Why bursts in thalamic nuclei receiving basal ganglia output fail to wake the cortex during sleep?

Why bursts in thalamic nuclei receiving basal ganglia output fail to wake the cortex during sleep? | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Why bursts in thalamic nuclei receiving basal ganglia output fail to wake the cortex during sleep? Xiaowei Liu, Jing Guang, Zvi Israel, Denise Wajnsztajn, Aeyal Raz, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7607905/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 23 Jan, 2026 Read the published version in Communications Biology → Version 1 posted You are reading this latest preprint version Abstract Thalamic neurons discharge tonically during wakefulness and rapid-eye-movement (REM) sleep, but burst during non-REM (NREM) sleep. It has been hypothesized that NREM thalamic bursts do not serve as a cortical "wake-up" signal due to their periodic and synchronized nature. To explore this hypothesis, we studied polysomnographic signals, field potentials, and spiking activity of multiple neurons in the ventral anterior and centromedian thalamic nuclei of two female non-human primates during naturally occurring vigilance states. These nuclei receive GABAergic output from the basal ganglia, with discharge rate decreasing during NREM sleep. Nevertheless, NREM bursting increased significantly as reported for glutamate-driven thalamic nuclei. NREM bursts were neither periodic nor tightly synchronized. However, EEG activity and thalamic field potentials time-locked to burst onset during NREM sleep markedly differed from those observed during wakefulness and REM sleep. These results suggest that NREM thalamic bursts do not awaken the cortex, due to unique state-dependent thalamocortical dynamics. Biological sciences/Neuroscience/Neural circuits Biological sciences/Neuroscience/Neuronal physiology Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction The thalamus is a key forebrain structure that gates peripheral, subcortical, and cortico-cortical communication. Thalamic neurons fire in tonic mode during wakefulness and REM sleep, but switch to burst mode during NREM sleep 1-4 . However, thalamic bursts were also observed in the awake state 5, 6 . The post-synaptic effects of bursts are more influential than those of single spikes 7, 8 . Therefore, bursts can be used to decode salient events and as “wake-up” calls for the cortex 5 . This naturally raises the question of why NREM thalamic bursts do not wake the cortex. The textbook answer is thatthe rhythmic and synchronized bursts of thalamocortical cells during NREM sleep 9, 10 reflect a null message for the cortex 1 . However, the periodicity and synchronization of thalamic bursts have never been analytically examined (for periodicity) or experimentally tested (for synchronization) in in vivo conditions. Here, we recorded the electrophysiological activity of ventral anterior (VA) and centromedian (CM) thalamic nuclei across the three vigilance states of non-human primates (NHPs). These thalamic nuclei are innervated by basal ganglia GABAergic output structures 11, 12 , and have been less explored in previous studies of the thalamus in sleep. The primary objectives of this study were to elucidate the tonic-burst activity of these thalamic nuclei and to understand why NREM thalamic bursts do not awaken the cortex. Results Two female African green monkeys (Chlorocebus aethiops sabaeus) contributed to the experiments (Fig. 1a-d). All experimental procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and the ethical guidelines of the Hebrew University of Jerusalem. Before the experimental recording, the animals were habituated to sleeping in the lab. A head holder, bilateral eye coils, two frontal skull-mounted EEG electrodes, and a 27 × 34 mm recording chamber located above a central frontoparietal craniectomy site were implanted across four separate surgeries. Surgeries were performed under general anaesthesia with appropriate antibiotic and analgesic administration. Each surgery was followed by a recovery and re-training period of 3-6 weeks. Finally, a 3T MRI scan verified the precise location of the recording chamber (Fig. 1a). Recordings were conducted in a noise-attenuated room. The experiment usually started at 4 PM. The light in the room was turned off at around 7-8 PM, and recording usually continued until 4-5 AM when the monkey returned to the open yard in the animal facility with her peers. We recorded the polysomnography signals (Fig. 1b-d), field potential (FP), and spiking activity from two sets of four microelectrodes targeting the right and left VA or CM thalamic nuclei. Figure 1. Experimental methods, sleep architecture, and general thalamic physiology across vigilance states. a, MRI and connectivity scheme of recording targets. b, One-night hypnogram and 20-minute (marked in red square) polysomnographic examples. The awake (W), NREM (N), REM (R), and unclassified/unavailable (UC/UA) epochs are marked in yellow, purple, pink, and light and dark grey, respectively. c, Sleep architecture of the two monkeys. d, top: proportion of W, N, and R states per session across sequential recording nights in both NHPs. Bottom: Violin plots of polysomnographic metric distributions across vigilance states. e, Examples of W, N, and R spiking activity. f, Firing rate. g, Inter spike intervals (ISI) histograms. h, Coefficient of variation of ISIs (CV-ISI). i, ISI return maps. j, Auto-correlation histograms. k, Power spectrum densities. l, Cross-correlation histograms. m, The distribution of R values at time zero. n, The average time-zero R values as a function of the vigilance state and the distance. The p-values indicating Bonferroni-corrected significant differences are marked in magenta; otherwise, they are in black. Abbreviations. BG-basal ganglia; BS-brain stem; CM-centromedian nucleus; Ctx-cortex; ContraHemi-contralateral hemisphere; EMG-electromyography; EEG-electroencephalogram; IpsiHemi-ipsilateral hemisphere; N, NREM-Non-rapid eye movement; R, REM-Rapid eye movement; RMS-root mean square; SameEle-same electrode; SpC-spinal cord; VA-Ventral anterior thalamic nucleus. We recorded the activity of 1,161 VA and CM neurons that emitted 1,254,441 bursts across 37 recording sessions in the two NHPs (Table 1). The results of a subclass of neurons with prominent isolation scores (Table S1) are reported in Fig. S1. These results align with those reported here. The results of the two NHPs did not differ (Fig. S2) and were therefore pooled for all subsequent analyses. Table 1. The neuronal database NHP/Structure VA CM VA+CM Nights Md 11 8 19 Wh 9 9 18 Md+Wh 20 17 37 Neurons Md 326 178 504 Wh 363 294 657 Md+Wh 689 472 1161 Bursts (MI detection method) Md 198,347 184,701 383,048 Wh 369,978 501,415 871,393 Md+Wh 568,325 686,116 1,254,441 This neuronal database shows the n umber of recording sessions (nights), well-isolated (isolation quality ≥ 0.7) recorded neurons, and detected (using the MI method) bursts in the ventral anterior and centromedian (VA and CM) thalamic nuclei of the two NHPs (Md and Wh) that participated in the study. Abbreviations. NHP – non-human primate; MI – maximum interval burst detection method. Sleep architecture Figure 1b illustrates a typical hypnogram and a 20-minute segment showing the changes in eye movements, EMG, and EEG across vigilance states. These states were semi-automatically detected 13 for non-overlapping 10-second periods. Figure 1c shows the average fraction of the vigilance states. The sleep patterns and their characteristics remained stable throughout the experiment duration (Fig. 1d). Spiking activity in the thalamus across vigilance states VA and CM neurons depict the expected tonic/burst/tonic activity during the awake, NREM, and REM sleep states, respectively (Fig. 1e). The average discharge rate of VA and CM neurons was approximately 15 spikes/s in the awake and REM sleep states. In line with previous thalamus studies 2-4, 14 and other brain structures 13 , the firing rate during NREM sleep significantly decreased (Fig. 1f). We used several methods to quantify the firing pattern of VA and CM neurons. Fig. 1g shows that the inter-spike interval (ISI) histograms during wakefulness and REM are relatively flat, while short ISIs during NREM are markedly more frequent. The coefficients of variation of the ISIs (CV-ISI) range from 1.5 to 2.5 (Fig. 1h), and are significantly higher during NREM (Fig. 1h and Fig. S1c). The ISI return maps (Fig. 1i and Fig. S1d), and the average autocorrelation histograms (Fig. 1j) reveal bursting activity during NREM. Finally, the average power spectrum densities (PSD) of thalamic spike trains (Fig. 1k) do not exhibit periodic peaks. Neuronal synchronization is best measured by cross-correlation analysis. In this study, neuron pairs were simultaneously recorded by the same electrode or by different electrodes in the ipsilateral or contralateral hemispheres. The distances between the corresponding neuron pairs in these situations are ≤ 0.2 15 , 0.5-4, and 5-10 mm, respectively. The average correlation coefficient (R) cross-correlation histograms of all spikes (Fig. 1l-n) depict a symmetric positive correlation around time zero in NREM, suggesting a “common-input” mechanism 16 . The conditional discharge rate correlation histograms (Fig. S3), and the correlation histograms of the multi-unit activity (i.e., raw signals filtered by a 300-9000Hz bandpass, Fig. S4) align with the above reports. Detection of thalamic bursts and their properties across vigilance states Our raw data (Fig. 1e) and discharge pattern analysis (Figs. 1g-k, S1b-d, i, j, S3a, and S4a) suggest that VA and CM neurons tend to burst more frequently during NREM. We, therefore, tested several burst detection methods to identify bursts and to investigate their properties (Fig. 2). We found that the maximal-interval method best matched the expert choice and used it hereafter. The Poisson surprise method yielded more conservative, but similar results (Fig. S5). Figs. 2a and S5a illustrate the raw spiking data with detected bursts marked in color. The burst properties are shown in Figs. 2b, S1e and S5b. The mean burst frequency ranges from 0.4 to 1 bursts/s, and is higher during NREM than during wakefulness and REM sleep. Accordingly, the mean inter-burst interval (IBI) varies between 1 and 2 seconds and is shorter during NREM sleep. This long IBI is in line with the low-threshold calcium mechanism and previous findings 2-4 . The mean duration of bursts is 15-20 ms. It is slightly longer in the VA relative to the CM, and tends to be shorter during NREM. The mean number of spikes per burst is between 4 and 5 spikes. Accordingly, the average frequency of the spikes in a burst is above 200 spikes/s, which is more than 10 times the tonic firing rate. Fig. S6 details the tonic activity (after burst removal) for comparison with the thalamic overall spiking activity (Fig. 1), and the burst activity (Figs. 2 and 3). Finally, the burst ordinal ISI relations decelerate (Fig. 2b, 2 nd and 4 th rows), in line with previous in vivo reports 17 and the associations of thalamic bursts with low-threshold calcium channels and spikes 9, 10 . Fig ure 2. Burst detection and features across vigilance states. a, Examples of burst detection. 2-second traces (left) and 100-ms segments around the burst marked by a red star (right). b, VA and CM burst features (two upper and two lower rows, respectively). 1 st Row, from left to right: Burst frequency, Inter-burst interval (IBI), burst duration, number of spikes per burst, intra-burst discharge rate. 2 nd Row, ordinal inter-spike intervals in bursts of different lengths. c, IBI-Time Histograms. d, Coefficient of variations of the IBIs (CV-IBI). e, IBI return map. White lines indicate log 10 (50) ms (the minimum silent time before a burst in the maximum interval burst detection methods). f, Average autocorrelation histograms of thalamic bursts. g, Power spectral density of thalamic bursts. h, Left - Histograms of burst spikes fraction in 10-second segments. Right - the distribution of burst spikes fraction per 10-second segments. I, fraction of firing modes (burst, tonic, and mixed) in different vigilance states. The p-values indicating Bonferroni-corrected significant differences are marked in magenta; otherwise, they are in black. Abbreviations and color coding as in Figure 1. To furtherly explore the characteristics of thalamic bursts, we proceeded to test their temporal patterns. We marked each detected burst as a single event and applied the same methods previously used for analysing individual spikes to the burst trains. The IBI histograms exhibit a predominance of relatively short IBI (Fig. 2c and S1f). The CV-IBI values (Figs. 2d and S1g) range between 0.8 and 1.2, suggesting that bursting timing more closely resembles a Poisson process than a periodic pattern. NREM exhibits the highest CV-IBI (>1) among the three vigilance states. Similarly, the IBI return map, the autocorrelation histogram, and the PSD of the bursts (Figs 2e-g, S1h-j, and S7a) did not show a robust signature of periodic activity 18, 19 . Rather, these metrics indicate a tendency for temporal clustering of the bursts (bursts of bursts), particularly during NREM sleep. The tendency for clustering of bursts may be related to the division of thalamic activity into tonic and burst mode states. The neural activity was segmented into 10-second epochs, aligned with the corresponding segments of polysomnography. For each segment, the fraction of burst spikes relative to total spikes was calculated (Figs. 2h, i, and S1k, l). Finally, the segments were classified into three discharge modes: burst (>30% of spikes in bursts), tonic (<3%), or mixed (3-30%). Figures 2h and S1k depict the distribution of these segments in our recordings. NREM differed markedly from the awake and REM states, with more than 65% of segments in burst mode (Figs. 2i and S1l). Synchronization patterns of thalamic bursts across vigilance states Although coupled periodic processes tend to synchronize, synchronization can be detected independently of periodicity 20 . We calculated the cross-correlation histograms of burst trains for neuron pairs simultaneously recorded by the same electrode (Fig. 3a), or by different electrodes in the ipsilateral or contralateral hemispheres (Fig. 1e). Figures 3b-d, S1m-o, S5j-l, and S6h-j illustrate the cross-correlation histograms, organized by distance and vigilance states. These histograms are calculated with a 0.7 ms bin width and smoothed using a moving Gaussian kernel with an SD of 60 bins. Most cross-correlation histograms demonstrate a central, symmetric, and broad peak. However, the magnitude of these peaks at time zero, indicating tight synchronicity, is extremely low. The correlation coefficient values do not exceed 0.001. Fig ure 3. Thalamic bursts during sleep exhibit prolonged timescales and loose synchronization . a, An example of bursts of two units (detected bursts are marked in blue and magenta stars, respectively) simultaneously recorded by the same electrode in the VA and CM nuclei. Left – 10-second trace, Right - higher magnification of the red box area (200-ms trace). b, Cross-correlation histograms of bursts from the same electrode, or different electrodes in the ipsilateral or contralateral hemisphere (left to right). c, The distribution of R values at time 0. The number of R values at time 0 is the same as that of the neuron pairs with repetition shown in b. Data is shown in mean ± SEM and the n indicates the number of neuron pairs with repetition in b . The Wilcoxon rank sum test was used to calculate the statistically significant difference of R values at time 0 among different vigilance stages (p < 0.05/3). P values indicating significant differences are marked in magenta. Otherwise, they are in black. d , 3D-bar plots of the average time-zero R values as a function of wake-sleep state and distance. Abbreviations and color coding as in Fig. 1. We also calculated the burst cross-correlation histogram as the conditional discharge rate (Fig. S7b) and found similar results. The conditional firing rate cross-correlation histogram enabled us to estimate the proportion of bursts attributable to a common input mechanism relative to the total number of bursts. We dubbed this metric the association index (AI) 21 . The bursts’ AI values are also low (AI < 0.15, Fig. S7c-f), even though their values are slightly higher than those of the all-inclusive (Fig. S3c-f) and burst-removed (tonic activity, Fig. S6k-o) spike trains. The AI values take into account all spikes in the central peak (from -0.5 to 0.5 s) of the cross-correlation histogram, thereby including both tightly and loosely synchronized burst pairs. We, therefore, recalculated the cross-correlation histograms with 1-s bins (and no smoothing). As expected, the R (t = 0) is higher (Fig. S8), but not exceeding 0.4 (R 2 < 0.16). Thus, both R and AI cross-correlation analyses indicate that thalamic bursts synchronize only loosely and over extended timescales. The relationship between t halamic bursts and the vigilance states Up to this point, our findings suggest that thalamic bursts during sleep are neither periodic nor tightly synchronized. We, therefore, tested whether sleep bursts triggered wakefulness or promoted transitions to micro-arousal states 22, 23 . The burst features of every 10-second segment were aligned to the vigilance state transition (e.g., from NREM to REM sleep, Fig. S9). For most vigilance state transitions, the burst frequency before and around the transition did not significantly exceed the expected values (Figs. 4a and S9a). Changes in other burst features qualitatively resemble those of burst frequency, except that the actual spike number per burst is significantly more than the predicted value before the transition from NREM to REM sleep (S9b, c). Additionally, bursts don’t help maintain wakefulness (Figs. 4b-Left and S9d), but rather significantly increase the probability of NREM during sleep (Figs. 4b-middle and S9e). Finally, they don’t play a role in the change of vigilance states during REM sleep (Figs. 4b-right and S9f). In a nutshell, bursts are not related to waking up our monkeys. Figure 4. The effect of bursts on awake-sleep stages, polysomnographic features, EEG, and thalamic FP across vigilance states . a, The difference between the actual and predicted burst frequencies aligned to the wake-sleep state transition. b, Comparison of the probability to stay in the same, or to switch to other vigilance states between segments with burst and tonic discharge. Positive values indicate this vigilance state is reinforced by bursts. c, Burst-triggered averages of eye open/close state. A zoomed-in subplot is displayed as an inset. d-h, Burst-triggered saccadic and drift eye movements, EMG, EEG, and FP. The colored circles in the subplots c-h indicate Bonferroni-corrected significant difference from the baseline. Abbreviations. W2N, N2R, N2W, R2N, and R2W - transitions from W to N, from N to R, from N to W, from R to N, and from R to W, respectively. WB vs WT – a neuron firing in burst or tonic mode in the W state. NB vs NT – a neuron firing in burst or tonic mode in the N state. RB vs RT – a neuron firing in burst or tonic mode in the R state. The remaining abbreviations and color coding are as shown in Fig. 1. The 10-second temporal resolution of our behavioral analysis might mask fast changes. Therefore, we examined the effect of a single burst on polysomnographic metrics with a smaller timescale (± 2 s) and high time resolution (0.7 or 1 ms). The percentage of eye closure increases around the bursts during NREM sleep (Fig. 4c). Additionally, the frequency of fast (saccade) and slow (drift) eye movements (Fig. 4d, e) as well as the EMG power (Fig. 4f) were not affected