A thalamic perspective of (un)consciousness in pharmacological and pathological states in humans.

A thalamic perspective of (un)consciousness in pharmacological and pathological states in humans - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Brain Commun . 2026 Feb 28;8(2):fcag021. doi: 10.1093/braincomms/fcag021 Search in PMC Search in PubMed View in NLM Catalog Add to search A thalamic perspective of (un)consciousness in pharmacological and pathological states in humans Dorottya Szocs Dorottya Szocs 1 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 2 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK Find articles by Dorottya Szocs 1, 2, ✉ , Dian Lyu Dian Lyu 3 Departments of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA 94304, USA Find articles by Dian Lyu 3 , Andrea I Luppi Andrea I Luppi 4 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 5 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 6 Montreal Neurological Institute, McGill University, Montreal, QC H3A 2B4, Canada Find articles by Andrea I Luppi 4, 5, 6 , Peter Coppola Peter Coppola 7 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 8 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK Find articles by Peter Coppola 7, 8 , Rebecca E Woodrow Rebecca E Woodrow 9 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 10 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK Find articles by Rebecca E Woodrow 9, 10 , Guy B Williams Guy B Williams 11 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 12 Wolfson Brain Imaging Centre, University of Cambridge, Cambridge CB2 0QQ, UK Find articles by Guy B Williams 11, 12 , Judith Allanson Judith Allanson 13 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK Find articles by Judith Allanson 13 , John D Pickard John D Pickard 14 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK Find articles by John D Pickard 14 , Adrian M Owen Adrian M Owen 15 The Brain and Mind Institute, Department of Psychology, The University of Western Ontario, London, ON N6A 5B7, Canada Find articles by Adrian M Owen 15 , Lorina Naci Lorina Naci 16 Institute for Neuroscience, Trinity College Dublin, Dublin D02, Ireland Find articles by Lorina Naci 16 , David K Menon David K Menon 17 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 18 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK Find articles by David K Menon 17, 18 , Emmanuel A Stamatakis Emmanuel A Stamatakis 19 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 20 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK Find articles by Emmanuel A Stamatakis 19, 20 Author information Article notes Copyright and License information 1 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 2 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 3 Departments of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA 94304, USA 4 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 5 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 6 Montreal Neurological Institute, McGill University, Montreal, QC H3A 2B4, Canada 7 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 8 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 9 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 10 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 11 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 12 Wolfson Brain Imaging Centre, University of Cambridge, Cambridge CB2 0QQ, UK 13 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 14 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 15 The Brain and Mind Institute, Department of Psychology, The University of Western Ontario, London, ON N6A 5B7, Canada 16 Institute for Neuroscience, Trinity College Dublin, Dublin D02, Ireland 17 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 18 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 19 Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK 20 Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK ✉ Correspondence to: Dorottya Szocs, Department of Clinical Neurosciences, University of Cambridge, Clare Hall, Herschel Rd, Cambridge CB3 9AL, UK E-mail: [email protected] Received 2024 Jul 1; Revised 2025 Nov 30; Accepted 2026 Feb 27; Collection date 2026. © The Author(s) 2026. Published by Oxford University Press on behalf of the Guarantors of Brain. This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( https://creativecommons.org/licenses/by/4.0/ ), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. PMC Copyright notice PMCID: PMC13084198  PMID: 42004009 Abstract Currently, there is substantial ongoing discussion around the functional role of the thalamus in consciousness. What is missing in the literature, however, is a systematic investigation of the relevance of specific thalamic nuclei in pharmacologically and pathologically altered states of consciousness in humans. Using resting-state functional magnetic resonance imaging in both healthy anaesthetized volunteers ( n = 16) and patients with disorders of consciousness ( n = 22), we sought to identify which specific thalamic subregions in both cohorts may be differentially significant for loss of consciousness. Our findings revealed that, among all nuclei, the pulvinar was found to have the strongest functional connectivity change with loss of consciousness following anaesthesia, while demonstrating distinct functional connectivity patterns related to higher-order default mode and executive control networks. Remarkably, in loss of consciousness in disorders of consciousness patients, the ventral-latero-ventral was found to have the strongest connectivity change in comparison with healthy controls, while exhibiting discrete functional connectivity patterns related to higher-order default mode, executive control and frontoparietal networks. Furthermore, we provide evidence that this neural connectivity biomarker in patients also mirrored the changes observed at the behavioural level, which could have clinical implications for targeted deep brain stimulation in therapy for disorders of consciousness. Keywords: disorders of consciousness, thalamus, unresponsive wakefulness syndrome, minimally conscious state, fMRI There is ongoing debate about thalamic contributions to consciousness. Using anaesthesia to experimentally model disorders of consciousness, Szocs et al ., 2025 demonstrate differential roles for thalamic nuclei and functional changes mirrored at the behavioural level, with therapeutic implications for targeted deep brain stimulation. Graphical Abstract Graphical Abstract. Open in a new tab Introduction The pursuit of identifying ‘neural correlates of consciousness’ has both scientific and clinical relevance, given the pressing challenge of detecting covert consciousness in uncommunicative individuals with disorders of consciousness (DOC). 1 We currently still lack a definitive understanding of which neural circuits—common or differential—are responsible for altered states of consciousness and how these work. 2 There is substantial evidence that functional neuroimaging can not only detect covert awareness in patients who appear entirely unresponsive at the bed side but can predict whether a patient may recover. 3 Consciousness has been investigated extensively from the perspective of large-scale cortical networks; notably, these major recurrent networks in both health and disease include the default mode (DMN), salience (SN), sensorimotor (SMN), frontoparietal (FPN), dorsal attention (DAN), executive control (ECN), auditory (AN) and visual (VN) networks. 