c and d). Finally, we be detected (Fig. 5c; two-way rANOVA, principal effect of cortical evaluatedthecorrelationsinconditionsoflowandhighketamine areas: P=0.0325, principal effect of frequency range: P=0.0744). dosages, when only the variation of perturbational complex- However, a clear interaction effect between cortical areas and ity occurred, within the same unresponsive behavioural state frequency ranges was identified ( Fig. 5c; two-way rANOVA, inter- (Fig. 4e and f). With this, we attempted to identify possible roles action effect: P=0.0056), thus indicating the relation between the that specific cortical regions might have in specific state transi- increments of HF and LF powers changed depending on the spe- tions.Withinthewakefulnesscondition,wecouldnotidentifyany cific cortical area. Indeed, LF power increased significantly more ST significant correlation between PCI and HF/LF ratio at the level than HF power selectively in the bilateral posteromedial cortex of any cortical area. Nevertheless, the correlations with higher (left and right RS/V2), conversely, HF power had the tendency R , which were also closer to the threshold for statistical signif- to increase more than LF power in parieto-frontal areas, even if icance, seemed to be clustered in the secondary motor cortex without reaching statistical significance ( Fig. 5c). (Fig. 4a and b; right M2 : linear fit, R =0.377, P=0.0538; left RS/V2: linear fit, R =0.077, P=1). Likewise, the HF/LF ratio could Discussion ST not clearly predict PCI between wakefulness and low ketamine dose, when only behavioural responsiveness changed, for any of We reanalysed multichannel EEG data from wakefulness (W) and the cortical areas (Fig. 4c and d; right M2 : linear fit, R =0.386, two levels of ketamine anaesthesia (K1, K2) in rats, based on pre- P=0.2173; left RS/V2: linear fit, R =0.271, P=0.4124). On the vious experiments (Arena et al. 2021). To assess the capacity for ST otherhand,byconsideringonlythevariationsinducedbyincreas- consciousness in each state, we computed PCI (Comolatti et al. ingketaminedosage(K1andK2),wefoundasignificantandstrong 2019) and compared these results to region-specific estimates of correlation selectively associated with the left RS/V2 cortex, thus cortical activation, assessed by HF/LF ratio of spontaneous EEG indicating that the state of activation or deactivation of the pos- activityforeachelectrode(Fernandezetal.2017;Siclarietal.2017; teromedialcortexcouldeffectivelylinearlypredictthecomplexity Poulet and Crochet 2018). level of the entire cortical network and its breakdown (Fig. 4e and Inhumans,ketamineanaesthesiahasbeenobservedtoinduce f; right M2 : linear fit, R =0.064, P=0.1; left RS/V2: linear fit, a dissociated state of behavioural unresponsiveness with dream- R =0.568, P=0.0370). Coherently with these results, by consider- like, vivid conscious experiences (Collier 1972; Sarasso et al. 2015). ingonlytheconditionsofwakefulnessandhighdoseofketamine, Wetookadvantageofthisandusedacontrolledintravenousinfu- Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 9 Figure 4. HF/LF ratio of spontaneous EEG from posteromedial cortex selectively correlated with PCIST in conditions of behavioural unresponsiveness, with light and high ketamine anaesthesia. (a, b) The putative correlations between PCIST and HF/LF ratio are shown in wakefulness across rats and 3 recording sessions (Wr1 Wr2 Wr3), during the presence of both behavioural responsiveness and high cortical complexity. (c, d) The same putative correlations are shown across rats and across conditions of wakefulness (W) and low ketamine anaesthesia (K1), when loss of behavioural responsiveness occurred, but high cortical complexity persisted. (e, f) Correlations between PCIST and HF/LF ratio are also computed and shown across rats and across conditions of low and high ketamine doses (K1 and K2, respectively) when reduction of cortical complexity occurred and behavioural unresponsiveness was unchanged. In panels A, C, E the colour maps show the spatial interpolation of R2 and t-scores (above and below respectively, for all panels) from the correlations of all channels. The horizontal black line in the colour bar indicates the threshold for statistical significance. The white arrowhead indicates channels with statistically significant correlation between PCIST and HF/LF ratio. In panels B, D, F the correlations and/or absence of correlation of right secondary motor cortex (M2C, left side of the panel) and left posteromedial cortex (RS/V2, right side of the panel) are reported with relative R2 and P-values, (corrected for multiple comparisons). Dashed line is used to indicate a correlation close to statistical significance (panel B, left), while a continuous line indicates a statistically significant linear fitting and correlation (panel F, right) 10 Arena et al. Figure 5. The reduction of HF/LF ratio from low to high dosage of ketamine in the posteromedial cortex was explained by a selective higher increment of LF powers with respect to HF. (a) The colour maps show the topographical distributions (R-C: rostral-caudal) of the 16 EEG electrodes (small green circles) and the spatial interpolations of the LF power (1–4 Hz, above) and HF power (20–80 Hz, below), averaged across time, trials and rats, for each condition. (b) Left, averaged HF (dashed line) and LF (continuous line) powers across hemispheres and rats are shown for each cortical area, during both low and high ketamine dosage (K1 and K2, respectively). On the right, the colour maps show the spatial interpolation of t-scores from comparing LF powers (up) and HF powers (bottom) between K1 and K2 conditions, for each channel. The horizontal black lines in the colour bar indicate the threshold for statistical significance (t5 =4.5258, Bonferroni–Holm corrected). White arrowheads indicate channels with statistically significant differences across conditions. (c) Left, the ratio between low and high doses of ketamine is reported for both HF and LF powers at the level of each cortical area, thus showing the power increments induced by deep ketamine anaesthesia (K2). Right, spatial interpolation of the t-scores from comparing LF power increment with HF power increments induced by increasing ketamine dosage, for each channel. The horizontal black line in the colour bar indicates the threshold for statistical significance (t5 =4.5258, Bonferroni–Holm corrected). White arrowheads indicate the channels with statistically significant differences between the two frequency ranges sionofketaminetodissociatethecapacityforconsciousnessfrom experiments (Arena et al. 2021) might be due to the smaller sam- ST responsiveness.InagreementwithaPCIstudyinhumans(Sarasso plesizehere.However,inbothWandK1conditions,PCI builtup ST et al. 2015), we found that PCI was not significantly changed again after an initial decay, reaching similar values between 200 during the unresponsive state caused by light ketamine anaes- and 300ms after the stimulus onset (Fig. 2). Indeed, our results thesia (K1) compared to wakefulness (Fig. 2), even if it tended to supporttheideathathighperturbationalcomplexityisassociated be lower in the K1 condition. This tendency is more in line with with a capacity to sustain long-lasting sequences of determinis- results from a larger dataset from rats (Arena et al. 2021) and tic activations (Mukovski et al. 2007; Pigorini et al. 2015; Rosanova ST in agreement with the time course of PCI , which showed both et al. 2018; Arena et al. 2021) as shown by the long-lasting phase- periods of similarities and differences between wakefulness and locked cortical ERPs (Fig. 2) and by the strong yet diverse global light ketamine anaesthesia. The difference in the statistical sig- connectivity observed here (Supplementary Fig. S3). Moreover, ST ST nificance of the PCI values reported here compared to previous the widespread responses to stimulation required for high PCI Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 11 may also indicate a capacity for the kind of global broadcast- behavioural responsiveness and capacity for consciousness seem ing required for consciousness according to GNW (Mashour et al. more strongly connected and difficult to dissociate ( Sarasso et al. 2020), aswellasforthebraintofunctionasanintegratedanddif- 2015; Arena et al. 2021). ferentiated whole as is required for consciousness in IIT (Casali Next, we explored the spectral properties of ongoing cortical et al. 2013; Tononi et al. 2016). Although a partially reduced level activity in order to identify possible associations with the differ- ST of consciousness might be inferred from the tendency of PCI to ent combinations of behavioural responsiveness and capacity for ST be lower with light ketamine anaesthesia compared to wakeful- consciousness, as assessed by PCI (Comolatti et al. 2019). The ness, we observed relatively high spatiotemporal complexity dur- spontaneous EEG activity during ketamine anaesthesia was char- ing both conditions, with long-lasting and well-integrated phase- acterized by a widespread, dose-dependent increase in HF power locked cortical activations. Thus, taken together, our findings are compared to wakefulness (Figs 1 and 5), in agreement with pre- compatible with a fully, or at least partially, preserved capacity vious findings ( Maksimow et al. 2006; Akeju et al. 2016; Li and for consciousness during both wakefulness and light ketamine Mashour 2019). HF oscillations are usually associated with neu- anaesthesia, independently of behavioural responsiveness. ronal firing ( Steriade et al. 1996, 2001; Mukovski et al. 2007) and However, ketamine has also been shown to produce fluctua- cortical activation (Fernandez et al. 2017; Siclari et al. 2017; Poulet tionsbetweenhighandlowspatiotemporalcomplexityofsponta- and Crochet 2018). Thus, the observed increase in HF power is neous EEG in humans, following bolus injection of an anaesthetic consistent with the enhanced presynaptic release occurring after dose (Li and Mashour 2019). This phenomenon may be caused ketamine administration (Ferro et al. 2017), and with the idea by the unstable pharmacokinetic of the bolus injection, suggest- that ketamine, via its antagonism of NMDA receptors, might ing a dose-dependent effect. Consistently, it was reported that mainly inhibit GABAergic interneurons, producing a state of over- soon after bolus injection, when ketamine plasma level is high, allcorticalexcitation(Seamans2008). Interestingly, LFpoweralso the spontaneous EEG can assume a gamma-burst dynamic, with increased from wakefulness to ketamine anaesthesia in a dose- slow oscillations that interrupted an enhanced HF, gamma activ- dependent manner (Figs 1 and 5), in agreement with previous ity (Akeju et al. 2016). Conversely, when ketamine plasma level reports (Akeju et al. 2016; Li and Mashour 2019). Thus, ketamine was reduced, about 10min after bolus injection, the sponta- induced a scaling up of the entire power spectrum of the sponta- neous EEG activity was characterized by uninterrupted, stable neous EEG, suggesting a maintained balance between the inhibi- gamma/beta activity, with reduced slow frequency power (Akeju tion and excitation of the overall cortical network underlying the et al. 2016), and the spatiotemporal complexity of spontaneous EEG signal (Gao et al. 2017). This was supported by the observa- EEG stabilized at wakefulness-like values (Li and Mashour 2019). tionofasimilarspectralexponentacrossconditions(Fig.1),which Importantly, the spatiotemporal complexity of the thalamocorti- has been hypothesized to indicate an aroused or conscious state cal system is thought to be a promising neuronal correlate for the (Colombo et al. 2019; Lendner et al. 2020; Arena et al. 2021). Why ST capacity for consciousness (Tononi and Edelman 1998; Dehaene thendidweobservethestrongreductionofPCI inthetransition 2014; Koch et al. 2016; Sarasso et al. 2021), and was empirically from low to high ketamine condition (Fig. 2)? found to correlate with conscious experience, in both sponta- Onehypothesisisthatspecificcorticalcircuitsmightbepartic- neous (Ferenets et al. 2006; Schartner et al. 2015; Demertzi et al. ularly relevant for sustaining complex neuronal interactions and 2019) and perturbed activity (Massimini et al. 2009; Casali that the activation state of these circuits might diverge from the et al. 2013; Sarasso et al. 2015; Casarotto et al. 2016; Rosanova averagedynamicoftheentirecorticalnetwork.Indeed,itisknown et al. 2018; Comolatti et al. 2019). Thus, it is conceivable that that transient and local cortical deactivations or activations can lowandhigh dosesofketamine, althoughboth causebehavioural occur and dissociate from the global brain state, such as with unresponsiveness, may still induce quite different states of con- local sleep during wakefulness (Murphy et al. 2011; Vyazovskiy sciousness: the low dose may allow vivid but covert dream-like et al. 2011; Fernandez et al. 2017), possibly modifying the capac- experiences to occur (Collier 1972; Sarasso et al. 2015), while ity for behaviour and/or conscious experience (Vyazovskiy et al. the high dose may cause dreamless unconsciousness, due to 2011; Fernandez et al. 2017; Siclari et al. 2017; Poulet and Crochet different, dose-dependent effects on cortical complexity (Li and 2018). For example, it has been shown that localized reduction of Mashour 2019). LF power and increased HF activity (high HF/LF ratio) within the We tested this hypothesis by repeating the same electrophysi- posterior cortex is strongly associated with dream experience in ological recording/stimulations in the same rats during the con- humans during deep stages of sleep (Siclari et al. 2017), a state stant intravenous infusion of ketamine at a higher rate, which dominated by LF activity that is often linked to unconsciousness gives a more constant systemic concentration than with bolus (Tononi and Massimini 2008). Thus, to uncover the role and the ST injection. In this high-dose condition (K2), we found that PCI state of activation of specific areas, we similarly measured the HF was strongly reduced along with an earlier interruption of phase- (20–80Hz)/LF(1–4Hz)powerratiofromthespontaneousactivityof locked response (Fig. 2) and with a drastic reduction of cortical all the 16 epidural electrodes and compared across experimental functional connectivity and diversity (Supplementary Fig. S3). conditions. These results are compatible with a reduced capacity to inte- Although both light and deep ketamine anaesthesia caused an grateorbroadcastinformationwithintheglobalcorticalnetwork, unresponsive behavioural state, we were able to identify regional thus suggesting a reduced capacity for consciousness during variationsinHF/LFratiothatwererelatedtochangesintheglobal ST deep ketamine anaesthesia (K2). This also indicates a clear dose- PCI value. Strikingly, the bilateral posteromedial cortex was the dependent effect, which was only suggested by previous experi- only region that showed a consistent reduction of HF/LF ratio, ST ments(Akejuetal. 