by thalamic bursts. T he relationship between t halamic bursts and EEG /FP activity across vigilance stages Finally, we examined the effects of thalamic bursts on the simultaneously recorded EEG and FP. The EEG activity time-locked to burst onset is shown in Figure 4g. It differs across vigilance stages. During NREM sleep, EEG activity begins to decline (indicating depolarization) one second before the burst. It is followed by a rapid rebound at the burst time and subsequent negative waves. The REM EEG shows a slight positive deflection before the thalamic bursts, which is more evident in CM. In the awake state, the burst triggers a subtle, slow positive EEG deflection, followed by a moderate, prolonged decrease after the burst. There are slight differences between the awake and REM EEG responses. However, the preceding robust NREM negative EEG deflection is unmatched. Cortical FP is often referred to as local EEG. Monopolar subcortical FP is probably affected by volume conductance from the cortex 24, 25 , and correlates with a non-linear summation of cortical activity. Figure 4h shows the burst-triggered FP recorded by the same electrode, or by different electrodes in the ipsilateral or contralateral hemispheres. The shape of the burst-triggered potential differs between the VA and CM and is slightly affected by the distance. In line with the EEG results, the vigilance stage has a robust effect on the burst-triggered FPs. The NREM evoked activity is much more pronounced. Notably, the difference between the effect of VA and CM bursts on FP is much more prominent than in the previous analysis of their properties. In any case, the burst-triggered FP and EEG indicate that the vigilance state robustly modulates thalamic-cortical dynamics. Discussion Neural activity in the thalamus is correlated with and probably plays a causal role in the generation of circadian and ultradian sleep rhythms. Thalamic neurons burst during NREM sleep. It has been proposed that these bursts are periodic and synchronized 1 , 5 , thereby transmitting a null signal to the cortex 26 . Our quantitative analysis reveals that the thalamic bursts were neither periodic nor tightly synchronized. However, burst-triggered EEG and FP responses varied across vigilance states, suggesting that NREM bursts do not wake the cortex due to altered dynamics of the thalamocortical network. Is the basal ganglia-thalamus chain broken at night? In 1989, Albin and colleagues published a groundbreaking model of the basal ganglia (BG)-thalamocortical network 11 . Over the years, BG researchers have proposed numerous modifications to this model 27 . Nevertheless, nearly all assumed that the output of the basal ganglia drives (inhibits or disinhibits) the thalamus and cortex 28 , 29 . Actions and movements are facilitated by reduced BG GABAergic flow in the thalamus, thalamic depolarization, and activation of the frontal cortex and descending motor pathways (Fig. 5 a). Parkinson’s akinesia is due to the opposite changes in BG, thalamic, and frontal-cortex activity. The researchers of the thalamus hold a different view. They classify the basal ganglia as modulators of thalamic activity. The current study suggests revising our BG models to align them with thalamic models 30 – 32 . The robust switch of VA and CM neurons to burst discharge during NREM does not align with the reduced discharge rate of their BG GABAergic afferents 13 , 33 . Notably, GABA release may be affected by the discharge pattern. The discharge pattern of basal ganglia output neurons tends to burst during NREM 13 , 33 , and this discharge pattern might lead to a larger release of GABA. Direct neurochemical studies of GABA flow in the thalamus during sleep are therefore needed. Additionally, the effects of GABA on the membrane potentials (and, consequently, on discharge rate) are complex 28 , 32 . Despite these limitations, we suggest that during sleep the basal ganglia modulate, but do not drive, their thalamic targets (Fig. 5 b). Finally, functional connectivity between the basal ganglia and thalamus may likewise depend on vigilance state. The BG may drive the thalamus in activated brain states, but modulate the thalamus during NREM sleep. Future studies may capitalize on the distinction between driving and modulation to investigate whether these mechanisms can be utilized to elucidate brain connectivity, including BG-to-brainstem connectivity, more effectively (Fig. 5 b). In vivo thalamic bursts are characterized by a prolonged refractory period but lack apparent periodicity . In vitro intracellular recording played a crucial role in understanding the role and mechanism of low-threshold calcium channels in the generation of thalamic bursts. However, brain slice preparations often exhibit more synchronous oscillations than those observed in the in vivo recordings 34 , 35 . Here, we recorded VA and CM spiking activity in vivo across natural vigilance states. Thalamic bursts were most common in the NREM stage. Notably, quantitative analysis of the burst discharge pattern, in both the temporal and the frequency domains, reveals no significant evidence of periodic thalamic bursts (Fig. 2 ). Previous in vivo studies reporting the periodic behavior of thalamic bursts have not employed quantitative analysis. Visual inspection of their raw data figures aligns with our analysis. In vivo recorded bursts of thalamic neuron pairs do not exhibit tight synchronization . As for periodicity, a synchronicity of neuronal activity is more commonly observed during low-arousal states and in vitro conditions. Physiological studies of thalamic brain slices 9 , 10 , as well as under anaesthesia 36 , often report synchronous activity. Our study of thalamic bursts reveals a broad and low-amplitude correlation peak (Fig. 3 ). Tight synchronicity would not enable simultaneous recording of two units by the same electrode. Again, our results, showing bursts of two distinct units recorded by a single microelectrode (Fig. 3 a), align with the raw data displayed in previous studies 3 , 4 . Our quantitative results indicate that the synchronicity of thalamic bursts during NREM is negligible, suggesting that, as for periodicity, synchronicity is probably not the hallmark of the thalamic null message. Why do thalamic bursts not wake the cortex during NREM? The EEG represents the summed activity of the underlying cortex. The subcortical FP is a good proxy for whole-cortex and thalamic activity 24 , 25 . Our burst-triggered EEG and FP (Fig. 4 g, h) reveal robust effects of the vigilance states. NREM EEG and FP activity exhibit an anticipatory negative deflection (indicating cellular depolarization) that begins one second before the thalamic burst. This probably suggests that during NREM sleep, cortical input is the primary driver of thalamic bursts. The post-burst EEG depolarization further suggests that the thalamic bursts are part of a closed thalamocortical loop that generates the delta rhythm 37 . In any case, the dynamics of the thalamocortical network during NREM sleep significantly differ from those during awake and REM sleep. We therefore suggest that this is why NREM bursts do not wake the cortex and the animal (Fig. 4 a-f). These distinct dynamics are probably due to shifts in cholinergic modulation of this network 38 , 39 , and underlie the varied effects of thalamic bursts on arousal and attention (Fig. 5 c,d). Finally, the impact of thalamic bursts on the cortex during REM sleep closely resembles that observed during wakefulness. The REM atonia prevents the cortical awake-like activity from resulting in dream enhancement. There may be additional reasons why NREM thalamic bursts fail to wake the cortex. The membrane potential of brain cells shifts dynamically between wakefulness and different sleep stages. In thalamic neurons, hyperpolarization during NREM promotes bursts. The NREM thalamic bursts fail to activate the cortex because cortical cells are hyperpolarized (down state). However, thalamic bursts can “wake up” the cortex during activated states since cortical cells are depolarized. Alternatively, the thalamic bursts perhaps play the “wake up” call function to the cortex when the bursts and tonic activity coexist. This means that the burst itself doesn’t transfer information but enhances the fidelity of information transfer of the tonic activity. Summary and Limitations . In this study, we recorded the activity of 1,161 VA and CM neurons that emitted 1,254,441 bursts across 37 recording sessions in two NHPs. VA and CM burst activity significantly increased during NREM sleep. However, these bursts were neither periodic nor tightly synchronized. Burst-triggered frontal EEG and thalamic FP varied across vigilance states. We, therefore, conclude that state-dependent thalamocortical dynamics support different functional roles of thalamic bursts during activated brain states and NREM sleep. Our readers should be aware of the study's limitations. The study is based on recordings from two NHPs. The results were consistent between these two NHPs. N = 2 is the standard practice in NHP research, and this is further justified by the 3R ethical rules for using animals in research. Nevertheless, further studies are needed. Similarly, our recordings are limited to only two thalamic nuclei out of more than twenty. The CM is part of the intralaminar thalamic nuclei. Therefore, the similarity between VA and CM results may justify generalizing our findings to other thalamic nuclei. Notably, the activity of the thalamic reticular nucleus, a key hub in the thalamocortical network, has not been explored. Third, our data were simultaneously recorded in left/right homologous structures, but not in serially connected structures (e.g., thalamus and cortex). A more comprehensive understanding of thalamocortical interactions requires simultaneous recordings from both structures and careful consideration of the immense diversity among cortical lamina and neurons 14 , 40 . We, therefore, hope that future multidisciplinary studies of different biological species, including humans, will support our results and conclusions. Materials and Methods Animals. Data were obtained from two female vervet monkeys (Cercopithecus aethiops, monkeys Md and Wh) weighing 4–5 kg. Care and surgical procedures followed the National Research Council Guide for the Care and Use of Laboratory Animals 41 and the Hebrew University guidelines for the care and use of animals in research. All experimental procedures were approved (MD-15-14412-5) and supervised by the Institutional Animal Care and Use Committee of the Hebrew University and Hadassah Medical Center, and the veterinary staff of the Hebrew University's primate facility. Training and surgery. The non-human primates (NHPs) were habituated to sitting and sleeping in a primate chair within a dark, double-walled, sound-attenuating experimental room. They were trained to perform a modified memory-guided saccade using a video eye-tracker (ISCAN, 21 Cabot Road, Woburn, MA 01801, USA). After the initial training, the NHPs underwent four surgical procedures over a six-month period. In the first surgery, a head holder and two cranial (ground) screws were implanted, and the behavioral training (with a video eye-tracker) was conducted with the head fixed. In the next two surgeries, eye coils were implanted in both eyes 42 . Finally, a 34*27 mm craniectomy was done, and a recording chamber and frontal EEG skull screws were implanted in the skull in the fourth surgery. The surgeries were performed by a board-certified neurosurgeon (ZI), an ophthalmologist (DW), and an anaesthesiologist (AR), with the support of the research team (JG and XL), and under veterinary supervision. All surgeries were performed under general anaesthesia with appropriate antibiotics and pain relief medicine. Following recovery from the last surgery, the precise location of the chamber was determined through a 3T MRI examination performed under moderate sedation (Medetomidine (Domitor) and Ketamine, i.m.). After the first (healthy condition) recording sessions, the NHPs were treated with MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine), and behavioral and neuronal activity were recorded in the Parkinsonian state. Here, we report only the results prior to the MPTP treatment. Upon completion of the experiment, all surgical attachments were removed from the NHPs. Monkey Md was then rehabilitated and placed at the Israeli Primate Sanctuary. Monkey Wh was euthanized to avoid suffering following discussions with the veterinary staff and the Institutional Animal Care and Use Committee. Recording Protocol : The NHPs were taken into the experimental room at around 4–5 PM during the week. They usually completed task performance within two hours and then slept for the entire night (from 7–8 PM until 4–5 AM, 5 nights per week), with the lights off but under infrared video and human supervision. They were food-restricted during the daytime and were fed during the task performance. Supplementary food was provided to the monkeys when they returned to the primate facility if their minimum daily caloric intake had not been met. They were housed in the monkey colony with their peers in a yard during the day and continuously on weekends. During the recording sessions, the eye open/close state was tracked by an infrared eye video tracker (49–50 frame/s) and the eye position was continuously recorded with the bilateral eye-coil X-Y signals (7-SSCP-JGASM CUSTOM 13 ‘’COIL 13’’, 7-MTS-4340 3D / 4 sense coil, and 6-YET-H3 EYE COIL TEMPLATE 12 TO 20mm; Crist Instrument, Hagerstown, MD, USA). The EMG signal (trapezius muscle) and two frontal skull EEG signals were also continuously recorded. The eye-coils, EMG, and EEG signals were sampled at 2,750 Hz (SnR, Alpha Omega Engineering, Nof Hagalil, Israel). The VA and CM thalamic nuclei were located based on the MRI examination (Fig. 1a), primate brain anatomic maps, and electrophysiological mapping 43 . During each recording session (night), up to four independently controlled microelectrodes were advanced separately into the targeted structures (VA or CM) in each hemisphere, for a total of up to eight electrodes per session. The microelectrodes were glass-coated tungsten, and their impedance ranged from 0.5 to 0.75 MΩ at 1000 Hz. Two experimenters (XL and JG) separately manipulate the Alpha Omega system (Electrode Positioning System, SnR, Alpha Omega Engineering, Nof Hagalil, Israel), each controlling four microelectrodes in one hemisphere. We only recorded the neuronal activity in homologous structures (e.g., left and right VA). The neural activity was hardware-filtered by broadband 0.075–9000Hz 4-pole Butterworth hardware filters and continuously sampled at 44 kHz. The raw neural activity was online filtered to visually display (and store) LFP (0.075–300 Hz, sampling rate of 1,375 Hz) and SPK (300–9000 Hz, sampling rate of 44 kHz). The single-unit activity was detected and sorted online by manually setting an amplitude threshold and the shape of the action potential (template), as well as the maximal allowed deviation between threshold-crossing signals and the template. Up to four templates can be generated for each electrode. The data was synchronized and collected by AlphaLab SnR (Alpha-Omega Engineering, Nof Hagalil, Israel). Polysomnography analysis . A detailed description of our polysomnography methods can be found here 44 . In the current research, EMG was digitally offline bandpass-filtered in the 10 to 500 Hz range (stopband at 0 to 5 Hz, 520 to 1375 Hz). To minimize phase distortions, the forward-backward filtering was performed (this zero-phase filtering was used for any signal filtered offline). Sleep staging was performed using a semiautomatic staging algorithm that clustered 10-second nonoverlapping segments. Different vigilance stages (wakefulness, NREM, REM, and ambiguous/unclassified) were identified based on the eye-open fraction, the root mean square of the EMG signal, and the high/low EEG power ratio (the average power at 15 to 25 Hz / the average power at 0.1 to 7 Hz). The segment would be classified as NA (Not Available) if the missed length of any signal (Eye open state, EEG, EMG, Coil eye-position, LFP, SPK) was longer than 11*(1/SR)*1000 ms (out of 10,000 ms); SR is the sampling rate of the corresponding signal. Before semiautomatic clustering, 10% of the night epochs were scored manually by a trained expert (JG). Both left and right EMGs were used separately for the staging analysis. The better staging results provided by the semiautomatic algorithm, which matched the expert staging in more than 85% of the tested segments, were accepted for further analysis. Eye-coil signals were converted from voltage to degrees (angular position) using calibration data obtained on the same day. The transform relationship was generated using the fit geometric transformation 45 . The 'projective' transformation type was used since the original calibration data demonstrated that the scene appeared tilted. This transformation maintained the straight lines straightness and converged parallel lines toward a vanishing point. The converted eye-coil signals were used to calculate the velocity and acceleration of the eyes, as well as to identify saccades and drifts. Only saccades/drifts with a magnitude larger than one degree were kept for further analysis. The basic sleep staging was refined based on the eye open percentage, EMG, and the saccade frequency of the right eye. Typically, the EMG RMS is relatively larger in NREM sleep and smallest in REM sleep (Fig. 1b, d and Fig. 4 f). Therefore, if a REM segment exhibits an EMG RMS value larger than the average EMG RMS of the NREM segments, or if an NREM segment shows an EMG RMS value smaller than the average EMG RMS of the REM segments, these segments will be considered unreasonable. This was done for both left and right EMG. The left/right EMG with relatively less unreasonable segments would be kept for further analysis. The NREM segments that occurred near REM epochs (two segments before and five after the previous REM segment) and additionally were characterized with eye-open ratio close to zero, more than 0.5/s eye saccades, and low EMG activity (the left/right EMG not larger than two times of the maximum EMG RMS during REM) were identified as REM candidates. For each candidate REM segment, if the duration of REM within the surrounding segments (three segments before and after the REM candidate) was longer than that of not-REM segments (NREM, wakefulness, or unclassified), this candidate REM would be classified as REM; otherwise, it would be identified as NREM. The awake or unclassified segments demonstrating an eye-open fraction between 0.4 and 0.7 were identified as wakeful candidates. These candidates would be classified as awake or unclassified segments using a method similar to the one used to determine the REM candidates. Finally, in the original classification of sleep stages, the eye open fraction must be larger than 0.6, 0.4–0.6, and smaller than 0.4 in the awake, unclassified, and sleep (NREM and REM) states, respectively. The NREM segments with eye-open fractions between 0.05 and 0.3 would be refined as unclassified if the length of unclassified time within the surroundings of this NREM candidate was longer than the NREM sleep time. An example of a one-night hypnogram and PSG example is shown in Fig. 1b. Spike Analysis. Spiking (300–9000 Hz) signals were filtered using an IIR comb notch filter with 881 (44,000/50 + 1) notches, which removed the 50 Hz power noise and its harmonics from the signal, given a sampling rate of 44,000 Hz. The Q factor for this filter was set to 35, namely, a highly sharp notch filter. The filtered spiking activities were rectified by absolute value, and the mean of the rectified vector was subtracted