4–11 Among these networks, the medial prefrontal cortex (mPFC) and PCC (posterior cingulate cortex)/precuneus nodes of the DMN have been found to be significant for predicting conscious awakening. 4 , 12 , 13 Furthermore, salience network (SN) connectivity was able to distinguish between consciouses and unconscious states, whereas the DMN connectivity strength, particularly between the PCC and left lateral parietal cortex (LLPC), instead predicted recovery of consciousness in DOC patients with unresponsive wakefulness syndrome (UWS). 14 Moreover, regions in the auditory (AN) network were more functionally connected in minimally conscious state (MCS) patients compared to UWS. 15 There may not be one but, rather, several pathways to unconsciousness, with perhaps some more prominent than others. The thalamus is well-positioned to mediate whole-brain system-level interactions between the cortex, basal ganglia and cerebellum, through inputs of unique thalamic populations—parvalbumin-staining core and calbindin-staining matrix cells—to sustain conscious states, 16 , 17 of which the nuclei of the ventral thalamus (specifically the ventral lateral, ventral anterior and mediodorsal nuclei) play a particularly important role in shaping the dynamics of frontal cortical areas. 16 Moreover, a breakdown of anterior–posterior connectivity has been attributed to the unconscious state induced by anaesthetics, specifically, a depression of the lateral frontoparietal network (FPN). 18 Precisely how this phenomenon may occur is still unknown, though the authors address possible corticocortical and thalamocortical involvement. On this note, unconsciousness induced by dexmedetomidine has been linked to reduced regional cerebral bloodflow (rCBF) in the default mode network (DMN), fronto-parietal network (FPN) and thalamus, 19 whereas propofol is known to act upon GABAergic circuits in the thalamus, brainstem and cortex, resulting in the phenomenon of anteriorization or emergence of frontal alpha oscillations. 20 , 21 In disorders of consciousness, there have been several studies which attempt to outline thalamic mechanisms in DOC cohorts following brain injury, which focus on the whole thalamus 22 or the ‘central thalamus’, notably central-lateral (CL) nucleus. 23 , 24 At a more subcortical level, coma-causing injuries to the pontine tegmentum in the brainstem have been associated to a functional disconnection of the ventral anterior insula (AI) and pregenual anterior cingulate cortex (pACC), 25 alluding to a specialized brain network specific to pathological loss of consciousness. Finally, other lines of evidence have suggested multiple pathways to unconsciousness, which highlight the importance of cerebellar and subcortical areas, 26 ventral tegmentum area 27 and posteromedial areas. 28 Emerging evidence points to thalamic involvement in modulating arousal states. 20 , 23 This complex subcortical structure has received particular attention as it is situated, both anatomically and functionally, at the convergence of the hypothalamus, telencephalon and brainstem, and most accounts of its functional role propose that it orchestrates dynamic interactions among the cortex, basal ganglia and cerebellum, which together underlie conscious experience. 20 , 29 Even more so, at a macroscopic level, it plays a crucial role in systems-level dynamics by integrating multimodal information across large-scale cortical networks. 30 On a microscopic level, a common classification based on histology divides the thalamus into distinct nuclei: anterior division [i.e. anteriomedial (AM), anteriolateral (AL), anterodorsal (AD), laterodorsal (LD) nuclei]; medial group with midline group [i.e. paratenial (PT), paraventricular (PV), centromedian (CM) nuclei], intralaminar group [i.e. centrolateral (CL), centromedial (CM), parafascicular (PFN) nuclei) and mediodorsal (MD) nuclei]; lateral division [i.e. ventral anterior (VA), ventrolateral (VL), ventral posteriomedial (VPM), ventral posterolateral (VPL) and ventromedial (VM) nuclei]; posterior group [i.e. posteromedial (PM), lateroposterior (LP), pulvinar (PU), medial (MGN) and lateral geniculate (LGN) nuclei]; and a coating of inhibitory neurons (thalamic reticular nucleus). 29 In an alternative classification scheme, nuclei have been grouped according to the type of input (sensory, limbic or motor), 31 though these divisions only broadly inform function. 32 Furthermore, a more nuanced classification model based on thalamocortical output recognizes that individual thalamic nuclei are comprised of an amalgam of cell types, 17 , 29 , 33 most notably the parvalbumin-staining core cells, which serve to drive excitatory activity via projections to cortical layers III and IV (receiving glutamatergic inputs from sensory nuclei, association cortical areas and the deep cerebellar nuclei) versus calbindin-staining matrix cells, which preferentially project to supra-granular and infragranular cortices in a much more diffuse manner (receive GABAergic inputs from the globus pallidus. 17 , 20 , 29 , 33 , 34 Further classification schemes are based on the origin and strength of inputs, i.e. distinguishing first-order from higher-order nuclei. 34 Importantly, specific thalamic nuclei mediate higher-order cognitive functions, 35 such as attention 36 , 37 and awareness. 38 , 39 Emerging animal evidence has demonstrated the involvement of distinct thalamic nuclei in pharmacologically induced unconsciousness, specifically demonstrating that central thalamic stimulation induced arousal in anaesthetized macaques. 38 , 40 Equally, human studies have highlighted thalamic interactions with the DMN in anaesthesia and DOC. 28 , 41 However, only few human studies have differentiated the cytoarchitecturally and functionally distinct thalamic nuclei, to elucidate their respective relevance in altered states of consciousness. 42–44 What is missing, therefore, is a thorough investigation of the differential role of individual thalamic nuclei with relevance to pathological impairment in consciousness following brain injury. This is particularly critical in light of the current debate around the functional role of the thalamus in conscious processing. 20 , 45 Furthermore, such understanding is crucial from a clinical standpoint, since deep brain stimulation (DBS) of the CL nuclei has been found to improve behavioural outcomes in patients with msTBI 46 and DOC. 24 Thalamic stimulation has been implemented in other brain disorders, notably obsessive-compulsive disorder, 47 Parkinson’s disease, essential tremor and dystonia. 48 More recently, a stimulation paradigm of both the anterior and medial pulvinar thalamic nuclei has been efficacious for treating drug-resistant epilepsy. 49 , 50 Even more so, with advances in MRI spatial resolution and improvements in thalamic parcellation atlases, 51–57 we are witnessing innovative attempts to functionally disentangle the thalamic ensemble. This has led to more investigations of the role that individual thalamic nuclei play in the context of arousal, most notably in sleep 43 and propofol anaesthesia, 42 though the complexity of the thalamus has, so far, been mostly overlooked in the literature in patient cohorts, particularly in disorders of consciousness. This study provides a systematic investigation of distinct thalamic nuclei, i.e. seven major clusters—pulvinar (Pu), anterior (Ant), medio-dorsal (MD), ventral-latero-dorsal (VLD), central-lateral, lateral-posterior, medial-pulvinar (CL-LP-MPu), ventral-anterior (VA) and ventral-latero-ventral (VLV)—and their functional relationship to cortical and subcortical areas, using functional neuroimaging datasets from cohorts of (i) healthy anaesthetized volunteers ( n = 16), i.e. pharmacologically induced and (ii) patients with disorders of consciousness ( n = 22), i.e. pathological, given that anaesthesia has been previously demonstrated to be an experimental model for DOC studies. 