2016; Liand Mashour 2019). Inagreementwith from low to high ketamine dosage, along with the drop in PCI Li and Mashour (2019), our results demonstrate that ketamine (Fig. 3). The reduced ratio within the posteromedial cortex indi- anaesthesia can be used to dissociate behavioural responsive- cated a local deactivation (Poulet and Crochet 2018), which was ness from cortical complexity, thus representing a ‘unique tool explainedbyalargerincreaseofLFthanHFpowerinducedbythe to probe different states of consciousness’ (Li and Mashour 2019). increased ketamine dosage (Fig. 5). This local power imbalance This is in contrast to other general anaesthetics, for which might indicate a particularly pronounced gamma-burst activity 12 Arena et al. pattern (Supplementary Fig. S4), typical of high ketamine plasma induce slow-wave oscillations and synchronized, rhythmic neu- levels, with slow waves interrupting enhanced HF activity (Akeju ronal silencing selectively in the retrosplenial cortex in mice, et al. 2016; Li and Mashour 2019). Consistently, the HF/LF ratio affecting behaviour (Vesuna et al. 2020). The retrosplenial cortex ST over the posteromedial cortex strongly correlated with the PCI is a particularly highly integrated area, within the medial cortical level during ketamine administration (Fig. 4) and across all con- subnetwork of rodents (Zingg et al. 2014). It receives information ditions (Fig. 3), but not within wakefulness alone or between from the claustrum, indirectly from the hippocampus through ST wakefulness and light ketamine anaesthesia (Fig. 4), where PCI the subiculum, and it is directly interconnected with several sen- did not change substantially. These results indicate that the local sory areas (visual, auditory, and somatosensory) and high-order state of the RS/V2 cortex is associated with the capacity for long- associative areas, including the medial frontal cortex. Thus, it lasting, broadly integrated and differentiated cortical activations is likely to play important roles in multisensory integration and ST asassessedbyPCI .Thuspossibly,theposteromedialcortexmay also integration with higher functions such as episodic memory, play an important role in sustaining the capacity for conscious- spatial navigation, and motor planning (Zingg et al. 2014). Given ness, in a general agreement with both IIT and GNW (Tononi et al. this, it is not surprising that deactivation of this area (low HF/LF 2016;Mashouretal.2020).Inotherwords,ourresultsmaysupport ratio, Fig. 3), due to enhanced LF activity (Fig. 5), could be associ- the hypothesis that a selective deactivation of the posteromedial atedwithdisruptionofwidespreadintegrationofcomplexcortical ST cortex—as indicated by the localized decrease in HF/LF power— interactions as seen here, with the drop of PCI at high ketamine is correlated with, and may even underlie, a sharp reduction of dosage (Fig. 2). In other words, there is reason to believe that the brain’s capacity to globally broadcast information or to func- the specific deactivation of a region in the posteromedial cortex tion as an integrated and differentiated whole that is capable of can be directly involved in breaking down the properties required sustaining consciousness. This is complementary to the occur- for sustaining a capacity for consciousness. However, the asso- rence of dreaming during deep stages of sleep in humans, which ciative frontal cortex is also highly integrated (Zingg et al. 2014; was associated with reduced LF activity and increased HF power BarthasandKwan2017)andseveralpiecesofevidencesuggestits in posterior cortical areas, indicating a local cortical activation involvement in conscious processing (Del Cul et al. 2009; Kapoor [high HF/LF ratio, (Siclari et al. 2017)]. Moreover, our findings are et al. 2020; Weilnhammer et al. 2021; Levinson et al. 2021). More- consistent with the reduced functional integration and diversity over, our results cannot exclude that a selective inhibition or that was seen in the posterior regions of the brain’s default mode lesioning of M2, within a global activated state, could also dis- network during unconsciousness, in humans (Luppi et al. 2019). rupt cortical complexity. Thus, in future experiments, it will be In contrast, light ketamine anaesthesia produced a signif- important to causally control the state of activation of single cor- ST icant reduction of HF/LF ratio compared to wakefulness only tical areas, with local intervention, in combination with PCI , to over the right primary motor and somatosensory cortex (Fig. 3). betteraddresstheroleofspecificcorticalregionsinsustainingthe This was consistent with the loss of behavioural responsiveness capacity for consciousness. inducedbythelowketaminedose, andpossiblywithananalgesic The main findings presented here are compatible with several effect. A correlation between the HF/LF ratio of the right sec- theories of consciousness, as it is widely agreed that some sort ST ondary motor cortex and PCI was also found across conditions of long-range interactions within the brain are required to sus- (Fig.3). However, thisrelationwasatleastpartiallyexplainedbya tain its capacity for consciousness. For example, GNW requires wakefulness-specific weak correlation ( Fig. 4), which could reflect information to be globally broadcast (Baars 2005; Dehaene et al. behaviouralvariationswithinthesamestate,suchasactive/quiet 2011; Mashour et al. 2020), IIT requires the physical substrate wakefulness or transient attentional loading. These findings are of consciousness to be integrated (Tononi and Edelman 1998; indeedconsistentwiththeconnectionbetweentheprefrontalcor- Tononietal. 2016), andatleastsomehigher-ordertheoriesrequire tex and behavioural state, which was recently demonstrated by long-range interaction to maintain the capacity to form repre- local cortical injections of carbachol, during general anaesthesia, sentations in associative cortices about first-order states in early in rats (Pal et al. 2018). In these experiments, during sevoflu- sensory regions (Lau and Rosenthal 2011; Brown et al. 2019). raneanaesthesia,localcholinergicmodulationofbothassociative Nonetheless, our results suggest that the kind of global corti- frontalandparietalcorticesproducedatransitionfromslow-wave calcomplexityassociatedwithconsciousexperiencebreaksdown EEG activity to low voltage, wakefulness-like, fast oscillations when ketamine specifically deactivates the posteromedial cortex (Pal et al. 2018). EEG temporal complexity also increased with (Figs 3 and 4). Moreover, the finding that the ketamine-induced both local cortical activations (Pal et al. 2020). However, only the deactivation of primary motor and somatosensory cortex was cholinergic activation of the prefrontal cortex was able to restore associated with loss of behavioural responsiveness, but not with ST wakefulness-like motor activity (Pal et al. 2018), showing how cor- significant changes in PCI (see Fig. 3), or in durable and well- ticalcomplexitycanbeeffectivelydissociatedfrombehaviour(Pal integrated cortical activations (Fig. 3, Supplementary Fig. S3), et al. 2020). Unfortunately, PCI was not tested, and the state of seems to provide an example that some cortices may be deacti- consciousness was only inferred by the simple motor activity (Pal vated without disrupting the normal capacity for consciousness. et al. 2018, 2020), while it is known that these phenomena can Thus, theories of consciousness should be able to explain why dissociate in humans, under several circumstances (Blumenfeld some regions may be deactivated without any apparent effect 2005; Owen et al. 2006; Ghoneim et al. 2009; Noreika et al. 2011; on the brain’s overall capacity for consciousness, while others Castelnovo et al. 2018; Linassi et al. 2018). are more critical. We also observed that changes in the HF/LF In our experimental setting, the electrodes over the postero- ratio of secondary motor regions during wakefulness had a ten- ST medial cortex cover both the medial part of the secondary visual dency of correlating with changes in PCI (Fig. 4). This weak cortex and the caudal part of the retrosplenial cortex (Supple- relation may represent a modulatory effect of frontal regions mentary Fig. S1). Interestingly, the dose-dependent deactiva- on the overall capacity for consciousness during wakefulness. tion of this cortical region is reminiscent of a recent finding, in As these variations did not result from controlled intervention, whichsub-anaestheticbolusinjectionsofketaminewerefoundto theysuggestthatspontaneouschangesinfrontalactivity(HF/LF), Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 13 likely involving changes in related cognitive functions [working authors participated in the interpretation of results and revi- memory, attention, cognitive control, planning, decision-making, sion of the manuscript, and approved the final version of the etc.; (Dalley et al. 2004)], might reflect spontaneous modulation manuscript. of cortical complexity within limits of normal wakefulness. This seems to be compatible with theories of consciousness that Conflict of interest statement also attribute an active role to the frontal cortices of selec- The authors declare that they have no financial competing tively modulating the contents of consciousness and attention interests. at any given moment (Baars 2005; Dehaene et al. 2011; Lau and Rosenthal 2011; Helfrich et al. 2018; Brown et al. 2019; Mashour References et al. 2020). Of course, these findings are not conclusive, as the specific Akeju O, Song AH, Hamilos AE et al. Electroencephalogram sig- regionalchangesinHF/LFobservedaredependentonfluctuations natures of ketamine anesthesia-induced unconsciousness. Clin in ongoing activity as opposed to interventional inactivation. Fur- Neurophysiol 2016;127:2414–22. thermore, it is not necessarily the case that the changes in HF/LF Arena A, Comolatti R, Thon S et al. General anesthesia disrupts observed in different regions were caused by the same underlying complex cortical dynamics in response to intracranial electrical processes, and it is also uncertain that they were always indica- stimulation in rats. eNeuro 2021;8:4. tive of a deactivation of the region. To address these issues, we Arena A, Thon S, Storm J. PCI-like measure in rodents. Hum Brain aim for future experiments with controlled, direct inactivation Project Neuroinf Platform 2019. ST of individual cortical regions while measuring PCI from awake Baars BJ. Global workspace theory of consciousness: toward a cog- rodents. nitive neuroscience of human experience. In: Laureys S (ed.), The Boundaries of Consciousness: Neurobiology and Neuropathology. Amsterdam: Elsevier, 2005, 45–53. Conclusion Barthas F, Kwan AC. Secondary motor cortex: where “sensory” By comparing EEG in light and deep ketamine anaesthesia in meets “motor” in the rodent frontal cortex. Trends Neurosci rats, we dissociated changes in global and local cortical dynam- 2017;40:181–93. ics related to loss of behavioural responsiveness from those likely Blumenfeld H. Consciousness and epilepsy: why are patients ST related to capacity for consciousness (assessed by PCI ). Light with absence seizures absent? Prog Brain Res 2005;150: ketamine anaesthesia induced an unresponsive state with appar- 271–86. ent deactivation of primary somatosensory and motor regions Boly M, Massimini M, Tsuchiya N et al. Are the neural correlates of (indicated by locally reduced HF/LF power ratio). The mean consciousness in the front or in the back of the cerebral cortex? ST PCI value was somewhat lower than in wakefulness, but long- Clinical and neuroimaging evidence. J Neurosci 2017;37:9603–13. lasting phase-locked responses with high and diversified func- Brown EN, Lydic R, Schiff ND. General anesthesia, sleep, and coma. tional connectivity suggested a dissociated state, with largely N Engl J Med 2010;363:2638–50. conserved consciousness. In contrast, deep ketamine anaesthe- BrownR,LauH,LeDouxJE.Understandingthehigher-orderapproach ST sia strongly reduced PCI , with early interruption of phase- to consciousness. Trends Cogn Sci 2019;23:754–68. locked response and disruption of functional connectivity, sug- Casali AG, Gosseries O, Rosanova M et al. A theoretically based gesting a more fully unconscious state. The latter was associated index of consciousness independent of sensory processing and with highly selective deactivation of the posteromedial cortex, behavior. Sci Transl Med 2013;5:198ra105. which was not seen during light ketamine anaesthesia, suggest- Casarotto S, Comanducci A, Rosanova M et al. Stratification of unre- ing a significant role for this cortical region in the capacity for sponsive patients by an independently validated index of brain consciousness. complexity. Ann Neurol 2016;80:718–29. Castelnovo A, Lopez R, Proserpio P et al. deep stages of sleep para- Supplementary data somnias as disorders of sleep-state dissociation. Nat Rev Neurol 2018;14:470–81. Supplementary data is available at NCONSC online. Chernik DA, Gillings D, Laine H et al. Validity and reliability of the Observer’s Assessment of Alertness/Sedation Scale: study with Data availability intravenous midazolam. J Clin Psychopharmacol 1990;10:244–51. Data will be available in the public repository EBRAINS upon CohenMX.AnalyzingNeuralTimeSeriesData:TheoryandPractice.1edn. publication at the following DOIs: 10.25493/S0DM-BK5; 10.25493/ Cambridge, MA: MIT Press, 2014. E9EF-6SB. Collier BB. Ketamine and the conscious mind. Anaesthesia 1972;27:120–34. Colombo MA, Napolitani M, Boly M et al. The spectral exponent Funding of the resting EEG indexes the presence of consciousness dur- Thisproject/researchreceivedfundingfromtheEuropeanUnion’s ing unresponsiveness induced by propofol, xenon, and ketamine. Horizon2020FrameworkProgrammeforResearchandInnovation NeuroImage 2019;189:631–44. under the Specific Grant Agreements No. 945539 (Human Brain Comanducci A, Boly M, Claassen J et al. Clinical and advanced neu- Project SGA3, JFS), No. 785907 (Human Brain Project SGA2; JFS), rophysiologyintheprognosticanddiagnosticevaluationofdisor- and No. 720270 (Human Brain Project SGA1; JFS). ders of consciousness: review of an IFCN-endorsed expert group. Clin Neurophysiol 2020;131:2736–65. Author contributions Comolatti R, Pigorini A, Casarotto S et al. A fast and gen- A.A. designed the experiments; A.A., S.T. performed experi- eral method to empirically estimate the complexity of brain ments and collected data; A.A. analysed data; R.C. performed responses to transcranial and intracranial stimulations. Brain PCIST analysis; A.A., B.E.J., J.F.S. wrote the manuscript; all Stimul 2019;12:1280–9. 14 Arena et al. Compte A, Sanchez-Vives MV, McCormick DA et al. Cellular and Laureys S, Celesia GG, Cohadon F et al. Unresponsive wakefulness network mechanisms of slow oscillatory activity (<1 Hz) and syndrome: a new name for the vegetative state or apallic syn- wave propagations in a cortical network model. J Neurophysiol drome. BMC Med 2010;8:68. 