to obtain multi-unit activities (MUA). A low-pass Butterworth filter (210 Hz passband frequency, 260 Hz stopband frequency, 1 dB passband ripple, and 5 dB stopband attenuation) was designed to filter the MUA. The filtered MUA was down-sampled to 1/32 of the original sampling rate (yielding a 1,375 Hz sampling rate) by averaging the amplitude of the surrounding 32 sampling points. The filtered and down-sampled MUA was used to calculate its auto- and cross-correlation histograms after being segmented into 10-second epochs corresponding to the sleep staging epochs. The mean value of this 10-second MUA segment was subtracted to minimize the DC (zero frequency) power. The lag range of the auto- and cross-correlation histograms was from − 2 to 2 seconds. The auto- and cross-correlation histograms were calculated using the built-in function of MATLAB 2020b (xcorr, the correlation coefficient method), and smoothed by convolution with a Gaussian kernel having a standard deviation of 3.6 ms and 43.6 ms (5 and 60 times the time resolution, i.e., 1/1375 s), respectively. The xcorr correlation coefficient method ensures that the correlation coefficients (R) values could range from − 1 to 1. The action potentials (spikes) of well-isolated neurons (Isolation Score > 0.7 or > 0.85) 46 were transformed into a continuous binary train of 0/1 values, representing single-unit activity (SUA). The SUA was down-sampled to 1/32 of the original sampling rate (i.e., from 44 kHz to 1,375 Hz) by summation of its trains. The spike train of every neuron was also separated into 10-second epochs based on the segments of sleep stages. The firing rate (FR, the number of spikes/s), the inter-spike intervals (ISI), and the coefficient of variation of the ISIs (CV-ISI, defined as STD(ISI)/Mean(ISI)) were calculated for every 10-second segment, and then were clustered into three groups (wakefulness, NREM, and REM) based on the awake-sleep staging. For each neuron, the FR, ISI, and CV-ISI were averaged within each group. The power spectrum density (PSD) of SUA was calculated using a 10-second moving window, with a 5-second moving step, a frequency range of 0.1 Hz to 100 Hz, and a frequency resolution of 0.1 Hz. For each 10-second binary (0/1) train, we subtracted the mean to minimize the DC (Frequency = 0 Hz) power. The PSD unit is therefore given as NormSpk 2 /Hz. The auto- and cross-correlation (R-values) histograms of spike trains were calculated using the built-in function of MATLAB 2020b (xcorr, the correlation coefficient method). For this method, the mean value of each 10-second spike train was subtracted to show the negative and positive correlation values. The lag range and normalization were the same as for the MUA correlation analysis. The conditional discharge rate correlation histograms were calculated as the number of spikes (spike count) of the reference (triggered) cell, normalized by the number of spikes of the trigger neuron and the bin duration. The lag range was also from − 2 to 2 seconds. Edge effects were corrected by normalizing with the actual number of valid trigger spikes and bin duration. For example, at specific lag time points, a triggered spike train may be absent when aligned to a trigger spike, resulting in one fewer valid trigger spike contributing to that time point. The valid trigger count was adjusted by subtracting one from the total number of trigger spikes in such cases. The results from the conditional discharge method were used to calculate the association index (AI), which indicates the fraction of spikes contributed by the common-input mechanism out of total spikes. AI can be calculated as the area of the original cross-correlation histogram (counts/bin) peak divided by the total number of spikes from the triggered neurons. We used two ways to calculate the AI: AI avg = ((Nci 12 / Nt 2 ) + (Nci 21 / Nt 1 )) /2 (1) AI sqrt = (Nci 12 + Nci 21 )/(2*sqrt(Nt 1 *Nt 2 )) (2) AI sqrt /AI avg is the fraction of the spikes of the two studied neurons (1 and 2) generated by the common (shared) input to the neurons. Nci 12 indicates the number of spikes in the central (common input) peak (neuron 1 is the trigger neuron, and neuron 2 is the triggered one). Nci 21 indicates the number of spikes in the central peak, but the trigger/triggered neurons are switched. Nt 1 or Nt 2 is the total number of spikes of the triggered neurons. The base for calculating the Nci is the average firing rate of the triggered neuron, and the Nci peak range is from − 0.5s to 0.5s. Neurons 1 and 2 were referred to as two neuron pairs with repetition or one neuron pair without repetition. In the ‘with repetition’ condition, the pair was treated directionally, such that neuron 1 triggering neuron 2 and neuron 2 triggering neuron 1 were considered as two distinct pairs; in the ‘without repetition’ condition, both directions were used for the same calculation, and thus counted as a single neuron pair. Before getting the R values at time 0 and calculating the AI, the cross-correlation histograms were smoothed by Gaussian convolution with 43.6 ms (60 times the time resolution (1/1375 s)) standard deviation. The same smoothing method was also applied to the autocorrelation histograms and PSD. The standard deviations of the smoothing kernel for the autocorrelation histograms and PSD smoothing are 0.36 ms (0.5*(1/1375)*1000) and 0.2 Hz (2 times the frequency resolution (0.1 Hz)), respectively. Such light smoothing was applied to auto-correlation to minimize the strong effect of maximal values (1 for normalized autocorrelation histograms) at time zero. Burst detection. Bursts were detected using the maximal interval method (MI, https://www.neuroexplorer.com/downloads/NeuroExplorerManual.pdf ) 47 and the Poisson surprise (PS) method 48 . In the MI method, we set the maximum length of the first inter-spike interval (ISI) of a candidate burst to be 10 ms. We added spikes until we reached the maximum ISI limit (12 ms). A burst should display a minimum silent period of 50 ms (with zero or no more than one spike) before the burst. Every burst should include at least three spikes. In the PS method, we identified the burst candidate as at least three consecutive spikes with ISIs shorter (each one of them) than 1/10 of the average ISI in a 10-second epoch (ISI limit). Spikes are added to the end of the candidate burst until the ISI of the added spike is larger than 1.5 times the ISI limit or the number of added spikes is 5. We then calculated the PS of the candidate burst as a basic reference. Spikes will be removed from the beginning of the candidate burst if this maximizes the PS value. The removal process will be repeated until five spikes are removed, or the PS value is maximized. Ultimately, a candidate burst would be considered as a valid burst only if its PS value was at least 10. Burst analysis. Burst features, i.e., burst frequency, inter-burst interval (IBI), the coefficient of variation of IBI (CV-IBI), burst duration, the number of spikes per burst, the intra-burst firing rate, and the burst ISI ordinal duration, were analysed. Burst frequency indicates the number of bursts per second. IBI is the duration between the beginning (first spike) of the current (n) burst and the beginning (first spike) of the next (n + 1) burst. Burst durations were calculated as the duration between the burst's first and last spikes, and the typical duration of a single spike (96 sampling points) was added. The number of spikes per burst was divided by its corresponding burst duration to yield the intra-burst firing rate. The burst spikes were removed from the SUA to get the tonic activity. The FR, ISI, and CV-ISI were calculated for tonic activity. The same down-sampling features as used for the SUA were applied to tonic activity and to burst trains, where each burst was represented as a binary (0/1) event occurring at the time of the first spike in the burst. We also applied the same auto- and cross-correlation analyses described above for single-unit activity to tonic and burst activity. The cross-correlation histogram of burst trains in a particular vigilance state (e.g., NREM sleep) calculated by the conditional discharge method were removed from the analysis database if the trigger or triggered neuron had only one burst/10 s, both trigger and triggered neurons had bursts only near the same 2-second edge of the 10-second burst trains, and only one segment was available in this sleep stage. A total of 67 burst pairs without repetition were removed from the database of 4,416 burst pairs without repetition. The auto- and cross-correlation histograms of tonic activity were smoothed by convolution with a Gaussian kernel having a standard deviation of 0.36 ms (0.5*(1/1375)*1000) and 43.6 ms (60*(1/1375)*1000), respectively. For the burst trains, the standard deviations of the Gaussian kernel were 14.5 ms (20*(1/1375)*1000) for the auto-correlation histograms, and 43.6ms (60*(1/1375)*1000) for cross-correlation histograms. The auto-correlation histogram of burst trains was smoothed relatively stronger than that of SUA and tonic activity, because it had a longer refractory period. The same methods for calculating the AI and PSD of spike trains were also applied to tonic activity and burst trains. Therefore, the units of the PSDs for tonic activity and burst train are NormSpk 2 /Hz and NormBst 2 /Hz, respectively. The PSD of SUA, tonic activity, and burst trains was analyzed over the 0.1–10 Hz frequency range. Additionally, cross-correlation (R-values) histograms of spike trains, burst trains, and tonic activity with 1-second time resolution were also calculated using the correlation coefficient method (xcorr, the built-in function of MATLAB 2020b, Fig. S8). The percentage of spikes within bursts among all spikes in the 10-second epoch is defined as the probability of burst spikes. The firing mode of every 10-second segment was identified based on the probability of spikes occurring in bursts (burst spikes). Therefore, three firing modes were defined: tonic (< 3%), mixed (3%-30%), and burst (≥ 30%) mode. Analysis of the relations of bursts with behavioral and other physiological metrics. The burst frequency of every 10-second epoch, the probability of burst spikes, and the number of spikes per burst were aligned with the transition of sleep stages (e.g., transitioning from NREM to REM). To overcome the confounding effects of the inherently different percentage of vigilance stages, the corresponding predicted number of bursts was calculated using the average burst frequency of each neuron in wakefulness, NREM, or REM, and the percentage of the three sleep stages: N PreBst = BstNum W * Per W + BstNum N * Per N + BstNum R * Per R (3) N PreBst is the predicted number of bursts. BstNum W , BstNum N , and BstNum R indicate the average number of bursts, the probability of burst spikes or the number of spikes per burst in the wakefulness, NREM, and REM, respectively. Per W , Per N , and Per R indicate the percentage of the three vigilance states, respectively. The difference between the real and the corresponding predicted number of bursts was calculated. The two firing modes (burst and tonic) were combined with the three vigilance states (wakefulness, NREM, and REM). Therefore, there are six mode-stage combinations (burst-wakefulness, tonic-wakefulness, burst-NREM, tonic-NREM, burst-REM, and tonic-REM). We aligned the sleep stages to these six mode-stage combinations to explore the relationship between sleep states and firing mode. Only neurons having both tonic and burst firing mode segments were included for further analysis. Neurons having fewer than three aligned segments of wake-sleep stages were excluded. The difference between the percentage of the sleep stage related to the burst and tonic firing mode segments was calculated. Therefore, positive values indicated that the burst boosts staying in the same wake-sleep state or switching to another state. The eye-open/close state, rapid eye movements (saccades) frequency, slow eye movement (drifts) frequency, and EMG were aligned to bursts to reveal fast changes, which might be masked by the 10-second temporal resolution of our behavioral analysis. The eye open-close state is represented by the percentage of eye closure, i.e., the number of eye closures per second divided by the average frame rate of the video. A 4th -order bandpass Butterworth filter with a cut-off frequency of 8 Hz to 750 Hz was used to obtain the 10–500 Hz EMG signal. This filtered EMG was down-sampled from 2,750 Hz to 1,375 Hz by averaging its amplitude. The absolute value of the EMG (rectification) was calculated before averaging the filtered, down-sampled, and burst-aligned EMG. The burst-aligned eye open-close state, saccade frequency, and eye-drift frequency were smoothed by convolution with a Gaussian kernel with 50 ms (50*(1/1000)*1000) standard deviation. A Gaussian kernel with a 36.4 ms (50*(1/1375)*1000) standard deviation was used for the EMG. The frontal EEG and thalamic LFP (VA and CM) were also aligned to the bursts to study the relationship between the thalamic bursts and these extra- and intra-cranial physiological signals. The 50Hz power artifact and its harmonics (100Hz, 150Hz, 200Hz, and 250Hz) of EEG and LFP were removed by a second-order IIR notch filter. A low-pass Butterworth filter (210 Hz passband frequency, 260 Hz stopband frequency, 1 dB passband ripple, and 5 dB stopband attenuation) was also designed to filter the EEG and LFP. The EEG signals were resampled as the EMG signal. For EEG and LFP, the mean value was subtracted from the filtered and down-sampled signals before alignment to remove the DC component of the signal. We used a convolution with a Gaussian kernel having a standard deviation of 36.4 ms (50*(1/1375)*1000) to smooth the EEG and LFP signals. Finally, they were normalized by z-score based on their baseline (from two to one second before the burst). Statistical analysis : Statistical analysis was performed using MATLAB R2020b. All population data are presented as the mean ± SEM (standard error of the mean). The p-value was calculated using the Wilcoxon rank sum test for the independent data and the Wilcoxon signed-rank test for the two matched samples. The statistical tests were two-tailed. A P-value threshold of 0.05 was used, and the results were corrected for multiple comparisons using the Bonferroni method. Declarations Competing Interests All authors declare no competing interests. Funding This study is supported by grants from the ISF Breakthrough Research program (Grant No.: 1738/22) and the Collaborative Research Center TRR295, Germany (Project number 424778381) to HB. Author contributions JG and HB conceived the research and designed the experiments. ZI, DW, and AR performed the surgical procedure. XL and JG supported the surgical procedure. They also performed the experiments, including electrophysiological and behavioral recordings, analyzed the data, and conducted the statistical analysis. XL, JG, and HB prepared the figures and wrote the manuscript. HB supervised the work. All authors read and approved the final manuscript. Acknowledgments The authors would like to thank Uri Werner-Reiss, PhD, for his valuable support of the surgical procedures and all aspects of monkey care, Tamar Ravins Yaish, DMD, and the HUJI-ELSC animal facility team for their assistance. We thank Ad Aertsen for the fruitful discussion of correlation analysis and the association index, and Andy Horn, Jackie Schiller, Pnina Rapel, Aric Agmon, and Yuval Nir for their discussions and comments on early versions of the manuscript. We acknowledge the use of large language model (LLM) tools for linguistic editing to improve the clarity and grammar of this manuscript; no scientific content was generated by these tools. Data availability Data will be available upon request from the corresponding authors. Code availability Matlab code will be available upon request from the corresponding authors. Please note that the code used in this study was developed by the researchers for data analysis and visualization. It is intended for research purposes and may not meet professional coding standards. References Usrey, W.M., Sherman, S.M.: The Cerebral Cortex and Thalamus. Oxford University Press (2024) Weyand, T.G., Boudreaux, M., Guido, W.: Burst and tonic response modes in thalamic neurons during sleep and wakefulness. J. Neurophysiol. 85 , 1107–1118 (2001) McCarley, R.W., Benoit, O., Barrionuevo, G.: Lateral geniculate nucleus unitary discharge in sleep and waking: state- and rate-specific aspects. J. Neurophysiol. 50 , 798–818 (1983) Glenn, L., Steriade, M.: Discharge rate and excitability of cortically projecting intralaminar thalamic neurons during waking and sleep states. J. Neurosci. 2 , 1387–1404 (1982) Sherman, S.M.: A wake-up call from the thalamus. Nat. Neurosci. 4 , 344–346 (2001) Ramcharan, E.J., Gnadt, J.W., Sherman, S.M.: Burst and tonic firing in thalamic cells of unanesthetized, behaving monkeys. Vis. Neurosci. 17 , 55–62 (2000) Lisman, J.E.: Bursts as a unit of neural information: making unreliable synapses reliable. Trends Neurosci. 20 , 38–43 (1997) Guido, W., Lu, S.-M., Vaughan, J., Godwin, D.W., Sherman, S.M.: Receiver operating characteristic (ROC) analysis of neurons in the cat's lateral geniculate nucleus during tonic and burst response mode. Vis. Neurosci. 12 , 723–741 (1995) Lüthi, A., McCormick, D.A.: Periodicity of thalamic synchronized oscillations: the role of Ca2+-mediated upregulation of Ih. Neuron. 20 , 553–563 (1998) Jahnsen, H., Llinás, R.: Ionic basis for the electro-responsiveness and oscillatory properties of guinea‐pig thalamic neurones in vitro. J. Physiol. 349 , 227–247 (1984) Albin, R.L., Young, A.B., Penney, J.B.: The functional anatomy of basal ganglia disorders. Trends Neurosci. 12 , 366–375 (1989) Bergman, H., Wichmann, T., DeLong, M.R.: Reversal of experimental parkinsonism by lesions of the subthalamic nucleus. Science. 249 , 1436–1438 (1990) Mizrahi-Kliger, A.D., Kaplan, A., Israel, Z., Bergman, H.: Desynchronization of slow oscillations in the basal ganglia during natural sleep. Proceedings of the National Academy of Sciences 115, E4274-E4283 (2018) Redinbaugh, M.J., et al.: Thalamus modulates consciousness via layer-specific control of cortex. Neuron. 106 , 66–75 (2020). e12 Bar-Gad, I., Ritov, Y., Vaadia, E., Bergman, H.: Failure in identification of overlapping spikes from multiple neuron activity causes artificial correlations. J. Neurosci. Methods. 107 , 1–13 (2001) Perkel, D.H., Gerstein, G.L., Moore, G.P.: Neuronal spike trains and stochastic point processes: II. Simultaneous spike trains. Biophys. J. 7 , 419–440 (1967) Zirh, T., Lenz, F., Reich, S., Dougherty, P.: Patterns of bursting occurring in thalamic cells during parkinsonian tremor. Neuroscience. 83 , 107–121 (1998) Rivlin-Etzion, M., Ritov, Y., Heimer, G., Bergman, H.: Bar-Gad, I. Local shuffling of spike trains boosts the accuracy of spike train spectral analysis. J. Neurophysiol. 95 , 3245–3256 (2006) Cox, K.M., Kase, D., Znati, T., Turner, R.S.: Detecting rhythmic spiking through the power spectra of point process model residuals. J. Neural Eng. 21 , 046041 (2024) Heimer, G., Bar-Gad, I., Goldberg, J.A., Bergman, H.: Synchronization Of pallidal activity in the MPTP primate model of parkinsonism is not limited to oscillatory activity. Basal Ganglia VII, 29–34 (2002) Wong, R.O., Meister, M., Shatz, C.J.: Transient period of correlated bursting activity during development of the mammalian retina. Neuron. 11 , 923–938 (1993) Nir, Y., de Lecea, L.: Sleep and vigilance states: Embracing spatiotemporal dynamics. Neuron. 111 , 1998–2011 (2023) dos Lima, S.: Hippocampal and cortical communication around micro-arousals in slow-wave sleep. Sci. Rep. 9 , 5876 (2019) Logothetis, N.K., Kayser, C., Oeltermann, A.: In vivo measurement of cortical impedance spectrum in monkeys: implications for signal propagation. Neuron. 