26 , 27 , 58 Crucially, we sought to pinpoint which specific nuclei within the thalamus are most strongly associated with loss of consciousness, in both cohorts, which in turn may shed light upon the neural mechanisms of conscious processing, and more crucially, indicate potential areas for targeted brain stimulation therapy in DOC. Materials and methods Data acquisition Anaesthesia London Ontario dataset The original study 59 acquired the anaesthesia dataset between May and November 2014 at the Robarts Research Institute in London, Ontario Canada, have been approved by the Western University Ethics board, and have been previously published. 26 , 28 , 60 , 61 Nineteen healthy, right-handed, English-speaking volunteers (13 males; age range: 18–40 years) with no reported neurological conditions provided written informed consent. Three participants were excluded due to equipment failure or anaesthetic complications, resulting in a final sample of 16 volunteers. 59 Resting-state fMRI data were obtained across three propofol-induced conditions: awake (no sedation), deep sedation (Ramsay = 5) and recovery following anaesthesia. Prior to data acquisition in each condition, two anaesthesiologists and one anaesthesia nurse independently verified the Ramsay score (the assessor could not be blinded to the condition as they were responsible for determining anaesthetic depth). 28 , 59 Intravenous propofol was delivered via a computer-controlled Baxter AS50 infusion system (Singapore). Dosage was increased in a stepwise manner under computer control until participants reached a Ramsay sedation level of 5, corresponding to an absence of responsiveness to verbal or visual stimulation. When necessary, additional manual adjustments were performed to achieve and maintain the desired propofol concentration, which was stabilized using the pharmacokinetic simulation software TIVA Trainer (eurosiva.eu). Predicted blood concentrations followed the Marsh three-compartment model. The initial target concentration was 0.6 µg/ml, with incremental increases of 0.3 µg/ml between assessments of the Ramsay level. This procedure continued until participants ceased to respond verbally and could be aroused only by physical stimulation, marking the onset of data acquisition. Oxygen titration maintained SpO2 above 96%. The mean estimated effect-site propofol concentration was 2.48 (range: 1.82–3.14) µg ml −1 ; the mean estimated plasma propofol concentration was 2.68 (range: 1.92–3.44) µg ml −1 and the mean total mass of propofol administered was 486.58 (range: 373.30–599.86) mg. 28 , 59 Resting-state fMRI data were collected over an 8-minute acquisition period using a 3-tesla (3T) Siemens Trio scanner. Functional data comprised 256 Echo-planar imaging (EPI) volumes with the following parameters: 33 slices; isotropic resolution = 3 mm; 25% inter-slice gap; TR = 2000ms; TE = 30 ms; flip-angle = 75°; matrix = 64 × 64. Slices were acquired in an interleaved bottom-up order. High-resolution anatomical T1-weighted images were obtained using a 3D MPRAGE sequence (32-channel coil, 1 mm isotropic voxels) with the following parameters: TA = 5 min; TE = 4.25 ms; flip angle = 9°, matrix = 240 × 256. 28 , 59 Disorders of consciousness dataset MRI data from 24 patients with DOC were obtained between January 2010 and July 2015 at the Wolfson Brain Imaging Centre Addenbrookes (Cambridge, UK). For this study, the patients were selected from a larger cohort ( n = 71) on the basis of relatively preserved neuroanatomy integrity. 26 , 27 , 58 , 60 Patients were admitted to research ward and scanned at Wolfson Brain Imaging Centre. Written informed assent was obtained from the referring clinical teams and a family member or other relevant close contact. Each patient was referred to a full neurological examination and daily behavioural observations. Coma Recovery Scale–Revised (CRS-R) evaluations were conducted at least once on the day of scanning, with additional assessments performed periodically throughout the patient’s hospital stay. For each scanning session, the CRS-R score corresponded to the highest value recorded by the attending physician. Based on these assessments, patients were classified as being in either an UWS or MCS. Individuals lacking any behavioural evidence of awareness throughout the assessment period were diagnosed with UWS. In contrast, those demonstrating clear behavioural markers of consciousness, such as simple automatic movements (scratching, repositioning bed sheets), sustained visual fixation or pursuit or localization to noxious stimulation, were classified as being in a MCS. 26 , 27 , 58 MCS patients were further classified depending on the presence or absence of language function, into MCS+ or MCS−, respectively. 62 Detailed demographics are found in Table 1 . Ethical approval for data collection was obtained from the National Research Ethics Service. 26 , 27 , 58 Table 1. Demographics for DOC patients Patient Gender Age Aetiology Months Post-Injury Diagnosis CRS Tennis Task 1 M 21 TBI 45 MCS+ 11 positive 2 M 57 TBI 14 MCS− 12 negative 3 M 47 TBI 4 MCS+ 10 negative 4 M 36 TBI 34 UWS 8 negative 5 M 17 Anoxic 46 UWS 11 negative 6 F 38 Anoxic 9 MCS− 10 negative 7 F 38 TBI 13 MCS+ 11 positive 8 M 29 TBI 68 MCS+ 10 positive 9 M 23 TBI 4 MCS+ 7 positive 10 F 70 Cerebral bleed 11 MCS+ 9 negative 11 F 30 Anoxic 6 MCS− 9 positive 12 F 34 Anoxic 6 UWS 8 negative 13 M 22 Anoxic 5 UWS 7 negative 14 M 37 Anoxic 14 UWS 7 negative 15 F 62 Anoxic 7 UWS 7 negative 16 M 46 Anoxic 10 UWS 5 negative 17 M 21 TBI 7 MCS+ 11 negative 18 M 67 TBI 14 MCS− 11 positive 19 F 55 Hypoxic Unknown UWS 12 negative 20 M 28 TBI Unknown MCS+ 8 positive 21 M 22 TBI Unknown MCS+ 10 negative 22 F 28 ADEM Unknown UWS 6 negative Open in a new tab Patients were further subdivided according to their capacity to engage in volitional mental imagery during fMRI scanning, based on a paradigm previously established for detecting covert awareness in patients with disorders of consciousness. 58 , 63–65 Two imagery tasks were employed: a motor imagery (‘tennis task’) condition, where patients were instructed to imagine themselves playing tennis, repeatedly swinging their arm to return a ball to a perceived opponent, and a spatial imagery (‘navigation task’) condition, where patients were asked to visualize moving through a familiar environment i.e. around rooms of their house or along well-known city streets, while visualizing the corresponding surroundings. Each paradigm consisted of five alternating 30-s blocks of imagery and rest. Verbal cues (‘tennis,’ ‘navigation,’ or ‘relax’) indicated the onset of each condition, with ‘relax’ signalling rest periods during which participants were instructed to remain still with eyes closed. Univariate fMRI analysis was performed for all patients across both tasks using FSL software ( https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/ ); for each functional run, a general linear model (GLM) was specified, comprising alternating blocks of active imagery and rest. 66 Statistical significance was determined at the cluster level using a threshold of z > 2.3 (cluster-corrected P < 0.05). 