2003;89:2707–25. Lendner JD, Helfrich RF, Mander BA et al. An electrophysiological Dalley JW, Cardinal RN, Robbins TW. Prefrontal executive and cogni- marker of arousal level in humans. eLife 2020;9:e55092. tive functions in rodents: neural and neurochemical substrates. LevinsonM,PodvalnyE,BaeteSHetal.Corticalandsubcorticalsigna- Neurosci Biobehav Rev 2004;28:771–84. tures of conscious object recognition. Nat Commun 2021;12:2930. David O, Kilner JM, Friston KJ. Mechanisms of evoked and induced Li D, Mashour GA. Cortical dynamics during psychedelic and anes- responses in MEG/EEG. NeuroImage 2006;31:1580–91. thetized states induced by ketamine. NeuroImage 2019;196:32–40. Dehaene S. Consciousness and the Brain: Deciphering How the Brain Linassi F, Zanatta P, Tellaroli P et al. Isolated forearm technique: a Codes Our Thoughts. Reprint edn. Westminster, London, England: meta-analysis of connected consciousness during different gen- Penguin Publishing Group, 2014. eral anaesthesia regimens. Br J Anaesth 2018;121:198–209. Dehaene S, Changeux J-P, Naccache L. The global neuronal Lombardi F, Herrmann HJ, de Arcangelis L. Balance of excitation and workspace model of conscious access: from neuronal architec- inhibition determines 1/f power spectrum in neuronal networks. tures to clinical applications. In: Dehaene S, Christen Y (eds), Chaos 2017;27:047402. Characterizing Consciousness: From Cognition to the Clinic? Berlin, Loomis AL, Harvey EN, Hobart GA. Cerebral states during sleep, as Heidelberg: Springer, 2011, 55–84. studied by human brain potentials. J Exp Psychol 1937;21:127. DelCulA,DehaeneS,ReyesPetal.Causalroleofprefrontalcortexin Luppi AI, Craig MM, Pappas I et al. Consciousness-specific dynamic thethresholdforaccesstoconsciousness.Brain2009;132:2531–40. interactions of brain integration and functional diversity. Nat Demertzi A, Tagliazucchi E, Dehaene S et al. Human conscious- Commun 2019;10:4616. ness is supported by dynamic complex patterns of brain signal Luppi AI, Golkowski D, Ranft A et al. Brain network integration coordination. Sci Adv 2019;5:eaat7603. dynamics are associated with loss and recovery of consciousness Esser SK, Hill SL, Tononi G. Sleep homeostasis and cortical synchro- induced by sevoflurane. Hum Brain Mapp 2021;42:2802–22. ¨ ¨ ¨ nization:I.Modelingtheeffectsofsynapticstrengthonsleepslow MaksimowA, Sarkela M, Långsjo JW et al. Increase in high frequency waves. Sleep 2007;30:1617–30. EEG activity explains the poor performance of EEG spectral Ferenets R, Lipping T, Anier A et al. Comparison of entropy and com- entropy monitor during S-ketamine anesthesia. Clin Neurophysiol plexity measures for the assessment of depth of sedation. IEEE 2006;117:1660–8. Trans Biomed Eng 2006;53:1067–77. Mashour GA, Roelfsema P, Changeux JP, Dehaene S. Conscious pro- FernandezLMJ,ComteJC,LeMerrePetal.Highlydynamicspatiotem- cessing and the global neuronal workspace hypothesis. Neuron poral organization of low-frequency activities during behav- 2020;105:776–98. ioral states in the mouse cerebral cortex. Cereb Cortex 2017;27: Massimini M, Huber R, Ferrarelli F et al. The sleep slow oscillation as 5444–62. a traveling wave. J Neurosci 2004;24:6862–70. Ferro M, Lamanna J, Ripamonti M et al. Functional mapping of Massimini M, Huber R, Ferrarelli F et al. A perturbational approach brain synapses by the enriching activity-marker SynaptoZip. Nat for evaluating the brain’s capacity for consciousness. Prog Brain Commun 2017;8:1229. Res 2009;177:201–14. Funk CM, Peelman K, Bellesi M et al. Role of somatostatin-positive Mukovski M, Chauvette S, Timofeev I et al. Detection of active and cortical interneurons in the generation of sleep slow waves. J silentstatesinneocorticalneuronsfromthefieldpotentialsignal Neurosci 2017;37:9132–48. during slow-wave sleep. Cereb Cortex 2007;17:400–14. Gao R, Peterson EJ, Voytek B. Inferring synaptic excitation/inhibition Murphy M, Huber R, Esser S et al. The cortical topography of local balance from field potentials. NeuroImage 2017;158:70–8. sleep. Curr Top Med Chem 2011;11:2438–46. GaoS,CalderonDP.Robustalternativetotherightingreflextoassess Nielsen TA. A review of mentation in REM and deep stages of sleep: arousal in rodents. Sci Rep 2020;10:20280. “covert” REM sleep as a possible reconciliation of two opposing GhoneimMM,BlockRI,HaffarnanMetal.Awarenessduringanesthe- models. Behav Brain Sci 2000;23:851–66. discussion 904–1121. ¨ ´ ´ sia: risk factors, causes and sequelae: a review of reported cases Noreika V, Jylhankangas L, Moro L et al. Consciousness lost and in the literature. Anesth Analg 2009;108:527–35. found:subjectiveexperiencesinanunresponsivestate.BrainCogn Giacino JT, Kalmar K, Whyte J. The JFK coma recovery scale-revised: 2011;77:327–34. measurementcharacteristicsanddiagnosticutility.ArchPhysMed Odegaard B, Knight RT, Lau H. Should a few null findings fal- Rehabil 2004;85:2020–9. sify prefrontal theories of conscious perception? J Neurosci Helfrich RF, Fiebelkorn IC, Szczepanski SM et al. Neural mechanisms 2017;37:9593–602. of sustained attention are rhythmic. Neuron 2018;99:854–65.e5. Owen AM, Coleman MR, Boly M et al. Detecting awareness in the Kapoor V, Dwarakanath A, Safavi S et al. Decoding the contents of vegetative state. Science 2006;313:1402. consciousness from prefrontal ensembles. BioRxiv 2020. https:// Pal D, Dean JG, Liu T et al. Differential role of prefrontal and www.biorxiv.org/content/10.1101/2020.01.28.921841v1.abstract. parietal cortices in controlling level of consciousness. Curr Biol Koch C, Massimini M, Boly M, Tononi G. Neural correlates 2018;28:2145–52.e5. of consciousness: progress and problems. Nat Rev Neurosci Pal D, Li D, Dean JG et al. Level of consciousness is dissociable from 2016;17:307–21. electroencephalographic measures of cortical connectivity, slow Lau H, Rosenthal D. Empirical support for higher-order the- oscillations, and complexity. J Neurosci 2020;40:605–18. ories of conscious awareness. Trends Cogn Sci 2011;15: Paxinos G, Watson C. The Rat Brain in Stereotaxic Coordinates. 6th edn. 365–73. Cambridge, MA: Academic Press, 2007. Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 15 PigoriniA,SarassoS,ProserpioPetal.Bistabilitybreaks-offdetermin- Steriade M, Timofeev I, Grenier F. Natural waking and sleep isticresponsestointracorticalstimulationduringnon-REMsleep. states: a view from inside neocortical neurons. J Neurophysiol NeuroImage 2015;112:105–13. 2001;85:1969–85. Poulet JFA, Crochet S. The cortical states of wakefulness. Front Syst Tononi G. An information integration theory of consciousness. BMC Neurosci 2018;12:64. Neurosci 2004;5:42. Rosanova M, Fecchio M, Casarotto S et al. Sleep-like cortical Tononi G, Boly M, Massimini M et al. Integrated information the- OFF-periods disrupt causality and complexity in the brain ory:fromconsciousnesstoitsphysicalsubstrate.NatRevNeurosci of unresponsive wakefulness syndrome patients. Nat Commun 2016;17:450–61. 2018;9:4427. Tononi G, Edelman GM. Consciousness and complexity. Science Sanders RD, Tononi G, Laureys S et al. Unresponsiveness ̸= uncon- 1998;282:1846–51. sciousness. Anesthesiology 2012;116:946–59. TononiG,MassiminiM.Whydoesconsciousnessfadeinearlysleep? SarassoS,CasaliAG,CasarottoSetal.Consciousnessandcomplexity: Ann N Y Acad Sci 2008;1129:330–4. a consilience of evidence. Neurosci Conscious 2021. Valomon A, Riedner BA, Jones SG et al. A high-density electroen- Sarasso S, Boly M, Napolitani M et al. Consciousness and complex- cephalography study reveals abnormal sleep homeostasis in ity during unresponsiveness induced by propofol, xenon, and patientswithrapideyemovementsleepbehaviordisorder.SciRep ketamine. Curr Biol 2015;25:3099–105. 2021;11:4758. Schartner MM, Seth A, Noirhomme Q et al. Complexity of multi- van Vugt B, Dagnino B, Vartak D et al. The threshold for conscious dimensionalspontaneousEEGdecreasesduringpropofolinduced report: signal loss and response bias in visual and frontal cortex. general anaesthesia. PLoS One 2015;10:e0133532. Science 2018;360:537–42. Seamans J. Losing inhibition with ketamine. Nat Chem Biol Vanhaudenhuyse A, Charland-Verville V, Thibaut A et al. Conscious 2008;4:91–3. while being considered in an unresponsive wakefulness syn- Siclari F, Baird B, Perogamvros L et al. The neural correlates of drome for 20 years. Front Neurol 2018;9:671. dreaming. Nat Neurosci 2017;20:872–8. Vesuna S, Kauvar IV, Richman E et al. Deep posteromedial cortical Siclari F, Bernardi G, Cataldi J et al. Dreaming in deep stages of sleep: rhythm in dissociation. Nature 2020;586:87–94. a high-density EEG study of slow waves and spindles. J Neurosci Volgushev M, Chauvette S, Mukovski M, Timofeev I. Precise 2018;38:9175–85. long-range synchronization of activity and silence in neo- Siegle JH, Lopez AC, Patel YA et al. Open Ephys: an open-source, cortical neurons during slow-wave sleep. J Neurosci 2006;26: plugin-based platform for multichannel electrophysiology. J Neu- 5665–72. ral Eng 2017;14:045003. Vyazovskiy VV, Olcese U, Hanlon EC et al. Local sleep in awake rats. Solms M. Dreaming and REM sleep are controlled by different brain Nature 2011;472:443–7. mechanisms.BehavBrainSci2000;23:843–50.discussion904–1121. Vyazovskiy VV, Olcese U, Lazimy YM et al. Cortical firing and sleep SteriadeM,AmzicaF,ContrerasD.Synchronizationoffast(30-40Hz) homeostasis. Neuron 2009;63:865–78. spontaneous cortical rhythms during brain activation. J Neurosci Weilnhammer V, Fritsch M, Chikermane M et al. An active role 1996;16:392–417. of inferior frontal cortex in conscious experience. Curr Biol Steriade M, Nuñez A, Amzica F. A novel slow (< 1 Hz) oscillation 2021;31:2868–80.e8. of neocortical neurons in vivo: depolarizing and hyperpolarizing Zingg B, Hintiryan H, Gou L et al. Neural networks of the mouse components. J Neurosci 1993;13:3252–65. neocortex. Cell 2014;156:1096–111. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Neuroscience of Consciousness Oxford University Press

Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats

Loading next page...
 
/lp/oxford-university-press/capacity-for-consciousness-under-ketamine-anaesthesia-is-selectively-9IN4JRIy30

References (106)

Publisher
Oxford University Press
Copyright
© The Author(s) 2022. Published by Oxford University Press.
eISSN
2057-2107
DOI
10.1093/nc/niac004
Publisher site
See Article on Publisher Site

Abstract

It remains unclear how specific cortical regions contribute to the brain’s overall capacity for consciousness. Clarifying this could help distinguish between theories of consciousness. Here, we investigate the association between markers of regionally specific (de)activation and the brain’s overall capacity for consciousness. We recorded electroencephalographic responses to cortical electrical ST stimulation in six rats and computed Perturbational Complexity Index state-transition (PCI ), which has been extensively validated as an index of the capacity for consciousness in humans. We also estimated the balance between activation and inhibition of specific cortical areas with the ratio between high and low frequency power from spontaneous electroencephalographic activity at each elec- trode. We repeated these measurements during wakefulness, and during two levels of ketamine anaesthesia: with the minimal dose ST needed to induce behavioural unresponsiveness and twice this dose. We found that PCI was only slightly reduced from wakefulness tolightketamineanaesthesia, butdroppedsignificantlywithdeeperanaesthesia. The high-doseeffectwasselectivelyassociatedwith ST reducedhighfrequency/lowfrequencyratiointheposteromedialcortex, whichstronglycorrelatedwithPCI . Conversely, behavioural unresponsiveness induced by light ketamine anaesthesia was associated with similar spectral changes in frontal, but not posterior ST cortical regions. Thus, activity in the posteromedial cortex correlates with the capacity for consciousness, as assessed by PCI , dur- ing different depths of ketamine anaesthesia, in rats, independently of behaviour. These results are discussed in relation to different theories of consciousness. Keywords: consciousness; ketamine anesthesia; EEG markers of consciousness; perturbational complexity index Introduction Highlights It is widely recognized that only a limited fraction of our brain activityisdirectlyinvolvedinspecifyingourconsciousexperience • We dissociate responsiveness from consciousness using (Kochetal.2016).Ideally,atheoryofconsciousnessshouldbeable a two-level ketamine protocol in rats. to precisely explain why only certain parts of the brain and types ST • We correlate activity in cortical regions with PCI , an of activity contribute to any particular experience. This reflects indicator of capacity for consciousness. two main aspects of consciousness science: understanding which • Cortical deactivation in the back, but not the front, was brain structures and activities are required for having a capacity ST associated with a significant drop in PCI . for consciousness, and which are required for particular contents Received: 15 January 2021; Revised: 9 December 2021; Accepted: 24 January 2022 © The Author(s) 2022. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 2 Arena et al. of conscious experience. The former can be identified by study- Recently, several measures aiming to objectively assess global ing how properties of brain activity change between brain states states of consciousness independently of motor or sensory where the level of consciousness is thought to change, such as functions have been developed (see for example Luppi et al. 2021), comparingwakefulnesswithdeep, generalanaesthesiaordream- with a notable convergence in evaluating the complexity of brain less sleep (e.g. Casali et al. 2013). The latter can be identified activity as indication of the capacity for consciousness (Sarasso by contrasting brain activity between conditions where particular et al. 2021). In particular, the perturbational complexity index ST stimuli are perceived or not, without otherwise altering the level [PCI; (Casali et al. 2013)], and the more general measure PCI of consciousness (e.g. van Vugt et al. 2018). [‘PCI-state transition’; (Comolatti et al. 2019)], have been shown Recently, the roles played by the frontal vs. posterior parts to reliably and consistently assess the capacity for consciousness of the neocortex in the consciousness of healthy humans have inhumansinaccordancewiththesubjects’immediateordelayed been discussed (Boly et al. 2017; Odegaard et al. 2017). In this reports of experience (Sarasso et al. 2015; Casarotto et al. 2016; ST ‘front vs. back debate’, some have argued that the ‘evidence for Rosanovaetal. 2018). Recently, PCI hasalso been shownto work a direct, content-specific involvement of the “front” of the cor- consistently in rodents undergoing propofol, sevoflurane, and ST tex, includingmostprefrontalregions, ismissingorunclear’(Boly ketamine anaesthesia (Arena et al. 2021). PCI and PCI quantify et al. 2017), while others argued that ‘the literature highlights pre- thespatiotemporalcomplexityofrepeatable,global,cortical,elec- frontalcortex’sessentialroleinenablingthesubjectiveexperience trophysiological responses to a local, direct cortical stimulation, in perception’ (Odegaard et al. 2017). Although this debate largely thus estimating how much the resulting deterministic neuronal concerned empirical data, and the issue remains unresolved, the activations are both integrated and differentiated across cortical debate illustrates that different theories of consciousness entail areas and time. PCI is inspired by the general idea that the joint differenthypothesesaboutwhichbrainregionscontributedirectly presenceofintegrationanddifferentiationisrequiredforasystem toexperienceandwhichprovidenecessarybackgroundconditions to be conscious, which is a central part of the Integrated Informa- for consciousness. Hence, evidence showing regional differences tion Theory (IIT; Tononi 2004; Massimini et al. 2009). Later on, PCI in contribution to the capacity for, or contents of, consciousness has also been considered to be compatible with the Global Neu- may provide specific empirical support for some theories over ronal Workspace (GNW) Theory, as the measure is sensitive to others. global and sustained patterns of cortical interaction (Baars 2005; Some authors have suggested that frontal parts of the cortex Dehaene et al. 2011; Mashour et al. 2020). More importantly, PCI are crucial for consciousness (Del Cul et al. 2009) or have decoded may be considered an index of capacity for consciousness even perceptual contents from frontal regions (Levinson et al. 2021). in cases where we do not know the ground truth (Casarotto et al. Thus, the prefrontal cortex was found to be causally involved 2016; Comanduccietal. 