55 , 809–823 (2007) Marmor, O., et al.: Local vs. volume conductance activity of field potentials in the human subthalamic nucleus. J. Neurophysiol. 117 , 2140–2151 (2017) Churchland, M.M., Shenoy, K.V.: Preparatory activity and the expansive null-space. Nat. Rev. Neurosci. 25 , 213–236 (2024) Vishwanath, A., Bartlett, M.J., Falk, T., Cowen, S.L.: Decoupling of motor cortex to movement in Parkinson’s dyskinesia rescued by sub-anaesthetic ketamine. Brain. 148 , 2135–2150 (2025) Ben-Ari, Y.: The GABA excitatory/inhibitory developmental sequence: a personal journey. Neuroscience. 279 , 187–219 (2014) Mink, J.W.: The basal ganglia: focused selection and inhibition of competing motor programs. Prog. Neurobiol. 50 , 381–425 (1996) Bosch-Bouju, C., Hyland, B.I., Parr-Brownlie, L.C.: Motor thalamus integration of cortical, cerebellar and basal ganglia information: implications for normal and parkinsonian conditions. Front. Comput. Neurosci. 7 , 163 (2013) Schwab, B.C., et al.: Neural activity during a simple reaching task in macaques is counter to gating and rebound in basal ganglia–thalamic communication. PLoS Biol. 18 , e3000829 (2020) Goldberg, J.H., Farries, M.A., Fee, M.S.: Basal ganglia output to the thalamus: still a paradox. Trends Neurosci. 36 , 695–705 (2013) Adler, A., et al.: Neurons in both pallidal segments change their firing properties similarly prior to closure of the eyes. J. Neurophysiol. 103 , 346–359 (2010) Timofeev, I., Grenier, F., Steriade, M.: Impact of intrinsic properties and synaptic factors on the activity of neocortical networks in vivo. J. Physiology-Paris. 94 , 343–355 (2000) Crunelli, V., et al.: Dual function of thalamic low-vigilance state oscillations: rhythm-regulation and plasticity. Nat. Rev. Neurosci. 19 , 107–118 (2018) Steriade, M., Contreras, D., Amzica, F.: Synchronized sleep oscillations and their paroxysmal developments. Trends Neurosci. 17 , 201–207 (1994) Brain, A.: Thalamocortical Oscillations in the Sleeping. SCIENCE. 262 , 679–685 (1993) Nir, Y., Tononi, G.: Dreaming and the brain: from phenomenology to neurophysiology. Trends Cogn. Sci. 14 , 88–100 (2010) Steriade, M., McCarley, R.W.: Neuronal activities in brainstem and basal forebrain structures controlling waking and sleep states. In: Brain Control of Wakefulness Sleep. Springer, Boston, MA. , 381–416 (2005) Scala, F., et al.: Phenotypic variation of transcriptomic cell types in mouse motor cortex. Nature. 598 , 144–150 (2021) Council, N.R., et al.: Guide for the Care and Use of Laboratory Animals: Eighth Edition. National Academies (2011) Robinson, D.A.: A method of measuring eye movemnent using a scieral search coil in a magnetic field. IEEE Trans. Bio-med. Electron. 10 , 137–145 (1963) Glowinsky, S., Israel, Z., Heymann, S., Bergman, H.J.I.T.o.N, S., Engineering, R.: Divide and Conquer: Automatic Detection of the Thalamus to Empower DBS Physiological Navigation to the Subthalamic Region. IEEE Trans. Neural Syst. Rehabilitation Eng. (2025) Guang, J., et al.: Toward asleep DBS: cortico-basal ganglia spectral and coherence activity during interleaved propofol/ketamine sedation mimics NREM/REM sleep activity. npj Parkinson's Disease. 7 , 67 (2021) Goshtasby, A.: Image registration by local approximation methods. Image Vis. Comput. 6 , 255–261 (1988) Joshua, M., Elias, S., Levine, O., Bergman, H.: Quantifying the isolation quality of extracellularly recorded action potentials. J. Neurosci. Methods. 163 , 267–282 (2007) Cotterill, E., Eglen, S.: Burst detection methods. In: Chiappalone, M., Pasquale, V., Frega, M. (eds). In Vitro Neuronal Networks. Advances in Neurobiology, vol 22. Springer, Cham , 185–206 (2019) Legendy, C., Salcman, M.: Bursts and recurrences of bursts in the spike trains of spontaneously active striate cortex neurons. J. Neurophysiol. 53 , 926–939 (1985) Additional Declarations There is NO Competing Interest. Supplementary Files Supplementary.docx Cite Share Download PDF Status: Published Journal Publication published 23 Jan, 2026 Read the published version in Communications Biology → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7607905","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":519727853,"identity":"8c4b6a36-446f-4486-8460-596e9767ec57","order_by":0,"name":"Xiaowei Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABCUlEQVRIiWNgGAWjYDACZjApwcDHzMD4AMiQAbIZDBLAUswNeLWwMTMwGwAZPEhaGLFrgQE2IAIqZgBrgQLsWgyOMz98zFNhwcDGzmNWdaPGgodBuseg4AGDnTwDO3Ytks1sxsY8Z0AO4zG7nXMM6DCZMwZAhyUbNuBwGD8zg5l0bhtMCxvILzkgLcwJuPzCxsz+TTr3H0RLcc4/uJZ6nFr4gSqlcxsgWpiB1sG0HMapRbKZp9j4D9ALbMxsxdK5fUCGRFqBQYLBccM2HFoMzh/f+HBGTZ0cP//hjZ9zvgEZEsnbDH9UVMsDRQ5g0wIDPAjfAZEBgwGYQTxgfkCK6lEwCkbBKBj2AAA5CEDZ71VP6AAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0009-0006-8716-5925","institution":"The Edmond and Lily Safra Center for Brain Science, The Hebrew University, Jerusalem, Israel.","correspondingAuthor":true,"prefix":"","firstName":"Xiaowei","middleName":"","lastName":"Liu","suffix":""},{"id":519727854,"identity":"0cb19158-6c82-4b10-88ed-083cc6154287","order_by":1,"name":"Jing Guang","email":"","orcid":"https://orcid.org/0000-0002-5623-0716","institution":"The Hebrew University of Jerusalem,","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Guang","suffix":""},{"id":519727855,"identity":"29c3fe70-6e29-4a6b-9c0c-63fcc01fcf7d","order_by":2,"name":"Zvi Israel","email":"","orcid":"","institution":"Hadassah Hospital","correspondingAuthor":false,"prefix":"","firstName":"Zvi","middleName":"","lastName":"Israel","suffix":""},{"id":519727856,"identity":"73182e29-0fab-48bd-8e54-dcde17900678","order_by":3,"name":"Denise Wajnsztajn","email":"","orcid":"","institution":"Department of Ophthalmology, Hadassah University Hospital, Jerusalem, Israel.","correspondingAuthor":false,"prefix":"","firstName":"Denise","middleName":"","lastName":"Wajnsztajn","suffix":""},{"id":519727857,"identity":"2f005173-178f-4dd5-b300-14a594194e7b","order_by":4,"name":"Aeyal Raz","email":"","orcid":"","institution":"Department of Anaesthesiology, Rambam Health Care Campus, Haifa, Israel","correspondingAuthor":false,"prefix":"","firstName":"Aeyal","middleName":"","lastName":"Raz","suffix":""},{"id":519727858,"identity":"b25f0f78-b5a9-4c9b-95a9-75dd56e58d26","order_by":5,"name":"Hagai Bergman","email":"","orcid":"https://orcid.org/0000-0002-2402-6673","institution":"Hebrew University","correspondingAuthor":false,"prefix":"","firstName":"Hagai","middleName":"","lastName":"Bergman","suffix":""}],"badges":[],"createdAt":"2025-09-13 14:10:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7607905/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7607905/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s42003-026-09565-3","type":"published","date":"2026-01-23T05:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":95534030,"identity":"b66418a1-a79c-40a7-9d21-d881f8087fd5","added_by":"auto","created_at":"2025-11-10 10:28:04","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":292863,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eExperimental methods, sleep architecture, and general thalamic physiology across vigilance states. a, MRI and connectivity scheme of recording targets. b, One-night hypnogram and 20-minute (marked in red square) polysomnographic examples. The awake (W), NREM (N), REM (R), and unclassified/unavailable (UC/UA) epochs are marked in yellow, purple, pink, and light and dark grey, respectively. c, Sleep architecture of the two monkeys. d, top: proportion of W, N, and R states per session across sequential recording nights in both NHPs. Bottom: Violin plots of polysomnographic metric distributions across vigilance states. e, Examples of W, N, and R spiking activity. f, Firing rate. g, Inter spike intervals (ISI) histograms. h, Coefficient of variation of ISIs (CV-ISI). i, ISI return maps. j, Auto-correlation histograms. k, Power spectrum densities. l, Cross-correlation histograms. m, The distribution of R values at time zero. n, The average time-zero R values as a function of the vigilance state and the distance. The p-values indicating Bonferroni-corrected significant differences are marked in magenta; otherwise, they are in black. Abbreviations. BG-basal ganglia; BS-brain stem; CM-centromedian nucleus; Ctx-cortex; ContraHemi-contralateral hemisphere; EMG-electromyography; EEG-electroencephalogram; IpsiHemi-ipsilateral hemisphere; N, NREM-Non-rapid eye movement; R, REM-Rapid eye movement; RMS-root mean square; SameEle-same electrode; SpC-spinal cord; VA-Ventral anterior thalamic nucleus.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7607905/v1/f74a2e1c35f7dea0e19bffe0.jpg"},{"id":95534028,"identity":"b6360ac7-3925-480c-8db6-58469e7a3de8","added_by":"auto","created_at":"2025-11-10 10:28:03","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":268904,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBurst detection and features across vigilance states.\u003c/strong\u003e \u003cstrong\u003ea,\u003c/strong\u003e Examples of burst detection. 2-second traces (left) and 100-ms segments around the burst marked by a red star (right). \u003cstrong\u003eb,\u003c/strong\u003e VA and CM burst features (two upper and two lower rows, respectively). 1\u003csup\u003est \u003c/sup\u003eRow, from left to right: Burst frequency, Inter-burst interval (IBI), burst duration, number of spikes per burst, intra-burst discharge rate. 2\u003csup\u003end \u003c/sup\u003eRow, ordinal inter-spike intervals in bursts of different lengths. \u003cstrong\u003ec,\u003c/strong\u003e IBI-Time Histograms. \u003cstrong\u003ed,\u003c/strong\u003e Coefficient of variations of the IBIs (CV-IBI). \u003cstrong\u003ee,\u003c/strong\u003e IBI return map. White lines indicate log\u003csub\u003e10\u003c/sub\u003e(50) ms (the minimum silent time before a burst in the maximum interval burst detection methods). \u003cstrong\u003ef,\u003c/strong\u003e Average autocorrelation histograms of thalamic bursts. \u003cstrong\u003eg,\u003c/strong\u003e Power spectral density of thalamic bursts. \u003cstrong\u003eh,\u003c/strong\u003e Left - Histograms of burst spikes fraction in 10-second segments. Right - the distribution of burst spikes fraction per 10-second segments. \u003cstrong\u003eI,\u003c/strong\u003e fraction of firing modes (burst, tonic, and mixed) in different vigilance states. The p-values indicating Bonferroni-corrected significant differences are marked in magenta; otherwise, they are in black. Abbreviations and color coding as in Figure 1.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7607905/v1/76c25caea66706934ee034b8.jpg"},{"id":95534021,"identity":"373e2073-08cf-4e64-b518-07cf2bdb0aa4","added_by":"auto","created_at":"2025-11-10 10:28:01","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":160774,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThalamic bursts during sleep exhibit prolonged timescales and loose synchronization\u003c/strong\u003e. \u003cstrong\u003ea,\u003c/strong\u003e An example of bursts of two units (detected bursts are marked in blue and magenta stars, respectively) simultaneously recorded by the same electrode in the VA and CM nuclei. Left – 10-second trace, Right - higher magnification of the red box area (200-ms trace). \u003cstrong\u003eb,\u003c/strong\u003e Cross-correlation histograms of bursts from the same electrode, or different electrodes in the ipsilateral or contralateral hemisphere (left to right). \u003cstrong\u003ec,\u003c/strong\u003e The distribution of R values at time 0. The number of R values at time 0 is the same as that of the neuron pairs with repetition shown in \u003cstrong\u003eb.\u003c/strong\u003e Data is shown in mean ± SEM and the n indicates the number of neuron pairs with repetition in \u003cstrong\u003eb\u003c/strong\u003e. The Wilcoxon rank sum test was used to calculate the statistically significant difference of R values at time 0 among different vigilance stages (p \u0026lt; 0.05/3). P values indicating significant differences are marked in magenta. Otherwise, they are in black. \u003cstrong\u003ed,\u003c/strong\u003e 3D-bar plots of the average time-zero R values as a function of wake-sleep state and distance. Abbreviations and color coding as in Fig. 1.\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7607905/v1/d0e9e571b56bc4facf76cf31.jpg"},{"id":95534029,"identity":"7ef81aa2-51c6-4292-97d5-fe3c524d3705","added_by":"auto","created_at":"2025-11-10 10:28:03","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":193379,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eThe effect of bursts on awake-sleep stages, polysomnographic features, EEG, and thalamic FP across vigilance states\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003ea,\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e The difference between the actual and predicted burst frequencies aligned to the wake-sleep state transition. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eb, \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eComparison of the probability to stay in the same, or to switch to other vigilance states between segments with burst and tonic discharge. Positive values indicate this vigilance state is reinforced by bursts. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003ec,\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Burst-triggered averages of eye open/close state. A zoomed-in subplot is displayed as an inset. d-h, Burst-triggered saccadic and drift eye movements, EMG, EEG, and FP. The colored circles in the subplots c-h indicate Bonferroni-corrected significant difference from the baseline. Abbreviations. W2N, N2R, N2W, R2N, and R2W - transitions from W to N, from N to R, from N to W, from R to N, and from R to W, respectively. WB vs WT – a neuron firing in burst or tonic mode in the W state. NB vs NT – a neuron firing in burst or tonic mode in the N state. RB vs RT – a neuron firing in burst or tonic mode in the R state. The remaining abbreviations and color coding are as shown in Fig. 1.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7607905/v1/2d8b0e6a92793fc9c788a705.jpg"},{"id":95534022,"identity":"74273d26-1303-4989-95b2-bd412c5d342e","added_by":"auto","created_at":"2025-11-10 10:28:02","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":116974,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic models of basal ganglia-thalamus-cortex network – revised. a, The D1/D2 direct/indirect model of the basal ganglia-thalamus-cortex network. b, Revised model of the basal ganglia-thalamus-cortex modulatory/driving connectivity. c, A schematic model of the thalamic cortical network. d, Suggested model of changes in cholinergic modulation of the cortex and their effects on cortical layer-4 connectivity during wake-sleep stages. Abbreviations. A1- primary auditory cortex; BS – brainstem; CM – Centromedian intra-laminar thalamic nucleus; CT HO Nuc– cortical thalamic tract to high-order thalamic nuclei; CT HO + FO Nuc – cortical thalamic tract to high- and first-order thalamic nuclei; D1/D2 – D1 and D2 striatal medium spiny neurons; DR – Dorsal Raphe (Serotonin); EPT – Extra-pyramidal tracts; FF – feed forward; FB – feedback; GPe, GPi – external and internal segments of the globus pallidus; IN -interneurons; IT – Intratelencephalic tracts (=\u0026gt; low-hierarchy cortex =\u0026gt; striatum =\u0026gt; contralateral cortex); L2/3 – cortex layer 2/3 ( intra telencephalic neurons); L4 – cortex layer 4 (stellate neuron); L5 – cortex layer 5 (pyramidal and corticothalamic (to first order thalamic nuclei) neurons); L6 - cortex layer 6 (corticothalamic neuron (to first and higher-order thalamic nuclei)); LC – Locus coeruleus (Noradrenaline); LGN – lateral geniculate thalamic nucleus; M1 – primary motor cortex; MGN – medial geniculate thalamic nucleus; NBM – nucleus basalis of Mynert (Acetylcholine); PPN – Pedunculopontine nucleus (Acetylcholine); PT - pyramidal tract (=\u0026gt; spinal cord =\u0026gt; brainstem =\u0026gt; high-hierarchy Cortex); Pf - parafascicular intra-laminar thalamic nucleus; S1 – primary somatosensory cortex; SNc, SNr – substantia nigra compacta and reticulata; STN – subthalamic nucleus; Str – striatum; SuC – superior colliculus ; TMN – tuberomammillary nucleus (histamine); TC – Thalamocortical neuron; TRN –Thalamic reticular nucleus neuron; Thal – thalamus; V1 – primary visual cortex; VA – ventral anterior thalamic nucleus; VL- ventral lateral thalamic nucleus; VP – ventral posterior; VTA – Ventral tegmental area (Dopamine). Colour and shape coding: Light green – excitatory (e.g., glutamate) connection; Black red – inhibitory (e.g., GABA) connection; Black – dopaminergic connections; Cyan – cholinergic connection; Full and dashed lines, driving and modulatory connectivity; End arrow size reflects the connection efficacy.\u003c/p\u003e","description":"","filename":"Picture5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7607905/v1/d9173ae81f5369478b3897e7.jpg"},{"id":103052784,"identity":"61497557-18f6-44f9-95e0-ec47cecd187e","added_by":"auto","created_at":"2026-02-20 08:07:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2522562,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7607905/v1/bb848f9d-dd5b-4a64-a42b-fd2e3f22f437.pdf"},{"id":95534007,"identity":"fb4ddc11-c253-4912-9108-ff9353259831","added_by":"auto","created_at":"2025-11-10 10:27:59","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":3117805,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-7607905/v1/2ac1cfc1b2c7ac088f7be245.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Why bursts in thalamic nuclei receiving basal ganglia output fail to wake the cortex during sleep?","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe thalamus is a key forebrain structure that gates peripheral, subcortical, and cortico-cortical communication. Thalamic neurons fire in tonic mode during wakefulness and REM sleep, but switch to burst mode during NREM sleep\u003csup\u003e1-4\u003c/sup\u003e. However, thalamic bursts were also observed in the awake state\u003csup\u003e5, 6\u003c/sup\u003e. The post-synaptic effects of bursts are more influential than those of single spikes\u003csup\u003e7, 8\u003c/sup\u003e. Therefore, bursts can be used to decode salient events and as “wake-up” calls for the cortex\u003csup\u003e5\u003c/sup\u003e. This naturally raises the question of why NREM thalamic bursts do not wake the cortex. The textbook answer is thatthe rhythmic and synchronized bursts of thalamocortical cells during NREM sleep\u003csup\u003e9, 10\u003c/sup\u003e reflect a null message for the cortex\u003csup\u003e1\u003c/sup\u003e. However, the periodicity and synchronization of thalamic bursts have never been analytically examined (for periodicity) or\u0026nbsp;experimentally tested (for synchronization) in \u003cem\u003ein vivo\u003c/em\u003e conditions.\u003c/p\u003e\n\u003cp\u003eHere, we recorded the electrophysiological activity of ventral anterior (VA) and centromedian (CM) thalamic nuclei across the three vigilance states\u0026nbsp;of non-human primates\u0026nbsp;(NHPs). These thalamic nuclei are innervated by basal ganglia GABAergic output structures\u003csup\u003e11, 12\u003c/sup\u003e, and have been less explored\u0026nbsp;in\u0026nbsp;previous studies\u0026nbsp;of\u0026nbsp;the thalamus in\u0026nbsp;sleep.\u0026nbsp;The primary objectives of this study were to elucidate the tonic-burst activity of these\u0026nbsp;thalamic nuclei and to understand why NREM thalamic bursts do not awaken the cortex.