58 , 66 In a subset of patients, significantly greater brain activation was found during either of the ‘tennis task’ (supplementary motor area) or ‘navigation task’ (parahippocampal place area) than rest, which was used as evidence for task-responsiveness. 58 , 66 These patients were categorized ( Table 1 ) as ‘fMRI+ positive task responders’ (labelled ‘positive’), whereas the rest were ‘fMRI- negative non-responders’ (labelled ‘negative’). Resting-state fMRI data were collected over a 10-min acquisition period using a 3-tesla (3T) Siemens Trio scanner. Functional data comprised 300 EPI volumes with the following parameters: 33 slices; voxel size = 3 × 3 × 3.75 mm; TR = 2000ms; TE = 30 ms; flip angle = 78°. High-resolution anatomical T1-weighted images were obtained using a 3D sequence (1 mm isotropic voxels) with the following parameters: TR = 2300 ms; TE = 2.47 ms; 150 slices. Patients were excluded from further analysis if they met any of the following criteria: (i) large focal lesions exceeding one-third of a hemisphere or evidence of thalamic injury confirmed by neuroanatomical inspection; (ii) excessive motion of the head during resting-state acquisition (more than 3 mm translation and/or 3° rotation) or (iii) pre-processing failure i.e. segmentation and normalization. 26–28 , 60 Data pre-processing All functional and structural MRI data were pre-processed using SPM12 ( https://www.fil.ion.ucl.ac.uk/spm/software/spm12/ ). The first five functional volumes were removed to allow for scanner equilibrium/steady-state magnetization. Slice-timing correction was performed on the fMRI volumes, followed by realignment to mean functional volume, which produced realignment parameters that were included in the first-level statistical models. Using the mean functional image, direct spatial normalization to EPI-template was conducted using the ‘old norm’ function in SPM due to reduced variability across subjects compared to other approaches in previous studies. 67 The volunteers’ high-resolution structural images were co-registered to mean functional image or mean EPI (resulting from realignment stage) and segmented into grey matter, white matter, cerebrospinal fluid masks, and finally spatially normalized to the MNI-152 template. 68 Visual inspection or quality control of images was conducted for normalization to standard space, which was essential for the DOC dataset due to the potential effect of lesions on spatial transformations. One subject was excluded due to the half of the brain missing from the volumes, and another subject was excluded due to complete failure of normalization. Denoising was then conducted in the CONN software ( https://web.conn-toolbox.org ), 69 and functional images were smoothed at a FWHM Gaussian kernel of 6 mm. Movement parameters and their first temporal derivative were incorporated as first-level covariate to eliminate residual motion-related noise. Furthermore, the aCompCorr algorithm was applied to regressed out CSF, white-matter and motion-related signals from the time series (using first five principal components), which is a technique known to effectively remove movement-, respiratory- and cardiac-related artefacts, 70 particularly in patients with disorders of consciousness, 71 and has been validated for these kinds of datasets. 26 , 28 , 72 Additionally, the Artefact Detection Tools pipeline ( https://www.nitrc.org/projects/artefact_detect ) in the CONN toolbox was also implemented to further remove motion-related artifacts in the time-series data by regressing out the effect of outlier scans (movement greater than 0.09 mm) in a first-level analysis; this procedure helps minimize localized motion artefacts not captured by the aCompCorr algorithm. 70 , 73 Finally, linear detrending and a band-pass filter between 0.008 and 0.009 Hz were applied. Thalamic parcellation atlas We adopted seed-based functional connectivity (FC) analysis to establish which brain regions specific thalamic nuclei functionally interact with, where seeds (thalamic nuclei) need to be defined in regions of interest (ROIs). 69 Although MRI offers superior visualization of soft biological tissue and enables the characterization of distinct structures, the intrinsic contrast of conventional T1- and T2-weighted MRI are insufficient to reliably distinguish individual thalamic nuclei, largely due to the relatively small volume of the thalamus (∼8 cm 3 /hemisphere), which highlights the need for developing thalamic parcellation methods. Existing digital atlases of the thalamus remain few in number and show limited cross-validation or anatomical correspondence with one another. 52 A recently developed probabilistic atlas of thalamic subdivisions, derived from diffusion-weighted MRI data of 70 healthy Human Connectome Project volunteers, provides detailed mapping of thalamic organization. 53 Diffusion-weighted MRI (DW-MRI) uniquely enables non-invasive characterization of white matter pathways within each thalamic nucleus in relation to their cortical projections. In this atlas 53 the thalamus is segmented into seven regions closely matching the anatomical subparts, where six clusters correspond to histologically defined greater thalamic nuclei and the seventh cluster is a conglomerate encompassing three nuclei ( Supplementary Fig. 1 ). 53 This parcellation was selected as it provides an optimal balance between the spatial localization of thalamic voxels and their local diffusion characteristics, while showing strong correspondence with thalamic anatomy as defined in Morel’s stereotactic atlas. 53 , 74 , 75 Atlas standard space of the thalamus probabilistic masks was transformed into the pre-processed standard space using nearest neighbour interpolation to avoid overlapping of thalamic masks and impact of results. 76 Thalamic masks used correspond to (bilateral) nuclei: Pu, Ant, MD, VLD, central-lateral, lateral-posterior, medial-pulvinar group (CL-LP-MPu), VA and VLV ( Supplementary Figs 2 and 3 ). Statistical analysis We analysed the fMRI data of N = 16 healthy controls under deep sedation (mean plasma propofol concentration of 2.68 μ g/ml) and N = 22 DOC patients. FC was calculated using CONN ( https://web.conn-toolbox.org ) using seed-to-voxel analyses from 7 seeded thalamic nuclei aiming to investigate thalamic nuclei interactions across the whole brain. Temporal correlations between each thalamic seed and all other brain voxels were computed using a GLM. For the anaesthesia dataset where the experimental design is within-subject, FC analyses were first conducted individually, then the individual seed-to-voxel parameter estimate images were entered into group-level analyses. Specifically, paired-sample t tests were used for differences in the participants among the conditions of deep sedation versus awake and recovery versus deep sedation. The cortical SPM- t maps from the individuals, both unthresholded and thresholded, were grouped together with a one-sample t -test. For the DOC dataset having a between-subject design, two-sample t -test (DOC versus control) was conducted at group-level. We compared the DOC group to the healthy awake participants (‘control’ group) from the anaesthesia experiment. The group-level inference of both datasets was corrected for ­multiple comparisons using random field theory, with a voxel-level threshold of P < 0.005 (uncorrected) and cluster level of P < 0.05 (family-wise error FWE-corrected for multiple comparisons). 