2020). Therefore, fortheremainderofthis in determining the contents of experience (Weilnhammer et al. paper, weinterpretasignificantdropinPCIfromwhatisobserved 2021), and frontal neuronal activity correlated with visual per- in wakefulness as indicative of a relative reduction in capacity for ception (even during no-report paradigms; Kapoor et al. 2020). consciousness. However, others have pointed to evidence that the apparent While PCI is an index of the capacity for consciousness based frontal involvement in consciousness may be confounded with on evoked cortical dynamics, measures that quantify spectral task-related processes such as working memory, attention, or properties of spontaneous cortical activity have also for long preparation for motor response (Koch et al. 2016; Boly et al. 2017). been used successfully to study brain states (Loomis et al. 1937; Whilethisdoesnotimplythatfrontalcorticalregionsarenotnec- Fernandez et al. 2017; Siclari et al. 2017; Colombo et al. 2019; essary for, or causally involved in, specifying human experiences, Lendner et al. 2020). In particular, during deep stages of sleep it does remind us that brain functions associated with normal and general anaesthesia, both electroencephalography (EEG) and behaviouralresponsivenesscanbeconfoundedwiththosedirectly local field potentials are characterized by high amplitude, low involved in specifying conscious experiences (Sanders et al. 2012). frequency (LF,≤4Hz) oscillations or slow-waves (Massimini et al. The ability to respond coherently to external stimuli is used as 2004; Vyazovskiy et al. 2009; Brown et al. 2010). This slow- the main criterion for determining whether non-communicating wave activity reflects a bistable network dynamic, where neurons patients and non-human animals are conscious (Chernik et al. synchronously alternate between an up-state, with depolarized 1990; Giacino et al. 2004; Gao and Calderon 2020). However, this membrane potential and firing, and a down-state with neuronal approach is based on the assumption that unresponsive states hyperpolarizationandsilence(Steriadeetal.1993,2001;Volgushev are always unconscious, which is at odds with evidence that etal.2006;Vyazovskiyetal.2009),possiblyduetoincreasedinhibi- vivid experiences can occur in unresponsive states. For instance, tion,adaptation,andsynapticfatigue(Steriadeetal.2001;Compte dreams can occur in all stages of sleep (Nielsen 2000; Solms et al. 2003; Esser et al. 2007; Funk et al. 2017). Conversely, during 2000; Siclari et al. 2018) and during general anaesthesia (Noreika wakefulness, the EEG is mainly characterized by low amplitude, et al. 2011). Furthermore, patients can be conscious but unre- high frequency (HF,≥20Hz) oscillations, which reflect tonic neu- sponsive for decades after brain damage(Vanhaudenhuyse et al. ronaldepolarizationandfiring( Steriadeetal.1996,2001;Mukovski 2018), or they can be painfully aware during general anaesthe- et al. 2007; Vyazovskiy et al. 2009). This has inspired several mea- sia for surgery, while assumed to be unconscious (Ghoneim et al. sures for quantifying the cortical state of activation based on 2009). Conversely, quite complex behaviours can be preserved the relation between high (HF) and low (LF) frequency EEG power during conditions that are typically assumed to be unconscious, (Mukovski et al. 2007; Fernandez et al. 2017; Gao et al. 2017; Siclari e.g. sleepwalking, sleep talking (Castelnovo et al. 2018; Valomon et al. 2017; Colombo et al. 2019; Lendner et al. 2020). Thus, the et al. 2021), and presumed unconscious behaviours during certain HF/LF (power) ratio tends to drop when there is more inhibition epileptic seizures, and in unresponsive wakefulness syndrome and deactivation, whereas a higher HF/LF ratio suggests neu- (Blumenfeld 2005; Laureys et al. 2010). Thus, it is essential to ronal activation (Mukovski et al. 2007; Gao et al. 2017; Lombardi distinguish brain regions necessary for consciousness from those et al. 2017; Poulet and Crochet 2018). Coherently, the spectral that are necessary for behavioural responsiveness. relation between HF and LF powers has been associated with Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 3 the excitation/inhibition balance (Gao et al. 2017) and the ratio In addition to new data, we also included a new analysis of between LF and HF powers was found to correlate with motor datafromasetofpreviouslypublishedexperiments(Arena, Thon activity in mice (Fernandez et al. 2017). Thus, local changes in and Storm, 2019; Arena et al. 2021) with a new analysis. Specif- HF/LFratiomayindicatechangesincorticalactivation(Pouletand ically, we reanalysed data from a subset of six rats, from which Crochet 2018). werecordedepiduralEEGcontinuouslyandinresponsetoelectri- While measures of consciousness typically evaluate global calstimulationofthesecondarymotorcortexduringwakefulness brain dynamics, specific and localized changes in the HF/LF ratio andlightketamineanaesthesia(Arenaetal. 2021). Thesedatawill mayoccurwithinthesamebrainstateandmightaffectbehaviour be compared to those from a new set of recordings/stimulations or conscious experience. For example, local cortical sleep, which that were performed on the same six rats during a subsequent can affect task performance in rats, involves localized slowwaves period of ketamine anaesthesia with increased dosage. and neuronal silence in an otherwise awake brain state, dom- Experimental procedure inated by low amplitude and fast oscillations, with underlying tonic neuronal firing ( Vyazovskiy et al. 2011). Furthermore, pos- Epidural EEG was recorded by a grid of 16 screw electrodes terior increases in HF/LF ratio during deep stages of sleep suc- (stainlesssteel,1.2mmcalibre),whichwerechronicallyimplanted cessfully predicted whether or not humans reported dreaming through the skull, in contact with the dura. The recording elec- and were explained by a simultaneous local reduction of 1–4Hz trodes were organized symmetrically with respect to the sagittal activity and enhanced 20–50Hz activity (Siclari et al. 2017). Thus, suture and spanned most of the cortical surface of both hemi- HF/LF can vary regionally within global states, but it is not known spheres (Supplementary Fig. S1). The bilateral frontal cortex was whether local reductions in HF/LF are associated with reduced coveredbysixelectrodes, namedM2 , M2 andM2/M1thatmedi- R C consciousness levels. More generally, it is still unknown whether ally overlaid the rostral and caudal part of the secondary motor any particular localized spectral properties are related to the cortex and part of the primary motor cortex, respectively. Other brain’s global capacity to sustain consciousness. six electrodes covered left and right parieto-occipital associa- In this study, we aimed at investigating this relation, asking tive cortices: PA electrodes covered lateral parietal cortex, RS/PA whether specific regional changes of cortical activation, assessed medially covered retrosplenial and parietal cortex, while RS/V2 by the HF/LF ratio, are associated with changes in capacity for electrodes covered the posteromedial cortex, over the caudal part ST consciousness, assessed by PCI , irrespectively on the specific ofretrosplenialcortexandmedialpartofsecondaryvisualcortex. content of experience. We also aimed to dissociate levels of The last four electrodes, bilateral S1 and V1, overlaid the primary behavioural (un)responsiveness (assessed by responses to pain somatosensory cortex and the primary visual cortex respectively. stimuli) from the capacity for consciousness by comparing wake- Event-related potentials (ERPs) were recorded in response to elec- fulnesswithtwodistinctlevelsofketamineanaesthesia.Indeed,it tricalstimulationoftherightsecondarymotorcortexbyabipolar has been shownthat the unresponsivestate induced by ketamine tungsten electrode (see Supplementary Fig. S1 for detailed elec- can subtend vivid conscious experiences, with wakefulness-like, trodelocationswithrespecttobregma;PaxinosandWatson2007). high PCI (Sarasso et al. 2015; Arena et al. 2021). While it was The standard surgical procedure under a regime of controlled also previously reported that when ketamine plasma level is par- general anaesthesia/analgesia was adopted for implantation of ticularly high, such as soon after bolus injection (Akeju et al. chronic electrodes, and after 3days of recovery, rats were habit- 2016), the EEG alternates between low and highly complex activ- uated to head and body restriction in at least 3 subsequent days, ity patterns (Li and Mashour 2019). In this condition, slow-wave as previously described (Arena et al. 2021). The electrophysiologi- oscillations also occur, interrupting an enhanced HF activation, cal recording/stimulation began only when rats did not show any resultinginthegamma-burstactivitypattern(Akejuetal. 2016; Li signofdistressandwerecalmwithintherecordingsetup,withthe and Mashour 2019), and suggesting transient moments of uncon- head connected to a fixed head-bar by two chronically implanted sciousness at high ketamine doses. Here, we carefully control clamps and with the body inserted in a transparent acrylic tube, ketamine dosage by adopting intravenous infusion at two differ- with a natural posture. The tail was left outside the tube to test ST entconstantratesandinferconsciousexperiencerelyingonPCI reflex motor responses to pain stimulation. level, in comparison with wakefulness condition. Then, we inves- The six rats were subjected to electrophysiological record- tigatewhetherregionalHF/LF ratioreliablycovarieswithchanges ing/stimulation sessions during wakefulness and subsequent ST in PCI , within and across conditions. The results may be used ketamine anaesthesia. Ketamine (Vetoquinol, Ittigen, Switzer- to validate predictions and explanations from theories of con- land) was infused at a constant rate via a 26GA catheter in sciousnessbyconstrainingwhichcorticalregionsmainlyunderlie the tail vein. Subcutaneous injection of glycopyrrolate 0.01mg/kg consciousness as opposed to behavioural responsiveness. was also performed to reduce the increased salivation. Since rats keep eyes open during general anaesthesia, eye ointment was applied to maintain eyes humid and body temperature was kept at 36.5–37.5 C by a heating pad, as previously described in Materials and methods detail. During the recording session, the stimulating electrode Animal model and experimental data was connected to an isolated current stimulator (Isolator HG203, Six adult, male, Sprague–Dawley rats (n=6; body weight ∼370g) High Medical, London, UK) triggered by a voltage pulse genera- were used in this study. All the experiments and animal care tor (2100, A-M System, Washington DC, USA), while the epidural procedures were conducted at the University of Oslo and were EEGelectrodeswereconnectedtoa16-channelunipolaramplifier approved by the Norwegian Authority, Mattilsynet (FOTS: 11812) referenced to ground (RHD2132, Intan Technologies, Los Angeles, in agreement with the Norwegian law of animal handling. Efforts CA,USA),andcontrolledbyOpenEphyssystem(Siegleetal.2017), were made to avoid or minimize animals’ pain and distress. Rats which acquired and digitized the electrophysiological signal at 10 were caged in enriched environments, with ad libitum access to or 30kHz, 16-bit resolution. food and water and were exposed to a 12:12 hour light–dark cycle The EEG activity was continuously recorded from all 16 at 23 C constant room temperature. channels, in a dark environment, in which all rats received 4 Arena et al. ∼100 electrical monophasic current pulses of 50µA, 1ms, deliv- baseline window, α=0.01) was used to conserve only the ITPC ered at 0.1Hz, at first during wakefulness (W), and during incrementsthatdifferedsignificantlyfromtherespectivebaseline subsequent ketamine anaesthesia. The anaesthesia was induced (from −500 to −200ms). The time point of the last significant ITPC by an intravenous (iv) bolus injection of ketamine 30mg/kg value within 800ms after stimulation and in a broad frequency and then maintained at the initial constant rate of ketamine range (5-80Hz) was detected for each channel and considered to 1.75mg/kg/min iv, by a syringe pump. After 10min from induc- be the duration of the phase-locked EEG response, which quan- tion,thetailwaspinchedthreetimesbyforcepstocheckthepres- tified the temporal extension of the deterministic effect of the enceofbehaviouralreactionto painstimulation. Thebehavioural electrical stimulation [ITPC drop time; (David et al. 2006; Pigorini ST responsetypicallyconsistedofalateralandwidemovementofthe et al. 2015; Rosanova et al. 2018; Arena et al. 2021)]. PCI was used tail and/or in the alteration of the respiratory rhythm, with the to estimate the capacity for consciousness (Comolatti et al. 2019; occurrence of a deeper breath, accompanied by a sudden chest Arena et al. 2021) and was assessed in the full ERP window, from movement. The infusion rate was stepwise increased by adding 0 to 600ms, and across time, in short, moving windows of 100ms, ST 4%oftheinitialdosageuntilthebehaviouralresponsewasabsent with 50ms of overlap. PCI was computed using the available (3min between each increment). The resulting minimal dosage code at [github.com/renzocom/PCIst] with the same parameters of ketamine that abolished behavioural reaction to pain stim- previously described (Comolatti et al. 2019; Arena et al. 2021). The ulation was 1.78±0.02mg/kg/min, iv (mean±SEM across rats; functional connectivity across cortical areas in response to stim- 1.8mg/kg/minforsimplicity)anddefinedthecondition‘ketamine ulation was also used to estimate the level of integration and 1’ or K1. In order to evaluate possible dose dependencies, the differentiation of the cortical network, as previously described in same recording/stimulation was also repeated afterwards, during detail (Arena et al. 2021). The inter-site phase clustering (ISPC) deeperketamineanaesthesia (‘ketamine 2’ or K2), at the constant was assessed as the consistency across trials of the phase differ- rate of 3.5mg/kg/min iv, which corresponded to two times the ence across channels for all time-frequency points (Cohen 2014; initial constant rate. Arena et al. 2021). ISPC was calculated for each channel pair, the respective mean ISPC of the baseline (from −500 to −200ms) was subtracted and bootstrap statistic (500 permutations; posi- Analysis of electrophysiological signal tiveandnegativethresholdsbasedontheobtaineddistributionof The acquired electrophysiological data were analysed in MAT- the maximum and minimum ISPC values in the baseline window, LAB2016a (Math Works, Natick, Massachusetts, USA) and Origin α=0.05) was adopted to conserve only the significant variations 9.1 (OriginLab, Northampton, Massachusetts, USA) and prepro- from baseline. The ISPC values that could be determined by vol- cessed as described before (Arena et al. 2021). Raw epidural EEG ume conduction (clustering around 0 or pi) were set to 0 and wasbandpassfiltered(0.5–80Hz, Butterworth, thirdorder), down- the remaining ISPC values were averaged in the frequency range sampled to 500Hz and ERP epochs of 10s were extracted for each 5–14Hz and between 180 and 400ms. The proportion of the num- channel, centred at the stimulus onset (from −5 to 5s). All epochs ber of mean positive ISPC values for each channel was defined as were offset corrected by subtracting the average voltage of their the connectivity degree of the electrode (Cohen 2014; Arena et al. respective baseline (from −1 to 0s), and trials with high voltage 2021). artefacts in their baseline were removed. The first 90 trials of pre- The spontaneous cortical activity associated with the differ- processed signals were used for analysis, maintaining the same ent experimental conditions was quantified from 90 epochs of temporal sequence across animals and conditions. ERPs were not 5s of epidural EEG signal that preceded the electrical stimula- normalized and the electrical noise was similar along with the tion (from −5 to 0 s). A Morlet wavelet convolution (6 cycles, whole duration of recordings. 