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eTwo female African green monkeys (Chlorocebus aethiops sabaeus) contributed to the experiments (Fig. 1a-d). All experimental procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and the ethical guidelines of the Hebrew University of Jerusalem. Before the experimental recording, the animals were habituated to sleeping in the lab. A head holder, bilateral eye coils, two frontal skull-mounted EEG electrodes, and a 27 \u0026times; 34 mm recording chamber located above a central frontoparietal craniectomy site were implanted across four separate surgeries. Surgeries were performed under general anaesthesia with appropriate antibiotic and analgesic administration. Each surgery was followed by a recovery and re-training period of 3-6 weeks. Finally, a 3T MRI scan verified the precise location of the recording chamber (Fig. 1a).\u003c/p\u003e\n\u003cp\u003eRecordings were conducted in a noise-attenuated room. The experiment usually started at 4 PM. The light in the room was turned off at around 7-8 PM, and recording usually continued until 4-5 AM when the monkey returned to the open yard in the animal facility with her peers. We recorded the polysomnography signals (Fig. 1b-d), field potential (FP), and spiking activity from two sets of four microelectrodes targeting the right and left VA or CM thalamic nuclei.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFigure 1. Experimental methods, sleep architecture, and general thalamic physiology across vigilance states. a, MRI and connectivity scheme of recording targets. b, One-night hypnogram and 20-minute (marked in red square) polysomnographic examples. The awake (W), NREM (N), REM (R), and unclassified/unavailable (UC/UA) epochs are marked in yellow, purple, pink, and light and dark grey, respectively. c, Sleep architecture of the two monkeys. d, top: proportion of W, N, and R states per session across sequential recording nights in both NHPs. Bottom: Violin plots of polysomnographic metric distributions across vigilance states. e, Examples of W, N, and R spiking activity. f, Firing rate. g, Inter spike intervals (ISI) histograms. h, Coefficient of variation of ISIs (CV-ISI). i, ISI return maps. \u0026nbsp;j, Auto-correlation histograms. k, Power spectrum densities. l, Cross-correlation histograms. m, The distribution of R values at time zero. n, The average time-zero R values as a function of the vigilance state and the distance. The p-values indicating Bonferroni-corrected significant differences are marked in magenta; otherwise, they are in black. Abbreviations. BG-basal ganglia; BS-brain stem; CM-centromedian nucleus; Ctx-cortex; ContraHemi-contralateral hemisphere; EMG-electromyography; EEG-electroencephalogram; IpsiHemi-ipsilateral hemisphere; N, NREM-Non-rapid eye movement; R, REM-Rapid eye movement; RMS-root mean square; SameEle-same electrode; SpC-spinal cord; VA-Ventral anterior thalamic nucleus.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWe recorded the activity of 1,161 VA and CM neurons that emitted 1,254,441 bursts across 37 recording sessions in the two NHPs (Table 1). The results of a subclass of neurons with prominent isolation scores (Table S1) are reported in Fig. S1. These results align with those reported here. The results of the two NHPs did not differ (Fig. S2) and were therefore pooled for all subsequent analyses.\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"602\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\"\u003e\n \u003cp\u003eTable 1. The neuronal database\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNHP/Structure\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eVA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eVA+CM\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNights\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMd\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eWh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e18\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMd+Wh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e37\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNeurons\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMd\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e326\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e178\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e504\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eWh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e363\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e294\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e657\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMd+Wh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e689\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e472\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1161\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eBursts (MI detection method)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMd\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e198,347\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e184,701\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e383,048\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eWh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e369,978\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e501,415\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e871,393\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMd+Wh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e568,325\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e686,116\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1,254,441\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cem\u003eThis neuronal database shows the n\u003c/em\u003e\u003cem\u003eumber of recording sessions (nights), well-isolated (isolation quality\u0026nbsp;\u003c/em\u003e\u003cem\u003e\u0026ge;\u003c/em\u003e\u003cem\u003e\u0026nbsp;0.7) recorded neurons, and detected (using the MI method) bursts in the ventral anterior and centromedian (VA and CM) thalamic nuclei of the two NHPs (Md and Wh) that participated in the study. Abbreviations. NHP \u0026ndash; non-human primate; MI \u0026ndash; maximum interval burst detection method.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSleep architecture\u003c/em\u003e\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFigure 1b illustrates a typical hypnogram and a 20-minute segment showing the changes in eye movements, EMG, and EEG across vigilance states. These states were semi-automatically detected\u003csup\u003e13\u003c/sup\u003e for non-overlapping 10-second periods. Figure 1c shows the average fraction of the vigilance states. The sleep patterns and their characteristics remained stable throughout the experiment duration (Fig. 1d).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSpiking activity in the thalamus\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003eacross\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;vigilance states\u003c/em\u003e\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eVA and CM neurons depict the expected tonic/burst/tonic activity during the awake, NREM, and REM sleep states, respectively (Fig. 1e). The average discharge rate of VA and CM neurons was approximately 15 spikes/s in the awake and REM sleep states. In line with previous thalamus studies\u003csup\u003e2-4, 14\u003c/sup\u003eand other brain structures\u003csup\u003e13\u003c/sup\u003e, the firing rate during NREM sleep significantly decreased \u0026nbsp;(Fig. 1f).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe used several methods to quantify the firing pattern of VA and CM neurons.\u0026nbsp;Fig. 1g shows that the inter-spike interval (ISI) histograms during wakefulness and REM are relatively flat, while short ISIs during NREM are markedly more frequent. The coefficients of variation of the ISIs (CV-ISI) range from 1.5 to 2.5 (Fig. 1h), and are significantly higher during NREM (Fig. 1h and Fig. S1c). The ISI return maps (Fig. 1i and Fig. S1d), and the average autocorrelation histograms (Fig. 1j) reveal bursting activity during NREM. Finally, the average power spectrum densities (PSD) of thalamic spike trains (Fig. 1k) do not exhibit periodic peaks.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNeuronal synchronization is best measured by cross-correlation analysis. In this study, neuron pairs were simultaneously recorded by the same electrode or by different electrodes in the ipsilateral or contralateral hemispheres. The distances between the corresponding neuron pairs in these situations are \u0026nbsp;\u0026le; 0.2\u003csup\u003e15\u003c/sup\u003e, 0.5-4, and 5-10 mm, respectively. The average correlation coefficient (R) cross-correlation histograms of all spikes (Fig. 1l-n) depict a symmetric positive correlation around time zero in NREM, suggesting a \u0026ldquo;common-input\u0026rdquo; mechanism\u003csup\u003e16\u003c/sup\u003e. The conditional discharge rate correlation histograms (Fig. S3), and the correlation histograms of the multi-unit activity (i.e., raw signals filtered by a 300-9000Hz bandpass, Fig. S4) align with the above reports.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eDetection of thalamic bursts and their properties\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;across vigilance states\u003c/em\u003e\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOur raw data (Fig. 1e) and discharge pattern analysis (Figs. 1g-k, S1b-d, i, j, S3a, and S4a) suggest that VA and CM neurons tend to burst more frequently during NREM. We, therefore, tested several burst detection methods to identify bursts and to investigate their properties (Fig. 2). We found that the maximal-interval method best matched the expert choice and used it hereafter. The Poisson surprise method yielded more conservative, but similar results (Fig. S5).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFigs. 2a and S5a illustrate the raw spiking data with detected bursts marked in color. The burst properties are shown in Figs. 2b, S1e and S5b. The mean burst frequency ranges from 0.4 to 1 bursts/s, and is higher during NREM than during wakefulness and REM sleep. Accordingly, the mean inter-burst interval (IBI) varies between 1 and 2 seconds and is shorter during NREM sleep. This long IBI is\u0026nbsp;in line with\u0026nbsp;the\u0026nbsp;low-threshold calcium mechanism and previous findings\u003csup\u003e2-4\u003c/sup\u003e.\u0026nbsp;The mean duration of bursts\u0026nbsp;is 15-20 ms. It\u0026nbsp;is slightly longer in the VA\u0026nbsp;relative to the CM,\u0026nbsp;and tends to be shorter during NREM. The mean number of spikes per burst\u0026nbsp;is between 4 and 5 spikes. Accordingly, the average frequency of the spikes\u0026nbsp;in\u0026nbsp;a burst\u0026nbsp;is\u0026nbsp;above\u0026nbsp;200 spikes/s, which is\u0026nbsp;more than\u0026nbsp;10 times\u0026nbsp;the tonic firing rate.\u0026nbsp;Fig. S6 details the tonic activity (after burst removal) for comparison with the thalamic overall spiking activity (Fig. 1), and the burst activity (Figs. 2 and 3). Finally, the burst ordinal ISI relations decelerate (Fig. 2b,\u0026nbsp;2\u003csup\u003end\u003c/sup\u003e and 4\u003csup\u003eth\u003c/sup\u003e rows), in line with previous \u003cem\u003ein vivo\u003c/em\u003e reports\u003csup\u003e17\u003c/sup\u003e and the associations of thalamic bursts with low-threshold calcium channels and spikes\u003csup\u003e9, 10\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFig\u003c/strong\u003e\u003cstrong\u003eure\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;2.\u003c/strong\u003e \u003cstrong\u003eBurst detection and\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003efeatures across\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;vigilance states.\u003c/strong\u003e \u003cstrong\u003ea,\u003c/strong\u003e Examples of burst detection. 2-second traces (left) and 100-ms segments around the burst marked by a red star (right). \u003cstrong\u003eb,\u003c/strong\u003e VA and CM burst features (two upper and two lower rows, respectively). 1\u003csup\u003est\u0026nbsp;\u003c/sup\u003eRow, from left to right: Burst frequency, Inter-burst interval (IBI), burst duration, number of spikes per burst, intra-burst discharge rate. 2\u003csup\u003end\u0026nbsp;\u003c/sup\u003eRow, ordinal inter-spike intervals in bursts of different lengths. \u003cstrong\u003ec,\u003c/strong\u003e IBI-Time Histograms. \u003cstrong\u003ed,\u003c/strong\u003e Coefficient of variations of the IBIs (CV-IBI). \u003cstrong\u003ee,\u003c/strong\u003e IBI return map. White lines indicate log\u003csub\u003e10\u003c/sub\u003e(50) ms (the minimum silent time before a burst in the maximum interval burst detection methods). \u003cstrong\u003ef,\u003c/strong\u003e Average autocorrelation histograms of thalamic bursts. \u003cstrong\u003eg,\u003c/strong\u003e Power spectral density of thalamic bursts. \u003cstrong\u003eh,\u003c/strong\u003e Left - Histograms of burst spikes fraction in 10-second segments. Right - the distribution of burst spikes fraction per 10-second segments. \u003cstrong\u003eI,\u003c/strong\u003e fraction of firing modes (burst, tonic, and mixed) in different vigilance states. The p-values indicating Bonferroni-corrected significant differences are marked in magenta; otherwise, they are in black. Abbreviations and color coding as in Figure 1.\u003c/p\u003e\n\u003cp\u003eTo furtherly explore the characteristics of thalamic bursts, we proceeded to test their temporal patterns. We marked each detected burst as a single event and applied the same methods previously used for analysing individual spikes to the burst trains. The IBI histograms exhibit a predominance of relatively short IBI (Fig. 2c and S1f). The CV-IBI values (Figs. 2d and S1g) range between 0.8 and 1.2, suggesting that bursting timing more closely resembles a Poisson process than a periodic pattern. NREM exhibits the highest CV-IBI (\u0026gt;1) among the three vigilance states. Similarly, the IBI return map, the autocorrelation histogram, and the PSD of the bursts (Figs 2e-g, S1h-j, and S7a) did not show a robust signature of periodic activity\u003csup\u003e18, 19\u003c/sup\u003e. Rather, these metrics indicate a tendency for temporal clustering of the bursts (bursts of bursts), particularly during NREM sleep.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe tendency for clustering of bursts may be related to the division of thalamic activity into tonic and burst mode states. The neural activity was segmented into 10-second epochs, aligned with the corresponding segments of polysomnography. For each segment, the fraction of burst spikes relative to total spikes was calculated (Figs. 2h, i, and S1k, l). Finally, the segments were classified into three discharge modes: burst (\u0026gt;30% of spikes in bursts), tonic (\u0026lt;3%), or mixed (3-30%). Figures 2h and S1k depict the distribution of these segments in our recordings.\u0026nbsp;NREM differed markedly from the awake and REM states, with more than 65% of segments in burst mode (Figs. 2i and S1l).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSynchronization\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;patterns\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;of thalamic bursts\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003eacross\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;vigilance states\u003c/em\u003e\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlthough coupled periodic processes tend to synchronize, synchronization can be detected independently of periodicity\u003csup\u003e20\u003c/sup\u003e. We calculated the cross-correlation histograms of burst trains for neuron pairs simultaneously recorded by the same electrode (Fig. 3a), or by different electrodes in the ipsilateral or contralateral hemispheres (Fig. 1e). Figures 3b-d, S1m-o, S5j-l, and S6h-j illustrate the cross-correlation histograms, organized by distance and vigilance states. These histograms are calculated with a 0.7 ms bin width and smoothed using a moving Gaussian kernel with an SD of 60 bins. Most cross-correlation histograms demonstrate a central, symmetric, and broad peak. However, the magnitude of these peaks at time zero, indicating tight synchronicity, is extremely low. The correlation coefficient values do not exceed 0.001.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFig\u003c/strong\u003e\u003cstrong\u003eure\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;3.\u003c/strong\u003e \u003cstrong\u003eThalamic bursts during sleep exhibit prolonged timescales and loose synchronization\u003c/strong\u003e. \u003cstrong\u003ea,\u003c/strong\u003e An example of bursts of two units (detected bursts are marked in blue and magenta stars, respectively) simultaneously recorded by the same electrode in the VA and CM nuclei. Left \u0026ndash; 10-second trace, Right - higher magnification of the red box area (200-ms trace). \u003cstrong\u003eb,\u003c/strong\u003e Cross-correlation histograms of bursts from the same electrode, or different electrodes in the ipsilateral or contralateral hemisphere (left to right). \u003cstrong\u003ec,\u003c/strong\u003e The distribution of R values at time 0. The number of R values at time 0 is the same as that of the neuron pairs with repetition shown in \u003cstrong\u003eb.\u003c/strong\u003e Data is shown in mean \u0026plusmn; SEM and the n indicates the number of neuron pairs with repetition in \u003cstrong\u003eb\u003c/strong\u003e. The Wilcoxon rank sum test was used to calculate the statistically significant difference of R values at time 0 among different vigilance stages (p \u0026lt; 0.05/3). P values indicating significant differences are marked in magenta. Otherwise, they are in black. \u003cstrong\u003ed\u003c/strong\u003e\u003cstrong\u003e,\u003c/strong\u003e 3D-bar plots of the average time-zero R values as a function of wake-sleep state and distance. Abbreviations and color coding as in Fig. 1.\u003c/p\u003e\n\u003cp\u003eWe also calculated the burst cross-correlation histogram as the conditional discharge rate (Fig. S7b) and found similar results. The conditional firing rate cross-correlation histogram enabled us to estimate the proportion of bursts attributable to a common input mechanism relative to the total number of bursts. We dubbed this metric the association index (AI)\u003csup\u003e21\u003c/sup\u003e. \u0026nbsp;The bursts\u0026rsquo; AI values are also low (AI \u0026lt; 0.15, Fig. S7c-f), even though their values are slightly higher than those of the all-inclusive (Fig. S3c-f) and burst-removed (tonic activity, Fig. S6k-o) spike trains. The AI values take into account all spikes in the central peak (from -0.5 to 0.5 s) of the cross-correlation histogram, thereby including both tightly and loosely synchronized burst pairs. We, therefore, recalculated the cross-correlation histograms with 1-s bins (and no smoothing). As expected, the R (t = 0) is higher (Fig. S8), but not exceeding 0.4 (R\u003csup\u003e2\u003c/sup\u003e \u0026lt; 0.16). Thus, both R and AI cross-correlation analyses indicate that thalamic bursts synchronize only loosely and over extended timescales.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eThe relationship between t\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003ehalamic bursts and the vigilance states\u003c/em\u003e\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eUp to this point, our findings suggest that thalamic bursts during sleep are neither periodic nor tightly synchronized. We, therefore, tested whether sleep bursts triggered wakefulness or promoted transitions to micro-arousal states\u003csup\u003e22, 23\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe burst features of every 10-second segment were aligned to the vigilance state transition (e.g., from NREM to REM sleep, Fig. S9). For most vigilance state transitions, the burst frequency before and around the transition did not significantly exceed the expected values (Figs. 4a and S9a). Changes in other burst features qualitatively resemble those of burst frequency, except that the actual spike number per burst is significantly more than the predicted value before the transition from NREM to REM sleep (S9b, c). Additionally, bursts don\u0026rsquo;t help maintain wakefulness (Figs. 4b-Left and S9d), but rather significantly increase the probability of NREM during sleep (Figs. 4b-middle and S9e). Finally, they don\u0026rsquo;t play a role in the change of vigilance states during REM sleep (Figs. 4b-right and S9f). In a nutshell, bursts are not related to waking up our monkeys.