77 Radar plots are additionally presented to indicate the intrinsic connectivity network (ICN) spatial involvement (ICNi) in the contrasts of deep sedation versus awake and DOC versus control i.e. shows the voxel overlap between the FC results and canonical ICNs. These canonical ICNs were defined by an atlas 78 containing 10 well-matched resting-state networks (RSN) from the ICN_atlas toolbox ( https://www.nitrc.org/projects/icn_atlas/ ). 79 Description of ICN-RSN atlas: VN 1, 2, and 3 (‘ visual network ’): medial, occipital pole, lateral visual areas; DMN (‘ default mode network ’): medial parietal (precuneus and posterior cingulate), bilateral inferior–lateral–parietal and ventromedial frontal cortex; CN (‘ cerebellum network ’): cerebellum; SMN (‘ sensorimotor network ’): supplementary motor area, sensorimotor cortex, secondary somatosensory cortex; AUDN (‘ auditory network ’): superior temporal gyrus, Heschl's gyrus, posterior insular, including primary and association auditory cortices; ECN (‘ executive control network ’): medial–frontal areas, such as anterior cingulate and paracingulate; FPN1 and 2 (‘ frontoparietal network ’): frontoparietal areas; which are the only maps to be strongly lateralized. In addition, FPN1 corresponds strongly to perception, somesthesis, and pain, and FPN2 to cognition and language paradigms, concomitant with Broca's and Wernicke's areas. 57 Furthermore, to classify patterns of thalamo-cortical rs-FC findings based on network involvement (ICNi) with respect to individual nuclei, we performed hierarchical clustering using the pheatmap package in RStudio. 80 We used Euclidean as the distance metric and complete linkage clustering for its sensitivity to outliers. To investigate which nucleus had the greatest magnitude of change in FC with loss of consciousness, we computed in Matlab the average difference (unsigned magnitude) in FC (ΔFC) across the whole brain between the baseline (awake condition) and experimental comparison (deep sedation or DOC condition). Mean and standard deviations were used to normalize the values for ΔFC (z-scores). The subsequent statistical testing was computed in R and visualized in a box plot, where the y-axis represents the z-scored differential FC, and the x-axis depicts the specific thalamic nuclei. For statistical models, we used mixed linear model design wherever possible to correct for interpersonal variabilities (with the package ‘lme4’ in R). In these cases, we used the main effect as the fixed effect, and the subject ID as the random effect. The main effects which differ in different statistical tests were reported: for the DOC group comparison, the main effect for the ANOVA comparing the nucleus with the greatest magnitude of ΔFC with the rest of the nuclei was the difference of the dFC between this specific nucleus and the rest. The comparison for DOC subgroups did not have a within-individual structure, hence we did not conduct mixed linear models but used two-sample t -tests for that. Results Our findings address, firstly, which regions—or nuclei— of the thalamus were most strongly associated with loss of consciousness in both anaesthesia and DOC, and secondly, what mechanisms—or cortical networks— accounted for the functional changes between the respective thalamic nuclei and the rest of the brain. We thus present a whole-brain analysis of thalamo-cortical (and sub-cortical) FC changes across pharmacological and pathological loss of consciousness. We analysed the fMRI data of N = 16 healthy controls under deep sedation (mean plasma propofol concentration of 2.68 μ g/ml) and N = 22 DOC patients ( N = 13 in MCS and N = 9 in UWS). We used a thalamic parcellation atlas based on DW-MRI, where the thalamus is segmented into 7 ROIs closely matching the anatomical subparts in Morel’s atlas. 53 The pulvinar nuclei demonstrated the strongest cortical functional connectivity change in anaesthetic-induced loss of consciousness in healthy subjects First and foremost, we sought to investigate which thalamic nucleus was most associated with the overall FC changes in the brain with pharmacological (anaesthetic-induced) loss of consciousness. Thus, to determine which nucleus had the greatest magnitude of change in FC with loss of consciousness, we computed the average change or differential FC (ΔFC) across the whole brain between the baseline (FC maps indexed by individual’s correlation coefficient Beta for ‘awake condition’) and experimental comparison (FC maps for ‘deep sedation condition’). Mean and standard deviations were used to normalize the values for ΔFC (z-scores) displayed in a box plot where the y-axis represents the z-scored ΔFC and the x-axis depicts the specific thalamic nuclei ( Fig. 1 ). Among all nuclei, the Pu was found to have the greatest magnitude of ΔFC in healthy subjects with pharmacological loss of consciousness. We then tested, based on our preliminary finding, whether the Pu had the strongest ΔFC—significantly greater change—than the ΔFC of the rest of the nuclei. Thus, a linear mixed model correcting for individual differences was implemented to test for the significance of the Pu connectivity. We compared Pu group against the group of all other nuclei, and indeed, under anaesthesia, the ΔFC for Pu was found to be significant in comparison to the rest of the nuclei ( t = 2.081, P = 0.039). [All other nuclei were found to not be significantly different from each other in anaesthesia ( P = 0.442) following an F -test (ANOVA)]. Figure 1. Open in a new tab Magnitude of change in functional connectivity (ΔFC) in anaesthesia group comparison. Box plot displays z-scores for the anaesthesia group comparison ( n = 16), where among all nuclei, the Pu had the greatest magnitude of ΔFC in anaesthesia. For the statistical model, we used mixed linear model design to correct for interpersonal variabilities (with the package ‘lme4’ in R), with the main effect as the fixed effect and the subject ID as the random effect, to compare the magnitude of change in FC. Indeed, under anaesthesia, the ΔFC for Pu was found to be statistically significant in comparison to the rest of the nuclei ( t = 2.08, P = 0.039*). The rest of the nuclei were found to not be significantly different from each other in anaesthesia ( P = 0.442) with an F -test (ANOVA). Functional interactions between individual thalamic nuclei and whole brain in healthy volunteers under deep anaesthesia We then investigated the spatial localization of FC changes between thalamic nuclei and the rest of the brain with anaesthetic-induced loss of consciousness in healthy volunteers. We computed whole-brain connectivity (normalized as SPM-t) maps following seed-based FC analysis (contrast of deep sedation versus awake). Visually, SPM-t maps revealed that there were differential FC or cortical network patterns of connectivity that correspond to the discrete thalamic nuclei ( Fig. 2Ai–Gi ), following pharmacological loss of consciousness. Correspondingly, these FC patterns were reversed with recovery of consciousness from anaesthesia ( Supplementary Fig. 4A–G ), validating our findings. Figure 2. Open in a new tab Whole-brain resting-state (rs) functional connectivity (FC) maps in heathy volunteers in pharmacological (propofol-induced) loss of consciousness. Temporal correlations for thalamic seeds were computed for all other voxels in the brain using a GLM. For the anaesthesia dataset ( n = 16) where the experimental design is within-subject, FC analyses were first conducted individually, then individual seed-to-voxel parameter estimate images were entered into group-level analysis. Specifically, paired-sample t tests were used for differences in the participants among the condition of deep sedation versus awake. The cortical SPM- t maps from the individuals, both unthresholded and thresholded, were grouped together with a one-sample t -test. The group-level inference of the dataset was corrected for ­multiple comparisons using random field theory [voxel level threshold of P < 0.005 (uncorrected) and cluster level P < 0.05 (FWE-corrected for multiple comparisons)]. 