80 wavelets, linearly spanning from 1 to 80Hz) was performed The cortical excitation in response to electrical stimulation on each epoch for all channels. Spectral powers were extracted was quantified by the root mean squared (rms) amplitude of and averaged across samples, trials and channels, obtaining a the first 50ms of the mean ERP for all electrodes, and then global estimation of the power of each frequency for each ani- averaged across electrodes. Spectral powers and phases of ERPs mal and condition. The resulting periodogram was linearly fitted were obtained from Morlet wavelet convolution (three cycles in Log-Log coordinates, in the frequency range 20–40Hz. The wavelets, linearly spanning from 1 to 80Hz, with 1Hz resolution), slope of the obtained linear function was the spectral exponent which was performed for each trial and channel. The spectral of the 1/f function and was used to quantify the distribution powers were normalized over the mean power of the baseline of frequency powers in the spontaneous EEG activity (Gao et al. (−500 to −200ms) across trials, for each respective frequency and 2017; Colombo et al. 2019; Lendner et al. 2020; Arena et al. channel. The mean relative powers across trials were then dB 2021). Instantaneous powers were normalized by 1 and con- converted and bootstrap statistic (500 permutations; positive and verted in dB, and also averaged in high and low frequency ranges negativethresholdsbasedontheobtaineddistributionofthemax- (HF 20–80Hz, LF 1–4Hz, respectively), across and for each sin- imum and minimum dB values in the baseline window, α=0.05) gle channel, before dB conversion. The ratio between HF power was applied for each frequency and channel to conserve only the and LF power (HF/LF ratio) was also computed to estimate the significant dB variations from respective baseline. The resulting level of activation for the cortical area underlying each channel relative spectral power was averaged across trials in the HF range (Fernandez et al. 2017; Poulet and Crochet 2018) and its contribu- (20–80Hz)toestimatethetemporaldynamicoftheneuronalacti- tion to cortical complexity and capacity for consciousness (Siclari vation underlying the EEG signal (Mukovski et al. 2007; Pigorini et al. 2017). et al. 2015; Rosanova et al. 2018; Arena et al. 2021). Late incre- ments of HF power (> 0dB) were detected in a time window from Statistics 80 to 800ms (Arena et al. 2021). The inter-trial phase clustering [ITPC; (Cohen 2014)] was computed for each frequency and chan- Allresultsareexpressedasmean±SEManderrorbarsandshades nel and bootstrap statistic (500 permutations; threshold based represent SEM in the figures. Shades in linear regressions rep- on the obtained distribution of the maximum ITPC values in the resent the 95% confidence band. The topographical plots in the Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 5 figures report the Laplace interpolation of a variable over the dor- of ketamine. We also repeated the electrophysiological experi- sal surface of a skull, anchored to the true electrode locations. ment with an increased administration rate, to test for a possible The function of the colour maps is only for better visualization dosage dependency. As reported previously (Arena et al. 2021), since all the analyses were performed at the level of single we adjusted the first, low ketamine infusion rate (K1) for every channels. Parametric statistics were adopted after assessing the single rat, to the minimal dose that induced behavioural unre- normality of distribution of the measured variables, by apply- sponsiveness (i.e. no motor response to pain stimulation). This ing the Shapiro–Wilk test, in a population of 12–14 rats during first dosage, K1, was approximately 1.8mg/kg/min, iv, while the wakefulness. All the variables were tested in a repeated mea- secondsubsequentketaminedose, K2, wassettoabouttwotimes sure design. Thus, principal and interaction effects were tested the first one, at 3.5mg/kg/min, iv. withone-wayortwo-wayrepeatedmeasuresANOVA(rANOVA),in During wakefulness, the spontaneous EEG activity was char- which Greenhouse–Geisser correction was applied when spheric- acterized by low amplitude, fast oscillations, typical of cortical ity could not be assumed, while group comparisons were tested activation (Fig. 1a). Interestingly, despite the loss of behavioural with Student’s paired-samples t-test. Because one channel was responsiveness, the fast oscillations persisted in both K1 and K2 removed from the analysis in two rats, two-way ANOVA was used conditions (Fig. 1a), and a general increase of EEG amplitude to compare the spatial distribution of variables between left and occurred with ketamine infusion, as shown by the averaged peri- right hemispheres. Linear fitting was performed with the least- odogram across channels, which scaled up from wakefulness to square method. To evaluate correlations and goodness of fit, the K1 and further to the K2 condition (Fig. 1b). In line with this, coefficient of determination, R , was computed and a t-test was the mean HF power (20–80Hz) across channels increased from performed to test the null hypothesis of slope=0, establishing wakefulness to K1 and further to K2 (Fig. 1c; W: 42.42±0.54dB, a P-value. When multiple hypotheses were tested across con- K1: 47.52±0.83dB, K2: 49.15±0.64dB; one-way rANOVA, princi- ditions or along cortical areas, the Bonferroni–Holm correction pal effect of condition, P=9.1942*10-8; paired-samples t-test, W −5 was adopted (number of conditions=3, corrected α=0.01666; vs. K1, P=0.0005, W vs. K2, P=6.8217*10 , K1 vs. K2, P=0.0032). numberofcorticalareas=8,corrected α=0.00625foreachhemi- Likewise, the mean LF (1–4Hz) power increased from wakeful- sphere). Gaussian v-test was used to test volume conduction in ness to K1 and further to K2 (Fig. 1d; W: 73.57±1.26dB, K1: connectivity analysis (Cohen 2014; Arena et al. 2021). All statistics 77.75±0.72dB, K2: 79.75±0.54dB; one-way rANOVA, principal are two-tailed. The statistical significance in the figures are repre- effect of condition, P=0.0021; paired-samples t-test, W vs. K1, sented as follows: P<0.05 *, P<0.01 **, P<0.001 ***, P≥0.05 ns (not P=0.0440, W vs. K2, P=0.0271, K1 vs. K2, P=0.0271). The significant). spectral exponent of the mean periodograms was similar across conditions, indicating a general scaling across all frequencies (Fig.1e;W:−1.82±0.17,K1:−1.41±0.12,K2:−1.57±0.12;one-way rANOVA, principal effect of condition, P=0.1959; paired-samples Results t-test, W vs. K1, P=0.1719, W vs. K2, P=0.3077, K1 vs. K2, Perturbational complexity was independent of P=0.2838). behavioural responsiveness but was reduced by Across conditions, the same electrical stimulation induced a increasing ketamine dosage similar initial cortical excitation, quantified by the RMS ampli- Inpreviousstudiesinhumans,ketaminehasbeenfoundtoinduce tude of the early ERP deflections ( Fig. 2a and b; W: 52.63±5.13µV, unresponsiveness combined with vivid, dream-like experience, K1: 47.65±3.91µV, K2: 44.84±4.31µV; one-way rANOVA, prin- and high cortical complexity (Sarasso et al. 2015). Thus, to dis- cipal effect of condition, P=0.1943; paired-samples t-test, W sociate cortical complexity from behavioural responsiveness, we vs. K1: P=0.2942, W vs. K2: P=0.1613, K1 vs. K2: P=0.1132). recorded spontaneous and evoked EEG activity in six rats, during Consistently, the electrical pulse evoked an early, broad- wakefulness (W) and subsequent constant intravenous infusion band increment in power that was similar across conditions. Figure 1. The averaged power spectrum of spontaneous EEG was scaled up from wakefulness to ketamine anaesthesia and by increasing ketamine dosage. (a) Example of spontaneous EEG (5s) from the retrosplenial/parietal cortex (RS/PA) of one rat during wakefulness (W), light ketamine anaesthesia (ketamine 1, K1; administration rate: 1.8 mg/kg/min i.v.) and deep ketamine anaesthesia (ketamine 2, K2; administration rate: 3.5 mg/kg/min i.v.). (b) Mean periodograms from 16 channels and 6 rats exposed to the same conditions of A (shades represent SEM across rats). (c) Variations of mean high frequency power (HF, 20-80 Hz) and (d) low frequency power (LF, 1-4 Hz) are shown for each rat across conditions and increased from W to K1 and from K1 to K2. (e) The spectral exponents of the averaged periodograms across channels are also reported for each rat and condition 6 Arena et al. Figure 2. The spatiotemporal dynamics of ERPs revealed a drop in complexity from low to high dosage of ketamine. (a) Top, superimposition of mean ERPs from all 16 electrodes in response to single pulse stimulation (1 ms, 50 µA; dashed line) of the right secondary motor cortex (M2), from the same rat during wakefulness (W, left) and light and deep ketamine anaesthesia (K1, middle and K2, right respectively). One averaged ERP from the same channel over the right primary somatosensory cortex (S1) is in bold for clarity. Middle, power spectrogram (dB) and, bottom, phase-locking across trials (ITPC) from the channel shown in bold above (right S1). The black arrows indicate the moment of relaxation of HF power (dB ≤ 0) that follows the first response to stimulation. The ITPC drop time is indicated by vertical continuous lines. (b) The rms amplitude of the early ERP (first 50 ms from stimulus onset) averaged across channels is shown for all rats and conditions. (c) The ITPC drop time (in frequency range 5–80 Hz) averaged across channels is plotted for each rat and condition. (d) The ITPC drop time and the onset of later increased HF power was averaged across channels from all rats and shown for each condition. (e) Left, time courses of mean PCIST (moving windows of 100 ms, 50 ms overlap) and standard errors (shaded) are plotted for all conditions (horizontal lines indicate periods of statistically significant difference between conditions, P<0.05). Right, PCIST in range 0-600 ms is shown for each rat and condition This activation was quickly followed by a period of relaxation, principal effect of condition, P=0.001; paired-samples t-test, W with HF activity similar to or below the baseline (dB≤0), which vs. K1: P=0.7068, W vs. K2: P=0.0014, K1 vs. K2: P=0.01). lasted 150.52±12.04ms, averaged across rats and conditions The long-lasting response in wakefulness and K1 also showed a (Fig. 2a, see also Supplementary Fig. S2 for spectrograms and later increase of HF power in most of the channels (W: 100% of phase-lockingplotsinthebroadfrequencyrange,from1to80Hz). channels; K1: 92.36±4.33% of channels). Such HF activation was However, after this first dynamic, the ERP developed differently largely associated with the deterministic response, since its onset throughtimeindifferentconditions.Weevaluatedthephasecon- preceded the ITPC drop, as shown by comparing the times of sistency of the ERPs across trials, by computing the ITPC at each these events from the electrodes with such late increase in HF timepoint(Cohen2014;Arenaetal.2021)andquantifiedthedura- power (Fig. 2a and d; paired-samples t-test, ITPC drop time vs. −45 tion of the deterministic effect of the stimulus as ITPC drop time LateHFpoweronset;inW:n=94channels,P=7.8199*10 ;inK1: −13 (David et al. 2006; Pigorini et al. 2015; Rosanova et al. 2018; Arena n=87channels, P=2.6807*10 ). Incontrast, withhighketamine et al. 2021). During wakefulness and with low ketamine dose, dose, a late HF power activation was still detected in some elec- the ERPs showed long-lasting waveforms that were phase-locked trodes(46.6±14.93%ofchannels), butwasnotphase-locked,asit across trials, thus still deterministically caused by the stimula- occurred after the ITPC drop (Fig. 2a and d; paired-samples t-test, tion.Conversely,withthehighketaminedosage,thephase-locked ITPC drop time vs. Late HF power onset, in K2: n=44 channels, −5 response quickly died out, indicated by an earlier ITPC drop, P=2.1199*10 ). These results were also associated with an over- averaged across channels (Fig. 2a and c; W: 347.35±18.85ms, all reduction of the functional connectivity across channels in K1: 369.76±59.86ms, K2: 152.32±24.17ms; one-way rANOVA, the K2 condition, which also corresponded to reduced diversity Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 7 of cortical connectivity (Supplementary Fig. S3). Coherently, K2 The strong reduction in perturbational ST changed the time course of PCI , which was initially high and complexity was associated with a selective quicklydecayedsoonafterthestimulationinallconditions. How- deactivation of bilateral posteromedial cortex ST ever, in wakefulness and with low ketamine dose, PCI built up We found that high-dose ketamine caused a non-linear, dose- ST ST again, reaching similar values after 200ms. Then PCI dropped dependent drop of PCI (Fig. 2), but this drop was not associated again, faster in K1 than W, until fading. Conversely, with high with a clear overall transition of spontaneous EEG towards slow ST ketamine dose, PCI never recovered after the early decay and waves and reduced HF power (Fig. 1). Thus, we investigated remainedcloseto0(Fig.2aande).Thetimecoursewasinlinewith the spectral features of spontaneous activity from single cortical the perturbational complexity calculated for the entire response areas with higher detail, by computing the local, instantaneous ST window (0–600ms), thus PCI showed slightly higher values in HF/LF ratio (Fernandez et al. 2017; Siclari et al. 2017; Poulet and wakefulness than in K1 condition, but the difference was not Crochet 2018; see Fig. 3a). By averaging across time (from −5 to statistically significant for this sample [although a significant dif- 0 s) and trials for each electrode, we obtained a topographical ference was found for a larger data set (Arena et al. 2021), see distribution of the HF/LF ratio, which was similar between left ST Discussion]. With a high ketamine dose, however, PCI was sig- and right hemispheres, and this symmetry persisted in all condi- nificantlylowerthantheotherconditions( Fig.2e; W:78.47±8.44, tions (Fig. 3b; two-way ANOVA, principal effect of lateralization, K1: 59.57±5.77, K2: 23.75±6.55; one-way rANOVA, principal in W: P=0.6934, in K1: P=0.9902, in K2: P=0.9789). By com- −5 effect of condition, P=4.3026*10 ; paired-samples t-test, W vs. paring the mean HF/LF ratio across conditions, for each cortical K1, P=0.0535, W vs. K2, P=0.0028, K1 vs. K2, P=0.0005). area, we found that it significantly decreased from wakefulness Figure 3. The HF/LF ratio of spontaneous EEG in bilateral posteromedial cortex was selectively reduced by increasing ketamine dosage and correlated