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFigure 4.\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003eThe effect of bursts on\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;awake-sleep stages, polysomnographic features, EEG, and thalamic FP across vigilance states\u003c/em\u003e\u003c/strong\u003e\u003cem\u003e. \u003cstrong\u003ea,\u003c/strong\u003e The difference between the actual and predicted burst frequencies aligned to the wake-sleep state transition. \u003cstrong\u003eb,\u0026nbsp;\u003c/strong\u003eComparison of the probability to stay in the same, or to switch to other vigilance states between segments with burst and tonic discharge. Positive values indicate this vigilance state is reinforced by bursts. \u003cstrong\u003ec,\u003c/strong\u003e Burst-triggered averages of eye open/close state. A zoomed-in subplot is displayed as an inset. d-h, Burst-triggered saccadic and drift eye movements, EMG, EEG, and FP. The colored circles in the subplots c-h indicate Bonferroni-corrected significant difference from the baseline. Abbreviations. W2N, N2R, N2W, R2N, and R2W - transitions from W to N, from N to R, from N to\u003c/em\u003e\u003cem\u003e\u0026nbsp;W, from R to N, and from R to W, respectively. WB vs WT \u0026ndash; a neuron firing in burst or tonic mode in the W state. NB vs NT \u0026ndash; a neuron firing in burst or tonic mode in the N state. RB vs RT \u0026ndash; a neuron firing in burst or tonic mode in the R state. The remaining abbreviations and color coding are as shown in Fig. 1.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe 10-second temporal resolution of our behavioral analysis might mask fast changes. Therefore, we examined the effect of a single burst on polysomnographic metrics with a smaller timescale (\u0026plusmn; 2 s) and high time resolution (0.7 or 1 ms). The percentage of eye closure increases around the bursts during NREM sleep (Fig. 4c). Additionally, the frequency of fast (saccade) and slow (drift) eye movements (Fig. 4d, e) as well as the EMG power (Fig. 4f) were not affected by thalamic bursts.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eT\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003ehe relationship between t\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003ehalamic bursts and EEG\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e/FP\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;activity\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003eacross\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;vigilance stages\u003c/em\u003e\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFinally, we examined the effects of thalamic bursts on the simultaneously recorded EEG and FP. The EEG activity time-locked to burst onset is shown in Figure 4g. It differs across vigilance stages. During NREM sleep, EEG activity begins to decline (indicating depolarization) one second before the burst. It is followed by a rapid rebound at the burst time and subsequent negative waves. The REM EEG shows a slight positive deflection before the thalamic bursts, which is more evident in CM. In the awake state, the burst triggers a subtle, slow positive EEG deflection, followed by a moderate, prolonged decrease after the burst. There are slight differences between the awake and REM EEG responses. However, the preceding robust NREM negative EEG deflection is unmatched.\u003c/p\u003e\n\u003cp\u003eCortical FP is often referred to as local EEG. Monopolar subcortical FP is probably affected by volume conductance from the cortex\u003csup\u003e24, 25\u003c/sup\u003e, and correlates with a non-linear summation of cortical activity. Figure 4h shows the burst-triggered FP recorded by the same electrode, or by different electrodes in the ipsilateral or contralateral hemispheres. The shape of the burst-triggered potential differs between the VA and CM and is slightly affected by the distance. In line with the EEG results, the vigilance stage has a robust effect on the burst-triggered FPs. The NREM evoked activity is much more pronounced. Notably, the difference between the effect of VA and CM bursts on FP is much more prominent than in the previous analysis of their properties. In any case, the burst-triggered FP and EEG indicate that the vigilance state robustly modulates thalamic-cortical dynamics.\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eNeural activity in the thalamus is correlated with and probably plays a causal role in the generation of circadian and ultradian sleep rhythms. Thalamic neurons burst during NREM sleep. It has been proposed that these bursts are periodic and synchronized\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e, thereby transmitting a null signal to the cortex\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. Our quantitative analysis reveals that the thalamic bursts were neither periodic nor tightly synchronized. However, burst-triggered EEG and FP responses varied across vigilance states, suggesting that NREM bursts do not wake the cortex due to altered dynamics of the thalamocortical network.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eIs the basal ganglia-thalamus chain broken at night?\u003c/span\u003e In 1989, Albin and colleagues published a groundbreaking model of the basal ganglia (BG)-thalamocortical network\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. Over the years, BG researchers have proposed numerous modifications to this model\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. Nevertheless, nearly all assumed that the output of the basal ganglia drives (inhibits or disinhibits) the thalamus and cortex\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. Actions and movements are facilitated by reduced BG GABAergic flow in the thalamus, thalamic depolarization, and activation of the frontal cortex and descending motor pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). Parkinson\u0026rsquo;s akinesia is due to the opposite changes in BG, thalamic, and frontal-cortex activity.\u003c/p\u003e\u003cp\u003eThe researchers of the thalamus hold a different view. They classify the basal ganglia as modulators of thalamic activity. The current study suggests revising our BG models to align them with thalamic models\u003csup\u003e\u003cspan additionalcitationids=\"CR31\" citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. The robust switch of VA and CM neurons to burst discharge during NREM does not align with the reduced discharge rate of their BG GABAergic afferents\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. Notably, GABA release may be affected by the discharge pattern. The discharge pattern of basal ganglia output neurons tends to burst during NREM\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e, and this discharge pattern might lead to a larger release of GABA. Direct neurochemical studies of GABA flow in the thalamus during sleep are therefore needed. Additionally, the effects of GABA on the membrane potentials (and, consequently, on discharge rate) are complex\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. Despite these limitations, we suggest that during sleep the basal ganglia modulate, but do not drive, their thalamic targets (Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e5\u003c/span\u003eb). Finally, functional connectivity between the basal ganglia and thalamus may likewise depend on vigilance state. The BG may drive the thalamus in activated brain states, but modulate the thalamus during NREM sleep. Future studies may capitalize on the distinction between driving and modulation to investigate whether these mechanisms can be utilized to elucidate brain connectivity, including BG-to-brainstem connectivity, more effectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e5\u003c/span\u003eb).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cspan type=\"ItalicUnderline\" class=\"ItalicUnderline\" name=\"Emphasis\"\u003eIn vivo\u003c/span\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ethalamic bursts are characterized by a prolonged refractory period but lack apparent periodicity\u003c/span\u003e. \u003cem\u003eIn vitro\u003c/em\u003e intracellular recording played a crucial role in understanding the role and mechanism of low-threshold calcium channels in the generation of thalamic bursts. However, brain slice preparations often exhibit more synchronous oscillations than those observed in the \u003cem\u003ein vivo\u003c/em\u003e recordings\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Here, we recorded VA and CM spiking activity \u003cem\u003ein vivo\u003c/em\u003e across natural vigilance states. Thalamic bursts were most common in the NREM stage. Notably, quantitative analysis of the burst discharge pattern, in both the temporal and the frequency domains, reveals no significant evidence of periodic thalamic bursts (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Previous \u003cem\u003ein vivo\u003c/em\u003e studies reporting the periodic behavior of thalamic bursts have not employed quantitative analysis. Visual inspection of their raw data figures aligns with our analysis.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"ItalicUnderline\" class=\"ItalicUnderline\" name=\"Emphasis\"\u003eIn vivo\u003c/span\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003erecorded bursts of thalamic neuron pairs do not exhibit tight synchronization\u003c/span\u003e. As for periodicity, a synchronicity of neuronal activity is more commonly observed during low-arousal states and \u003cem\u003ein vitro\u003c/em\u003e conditions. Physiological studies of thalamic brain slices\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e, as well as under anaesthesia\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e, often report synchronous activity. Our study of thalamic bursts reveals a broad and low-amplitude correlation peak (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Tight synchronicity would not enable simultaneous recording of two units by the same electrode. Again, our results, showing bursts of two distinct units recorded by a single microelectrode (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003ea), align with the raw data displayed in previous studies\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. Our quantitative results indicate that the synchronicity of thalamic bursts during NREM is negligible, suggesting that, as for periodicity, synchronicity is probably not the hallmark of the thalamic null message.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eWhy do thalamic bursts not wake the cortex during NREM?\u003c/span\u003e The EEG represents the summed activity of the underlying cortex. The subcortical FP is a good proxy for whole-cortex and thalamic activity\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. Our burst-triggered EEG and FP (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eg, h) reveal robust effects of the vigilance states. NREM EEG and FP activity exhibit an anticipatory negative deflection (indicating cellular depolarization) that begins one second before the thalamic burst. This probably suggests that during NREM sleep, cortical input is the primary driver of thalamic bursts. The post-burst EEG depolarization further suggests that the thalamic bursts are part of a closed thalamocortical loop that generates the delta rhythm\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. In any case, the dynamics of the thalamocortical network during NREM sleep significantly differ from those during awake and REM sleep. We therefore suggest that this is why NREM bursts do not wake the cortex and the animal (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003ea-f). These distinct dynamics are probably due to shifts in cholinergic modulation of this network\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e, and underlie the varied effects of thalamic bursts on arousal and attention (Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e5\u003c/span\u003ec,d). Finally, the impact of thalamic bursts on the cortex during REM sleep closely resembles that observed during wakefulness. The REM atonia prevents the cortical awake-like activity from resulting in dream enhancement.\u003c/p\u003e\u003cp\u003eThere may be additional reasons why NREM thalamic bursts fail to wake the cortex. The membrane potential of brain cells shifts dynamically between wakefulness and different sleep stages. In thalamic neurons, hyperpolarization during NREM promotes bursts. The NREM thalamic bursts fail to activate the cortex because cortical cells are hyperpolarized (down state). However, thalamic bursts can \u0026ldquo;wake up\u0026rdquo; the cortex during activated states since cortical cells are depolarized. Alternatively, the thalamic bursts perhaps play the \u0026ldquo;wake up\u0026rdquo; call function to the cortex when the bursts and tonic activity coexist. This means that the burst itself doesn\u0026rsquo;t transfer information but enhances the fidelity of information transfer of the tonic activity.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eSummary and Limitations\u003c/span\u003e. In this study, we recorded the activity of 1,161 VA and CM neurons that emitted 1,254,441 bursts across 37 recording sessions in two NHPs. VA and CM burst activity significantly increased during NREM sleep. However, these bursts were neither periodic nor tightly synchronized. Burst-triggered frontal EEG and thalamic FP varied across vigilance states. We, therefore, conclude that state-dependent thalamocortical dynamics support different functional roles of thalamic bursts during activated brain states and NREM sleep.\u003c/p\u003e\u003cp\u003eOur readers should be aware of the study's limitations. The study is based on recordings from two NHPs. The results were consistent between these two NHPs. N\u0026thinsp;=\u0026thinsp;2 is the standard practice in NHP research, and this is further justified by the 3R ethical rules for using animals in research. Nevertheless, further studies are needed. Similarly, our recordings are limited to only two thalamic nuclei out of more than twenty. The CM is part of the intralaminar thalamic nuclei. Therefore, the similarity between VA and CM results may justify generalizing our findings to other thalamic nuclei. Notably, the activity of the thalamic reticular nucleus, a key hub in the thalamocortical network, has not been explored. Third, our data were simultaneously recorded in left/right homologous structures, but not in serially connected structures (e.g., thalamus and cortex). A more comprehensive understanding of thalamocortical interactions requires simultaneous recordings from both structures and careful consideration of the immense diversity among cortical lamina and neurons\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. We, therefore, hope that future multidisciplinary studies of different biological species, including humans, will support our results and conclusions.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eAnimals.\u003c/span\u003e Data were obtained from two female vervet monkeys (Cercopithecus aethiops, monkeys Md and Wh) weighing 4\u0026ndash;5 kg. Care and surgical procedures followed the National Research Council Guide for the Care and Use of Laboratory Animals\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e and the Hebrew University guidelines for the care and use of animals in research. All experimental procedures were approved (MD-15-14412-5) and supervised by the Institutional Animal Care and Use Committee of the Hebrew University and Hadassah Medical Center, and the veterinary staff of the Hebrew University's primate facility.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eTraining and surgery.\u003c/span\u003e The non-human primates (NHPs) were habituated to sitting and sleeping in a primate chair within a dark, double-walled, sound-attenuating experimental room. They were trained to perform a modified memory-guided saccade using a video eye-tracker (ISCAN, 21 Cabot Road, Woburn, MA 01801, USA). After the initial training, the NHPs underwent four surgical procedures over a six-month period. In the first surgery, a head holder and two cranial (ground) screws were implanted, and the behavioral training (with a video eye-tracker) was conducted with the head fixed. In the next two surgeries, eye coils were implanted in both eyes\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. Finally, a 34*27 mm craniectomy was done, and a recording chamber and frontal EEG skull screws were implanted in the skull in the fourth surgery. The surgeries were performed by a board-certified neurosurgeon (ZI), an ophthalmologist (DW), and an anaesthesiologist (AR), with the support of the research team (JG and XL), and under veterinary supervision. All surgeries were performed under general anaesthesia with appropriate antibiotics and pain relief medicine.\u003c/p\u003e\u003cp\u003eFollowing recovery from the last surgery, the precise location of the chamber was determined through a 3T MRI examination performed under moderate sedation (Medetomidine (Domitor) and Ketamine, i.m.). After the first (healthy condition) recording sessions, the NHPs were treated with MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine), and behavioral and neuronal activity were recorded in the Parkinsonian state. Here, we report only the results prior to the MPTP treatment. Upon completion of the experiment, all surgical attachments were removed from the NHPs. Monkey Md was then rehabilitated and placed at the Israeli Primate Sanctuary. Monkey Wh was euthanized to avoid suffering following discussions with the veterinary staff and the Institutional Animal Care and Use Committee.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eRecording Protocol\u003c/span\u003e: The NHPs were taken into the experimental room at around 4\u0026ndash;5 PM during the week. They usually completed task performance within two hours and then slept for the entire night (from 7\u0026ndash;8 PM until 4\u0026ndash;5 AM, 5 nights per week), with the lights off but under infrared video and human supervision. They were food-restricted during the daytime and were fed during the task performance. Supplementary food was provided to the monkeys when they returned to the primate facility if their minimum daily caloric intake had not been met. They were housed in the monkey colony with their peers in a yard during the day and continuously on weekends.\u003c/p\u003e\u003cp\u003eDuring the recording sessions, the eye open/close state was tracked by an infrared eye video tracker (49\u0026ndash;50 frame/s) and the eye position was continuously recorded with the bilateral eye-coil X-Y signals (7-SSCP-JGASM CUSTOM 13 \u0026lsquo;\u0026rsquo;COIL 13\u0026rsquo;\u0026rsquo;, 7-MTS-4340 3D / 4 sense coil, and 6-YET-H3 EYE COIL TEMPLATE 12 TO 20mm; Crist Instrument, Hagerstown, MD, USA). The EMG signal (trapezius muscle) and two frontal skull EEG signals were also continuously recorded. The eye-coils, EMG, and EEG signals were sampled at 2,750 Hz (SnR, Alpha Omega Engineering, Nof Hagalil, Israel).\u003c/p\u003e\u003cp\u003eThe VA and CM thalamic nuclei were located based on the MRI examination (Fig.