77 Results were computed from seed-to-voxel rs-FC analysis across 7 respective thalamic nuclei [bilateral] seeds — Pu, Ant, MD, VLD, CL-LP-MPu, VA, VLV — and the rest of the brain. Colour bar denotes the strength of the t-statistic. Unthresholded t-maps of the deep sedation versus awake contrast are shown ( Ai, Bi, Ci, Di, Ei, Fi, Gi ). Radar plots indicate the intrinsic connectivity network (ICN) spatial involvement (ICNi) of brain regions in the deep sedation versus awake contrast i.e. shows the voxel overlap between the FC results and canonical ICNs ( Aii, Bii, Cii, Dii, Eii, Fii, Gii ). These canonical ICNs were defined by an atlas 78 containing 10 well-matched resting-state networks (RSN) from the ICN_atlas toolbox ( https://www.nitrc.org/projects/icn_atlas/ ). 79 Description of ICN-RSN atlas: VN 1, 2, and 3 (‘ visual network ’): medial, occipital pole, and lateral visual areas; DMN (‘ default mode network ’): medial parietal (precuneus and posterior cingulate), bilateral inferior–lateral–parietal, and ventromedial frontal cortex; CN (‘ cerebellum network ’): cerebellum; SMN (‘ sensorimotor network ’): supplementary motor area, sensorimotor cortex, and secondary somatosensory cortex; AUDN (‘ auditory network ’): superior temporal gyrus, Heschl's gyrus, and posterior insular. It includes primary and association auditory cortices; ECN (‘ executive control network ’): medial–frontal areas, including anterior cingulate and paracingulate; FPN1 and 2 (‘ frontoparietal network ’): frontoparietal areas; these are the only maps to be strongly lateralized. In addition, FPN1 corresponds strongly to perception–somesthesis–pain, and FPN2 to cognition–language paradigms, consistent with Broca's and Wernicke's areas. 57 Then, to quantify more precisely the cortical network involvement in the SPM-t maps, we overlapped our maps with ICNs and calculated the ICN involvement with a resting-state brain network atlas 78 using the ICN_atlas toolbox. 79 We were able to further validate numerically that, indeed, there was prominent cortical network involvement that varied across the different thalamic nuclei ( Fig. 2Aii–Gii ). Finally, we used hierarchical clustering to identify these distinctive patterns of functional connectivity we found for each thalamic nucleus ( Fig. 3 ). This complemented our visual and quantitative assessments, as it was evident there were commonalities but also differential functional connectivity profiles across thalamic nuclei. With anaesthetic-induced loss of consciousness, the Pu and VLV were grouped into two separate clusters; the Ant and MD were grouped into a third cluster; and the CL-LP-MPu, VLD and VA were grouped into a fourth cluster. Additionally, the algorithm computed clusters based on cortical networks, and found that the higher-order networks were grouped separately (i.e. FPN1 and FPN2 into one cluster, and DMN and ECN into a separate cluster) from the lower-order networks (i.e. all other networks, including SM, VN, AUDN and CB, into additional clusters). Thus, as a follow-up to our previous findings ( Fig. 1 ), we identified that the Pu demonstrated strong FC changes with the DMN, accompanied by even stronger FC changes with the ECN, providing substantiation to the proposition that the Pu may manifest more acute contributions than other nuclei in anaesthetic-induced loss of consciousness, through interactions with the higher-order DMN and ECN networks. (With all other cortical networks—which were exclusively lower order networks except for FPN1—the Pu revealed opposite FC changes). Remarkably, therefore, the nucleus with the strongest delta FC (quantitative change) was the same one that also demonstrated the most distinct spatial pattern of FC changes across the cortex (qualitative change). Figure 3. Open in a new tab Hierarchical clustering of thalamo-cortical resting-state (rs) functional connectivity (FC) in heathy volunteers in pharmacological (propofol-induced) loss of consciousness. To classify patterns of thalamo-cortical rs-FC findings based on network involvement (ICNi) with respect to individual nuclei, we performed hierarchical clustering using the pheatmap package in RStudio. 80 We used Euclidean as the distance metric and complete linkage clustering for its sensitivity to outliers. The algorithm computed clusters based on thalamic nuclei (y-axis), where Pu and VLV were grouped into two separate clusters; the Ant and MD were grouped into a third cluster; and the CL-LP-MPu, VLD, and VA were grouped into a fourth cluster. Additionally, the algorithm computed clusters based on cortical networks (x-axis), and found that the higher-order networks were grouped separately (i.e. FPN1 and FPN2 into one cluster, and DMN and ECN into a separate cluster) from the lower-order networks (i.e. all other networks, including SM, VN, AUDN, and CB, into additional clusters). The ventral-latero-ventral nuclei exerted the strongest cortical functional connectivity change in pathological loss of consciousness in disorders of consciousness patients Analogous to our previous approach, we sought to identify which thalamic nucleus was most associated with the overall FC changes with pathological (DOC-induced) loss of consciousness. Thus, to determine which nucleus had the greatest magnitude of change in FC with loss of consciousness, we computed the average difference in FC (ΔFC) across the whole-brain between the baseline (‘awake condition’) and experimental comparison (‘DOC condition’). Mean and standard deviations were used to normalize the values for ΔFC (z-scores) displayed in a box plot where the y-axis represents the z-scored ΔFC and the x-axis depicts the specific thalamic nuclei ( Fig. 4 ). Among all thalamic nuclei, VLV was found to have the greatest magnitude of ΔFC with pathological loss of consciousness. We then wanted to test whether our preliminary finding that the VLV had the strongest ΔFC is statistically more significant than the ΔFC of the rest of the nuclei. Hence, a linear mixed model correcting for individual differences was implemented to test for the significance of the VLV connectivity. Indeed, in DOC, the ΔFC for VLV was found to be statistically significant in comparison to the rest of the nuclei ( t = 12.336, P = 2.2e-16***) with an F -test (ANOVA). Owing to significant differences found among the rest of the nuclei in DOC ( P = 1.703e-13), we made multiple comparisons between VLV with each of the rest of the nuclei. After correcting significance level for the multiple comparisons, VLV was shown significantly the highest among all. We suggest that the generality of significant alterations in the thalamic FC was partly due to the between-subject design of the dataset where the DOC group has considerable individual variations owing to various aetiologies, including TBI, global hypoxic-ischaemic encephalopathy, ischaemic stroke and other causes. 