with the level of PCIST. (a) Up, Example of spontaneous EEG (5s) from the posteromedial cortex (RS/V2) of one rat during wakefulness (W), light ketamine anaesthesia (ketamine 1, K1) and deep ketamine anaesthesia (ketamine 2, K2) and below, the relative spectrogram and the ratio between high frequency (HF, 20–80 Hz) and low frequency (LF, 1–4 Hz) powers (HF/LF ratio) in time. (b) The colour maps show the topographical distributions (R-C: rostral-caudal) of the 16 EEG electrodes (small green circles) and, above, the spatial interpolations of the HF/LF ratio, averaged across time, trials and rats, for each condition. Below, the colour maps report the spatial interpolation of the t-scores (paired-samples t-test) from comparing the HF/LF ratio across conditions for each channel (left, wakefulness vs. ketamine 1; middle, wakefulness vs. ketamine 2; right, ketamine 1 vs. ketamine 2). The horizontal black line in the colour bar indicates the threshold for statistical significance (t5 =4.5258, Bonferroni–Holm corrected). White arrowheads indicate the channels with statistically significant differences across conditions. (c) Above, the colour map shows the spatial distribution of the coefficient of determination R2 from the correlation between PCIST and HF/LF ratio, across rats and conditions, for each channel. Below, the colour map reports the spatial interpolation of t-scores, assessing the statistical significance of the correlation for each channel. The horizontal black line in the colour bar indicates the threshold for statistical significance (t16 =3.1458, Bonferroni–Holm corrected). White arrowheads indicate the channels showing statistically significant correlations. The correlations of right M2C and left RS/V2 are reported in (d) with respective R2 and P-values 8 Arena et al. −3 to low ketamine dosage only in right M2/M1 cortex (HF/LF*10 , the variation of behavioural responsiveness could not be sepa- paired-samples t-test; W: 4.28±0.60, K1: 1.13±0.25, P=0.0372) rated from changes in perturbational complexity, hence a signif- −3 ST and S1 cortex (HF/LF*10 , paired-samples t-test; W: 6.48±0.69, icant correlation between PCI and the HF/LF ratio was detected K1: 3.48±0.51, P=0.0372), although weak, non-significant reduc- in both frontal cortex and posteromedial cortex (Supplementary tions were also seen elsewhere (Fig. 3b). In contrast, high doses of Fig. S5). ketamineselectivelyreducedthemeanHF/LFratiointhebilateral Inprinciple,thereductionoftheHF/LFratiocanbedetermined posteromedial cortex (left and right RS/V2 channels), indicating by an increase of LF powers, by a decrease of HF powers, or by a specific deactivation of this part of the cortex, with respect to a combination of the two events. Thus, in order to explain the the K1 condition (Fig. 3a and b, Supplementary Fig. S4; left RS/V2, reduction seen in the posteromedial cortex, we assessed the LF −3 HF/LF*10 , paired-samples t-test; K1: 1.82±0.19, K2: 1.09±0.17, andHFpowersatthelevelofeachcorticalarea, acrossconditions −3 P=0.0088; right RS/V2, HF/LF*10 , paired-samples t-test; K1: (Fig. 5a). At first, we tested for possible lateralization, without 1.50±0.20, K2: 0.91±0.14, P=0.0431). In line with these results, finding any clear difference between left and right hemispheres the HF/LF ratios at both bilateral RS/V2 and right M2/M1 cortex in each experimental condition, for both LF powers (Fig. 5a; two- were also reduced from wakefulness to the K2 condition (Fig. 3b). way ANOVA, principal effect of lateralization, in W: P=0.7581, To test whether the level of activation of any cortical area was in K1: P=0.9459, in K2: P=0.9589;), as well as for HF power effectivelyabletopredictthecomplexityofglobalcorticaldynam- (Fig. 5a; two-way ANOVA, principal effect of lateralization, in W: ics,weassessedpossibleregionalcorrelationsbetweenHF/LFratio P=0.8284,inK1:P=0.9664,inK2:P=0.6748).However,LFpowers ST andPCI (Fig.3c).WefoundthattheHF/LFratiofromthesponta- weredifferentiallydistributedacrosscorticalareas,andanoverall neous activity of bilateral posteromedial cortex (left, right RS/V2) increase in power could be detected in relation to the increment ST was highly, linearly correlated with the PCI value across con- of ketamine dosage (Fig. 5b; two-way rANOVA, principal effect ditions (Fig. 3c and d; left RS/V2: linear fit, R =0.590, P=0.0016; of cortical areas: P=0.0009, principal effect of ketamine dosage: right RS/V2: linear fit, R =0.568, P=0.0024). A weaker but sig- P=0.0132). Similar effects were also found for HF powers (Fig. 5b; nificant correlation was also identified only at the level of the two-way rANOVA, principal effect of cortical areas: P=0.0109, right secondary motor cortex (Fig. 3c and d; right M2 : linear fit, principal effect of ketamine dosage: P=0.0041), and were in line R =0.417, P=0.0264). with the scaling up of the mean periodograms, seen by averag- Next, we assessed the same correlations between local HF/LF ing across electrodes (Fig. 1). Nevertheless, by comparing powers ST ratio and global PCI , this time distinguishing between brain between K1 and K2 conditions for every single channel, a signifi- states and state transitions. We first evaluated the correla- cant increase of LF power was only found at the level of bilateral tion in wakefulness condition alone, where both behavioural posteromedial cortex (left and right RS/V2), while the increase of responsiveness and high perturbational complexity were present HF power was more spatially sparse, without a clear clusteriza- (Fig. 4a and b; data from three different recordings, performed in tion (Fig. 5b). To more directly compare the relative increment three different days on the same rats were considered only in of powers that occurred by increasing ketamine dosage, we com- this condition, to increase the number of observations). Then, putedtheratiobetweenK2andK1conditions(K2/K1), forbothHF we repeated the estimation of the correlations by considering the and LF powers, at the level of each cortical area (Fig. 5c). Overall, conditions of wakefulness and low dosage of ketamine, when per- the power increments were differentially distributed across corti- turbationalcomplexityisstillhigh,butbehaviourtransitionsfrom cal areas, and an overall difference between HF and LF could not responsiveness to unresponsiveness (Fig. 4c and d). Finally, we be detected (Fig. 5c; two-way rANOVA, principal effect of cortical evaluatedthecorrelationsinconditionsoflowandhighketamine areas: P=0.0325, principal effect of frequency range: P=0.0744). dosages, when only the variation of perturbational complex- However, a clear interaction effect between cortical areas and ity occurred, within the same unresponsive behavioural state frequency ranges was identified ( Fig. 5c; two-way rANOVA, inter- (Fig. 4e and f). With this, we attempted to identify possible roles action effect: P=0.0056), thus indicating the relation between the that specific cortical regions might have in specific state transi- increments of HF and LF powers changed depending on the spe- tions.Withinthewakefulnesscondition,wecouldnotidentifyany cific cortical area. Indeed, LF power increased significantly more ST significant correlation between PCI and HF/LF ratio at the level than HF power selectively in the bilateral posteromedial cortex of any cortical area. Nevertheless, the correlations with higher (left and right RS/V2), conversely, HF power had the tendency R , which were also closer to the threshold for statistical signif- to increase more than LF power in parieto-frontal areas, even if icance, seemed to be clustered in the secondary motor cortex without reaching statistical significance ( Fig. 5c). (Fig. 4a and b; right M2 : linear fit, R =0.377, P=0.0538; left RS/V2: linear fit, R =0.077, P=1). Likewise, the HF/LF ratio could Discussion ST not clearly predict PCI between wakefulness and low ketamine dose, when only behavioural responsiveness changed, for any of We reanalysed multichannel EEG data from wakefulness (W) and the cortical areas (Fig. 4c and d; right M2 : linear fit, R =0.386, two levels of ketamine anaesthesia (K1, K2) in rats, based on pre- P=0.2173; left RS/V2: linear fit, R =0.271, P=0.4124). On the vious experiments (Arena et al. 2021). To assess the capacity for ST otherhand,byconsideringonlythevariationsinducedbyincreas- consciousness in each state, we computed PCI (Comolatti et al. ingketaminedosage(K1andK2),wefoundasignificantandstrong 2019) and compared these results to region-specific estimates of correlation selectively associated with the left RS/V2 cortex, thus cortical activation, assessed by HF/LF ratio of spontaneous EEG indicating that the state of activation or deactivation of the pos- activityforeachelectrode(Fernandezetal.2017;Siclarietal.2017; teromedialcortexcouldeffectivelylinearlypredictthecomplexity Poulet and Crochet 2018). level of the entire cortical network and its breakdown (Fig. 4e and Inhumans,ketamineanaesthesiahasbeenobservedtoinduce f; right M2 : linear fit, R =0.064, P=0.1; left RS/V2: linear fit, a dissociated state of behavioural unresponsiveness with dream- R =0.568, P=0.0370). Coherently with these results, by consider- like, vivid conscious experiences (Collier 1972; Sarasso et al. 2015). ingonlytheconditionsofwakefulnessandhighdoseofketamine, Wetookadvantageofthisandusedacontrolledintravenousinfu- Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 9 Figure 4. HF/LF ratio of spontaneous EEG from posteromedial cortex selectively correlated with PCIST in conditions of behavioural unresponsiveness, with light and high ketamine anaesthesia. (a, b) The putative correlations between PCIST and HF/LF ratio are shown in wakefulness across rats and 3 recording sessions (Wr1 Wr2 Wr3), during the presence of both behavioural responsiveness and high cortical complexity. (c, d) The same putative correlations are shown across rats and across conditions of wakefulness (W) and low ketamine anaesthesia (K1), when loss of behavioural responsiveness occurred, but high cortical complexity persisted. (e, f) Correlations between PCIST and HF/LF ratio are also computed and shown across rats and across conditions of low and high ketamine doses (K1 and K2, respectively) when reduction of cortical complexity occurred and behavioural unresponsiveness was unchanged. In panels A, C, E the colour maps show the spatial interpolation of R2 and t-scores (above and below respectively, for all panels) from the correlations of all channels. The horizontal black line in the colour bar indicates the threshold for statistical significance. The white arrowhead indicates channels with statistically significant correlation between PCIST and HF/LF ratio. In panels B, D, F the correlations and/or absence of correlation of right secondary motor cortex (M2C, left side of the panel) and left posteromedial cortex (RS/V2, right side of the panel) are reported with relative R2 and P-values, (corrected for multiple comparisons). Dashed line is used to indicate a correlation close to statistical significance (panel B, left), while a continuous line indicates a statistically significant linear fitting and correlation (panel F, right) 10 Arena et al. Figure 5. The reduction of HF/LF ratio from low to high dosage of ketamine in the posteromedial cortex was explained by a selective higher increment of LF powers with respect to HF. (a) The colour maps show the topographical distributions (R-C: rostral-caudal) of the 16 EEG electrodes (small green circles) and the spatial interpolations of the LF power (1–4 Hz, above) and HF power (20–80 Hz, below), averaged across time, trials and rats, for each condition. (b) Left, averaged HF (dashed line) and LF (continuous line) powers across hemispheres and rats are shown for each cortical area, during both low and high ketamine dosage (K1 and K2, respectively). On the right, the colour maps show the spatial interpolation of t-scores from comparing LF powers (up) and HF powers (bottom) between K1 and K2 conditions, for each channel. The horizontal black lines in the colour bar indicate the threshold for statistical significance (t5 =4.5258, Bonferroni–Holm corrected). White arrowheads indicate channels with statistically significant differences across conditions. (c) Left, the ratio between low and high doses of ketamine is reported for both HF and LF powers at the level of each cortical area, thus showing the power increments induced by deep ketamine anaesthesia (K2). Right, spatial interpolation of the t-scores from comparing LF power increment with HF power increments induced by increasing ketamine dosage, for each channel. The horizontal black line in the colour bar indicates the threshold for statistical significance (t5 =4.5258, Bonferroni–Holm corrected). White arrowheads indicate the channels with statistically significant differences between the two frequency ranges sionofketaminetodissociatethecapacityforconsciousnessfrom experiments (Arena et al. 2021) might be due to the smaller sam- ST responsiveness.InagreementwithaPCIstudyinhumans(Sarasso plesizehere.However,inbothWandK1conditions,PCI builtup ST et al. 2015), we found that PCI was not significantly changed again after an initial decay, reaching similar values between 200 during the unresponsive state caused by light ketamine anaes- and 300ms after the stimulus onset (Fig. 2). Indeed, our results thesia (K1) compared to wakefulness (Fig. 2), even if it tended to supporttheideathathighperturbationalcomplexityisassociated be lower in the K1 condition. This tendency is more in line with with a capacity to sustain long-lasting sequences of determinis- results from a larger dataset from rats (Arena et al. 2021) and tic activations (Mukovski et al. 2007; Pigorini et al. 2015; Rosanova ST in agreement with the time course of PCI , which showed both et al. 2018; Arena et al. 2021) as shown by the long-lasting phase- periods of similarities and differences between wakefulness and locked cortical ERPs (Fig. 2) and by the strong yet diverse global light ketamine anaesthesia. The difference in the statistical sig- connectivity observed here (Supplementary Fig. S3). Moreover, ST ST nificance of the PCI values reported here compared to previous the widespread responses to stimulation required for high PCI Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 11 may also indicate a capacity for the kind of global broadcast- behavioural responsiveness and capacity for consciousness seem ing required for consciousness according to GNW (Mashour et al. more strongly connected and difficult to dissociate ( Sarasso et al. 2020), aswellasforthebraintofunctionasanintegratedanddif- 2015; Arena et al. 2021). ferentiated whole as is required for consciousness in IIT (Casali Next, we explored the spectral properties of ongoing cortical et al. 2013; Tononi et al. 2016). Although a partially reduced level activity in order to identify possible associations with the differ- ST of consciousness might be inferred from the tendency of PCI to ent combinations of behavioural responsiveness and capacity for ST be lower with light ketamine anaesthesia compared to wakeful- consciousness, as assessed by PCI (Comolatti et al. 2019). The ness, we observed relatively high spatiotemporal complexity dur- spontaneous EEG activity during ketamine anaesthesia was char- ing both conditions, with long-lasting and well-integrated phase- acterized by a widespread, dose-dependent increase in HF power locked cortical activations. Thus, taken together, our findings are compared to wakefulness (Figs 1 and 5), in agreement with pre- compatible with a fully, or at least partially, preserved capacity vious findings ( Maksimow et al. 2006; Akeju et al. 2016; Li and for consciousness during both wakefulness and light ketamine Mashour 2019). HF oscillations are usually associated with neu- anaesthesia, independently of behavioural responsiveness. ronal firing ( Steriade et al. 1996, 2001; Mukovski et al. 2007) and However, ketamine has also been shown to produce fluctua- cortical activation (Fernandez et al. 2017; Siclari et al. 2017; Poulet tionsbetweenhighandlowspatiotemporalcomplexityofsponta- and Crochet 2018). Thus, the observed increase in HF power is neous EEG in humans, following bolus injection of an anaesthetic consistent with the enhanced presynaptic release occurring after dose (Li and Mashour 2019). This phenomenon may be caused ketamine administration (Ferro et al. 2017), and with the idea by the unstable pharmacokinetic of the bolus injection, suggest- that ketamine, via its antagonism of NMDA receptors, might ing a dose-dependent effect. Consistently, it was reported that mainly inhibit GABAergic interneurons, producing a state of over- soon after bolus injection, when ketamine plasma level is high, allcorticalexcitation(Seamans2008). Interestingly, LFpoweralso the spontaneous EEG can assume a gamma-burst dynamic, with increased from wakefulness to ketamine anaesthesia in a dose- slow oscillations that interrupted an enhanced HF, gamma activ- dependent manner (Figs 1 and 5), in agreement with previous ity (Akeju et al. 2016). Conversely, when ketamine plasma level reports (Akeju et al. 2016; Li and Mashour 2019). Thus, ketamine was reduced, about 10min after bolus injection, the sponta- induced a scaling up of the entire power spectrum of the sponta- neous EEG activity was characterized by uninterrupted, stable neous EEG, suggesting a maintained balance between the inhibi- gamma/beta activity, with reduced slow frequency power (Akeju tion and excitation of the overall cortical network underlying the et al. 2016), and the spatiotemporal complexity of spontaneous EEG signal (Gao et al. 2017). This was supported by the observa- EEG stabilized at wakefulness-like values (Li and Mashour 2019). tionofasimilarspectralexponentacrossconditions(Fig.1),which Importantly, the spatiotemporal complexity of the thalamocorti- has been hypothesized to indicate an aroused or conscious state cal system is thought to be a promising neuronal correlate for the (Colombo et al. 2019; Lendner et al. 2020; Arena et al. 2021). Why ST capacity for consciousness (Tononi and Edelman 1998; Dehaene thendidweobservethestrongreductionofPCI inthetransition 2014; Koch et al. 2016; Sarasso et al. 2021), and was empirically from low to high ketamine condition (Fig. 2)? found to correlate with conscious experience, in both sponta- Onehypothesisisthatspecificcorticalcircuitsmightbepartic- neous (Ferenets et al. 2006; Schartner et al. 2015; Demertzi et al. ularly relevant for sustaining complex neuronal interactions and 2019) and perturbed activity (Massimini et al. 2009; Casali that the activation state of these circuits might diverge from the et al. 2013; Sarasso et al. 2015; Casarotto et al. 2016; Rosanova averagedynamicoftheentirecorticalnetwork.Indeed,itisknown et al. 2018; Comolatti et al. 2019). Thus, it is conceivable that that transient and local cortical deactivations or activations can lowandhigh dosesofketamine, althoughboth causebehavioural occur and dissociate from the global brain state, such as with unresponsiveness, may still induce quite different states of con- local sleep during wakefulness (Murphy et al. 2011; Vyazovskiy sciousness: the low dose may allow vivid but covert dream-like et al. 2011; Fernandez et al. 2017), possibly modifying the capac- experiences to occur (Collier 1972; Sarasso et al. 2015), while ity for behaviour and/or conscious experience (Vyazovskiy et al. the high dose may cause dreamless unconsciousness, due to 2011; Fernandez et al. 2017; Siclari et al. 2017; Poulet and Crochet different, dose-dependent effects on cortical complexity (Li and 2018). For example, it has been shown that localized reduction of Mashour 2019). LF power and increased HF activity (high HF/LF ratio) within the We tested this hypothesis by repeating the same electrophysi- posterior cortex is strongly associated with dream experience in ological recording/stimulations in the same rats during the con- humans during deep stages of sleep (Siclari et al. 2017), a state stant intravenous infusion of ketamine at a higher rate, which dominated by LF activity that is often linked to unconsciousness gives a more constant systemic concentration than with bolus (Tononi and Massimini 2008). Thus, to uncover the role and the ST injection. In this high-dose condition (K2), we found that PCI state of activation of specific areas, we similarly measured the HF was strongly reduced along with an earlier interruption of phase- (20–80Hz)/LF(1–4Hz)powerratiofromthespontaneousactivityof locked response (Fig. 2) and with a drastic reduction of cortical all the 16 epidural electrodes and compared across experimental functional connectivity and diversity (Supplementary Fig. S3). conditions. These results are compatible with a reduced capacity to inte- Although both light and deep ketamine anaesthesia caused an grateorbroadcastinformationwithintheglobalcorticalnetwork, unresponsive behavioural state, we were able to identify regional thus suggesting a reduced capacity for consciousness during variationsinHF/LFratiothatwererelatedtochangesintheglobal ST deep ketamine anaesthesia (K2). This also indicates a clear dose- PCI value. Strikingly, the bilateral posteromedial cortex was the dependent effect, which was only suggested by previous experi- only region that showed a consistent reduction of HF/LF ratio, ST ments(Akejuetal. 2016; Liand Mashour 2019). Inagreementwith from low to high ketamine dosage, along with the drop in PCI Li and Mashour (2019), our results demonstrate that ketamine (Fig. 3). The reduced ratio within the posteromedial cortex indi- anaesthesia can be used to dissociate behavioural responsive- cated a local deactivation (Poulet and Crochet 2018), which was ness from cortical complexity, thus representing a ‘unique tool explainedbyalargerincreaseofLFthanHFpowerinducedbythe to probe different states of consciousness’ (Li and Mashour 2019). increased ketamine dosage (Fig. 5). This local power imbalance This is in contrast to other general anaesthetics, for which might indicate a particularly pronounced gamma-burst activity 12 Arena et al. pattern (Supplementary Fig. S4), typical of high ketamine plasma induce slow-wave oscillations and synchronized, rhythmic neu- levels, with slow waves interrupting enhanced HF activity (Akeju ronal silencing selectively in the retrosplenial cortex in mice, et al. 2016; Li and Mashour 2019). Consistently, the HF/LF ratio affecting behaviour (Vesuna et al. 2020). The retrosplenial cortex ST over the posteromedial cortex strongly correlated with the PCI is a particularly highly integrated area, within the medial cortical level during ketamine administration (Fig. 4) and across all con- subnetwork of rodents (Zingg et al. 2014). It receives information ditions (Fig. 3), but not within wakefulness alone or between from the claustrum, indirectly from the hippocampus through ST wakefulness and light ketamine anaesthesia (Fig. 4), where PCI the subiculum, and it is directly interconnected with several sen- did not change substantially. These results indicate that the local sory areas (visual, auditory, and somatosensory) and high-order state of the RS/V2 cortex is associated with the capacity for long- associative areas, including the medial frontal cortex. Thus, it lasting, broadly integrated and differentiated cortical activations is likely to play important roles in multisensory integration and ST asassessedbyPCI .Thuspossibly,theposteromedialcortexmay also integration with higher functions such as episodic memory, play an important role in sustaining the capacity for conscious- spatial navigation, and motor planning (Zingg et al. 2014). Given ness, in a general agreement with both IIT and GNW (Tononi et al. this, it is not surprising that deactivation of this area (low HF/LF 2016;Mashouretal.2020).Inotherwords,ourresultsmaysupport ratio, Fig. 3), due to enhanced LF activity (Fig. 5), could be associ- the hypothesis that a selective deactivation of the posteromedial atedwithdisruptionofwidespreadintegrationofcomplexcortical ST cortex—as indicated by the localized decrease in HF/LF power— interactions as seen here, with the drop of PCI at high ketamine is correlated with, and may even underlie, a sharp reduction of dosage (Fig. 2). In other words, there is reason to believe that the brain’s capacity to globally broadcast information or to func- the specific deactivation of a region in the posteromedial cortex tion as an integrated and differentiated whole that is capable of can be directly involved in breaking down the properties required sustaining consciousness. This is complementary to the occur- for sustaining a capacity for consciousness. However, the asso- rence of dreaming during deep stages of sleep in humans, which ciative frontal cortex is also highly integrated (Zingg et al. 2014; was associated with reduced LF activity and increased HF power BarthasandKwan2017)andseveralpiecesofevidencesuggestits in posterior cortical areas, indicating a local cortical activation involvement in conscious processing (Del Cul et al. 2009; Kapoor [high HF/LF ratio, (Siclari et al. 2017)]. Moreover, our findings are et al. 2020; Weilnhammer et al. 2021; Levinson et al. 2021). More- consistent with the reduced functional integration and diversity over, our results cannot exclude that a selective inhibition or that was seen in the posterior regions of the brain’s default mode lesioning of M2, within a global activated state, could also dis- network during unconsciousness, in humans (Luppi et al. 2019). rupt cortical complexity. Thus, in future experiments, it will be In contrast, light ketamine anaesthesia produced a signif- important to causally control the state of activation of single cor- ST icant reduction of HF/LF ratio compared to wakefulness only tical areas, with local intervention, in combination with PCI , to over the right primary motor and somatosensory cortex (Fig. 3). betteraddresstheroleofspecificcorticalregionsinsustainingthe This was consistent with the loss of behavioural responsiveness capacity for consciousness. inducedbythelowketaminedose, andpossiblywithananalgesic The main findings presented here are compatible with several effect. A correlation between the HF/LF ratio of the right sec- theories of consciousness, as it is widely agreed that some sort ST ondary motor cortex and PCI was also found across conditions of long-range interactions within the brain are required to sus- (Fig.3). However, thisrelationwasatleastpartiallyexplainedbya tain its capacity for consciousness. For example, GNW requires wakefulness-specific weak correlation ( Fig. 4), which could reflect information to be globally broadcast (Baars 2005; Dehaene et al. behaviouralvariationswithinthesamestate,suchasactive/quiet 2011; Mashour et al. 2020), IIT requires the physical substrate wakefulness or transient attentional loading. These findings are of consciousness to be integrated (Tononi and Edelman 1998; indeedconsistentwiththeconnectionbetweentheprefrontalcor- Tononietal. 2016), andatleastsomehigher-ordertheoriesrequire tex and behavioural state, which was recently demonstrated by long-range interaction to maintain the capacity to form repre- local cortical injections of carbachol, during general anaesthesia, sentations in associative cortices about first-order states in early in rats (Pal et al. 2018). In these experiments, during sevoflu- sensory regions (Lau and Rosenthal 2011; Brown et al. 2019). raneanaesthesia,localcholinergicmodulationofbothassociative Nonetheless, our results suggest that the kind of global corti- frontalandparietalcorticesproducedatransitionfromslow-wave calcomplexityassociatedwithconsciousexperiencebreaksdown EEG activity to low voltage, wakefulness-like, fast oscillations when ketamine specifically deactivates the posteromedial cortex (Pal et al. 2018). EEG temporal complexity also increased with (Figs 3 and 4). Moreover, the finding that the ketamine-induced both local cortical activations (Pal et al. 2020). However, only the deactivation of primary motor and somatosensory cortex was cholinergic activation of the prefrontal cortex was able to restore associated with loss of behavioural responsiveness, but not with ST wakefulness-like motor activity (Pal et al. 2018), showing how cor- significant changes in PCI (see Fig. 3), or in durable and well- ticalcomplexitycanbeeffectivelydissociatedfrombehaviour(Pal integrated cortical activations (Fig. 3, Supplementary Fig. S3), et al. 2020). Unfortunately, PCI was not tested, and the state of seems to provide an example that some cortices may be deacti- consciousness was only inferred by the simple motor activity (Pal vated without disrupting the normal capacity for consciousness. et al. 2018, 2020), while it is known that these phenomena can Thus, theories of consciousness should be able to explain why dissociate in humans, under several circumstances (Blumenfeld some regions may be deactivated without any apparent effect 2005; Owen et al. 2006; Ghoneim et al. 2009; Noreika et al. 2011; on the brain’s overall capacity for consciousness, while others Castelnovo et al. 2018; Linassi et al. 2018). are more critical. We also observed that changes in the HF/LF In our experimental setting, the electrodes over the postero- ratio of secondary motor regions during wakefulness had a ten- ST medial cortex cover both the medial part of the secondary visual dency of correlating with changes in PCI (Fig. 4). This weak cortex and the caudal part of the retrosplenial cortex (Supple- relation may represent a modulatory effect of frontal regions mentary Fig. S1). Interestingly, the dose-dependent deactiva- on the overall capacity for consciousness during wakefulness. tion of this cortical region is reminiscent of a recent finding, in As these variations did not result from controlled intervention, whichsub-anaestheticbolusinjectionsofketaminewerefoundto theysuggestthatspontaneouschangesinfrontalactivity(HF/LF), Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 13 likely involving changes in related cognitive functions [working authors participated in the interpretation of results and revi- memory, attention, cognitive control, planning, decision-making, sion of the manuscript, and approved the final version of the etc.; (Dalley et al. 2004)], might reflect spontaneous modulation manuscript. of cortical complexity within limits of normal wakefulness. This seems to be compatible with theories of consciousness that Conflict of interest statement also attribute an active role to the frontal cortices of selec- The authors declare that they have no financial competing tively modulating the contents of consciousness and attention interests. at any given moment (Baars 2005; Dehaene et al. 2011; Lau and Rosenthal 2011; Helfrich et al. 2018; Brown et al. 2019; Mashour References et al. 2020). Of course, these findings are not conclusive, as the specific Akeju O, Song AH, Hamilos AE et al. Electroencephalogram sig- regionalchangesinHF/LFobservedaredependentonfluctuations