\u0026nbsp;1a), primate brain anatomic maps, and electrophysiological mapping\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e. During each recording session (night), up to four independently controlled microelectrodes were advanced separately into the targeted structures (VA or CM) in each hemisphere, for a total of up to eight electrodes per session. The microelectrodes were glass-coated tungsten, and their impedance ranged from 0.5 to 0.75 MΩ at 1000 Hz. Two experimenters (XL and JG) separately manipulate the Alpha Omega system (Electrode Positioning System, SnR, Alpha Omega Engineering, Nof Hagalil, Israel), each controlling four microelectrodes in one hemisphere. We only recorded the neuronal activity in homologous structures (e.g., left and right VA). The neural activity was hardware-filtered by broadband 0.075\u0026ndash;9000Hz 4-pole Butterworth hardware filters and continuously sampled at 44 kHz. The raw neural activity was online filtered to visually display (and store) LFP (0.075\u0026ndash;300 Hz, sampling rate of 1,375 Hz) and SPK (300\u0026ndash;9000 Hz, sampling rate of 44 kHz). The single-unit activity was detected and sorted online by manually setting an amplitude threshold and the shape of the action potential (template), as well as the maximal allowed deviation between threshold-crossing signals and the template. Up to four templates can be generated for each electrode. The data was synchronized and collected by AlphaLab SnR (Alpha-Omega Engineering, Nof Hagalil, Israel).\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ePolysomnography analysis\u003c/span\u003e. A detailed description of our polysomnography methods can be found here\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e. In the current research, EMG was digitally offline bandpass-filtered in the 10 to 500 Hz range (stopband at 0 to 5 Hz, 520 to 1375 Hz). To minimize phase distortions, the forward-backward filtering was performed (this zero-phase filtering was used for any signal filtered offline). Sleep staging was performed using a semiautomatic staging algorithm that clustered 10-second nonoverlapping segments. Different vigilance stages (wakefulness, NREM, REM, and ambiguous/unclassified) were identified based on the eye-open fraction, the root mean square of the EMG signal, and the high/low EEG power ratio (the average power at 15 to 25 Hz / the average power at 0.1 to 7 Hz). The segment would be classified as NA (Not Available) if the missed length of any signal (Eye open state, EEG, EMG, Coil eye-position, LFP, SPK) was longer than 11*(1/SR)*1000 ms (out of 10,000 ms); SR is the sampling rate of the corresponding signal. Before semiautomatic clustering, 10% of the night epochs were scored manually by a trained expert (JG). Both left and right EMGs were used separately for the staging analysis. The better staging results provided by the semiautomatic algorithm, which matched the expert staging in more than 85% of the tested segments, were accepted for further analysis.\u003c/p\u003e\u003cp\u003eEye-coil signals were converted from voltage to degrees (angular position) using calibration data obtained on the same day. The transform relationship was generated using the fit geometric transformation\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e. The 'projective' transformation type was used since the original calibration data demonstrated that the scene appeared tilted. This transformation maintained the straight lines straightness and converged parallel lines toward a vanishing point. The converted eye-coil signals were used to calculate the velocity and acceleration of the eyes, as well as to identify saccades and drifts. Only saccades/drifts with a magnitude larger than one degree were kept for further analysis.\u003c/p\u003e\u003cp\u003eThe basic sleep staging was refined based on the eye open percentage, EMG, and the saccade frequency of the right eye. Typically, the EMG RMS is relatively larger in NREM sleep and smallest in REM sleep (Fig.\u0026nbsp;1b, d and Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003ef). Therefore, if a REM segment exhibits an EMG RMS value larger than the average EMG RMS of the NREM segments, or if an NREM segment shows an EMG RMS value smaller than the average EMG RMS of the REM segments, these segments will be considered unreasonable. This was done for both left and right EMG. The left/right EMG with relatively less unreasonable segments would be kept for further analysis. The NREM segments that occurred near REM epochs (two segments before and five after the previous REM segment) and additionally were characterized with eye-open ratio close to zero, more than 0.5/s eye saccades, and low EMG activity (the left/right EMG not larger than two times of the maximum EMG RMS during REM) were identified as REM candidates. For each candidate REM segment, if the duration of REM within the surrounding segments (three segments before and after the REM candidate) was longer than that of not-REM segments (NREM, wakefulness, or unclassified), this candidate REM would be classified as REM; otherwise, it would be identified as NREM. The awake or unclassified segments demonstrating an eye-open fraction between 0.4 and 0.7 were identified as wakeful candidates. These candidates would be classified as awake or unclassified segments using a method similar to the one used to determine the REM candidates. Finally, in the original classification of sleep stages, the eye open fraction must be larger than 0.6, 0.4\u0026ndash;0.6, and smaller than 0.4 in the awake, unclassified, and sleep (NREM and REM) states, respectively. The NREM segments with eye-open fractions between 0.05 and 0.3 would be refined as unclassified if the length of unclassified time within the surroundings of this NREM candidate was longer than the NREM sleep time. An example of a one-night hypnogram and PSG example is shown in Fig.\u0026nbsp;1b.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eSpike Analysis.\u003c/span\u003e Spiking (300\u0026ndash;9000 Hz) signals were filtered using an IIR comb notch filter with 881 (44,000/50\u0026thinsp;+\u0026thinsp;1) notches, which removed the 50 Hz power noise and its harmonics from the signal, given a sampling rate of 44,000 Hz. The Q factor for this filter was set to 35, namely, a highly sharp notch filter. The filtered spiking activities were rectified by absolute value, and the mean of the rectified vector was subtracted to obtain multi-unit activities (MUA). A low-pass Butterworth filter (210 Hz passband frequency, 260 Hz stopband frequency, 1 dB passband ripple, and 5 dB stopband attenuation) was designed to filter the MUA. The filtered MUA was down-sampled to 1/32 of the original sampling rate (yielding a 1,375 Hz sampling rate) by averaging the amplitude of the surrounding 32 sampling points. The filtered and down-sampled MUA was used to calculate its auto- and cross-correlation histograms after being segmented into 10-second epochs corresponding to the sleep staging epochs. The mean value of this 10-second MUA segment was subtracted to minimize the DC (zero frequency) power. The lag range of the auto- and cross-correlation histograms was from \u0026minus;\u0026thinsp;2 to 2 seconds. The auto- and cross-correlation histograms were calculated using the built-in function of MATLAB 2020b (xcorr, the correlation coefficient method), and smoothed by convolution with a Gaussian kernel having a standard deviation of 3.6 ms and 43.6 ms (5 and 60 times the time resolution, i.e., 1/1375 s), respectively. The xcorr correlation coefficient method ensures that the correlation coefficients (R) values could range from \u0026minus;\u0026thinsp;1 to 1.\u003c/p\u003e\u003cp\u003eThe action potentials (spikes) of well-isolated neurons (Isolation Score\u0026thinsp;\u0026gt;\u0026thinsp;0.7 or \u0026gt;\u0026thinsp;0.85)\u003csup\u003e46\u003c/sup\u003e were transformed into a continuous binary train of 0/1 values, representing single-unit activity (SUA). The SUA was down-sampled to 1/32 of the original sampling rate (i.e., from 44 kHz to 1,375 Hz) by summation of its trains. The spike train of every neuron was also separated into 10-second epochs based on the segments of sleep stages. The firing rate (FR, the number of spikes/s), the inter-spike intervals (ISI), and the coefficient of variation of the ISIs (CV-ISI, defined as STD(ISI)/Mean(ISI)) were calculated for every 10-second segment, and then were clustered into three groups (wakefulness, NREM, and REM) based on the awake-sleep staging. For each neuron, the FR, ISI, and CV-ISI were averaged within each group.\u003c/p\u003e\u003cp\u003eThe power spectrum density (PSD) of SUA was calculated using a 10-second moving window, with a 5-second moving step, a frequency range of 0.1 Hz to 100 Hz, and a frequency resolution of 0.1 Hz. For each 10-second binary (0/1) train, we subtracted the mean to minimize the DC (Frequency\u0026thinsp;=\u0026thinsp;0 Hz) power. The PSD unit is therefore given as NormSpk\u003csup\u003e2\u003c/sup\u003e/Hz.\u003c/p\u003e\u003cp\u003eThe auto- and cross-correlation (R-values) histograms of spike trains were calculated using the built-in function of MATLAB 2020b (xcorr, the correlation coefficient method). For this method, the mean value of each 10-second spike train was subtracted to show the negative and positive correlation values. The lag range and normalization were the same as for the MUA correlation analysis.\u003c/p\u003e\u003cp\u003eThe conditional discharge rate correlation histograms were calculated as the number of spikes (spike count) of the reference (triggered) cell, normalized by the number of spikes of the trigger neuron and the bin duration. The lag range was also from \u0026minus;\u0026thinsp;2 to 2 seconds. Edge effects were corrected by normalizing with the actual number of valid trigger spikes and bin duration. For example, at specific lag time points, a triggered spike train may be absent when aligned to a trigger spike, resulting in one fewer valid trigger spike contributing to that time point. The valid trigger count was adjusted by subtracting one from the total number of trigger spikes in such cases.\u003c/p\u003e\u003cp\u003eThe results from the conditional discharge method were used to calculate the association index (AI), which indicates the fraction of spikes contributed by the common-input mechanism out of total spikes. AI can be calculated as the area of the original cross-correlation histogram (counts/bin) peak divided by the total number of spikes from the triggered neurons. We used two ways to calculate the AI:\u003c/p\u003e\u003cp\u003eAI\u003csub\u003eavg\u003c/sub\u003e = ((Nci\u003csub\u003e12\u003c/sub\u003e/ Nt\u003csub\u003e2\u003c/sub\u003e) + (Nci\u003csub\u003e21\u003c/sub\u003e/ Nt\u003csub\u003e1\u003c/sub\u003e)) /2 (1)\u003c/p\u003e\u003cp\u003eAI\u003csub\u003esqrt\u003c/sub\u003e = (Nci\u003csub\u003e12\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;Nci\u003csub\u003e21\u003c/sub\u003e)/(2*sqrt(Nt\u003csub\u003e1\u003c/sub\u003e*Nt\u003csub\u003e2\u003c/sub\u003e)) (2)\u003c/p\u003e\u003cp\u003eAI\u003csub\u003esqrt\u003c/sub\u003e /AI\u003csub\u003eavg\u003c/sub\u003e is the fraction of the spikes of the two studied neurons (1 and 2) generated by the common (shared) input to the neurons. Nci\u003csub\u003e12\u003c/sub\u003e indicates the number of spikes in the central (common input) peak (neuron 1 is the trigger neuron, and neuron 2 is the triggered one). Nci\u003csub\u003e21\u003c/sub\u003e indicates the number of spikes in the central peak, but the trigger/triggered neurons are switched. Nt\u003csub\u003e1\u003c/sub\u003e or Nt\u003csub\u003e2\u003c/sub\u003e is the total number of spikes of the triggered neurons. The base for calculating the Nci is the average firing rate of the triggered neuron, and the Nci peak range is from \u0026minus;\u0026thinsp;0.5s to 0.5s. Neurons 1 and 2 were referred to as two neuron pairs with repetition or one neuron pair without repetition. In the \u0026lsquo;with repetition\u0026rsquo; condition, the pair was treated directionally, such that neuron 1 triggering neuron 2 and neuron 2 triggering neuron 1 were considered as two distinct pairs; in the \u0026lsquo;without repetition\u0026rsquo; condition, both directions were used for the same calculation, and thus counted as a single neuron pair.\u003c/p\u003e\u003cp\u003eBefore getting the R values at time 0 and calculating the AI, the cross-correlation histograms were smoothed by Gaussian convolution with 43.6 ms (60 times the time resolution (1/1375 s)) standard deviation. The same smoothing method was also applied to the autocorrelation histograms and PSD. The standard deviations of the smoothing kernel for the autocorrelation histograms and PSD smoothing are 0.36 ms (0.5*(1/1375)*1000) and 0.2 Hz (2 times the frequency resolution (0.1 Hz)), respectively. Such light smoothing was applied to auto-correlation to minimize the strong effect of maximal values (1 for normalized autocorrelation histograms) at time zero.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eBurst detection.\u003c/span\u003e Bursts were detected using the maximal interval method (MI, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.neuroexplorer.com/downloads/NeuroExplorerManual.pdf\u003c/span\u003e\u003cspan address=\"https://www.neuroexplorer.com/downloads/NeuroExplorerManual.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e)\u003csup\u003e47\u003c/sup\u003e and the Poisson surprise (PS) method\u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. In the MI method, we set the maximum length of the first inter-spike interval (ISI) of a candidate burst to be 10 ms. We added spikes until we reached the maximum ISI limit (12 ms). A burst should display a minimum silent period of 50 ms (with zero or no more than one spike) before the burst. Every burst should include at least three spikes.\u003c/p\u003e\u003cp\u003eIn the PS method, we identified the burst candidate as at least three consecutive spikes with ISIs shorter (each one of them) than 1/10 of the average ISI in a 10-second epoch (ISI limit). Spikes are added to the end of the candidate burst until the ISI of the added spike is larger than 1.5 times the ISI limit or the number of added spikes is 5. We then calculated the PS of the candidate burst as a basic reference. Spikes will be removed from the beginning of the candidate burst if this maximizes the PS value. The removal process will be repeated until five spikes are removed, or the PS value is maximized. Ultimately, a candidate burst would be considered as a valid burst only if its PS value was at least 10.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eBurst analysis.\u003c/span\u003e Burst features, i.e., burst frequency, inter-burst interval (IBI), the coefficient of variation of IBI (CV-IBI), burst duration, the number of spikes per burst, the intra-burst firing rate, and the burst ISI ordinal duration, were analysed. Burst frequency indicates the number of bursts per second. IBI is the duration between the beginning (first spike) of the current (n) burst and the beginning (first spike) of the next (n\u0026thinsp;+\u0026thinsp;1) burst. Burst durations were calculated as the duration between the burst's first and last spikes, and the typical duration of a single spike (96 sampling points) was added. The number of spikes per burst was divided by its corresponding burst duration to yield the intra-burst firing rate.\u003c/p\u003e\u003cp\u003eThe burst spikes were removed from the SUA to get the tonic activity. The FR, ISI, and CV-ISI were calculated for tonic activity. The same down-sampling features as used for the SUA were applied to tonic activity and to burst trains, where each burst was represented as a binary (0/1) event occurring at the time of the first spike in the burst. We also applied the same auto- and cross-correlation analyses described above for single-unit activity to tonic and burst activity. The cross-correlation histogram of burst trains in a particular vigilance state (e.g., NREM sleep) calculated by the conditional discharge method were removed from the analysis database if the trigger or triggered neuron had only one burst/10 s, both trigger and triggered neurons had bursts only near the same 2-second edge of the 10-second burst trains, and only one segment was available in this sleep stage. A total of 67 burst pairs without repetition were removed from the database of 4,416 burst pairs without repetition. The auto- and cross-correlation histograms of tonic activity were smoothed by convolution with a Gaussian kernel having a standard deviation of 0.36 ms (0.5*(1/1375)*1000) and 43.6 ms (60*(1/1375)*1000), respectively. For the burst trains, the standard deviations of the Gaussian kernel were 14.5 ms (20*(1/1375)*1000) for the auto-correlation histograms, and 43.6ms (60*(1/1375)*1000) for cross-correlation histograms. The auto-correlation histogram of burst trains was smoothed relatively stronger than that of SUA and tonic activity, because it had a longer refractory period. The same methods for calculating the AI and PSD of spike trains were also applied to tonic activity and burst trains. Therefore, the units of the PSDs for tonic activity and burst train are NormSpk\u003csup\u003e2\u003c/sup\u003e/Hz and NormBst\u003csup\u003e2\u003c/sup\u003e/Hz, respectively. The PSD of SUA, tonic activity, and burst trains was analyzed over the 0.1\u0026ndash;10 Hz frequency range. Additionally, cross-correlation (R-values) histograms of spike trains, burst trains, and tonic activity with 1-second time resolution were also calculated using the correlation coefficient method (xcorr, the built-in function of MATLAB 2020b, Fig. S8).\u003c/p\u003e\u003cp\u003eThe percentage of spikes within bursts among all spikes in the 10-second epoch is defined as the probability of burst spikes. The firing mode of every 10-second segment was identified based on the probability of spikes occurring in bursts (burst spikes). Therefore, three firing modes were defined: tonic (\u0026lt;\u0026thinsp;3%), mixed (3%-30%), and burst (\u0026ge;\u0026thinsp;30%) mode.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eAnalysis of the relations of bursts with behavioral and other physiological metrics.\u003c/span\u003e The burst frequency of every 10-second epoch, the probability of burst spikes, and the number of spikes per burst were aligned with the transition of sleep stages (e.g., transitioning from NREM to REM). To overcome the confounding effects of the inherently different percentage of vigilance stages, the corresponding predicted number of bursts was calculated using the average burst frequency of each neuron in wakefulness, NREM, or REM, and the percentage of the three sleep stages:\u003c/p\u003e\u003cp\u003eN\u003csub\u003ePreBst\u003c/sub\u003e = BstNum\u003csub\u003eW\u003c/sub\u003e * Per\u003csub\u003eW\u003c/sub\u003e + BstNum\u003csub\u003eN\u003c/sub\u003e * Per\u003csub\u003eN\u003c/sub\u003e + BstNum\u003csub\u003eR\u003c/sub\u003e * Per\u003csub\u003eR\u003c/sub\u003e (3)\u003c/p\u003e\u003cp\u003eN\u003csub\u003ePreBst\u003c/sub\u003e is the predicted number of bursts. BstNum\u003csub\u003eW\u003c/sub\u003e, BstNum\u003csub\u003eN\u003c/sub\u003e, and BstNum\u003csub\u003eR\u003c/sub\u003e indicate the average number of bursts, the probability of burst spikes or the number of spikes per burst in the wakefulness, NREM, and REM, respectively. Per\u003csub\u003eW\u003c/sub\u003e, Per\u003csub\u003eN\u003c/sub\u003e, and Per\u003csub\u003eR\u003c/sub\u003e indicate the percentage of the three vigilance states, respectively. The difference between the real and the corresponding predicted number of bursts was calculated.