7 , 62 Figure 4. Open in a new tab Magnitude of change in functional connectivity (ΔFC) in DOC group comparison. Box plot displays z-scores for the DOC group comparison ( n = 22), where among all nuclei, the VLV had the greatest magnitude of ΔFC in DOC. Linear mixed models for within-subject analysis were used to compare the magnitude of change in FC. The ΔFC for VLV was found to be statistically significant in comparison to the rest of the nuclei (t = 12.336, P = 2.2e-16***) with an F-test (ANOVA). Owing to significant differences found among the rest of the nuclei in DOC ( P = 1.703e-13), we made multiple comparisons between VLV with each of the rest of the nuclei (where the main effect for the ANOVA comparing VLV with the rest of the nuclei was the difference of the dFC between the VLV and the rest of the nuclei.) After multiple comparison, VLV was significantly the highest among all. Functional interactions between individual thalamic nuclei and whole brain in disorders of consciousness patients We then explored the spatial localization of FC changes between thalamic nuclei and the rest of the brain with pathological loss of consciousness in DOC patients. We computed whole-brain connectivity (normalized as SPM-t) maps following seed-based FC analysis (contrast of DOC versus control). Visually, SPM-t maps reveal that there were differential FC or cortical network patterns of connectivity that correspond to the discrete thalamic nuclei ( Fig. 5Ai–Gi ), with pathological loss of consciousness. Figure 5. Open in a new tab Whole-brain resting-state (rs) functional connectivity (FC) maps in DOC patients in loss of consciousness. Temporal correlations for thalamic seeds were computed for all other voxels in the brain using a GLM. For the DOC dataset ( n = 22) having a between-subject design, two-sample t -test (DOC versus control) was conducted at the group level. The group-level inference of the dataset was corrected for ­multiple comparisons using random field theory [voxel level threshold of P < 0.005 (uncorrected) and cluster level P < 0.05 (FWE-corrected for multiple comparisons)]. 77 Results were computed from seed-to-voxel rs-FC analysis across 7 respective thalamic nuclei [bilateral] seeds—Pu, Ant, MD, VLD, CL-LP-MPu, VA, VLV—and rest of the brain. Colour bar denotes the strength of the t-statistic. Unthresholded t-maps of the DOC versus control contrast are shown ( Ai, Bi, Ci, Di, Ei, Fi, Gi) . Radar plots indicate the intrinsic connectivity network (ICN) spatial involvement (ICNi) of brain regions in the DOC versus control contrast i.e. shows the voxel overlap between the FC results and canonical ICNs ( Aii, Bii, Cii, Dii, Eii, Fii, Gii ). These canonical ICNs were defined by an atlas 78 containing 10 well-matched resting-state networks (RSN) from the ICN_atlas toolbox ( https://www.nitrc.org/projects/icn_atlas/ ). 79 Description of ICN-RSN atlas: VN 1, 2, and 3 (‘ visual network ’): medial, occipital pole, and lateral visual areas; DMN (‘ default mode network ’): medial parietal (precuneus and posterior cingulate), bilateral inferior–lateral–parietal, and ventromedial frontal cortex; CN (‘ cerebellum network ’): cerebellum; SMN (‘ sensorimotor network ’): supplementary motor area, sensorimotor cortex, and secondary somatosensory cortex; AUDN (‘ auditory network ’): superior temporal gyrus, Heschl's gyrus, and posterior insular. It includes primary and association auditory cortices; ECN (‘ executive control network ’): medial–frontal areas, including anterior cingulate and paracingulate; FPN1 and 2 (‘ frontoparietal network ’): frontoparietal areas; these are the only maps to be strongly lateralized. In addition, FPN1 corresponds strongly to perception–somesthesis–pain, and FPN2 to cognition–language paradigms, consistent with Broca's and Wernicke's areas. 57 Next, to quantify the cortical network involvement in the SPM-t maps, we overlapped our maps with ICNs and calculated the ICN involvement with a resting-state brain network atlas 78 using the ICN_atlas toolbox. 79 We were able to further validate numerically that, indeed, there was prominent cortical network involvement that varied across the different thalamic nuclei ( Fig. 5Aii–Gii ). Finally, we used hierarchical clustering to identify these distinctive patterns of functional connectivity we found for each thalamic nucleus ( Fig. 6 ). This complemented our visual and quantitative assessments, as it was evident there were commonalities but also differential functional connectivity profiles across thalamic nuclei. With loss of consciousness in DOC, the VLD and CL-LP-MPu were grouped into one cluster; the MD, Ant and VA were grouped into a second cluster and the Pu and VLV were grouped into a third and fourth cluster respectively. Furthermore, the algorithm computed clusters based on cortical networks, and found that the higher-order networks were grouped separately (i.e. FPN1 and FPN2 into one cluster, and DMN and ECN into a separate cluster) from the lower-order networks (i.e. all other networks, including SM, VN, AUDN and CB, into additional clusters). Hence, as a follow-up to our previous findings presented earlier in Fig. 4 , we identified that the VLV demonstrated strong FC changes with all higher-order networks, including the FPN1, FPN2, DMN and ECN. (With all other cortical networks—which were solely lower-order networks—the VLV revealed opposite FC changes). Crucially, therefore, the nucleus with the strongest delta FC (quantitative change) was the same one that also demonstrated the most distinct spatial pattern of FC changes across the cortex (qualitative change). This consolidates our premise that the VLV, through interactions with higher-order networks, may play a more pivotal role in pathological perturbations of consciousness than previously explored. Our findings complement previous lines of evidence highlighting the role of VLV as an integrative centre for motor control, receiving cortical-cerebellar-basal ganglia input and sending core-like projections to primary motor areas. 81 , 82 Figure 6. Open in a new tab Hierarchical clustering of thalamo-cortical resting-state (rs) functional connectivity (FC) in DOC patients in loss of consciousness. To classify patterns of thalamo-cortical rs-FC findings based on network involvement (ICNi) with respect to individual nuclei, we performed hierarchical clustering using the pheatmap package in RStudio. 80 We used Euclidean as the distance metric and complete linkage clustering for its sensitivity to outliers. The algorithm computed clusters based on thalamic nuclei (y-axis), where the VLD and CL-LP-MPu were grouped into one cluster; the MD, Ant, and VA were grouped into a second cluster; and the Pu and VLV were grouped into a third and fourth cluster respectively. Furthermore, the algorithm computed clusters based on cortical networks (x-axis), and found that the higher-order networks were grouped separately (i.e. FPN1 and FPN2 into one cluster, and DMN and ECN into a separate cluster) from the lower-order networks (i.e. all other networks, including SM, VN, AUDN, and CB, into additional clusters). Behavioural correspondence with the thalamic functional connectivity changes In our final analysis ( Fig. 7 ), we sought to further establish thalamic changes in the DOC patients by subjecting them to a test, which assessed their levels of covert consciousness. This test involved tasks of motor imagery (‘tennis task’) and spatial imagery (‘navigation task’), which allowed us to split DOC patients into two categories: those with positive responses to the task (‘fMRI+ or task responders’) and those with negative responses to the task (‘fMRI- or task non-responders’). All positive task-responders ( n = 7) were TBI patients (except for one patient) and were exclusively MCS patients. Among those who did not respond to the task ( n = 15), six patients were MCS and nine patients were UWS. We used this test as it has been demonstrated to be a more reliable proxy for consciousness because it provides a more sensitive evaluation of the cognitive functions of DOC patients that could not be readily observable with standard behavioural diagnosis such as MCS and UWS. 64 , 65 , 83 , 84 This is especially vital given the high rate of misdiagnosis in DOC patients. 