natures of ketamine anesthesia-induced unconsciousness. Clin in ongoing activity as opposed to interventional inactivation. Fur- Neurophysiol 2016;127:2414–22. thermore, it is not necessarily the case that the changes in HF/LF Arena A, Comolatti R, Thon S et al. General anesthesia disrupts observed in different regions were caused by the same underlying complex cortical dynamics in response to intracranial electrical processes, and it is also uncertain that they were always indica- stimulation in rats. eNeuro 2021;8:4. tive of a deactivation of the region. To address these issues, we Arena A, Thon S, Storm J. PCI-like measure in rodents. Hum Brain aim for future experiments with controlled, direct inactivation Project Neuroinf Platform 2019. ST of individual cortical regions while measuring PCI from awake Baars BJ. Global workspace theory of consciousness: toward a cog- rodents. nitive neuroscience of human experience. In: Laureys S (ed.), The Boundaries of Consciousness: Neurobiology and Neuropathology. Amsterdam: Elsevier, 2005, 45–53. Conclusion Barthas F, Kwan AC. Secondary motor cortex: where “sensory” By comparing EEG in light and deep ketamine anaesthesia in meets “motor” in the rodent frontal cortex. Trends Neurosci rats, we dissociated changes in global and local cortical dynam- 2017;40:181–93. ics related to loss of behavioural responsiveness from those likely Blumenfeld H. Consciousness and epilepsy: why are patients ST related to capacity for consciousness (assessed by PCI ). Light with absence seizures absent? Prog Brain Res 2005;150: ketamine anaesthesia induced an unresponsive state with appar- 271–86. ent deactivation of primary somatosensory and motor regions Boly M, Massimini M, Tsuchiya N et al. Are the neural correlates of (indicated by locally reduced HF/LF power ratio). The mean consciousness in the front or in the back of the cerebral cortex? ST PCI value was somewhat lower than in wakefulness, but long- Clinical and neuroimaging evidence. J Neurosci 2017;37:9603–13. lasting phase-locked responses with high and diversified func- Brown EN, Lydic R, Schiff ND. General anesthesia, sleep, and coma. tional connectivity suggested a dissociated state, with largely N Engl J Med 2010;363:2638–50. conserved consciousness. In contrast, deep ketamine anaesthe- BrownR,LauH,LeDouxJE.Understandingthehigher-orderapproach ST sia strongly reduced PCI , with early interruption of phase- to consciousness. Trends Cogn Sci 2019;23:754–68. locked response and disruption of functional connectivity, sug- Casali AG, Gosseries O, Rosanova M et al. A theoretically based gesting a more fully unconscious state. The latter was associated index of consciousness independent of sensory processing and with highly selective deactivation of the posteromedial cortex, behavior. Sci Transl Med 2013;5:198ra105. which was not seen during light ketamine anaesthesia, suggest- Casarotto S, Comanducci A, Rosanova M et al. Stratification of unre- ing a significant role for this cortical region in the capacity for sponsive patients by an independently validated index of brain consciousness. complexity. Ann Neurol 2016;80:718–29. Castelnovo A, Lopez R, Proserpio P et al. deep stages of sleep para- Supplementary data somnias as disorders of sleep-state dissociation. Nat Rev Neurol 2018;14:470–81. Supplementary data is available at NCONSC online. Chernik DA, Gillings D, Laine H et al. Validity and reliability of the Observer’s Assessment of Alertness/Sedation Scale: study with Data availability intravenous midazolam. J Clin Psychopharmacol 1990;10:244–51. Data will be available in the public repository EBRAINS upon CohenMX.AnalyzingNeuralTimeSeriesData:TheoryandPractice.1edn. publication at the following DOIs: 10.25493/S0DM-BK5; 10.25493/ Cambridge, MA: MIT Press, 2014. E9EF-6SB. Collier BB. Ketamine and the conscious mind. Anaesthesia 1972;27:120–34. Colombo MA, Napolitani M, Boly M et al. The spectral exponent Funding of the resting EEG indexes the presence of consciousness dur- Thisproject/researchreceivedfundingfromtheEuropeanUnion’s ing unresponsiveness induced by propofol, xenon, and ketamine. Horizon2020FrameworkProgrammeforResearchandInnovation NeuroImage 2019;189:631–44. under the Specific Grant Agreements No. 945539 (Human Brain Comanducci A, Boly M, Claassen J et al. Clinical and advanced neu- Project SGA3, JFS), No. 785907 (Human Brain Project SGA2; JFS), rophysiologyintheprognosticanddiagnosticevaluationofdisor- and No. 720270 (Human Brain Project SGA1; JFS). ders of consciousness: review of an IFCN-endorsed expert group. Clin Neurophysiol 2020;131:2736–65. Author contributions Comolatti R, Pigorini A, Casarotto S et al. A fast and gen- A.A. designed the experiments; A.A., S.T. performed experi- eral method to empirically estimate the complexity of brain ments and collected data; A.A. analysed data; R.C. performed responses to transcranial and intracranial stimulations. Brain PCIST analysis; A.A., B.E.J., J.F.S. wrote the manuscript; all Stimul 2019;12:1280–9. 14 Arena et al. Compte A, Sanchez-Vives MV, McCormick DA et al. Cellular and Laureys S, Celesia GG, Cohadon F et al. Unresponsive wakefulness network mechanisms of slow oscillatory activity (<1 Hz) and syndrome: a new name for the vegetative state or apallic syn- wave propagations in a cortical network model. J Neurophysiol drome. BMC Med 2010;8:68. 2003;89:2707–25. Lendner JD, Helfrich RF, Mander BA et al. An electrophysiological Dalley JW, Cardinal RN, Robbins TW. Prefrontal executive and cogni- marker of arousal level in humans. eLife 2020;9:e55092. tive functions in rodents: neural and neurochemical substrates. LevinsonM,PodvalnyE,BaeteSHetal.Corticalandsubcorticalsigna- Neurosci Biobehav Rev 2004;28:771–84. tures of conscious object recognition. Nat Commun 2021;12:2930. David O, Kilner JM, Friston KJ. Mechanisms of evoked and induced Li D, Mashour GA. Cortical dynamics during psychedelic and anes- responses in MEG/EEG. NeuroImage 2006;31:1580–91. thetized states induced by ketamine. NeuroImage 2019;196:32–40. Dehaene S. Consciousness and the Brain: Deciphering How the Brain Linassi F, Zanatta P, Tellaroli P et al. Isolated forearm technique: a Codes Our Thoughts. Reprint edn. Westminster, London, England: meta-analysis of connected consciousness during different gen- Penguin Publishing Group, 2014. eral anaesthesia regimens. Br J Anaesth 2018;121:198–209. Dehaene S, Changeux J-P, Naccache L. The global neuronal Lombardi F, Herrmann HJ, de Arcangelis L. Balance of excitation and workspace model of conscious access: from neuronal architec- inhibition determines 1/f power spectrum in neuronal networks. tures to clinical applications. In: Dehaene S, Christen Y (eds), Chaos 2017;27:047402. Characterizing Consciousness: From Cognition to the Clinic? Berlin, Loomis AL, Harvey EN, Hobart GA. Cerebral states during sleep, as Heidelberg: Springer, 2011, 55–84. studied by human brain potentials. J Exp Psychol 1937;21:127. DelCulA,DehaeneS,ReyesPetal.Causalroleofprefrontalcortexin Luppi AI, Craig MM, Pappas I et al. Consciousness-specific dynamic thethresholdforaccesstoconsciousness.Brain2009;132:2531–40. interactions of brain integration and functional diversity. Nat Demertzi A, Tagliazucchi E, Dehaene S et al. Human conscious- Commun 2019;10:4616. ness is supported by dynamic complex patterns of brain signal Luppi AI, Golkowski D, Ranft A et al. Brain network integration coordination. Sci Adv 2019;5:eaat7603. dynamics are associated with loss and recovery of consciousness Esser SK, Hill SL, Tononi G. Sleep homeostasis and cortical synchro- induced by sevoflurane. Hum Brain Mapp 2021;42:2802–22. ¨ ¨ ¨ nization:I.Modelingtheeffectsofsynapticstrengthonsleepslow MaksimowA, Sarkela M, Långsjo JW et al. Increase in high frequency waves. Sleep 2007;30:1617–30. EEG activity explains the poor performance of EEG spectral Ferenets R, Lipping T, Anier A et al. Comparison of entropy and com- entropy monitor during S-ketamine anesthesia. Clin Neurophysiol plexity measures for the assessment of depth of sedation. IEEE 2006;117:1660–8. Trans Biomed Eng 2006;53:1067–77. Mashour GA, Roelfsema P, Changeux JP, Dehaene S. Conscious pro- FernandezLMJ,ComteJC,LeMerrePetal.Highlydynamicspatiotem- cessing and the global neuronal workspace hypothesis. Neuron poral organization of low-frequency activities during behav- 2020;105:776–98. ioral states in the mouse cerebral cortex. Cereb Cortex 2017;27: Massimini M, Huber R, Ferrarelli F et al. The sleep slow oscillation as 5444–62. a traveling wave. J Neurosci 2004;24:6862–70. Ferro M, Lamanna J, Ripamonti M et al. Functional mapping of Massimini M, Huber R, Ferrarelli F et al. A perturbational approach brain synapses by the enriching activity-marker SynaptoZip. Nat for evaluating the brain’s capacity for consciousness. Prog Brain Commun 2017;8:1229. Res 2009;177:201–14. Funk CM, Peelman K, Bellesi M et al. Role of somatostatin-positive Mukovski M, Chauvette S, Timofeev I et al. Detection of active and cortical interneurons in the generation of sleep slow waves. J silentstatesinneocorticalneuronsfromthefieldpotentialsignal Neurosci 2017;37:9132–48. during slow-wave sleep. Cereb Cortex 2007;17:400–14. Gao R, Peterson EJ, Voytek B. Inferring synaptic excitation/inhibition Murphy M, Huber R, Esser S et al. The cortical topography of local balance from field potentials. NeuroImage 2017;158:70–8. sleep. Curr Top Med Chem 2011;11:2438–46. GaoS,CalderonDP.Robustalternativetotherightingreflextoassess Nielsen TA. A review of mentation in REM and deep stages of sleep: arousal in rodents. Sci Rep 2020;10:20280. “covert” REM sleep as a possible reconciliation of two opposing GhoneimMM,BlockRI,HaffarnanMetal.Awarenessduringanesthe- models. Behav Brain Sci 2000;23:851–66. discussion 904–1121. ¨ ´ ´ sia: risk factors, causes and sequelae: a review of reported cases Noreika V, Jylhankangas L, Moro L et al. Consciousness lost and in the literature. Anesth Analg 2009;108:527–35. found:subjectiveexperiencesinanunresponsivestate.BrainCogn Giacino JT, Kalmar K, Whyte J. The JFK coma recovery scale-revised: 2011;77:327–34. measurementcharacteristicsanddiagnosticutility.ArchPhysMed Odegaard B, Knight RT, Lau H. Should a few null findings fal- Rehabil 2004;85:2020–9. sify prefrontal theories of conscious perception? J Neurosci Helfrich RF, Fiebelkorn IC, Szczepanski SM et al. Neural mechanisms 2017;37:9593–602. of sustained attention are rhythmic. Neuron 2018;99:854–65.e5. Owen AM, Coleman MR, Boly M et al. Detecting awareness in the Kapoor V, Dwarakanath A, Safavi S et al. Decoding the contents of vegetative state. Science 2006;313:1402. consciousness from prefrontal ensembles. BioRxiv 2020. https:// Pal D, Dean JG, Liu T et al. Differential role of prefrontal and www.biorxiv.org/content/10.1101/2020.01.28.921841v1.abstract. parietal cortices in controlling level of consciousness. Curr Biol Koch C, Massimini M, Boly M, Tononi G. Neural correlates 2018;28:2145–52.e5. of consciousness: progress and problems. Nat Rev Neurosci Pal D, Li D, Dean JG et al. Level of consciousness is dissociable from 2016;17:307–21. electroencephalographic measures of cortical connectivity, slow Lau H, Rosenthal D. Empirical support for higher-order the- oscillations, and complexity. J Neurosci 2020;40:605–18. ories of conscious awareness. Trends Cogn Sci 2011;15: Paxinos G, Watson C. The Rat Brain in Stereotaxic Coordinates. 6th edn. 365–73. Cambridge, MA: Academic Press, 2007. Capacity for consciousness under ketamine anaesthesia is selectively associated with activity in posteromedial cortex in rats 15 PigoriniA,SarassoS,ProserpioPetal.Bistabilitybreaks-offdetermin- Steriade M, Timofeev I, Grenier F. Natural waking and sleep isticresponsestointracorticalstimulationduringnon-REMsleep. states: a view from inside neocortical neurons. J Neurophysiol NeuroImage 2015;112:105–13. 2001;85:1969–85. Poulet JFA, Crochet S. The cortical states of wakefulness. Front Syst Tononi G. An information integration theory of consciousness. BMC Neurosci 2018;12:64. Neurosci 2004;5:42. Rosanova M, Fecchio M, Casarotto S et al. Sleep-like cortical Tononi G, Boly M, Massimini M et al. Integrated information the- OFF-periods disrupt causality and complexity in the brain ory:fromconsciousnesstoitsphysicalsubstrate.NatRevNeurosci of unresponsive wakefulness syndrome patients. Nat Commun 2016;17:450–61. 2018;9:4427. Tononi G, Edelman GM. Consciousness and complexity. Science Sanders RD, Tononi G, Laureys S et al. Unresponsiveness ̸= uncon- 1998;282:1846–51. sciousness. Anesthesiology 2012;116:946–59. TononiG,MassiminiM.Whydoesconsciousnessfadeinearlysleep? SarassoS,CasaliAG,CasarottoSetal.Consciousnessandcomplexity: Ann N Y Acad Sci 2008;1129:330–4. a consilience of evidence. Neurosci Conscious 2021. Valomon A, Riedner BA, Jones SG et al. A high-density electroen- Sarasso S, Boly M, Napolitani M et al. Consciousness and complex- cephalography study reveals abnormal sleep homeostasis in ity during unresponsiveness induced by propofol, xenon, and patientswithrapideyemovementsleepbehaviordisorder.SciRep ketamine. Curr Biol 2015;25:3099–105. 2021;11:4758. Schartner MM, Seth A, Noirhomme Q et al. Complexity of multi- van Vugt B, Dagnino B, Vartak D et al. The threshold for conscious dimensionalspontaneousEEGdecreasesduringpropofolinduced report: signal loss and response bias in visual and frontal cortex. general anaesthesia. PLoS One 2015;10:e0133532. Science 2018;360:537–42. Seamans J. Losing inhibition with ketamine. Nat Chem Biol Vanhaudenhuyse A, Charland-Verville V, Thibaut A et al. Conscious 2008;4:91–3. while being considered in an unresponsive wakefulness syn- Siclari F, Baird B, Perogamvros L et al. The neural correlates of drome for 20 years. Front Neurol 2018;9:671. dreaming. Nat Neurosci 2017;20:872–8. Vesuna S, Kauvar IV, Richman E et al. Deep posteromedial cortical Siclari F, Bernardi G, Cataldi J et al. Dreaming in deep stages of sleep: rhythm in dissociation. Nature 2020;586:87–94. a high-density EEG study of slow waves and spindles. J Neurosci Volgushev M, Chauvette S, Mukovski M, Timofeev I. Precise 2018;38:9175–85. long-range synchronization of activity and silence in neo- Siegle JH, Lopez AC, Patel YA et al. Open Ephys: an open-source, cortical neurons during slow-wave sleep. J Neurosci 2006;26: plugin-based platform for multichannel electrophysiology. J Neu- 5665–72. ral Eng 2017;14:045003. Vyazovskiy VV, Olcese U, Hanlon EC et al. Local sleep in awake rats. Solms M. Dreaming and REM sleep are controlled by different brain Nature 2011;472:443–7. mechanisms.BehavBrainSci2000;23:843–50.discussion904–1121. Vyazovskiy VV, Olcese U, Lazimy YM et al. Cortical firing and sleep SteriadeM,AmzicaF,ContrerasD.Synchronizationoffast(30-40Hz) homeostasis. Neuron 2009;63:865–78. spontaneous cortical rhythms during brain activation. J Neurosci Weilnhammer V, Fritsch M, Chikermane M et al. An active role 1996;16:392–417. of inferior frontal cortex in conscious experience. Curr Biol Steriade M, Nuñez A, Amzica F. A novel slow (< 1 Hz) oscillation 2021;31:2868–80.e8. of neocortical neurons in vivo: depolarizing and hyperpolarizing Zingg B, Hintiryan H, Gou L et al. Neural networks of the mouse components. J Neurosci 1993;13:3252–65. neocortex. Cell 2014;156:1096–111.

Journal

Neuroscience of ConsciousnessOxford University Press

Published: Mar 4, 2022

Keywords: consciousness; ketamine anesthesia; EEG markers of consciousness; perturbational complexity index

There are no references for this article.