\u003c/p\u003e\u003cp\u003eThe two firing modes (burst and tonic) were combined with the three vigilance states (wakefulness, NREM, and REM). Therefore, there are six mode-stage combinations (burst-wakefulness, tonic-wakefulness, burst-NREM, tonic-NREM, burst-REM, and tonic-REM). We aligned the sleep stages to these six mode-stage combinations to explore the relationship between sleep states and firing mode. Only neurons having both tonic and burst firing mode segments were included for further analysis. Neurons having fewer than three aligned segments of wake-sleep stages were excluded. The difference between the percentage of the sleep stage related to the burst and tonic firing mode segments was calculated. Therefore, positive values indicated that the burst boosts staying in the same wake-sleep state or switching to another state.\u003c/p\u003e\u003cp\u003eThe eye-open/close state, rapid eye movements (saccades) frequency, slow eye movement (drifts) frequency, and EMG were aligned to bursts to reveal fast changes, which might be masked by the 10-second temporal resolution of our behavioral analysis. The eye open-close state is represented by the percentage of eye closure, i.e., the number of eye closures per second divided by the average frame rate of the video. A 4th -order bandpass Butterworth filter with a cut-off frequency of 8 Hz to 750 Hz was used to obtain the 10\u0026ndash;500 Hz EMG signal. This filtered EMG was down-sampled from 2,750 Hz to 1,375 Hz by averaging its amplitude. The absolute value of the EMG (rectification) was calculated before averaging the filtered, down-sampled, and burst-aligned EMG. The burst-aligned eye open-close state, saccade frequency, and eye-drift frequency were smoothed by convolution with a Gaussian kernel with 50 ms (50*(1/1000)*1000) standard deviation. A Gaussian kernel with a 36.4 ms (50*(1/1375)*1000) standard deviation was used for the EMG.\u003c/p\u003e\u003cp\u003eThe frontal EEG and thalamic LFP (VA and CM) were also aligned to the bursts to study the relationship between the thalamic bursts and these extra- and intra-cranial physiological signals. The 50Hz power artifact and its harmonics (100Hz, 150Hz, 200Hz, and 250Hz) of EEG and LFP were removed by a second-order IIR notch filter. A low-pass Butterworth filter (210 Hz passband frequency, 260 Hz stopband frequency, 1 dB passband ripple, and 5 dB stopband attenuation) was also designed to filter the EEG and LFP. The EEG signals were resampled as the EMG signal. For EEG and LFP, the mean value was subtracted from the filtered and down-sampled signals before alignment to remove the DC component of the signal. We used a convolution with a Gaussian kernel having a standard deviation of 36.4 ms (50*(1/1375)*1000) to smooth the EEG and LFP signals. Finally, they were normalized by z-score based on their baseline (from two to one second before the burst).\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eStatistical analysis\u003c/span\u003e: Statistical analysis was performed using MATLAB R2020b. All population data are presented as the mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SEM (standard error of the mean). The p-value was calculated using the Wilcoxon rank sum test for the independent data and the Wilcoxon signed-rank test for the two matched samples. The statistical tests were two-tailed. A P-value threshold of 0.05 was used, and the results were corrected for multiple comparisons using the Bonferroni method.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eCompeting Interests\u003c/h2\u003e\u003cp\u003eAll authors declare no competing interests.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e\u003cp\u003eThis study is supported by grants from the ISF Breakthrough Research program (Grant No.: 1738/22) and the Collaborative Research Center TRR295, Germany (Project number 424778381) to HB.\u003c/p\u003e\u003ch2\u003eAuthor contributions\u003c/h2\u003e\u003cp\u003eJG and HB conceived the research and designed the experiments. ZI, DW, and AR performed the surgical procedure. XL and JG supported the surgical procedure. They also performed the experiments, including electrophysiological and behavioral recordings, analyzed the data, and conducted the statistical analysis. XL, JG, and HB prepared the figures and wrote the manuscript. HB supervised the work. All authors read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgments\u003c/h2\u003e\u003cp\u003eThe authors would like to thank Uri Werner-Reiss, PhD, for his valuable support of the surgical procedures and all aspects of monkey care, Tamar Ravins Yaish, DMD, and the HUJI-ELSC animal facility team for their assistance. We thank Ad Aertsen for the fruitful discussion of correlation analysis and the association index, and Andy Horn, Jackie Schiller, Pnina Rapel, Aric Agmon, and Yuval Nir for their discussions and comments on early versions of the manuscript. We acknowledge the use of large language model (LLM) tools for linguistic editing to improve the clarity and grammar of this manuscript; no scientific content was generated by these tools.\u003c/p\u003e\u003ch2\u003eData availability\u003c/h2\u003e\u003cp\u003eData will be available upon request from the corresponding authors.\u003c/p\u003e\u003ch2\u003eCode availability\u003c/h2\u003e\u003cp\u003eMatlab code will be available upon request from the corresponding authors. Please note that the code used in this study was developed by the researchers for data analysis and visualization. It is intended for research purposes and may not meet professional coding standards.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eUsrey, W.M., Sherman, S.M.: The Cerebral Cortex and Thalamus. Oxford University Press (2024)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWeyand, T.G., Boudreaux, M., Guido, W.: Burst and tonic response modes in thalamic neurons during sleep and wakefulness. J. Neurophysiol. \u003cb\u003e85\u003c/b\u003e, 1107\u0026ndash;1118 (2001)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMcCarley, R.W., Benoit, O., Barrionuevo, G.: Lateral geniculate nucleus unitary discharge in sleep and waking: state- and rate-specific aspects. J. Neurophysiol. \u003cb\u003e50\u003c/b\u003e, 798\u0026ndash;818 (1983)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGlenn, L., Steriade, M.: Discharge rate and excitability of cortically projecting intralaminar thalamic neurons during waking and sleep states. J. Neurosci. \u003cb\u003e2\u003c/b\u003e, 1387\u0026ndash;1404 (1982)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSherman, S.M.: A wake-up call from the thalamus. Nat. Neurosci. \u003cb\u003e4\u003c/b\u003e, 344\u0026ndash;346 (2001)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRamcharan, E.J., Gnadt, J.W., Sherman, S.M.: Burst and tonic firing in thalamic cells of unanesthetized, behaving monkeys. Vis. Neurosci. \u003cb\u003e17\u003c/b\u003e, 55\u0026ndash;62 (2000)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLisman, J.E.: Bursts as a unit of neural information: making unreliable synapses reliable. Trends Neurosci. \u003cb\u003e20\u003c/b\u003e, 38\u0026ndash;43 (1997)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuido, W., Lu, S.-M., Vaughan, J., Godwin, D.W., Sherman, S.M.: Receiver operating characteristic (ROC) analysis of neurons in the cat's lateral geniculate nucleus during tonic and burst response mode. Vis. Neurosci. \u003cb\u003e12\u003c/b\u003e, 723\u0026ndash;741 (1995)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eL\u0026uuml;thi, A., McCormick, D.A.: Periodicity of thalamic synchronized oscillations: the role of Ca2+-mediated upregulation of Ih. Neuron. \u003cb\u003e20\u003c/b\u003e, 553\u0026ndash;563 (1998)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJahnsen, H., Llin\u0026aacute;s, R.: Ionic basis for the electro-responsiveness and oscillatory properties of guinea‐pig thalamic neurones in vitro. J. Physiol. \u003cb\u003e349\u003c/b\u003e, 227\u0026ndash;247 (1984)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlbin, R.L., Young, A.B., Penney, J.B.: The functional anatomy of basal ganglia disorders. Trends Neurosci. \u003cb\u003e12\u003c/b\u003e, 366\u0026ndash;375 (1989)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBergman, H., Wichmann, T., DeLong, M.R.: Reversal of experimental parkinsonism by lesions of the subthalamic nucleus. Science. \u003cb\u003e249\u003c/b\u003e, 1436\u0026ndash;1438 (1990)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMizrahi-Kliger, A.D., Kaplan, A., Israel, Z., Bergman, H.: Desynchronization of slow oscillations in the basal ganglia during natural sleep. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e 115, E4274-E4283 (2018)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRedinbaugh, M.J., et al.: Thalamus modulates consciousness via layer-specific control of cortex. Neuron. \u003cb\u003e106\u003c/b\u003e, 66\u0026ndash;75 (2020). e12\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBar-Gad, I., Ritov, Y., Vaadia, E., Bergman, H.: Failure in identification of overlapping spikes from multiple neuron activity causes artificial correlations. J. Neurosci. Methods. \u003cb\u003e107\u003c/b\u003e, 1\u0026ndash;13 (2001)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePerkel, D.H., Gerstein, G.L., Moore, G.P.: Neuronal spike trains and stochastic point processes: II. Simultaneous spike trains. Biophys. J. \u003cb\u003e7\u003c/b\u003e, 419\u0026ndash;440 (1967)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZirh, T., Lenz, F., Reich, S., Dougherty, P.: Patterns of bursting occurring in thalamic cells during parkinsonian tremor. Neuroscience. \u003cb\u003e83\u003c/b\u003e, 107\u0026ndash;121 (1998)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRivlin-Etzion, M., Ritov, Y., Heimer, G., Bergman, H.: Bar-Gad, I. Local shuffling of spike trains boosts the accuracy of spike train spectral analysis. J. Neurophysiol. \u003cb\u003e95\u003c/b\u003e, 3245\u0026ndash;3256 (2006)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCox, K.M., Kase, D., Znati, T., Turner, R.S.: Detecting rhythmic spiking through the power spectra of point process model residuals. J. Neural Eng. \u003cb\u003e21\u003c/b\u003e, 046041 (2024)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHeimer, G., Bar-Gad, I., Goldberg, J.A., Bergman, H.: Synchronization Of pallidal activity in the MPTP primate model of parkinsonism is not limited to oscillatory activity. Basal Ganglia VII, 29\u0026ndash;34 (2002)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWong, R.O., Meister, M., Shatz, C.J.: Transient period of correlated bursting activity during development of the mammalian retina. Neuron. \u003cb\u003e11\u003c/b\u003e, 923\u0026ndash;938 (1993)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNir, Y., de Lecea, L.: Sleep and vigilance states: Embracing spatiotemporal dynamics. Neuron. \u003cb\u003e111\u003c/b\u003e, 1998\u0026ndash;2011 (2023)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003edos Lima, S.: Hippocampal and cortical communication around micro-arousals in slow-wave sleep. Sci. Rep. \u003cb\u003e9\u003c/b\u003e, 5876 (2019)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLogothetis, N.K., Kayser, C., Oeltermann, A.: In vivo measurement of cortical impedance spectrum in monkeys: implications for signal propagation. Neuron. \u003cb\u003e55\u003c/b\u003e, 809\u0026ndash;823 (2007)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMarmor, O., et al.: Local vs. volume conductance activity of field potentials in the human subthalamic nucleus. J. Neurophysiol. \u003cb\u003e117\u003c/b\u003e, 2140\u0026ndash;2151 (2017)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChurchland, M.M., Shenoy, K.V.: Preparatory activity and the expansive null-space. Nat. Rev. Neurosci. \u003cb\u003e25\u003c/b\u003e, 213\u0026ndash;236 (2024)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVishwanath, A., Bartlett, M.J., Falk, T., Cowen, S.L.: Decoupling of motor cortex to movement in Parkinson\u0026rsquo;s dyskinesia rescued by sub-anaesthetic ketamine. Brain. \u003cb\u003e148\u003c/b\u003e, 2135\u0026ndash;2150 (2025)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBen-Ari, Y.: The GABA excitatory/inhibitory developmental sequence: a personal journey. Neuroscience. \u003cb\u003e279\u003c/b\u003e, 187\u0026ndash;219 (2014)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMink, J.W.: The basal ganglia: focused selection and inhibition of competing motor programs. Prog. Neurobiol. \u003cb\u003e50\u003c/b\u003e, 381\u0026ndash;425 (1996)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBosch-Bouju, C., Hyland, B.I., Parr-Brownlie, L.C.: Motor thalamus integration of cortical, cerebellar and basal ganglia information: implications for normal and parkinsonian conditions. Front. Comput. Neurosci. \u003cb\u003e7\u003c/b\u003e, 163 (2013)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSchwab, B.C., et al.: Neural activity during a simple reaching task in macaques is counter to gating and rebound in basal ganglia\u0026ndash;thalamic communication. PLoS Biol. \u003cb\u003e18\u003c/b\u003e, e3000829 (2020)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGoldberg, J.H., Farries, M.A., Fee, M.S.: Basal ganglia output to the thalamus: still a paradox. Trends Neurosci. \u003cb\u003e36\u003c/b\u003e, 695\u0026ndash;705 (2013)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAdler, A., et al.: Neurons in both pallidal segments change their firing properties similarly prior to closure of the eyes. J. Neurophysiol. \u003cb\u003e103\u003c/b\u003e, 346\u0026ndash;359 (2010)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTimofeev, I., Grenier, F., Steriade, M.: Impact of intrinsic properties and synaptic factors on the activity of neocortical networks in vivo. J. Physiology-Paris. \u003cb\u003e94\u003c/b\u003e, 343\u0026ndash;355 (2000)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCrunelli, V., et al.: Dual function of thalamic low-vigilance state oscillations: rhythm-regulation and plasticity. Nat. Rev. Neurosci. \u003cb\u003e19\u003c/b\u003e, 107\u0026ndash;118 (2018)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSteriade, M., Contreras, D., Amzica, F.: Synchronized sleep oscillations and their paroxysmal developments. Trends Neurosci. \u003cb\u003e17\u003c/b\u003e, 201\u0026ndash;207 (1994)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBrain, A.: Thalamocortical Oscillations in the Sleeping. SCIENCE. \u003cb\u003e262\u003c/b\u003e, 679\u0026ndash;685 (1993)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNir, Y., Tononi, G.: Dreaming and the brain: from phenomenology to neurophysiology. Trends Cogn. Sci. \u003cb\u003e14\u003c/b\u003e, 88\u0026ndash;100 (2010)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSteriade, M., McCarley, R.W.: Neuronal activities in brainstem and basal forebrain structures controlling waking and sleep states. \u003cem\u003eIn: Brain Control of Wakefulness Sleep. Springer, Boston, MA.\u003c/em\u003e, 381\u0026ndash;416 (2005)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eScala, F., et al.: Phenotypic variation of transcriptomic cell types in mouse motor cortex. Nature. \u003cb\u003e598\u003c/b\u003e, 144\u0026ndash;150 (2021)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCouncil, N.R., et al.: Guide for the Care and Use of Laboratory Animals: Eighth Edition. National Academies (2011)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRobinson, D.A.: A method of measuring eye movemnent using a scieral search coil in a magnetic field. IEEE Trans. Bio-med. Electron. \u003cb\u003e10\u003c/b\u003e, 137\u0026ndash;145 (1963)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGlowinsky, S., Israel, Z., Heymann, S., Bergman, H.J.I.T.o.N, S., Engineering, R.: Divide and Conquer: Automatic Detection of the Thalamus to Empower DBS Physiological Navigation to the Subthalamic Region. IEEE Trans. Neural Syst. Rehabilitation Eng. (2025)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuang, J., et al.: Toward asleep DBS: cortico-basal ganglia spectral and coherence activity during interleaved propofol/ketamine sedation mimics NREM/REM sleep activity. npj Parkinson's Disease. \u003cb\u003e7\u003c/b\u003e, 67 (2021)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGoshtasby, A.: Image registration by local approximation methods. Image Vis. Comput. \u003cb\u003e6\u003c/b\u003e, 255\u0026ndash;261 (1988)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJoshua, M., Elias, S., Levine, O., Bergman, H.: Quantifying the isolation quality of extracellularly recorded action potentials. J. Neurosci. Methods. \u003cb\u003e163\u003c/b\u003e, 267\u0026ndash;282 (2007)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCotterill, E., Eglen, S.: Burst detection methods. In: Chiappalone, M., Pasquale, V., Frega, M. (eds). \u003cem\u003eIn Vitro Neuronal Networks. Advances in Neurobiology, vol 22. Springer, Cham\u003c/em\u003e, 185\u0026ndash;206 (2019)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLegendy, C., Salcman, M.: Bursts and recurrences of bursts in the spike trains of spontaneously active striate cortex neurons. J. Neurophysiol. \u003cb\u003e53\u003c/b\u003e, 926\u0026ndash;939 (1985)\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7607905/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7607905/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eThalamic neurons discharge tonically during wakefulness and rapid-eye-movement (REM) sleep, but burst during non-REM (NREM) sleep. It has been hypothesized that NREM thalamic bursts do not serve as a cortical \"wake-up\" signal due to their periodic and synchronized nature. To explore this hypothesis, we studied polysomnographic signals, field potentials, and spiking activity of multiple neurons in the ventral anterior and centromedian thalamic nuclei of two female non-human primates during naturally occurring vigilance states. These nuclei receive GABAergic output from the basal ganglia, with discharge rate decreasing during NREM sleep. Nevertheless, NREM bursting increased significantly as reported for glutamate-driven thalamic nuclei. NREM bursts were neither periodic nor tightly synchronized. However, EEG activity and thalamic field potentials time-locked to burst onset during NREM sleep markedly differed from those observed during wakefulness and REM sleep. These results suggest that NREM thalamic bursts do not awaken the cortex, due to unique state-dependent thalamocortical dynamics.\u003c/strong\u003e\u003c/p\u003e","manuscriptTitle":"Why bursts in thalamic nuclei receiving basal ganglia output fail to wake the cortex during sleep?","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-10 10:19:41","doi":"10.21203/rs.3.rs-7607905/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"communications-biology","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"commsbio","sideBox":"Learn more about [Communications Biology](http://www.nature.com/commsbio/)","snPcode":"","submissionUrl":"","title":"Communications Biology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Communications Series","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"be5b3f97-034e-468a-b7a2-8159022a8098","owner":[],"postedDate":"November 10th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":55236250,"name":"Biological sciences/Neuroscience/Neural circuits"},{"id":55236251,"name":"Biological sciences/Neuroscience/Neuronal physiology"}],"tags":[],"updatedAt":"2026-02-20T08:07:06+00:00","versionOfRecord":{"articleIdentity":"rs-7607905","link":"https://doi.org/10.1038/s42003-026-09565-3","journal":{"identity":"communications-biology","isVorOnly":false,"title":"Communications Biology"},"publishedOn":"2026-01-23 05:00:00","publishedOnDateReadable":"January 23rd, 2026"},"versionCreatedAt":"2025-11-10 10:19:41","video":"","vorDoi":"10.1038/s42003-026-09565-3","vorDoiUrl":"https://doi.org/10.1038/s42003-026-09565-3","workflowStages":[]},"version":"v1","identity":"rs-7607905","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7607905","identity":"rs-7607905","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}