85 , 86 Concretely, this implies that the subgroup of positive task responders are less severely impaired on the DOC continuum, whereas the negative non-responders consists of more severe patients. Figure 7. Open in a new tab Magnitude of change in functional connectivity (ΔFC) in DOC subgroup comparison. Box plot displays ΔFC scores corresponding to each DOC subgroup (i. fMRI-/task non-responders and ii. fMRI+ /task responders) ( n = 22) differential FC from healthy controls. For statistical models, as this comparison does not have a within-individual structure, we used two-sample t-tests. Among all nuclei, the VLV had the greatest ΔFC between task non-responders and task responders after FDR correction. We proceeded with testing the behavioural relevance of FC alterations found before by investigating the differential FC (ΔFC) of thalamic nuclei between each distinct subgroup of DOC patients and healthy controls. The subgroups include those with negative fMRI responses to the Tennis Task (‘task non-responders’, fMRI-) and those with positive fMRI responses (‘task responders’, fMRI+). Across all 7 thalamic nuclei, the ΔFC was found to be statistically significant following a two-sample t -test between the fMRI+ and fMRI- subgroups, corrected for multiple comparisons (FDR-correction). In line with our previous findings ( Fig. 4 ), the greatest ΔFC across the two subgroups was found for the VLV ( P = 0.0083). Crucially, the VLV is significantly responsible for FC changes in pathological loss of consciousness overall, and it is even more prominent in the cohort of more severely impaired DOC patients (negative non-responders). This result reinforces our previous analysis and suggests the utility of VLV, an integral part of the motor thalamus, as not merely a reliable neural marker for consciousness but also, crucially, a behavioural marker that is specific to patients with DOC, particularly those at the more severe end of the spectrum. Discussion We sought to identify the contributions of major subdivisions of the thalamus to perturbed states of consciousness in pharmacological (propofol-induced anaesthesia) and pathological (DOC) models in humans. Using fMRI, we aimed to explore multiple distinct nuclei of the thalamus in loss of consciousness by revealing whole-brain FC changes. Specifically, we examined not only where the largest in magnitude, thalamo-cortical FC alterations occurred but also in which specific thalamic nucleus these changes were most prominent. We found that, among all thalamic nuclei, the Pu was found to have the biggest magnitude of change or strongest differential FC (ΔFC) with anaesthetic-induced loss of consciousness in healthy volunteers, which demonstrated strong FC changes with higher-order ECN and DMN and weaker FC changes with FPN2 cortical networks (and opposite FC changes with all other cortical networks, particularly lower-order). Remarkably, by contrast, in the patient cohort where loss of consciousness was induced by pathological reasons (i.e. DOC), the VLV was found to have the strongest ΔFC among thalamic nuclei, which demonstrated strong FC changes with all higher-order cortical networks, including the FPN1, FPN2, DMN and ECN (whilst revealed opposite FC changes with all other, exclusively lower-order, cortical networks). Crucially, we highlight that this finding is not only a prominent neural marker but also reflects changes at the behavioural level, which evidently has important clinical implications. Our findings of the Pu demonstrating the strongest impact on overall ΔFC in anaesthetic-induced loss of consciousness are in line with previous work demonstrating that particularly the medial pulvinar region is involved not only in visuomotor and auditory processing but also associative and higher cognitive processing, and even more so, abnormal medial pulvinar connectivity was associated to neurodevelopmental disorders. 87 The medial pulvinar was shown to be involved in altered states of consciousness, notably focal seizures, 88 and, more recently, stimulation of this region led to improvements in awareness in pharmaco-resistant temporal lobe epilepsy. 89 Furthermore, studies demonstrated that pulvinar lesions led to changes in conscious content—impaired feature binding 90 and hemineglect 91 —in humans. On the other hand, in the DOC patient cohort, the VLV was found to have the strongest (ΔFC) among all nuclei in pathological loss of consciousness. The VLV, known to be part of the motor thalamus, is involved in action control in a circuitry with basal ganglia, motor, premotor and prefrontal cortices. 57 , 82 , 92 More specifically, VL nuclei have connections to the globus pallidum internal (GPi) of the basal ganglia, 22 , 93 and atrophy in this area has been associated with decreased behavioural arousal patients with DOC. 22 Additionally, the VL has been implicated in higher-level cognition, particularly in the integration of attention and memory, has been demonstrated by evidence of VL lesions leading to impairments in stroke patients. 94 Importantly, the involvement of distinct cortical networks may account for the strongest whole-brain ΔFC identified for the Pu in pharmacological and VLV in pathological loss of consciousness. Higher-order networks exhibited strong FC changes with the Pu and VLV, in the healthy anaesthetized and DOC patient cohorts, respectively. In both types of loss of consciousness, the DMN and ECN, considered to be core consciousness hubs, were highly involved, though stemming from different thalamic hubs. Importantly, the strongest FC changes for the VLV were demonstrated with the ECN. Additionally, other higher-order networks, i.e. FPN were found to have the strongest FC changes with VLV in pathological loss of consciousness, confirming the importance of such consciousness-related networks. Thus, there may be shared mechanisms involving cortical network interactions in pharmacological and pathological loss of consciousness, though they have associations with different thalamic loci. Our results confirm previous literature surrounding the importance of large-scale consciousness-supporting networks, both in health and disease. 4–6 , 8–11 , 95 Among these, the default mode network, DMN, well-known to be largely linked to conscious states, 5 , 96–98 is a widely distributed network consisting of medial frontal and posterior medial parietal cortices, angular gyrus and hippocampus, of which the posteromedial cortex is a major connectivity hub. 99 It has been increasingly associated with mediating internal (self-related) and external (environmental) information processing. 98 , 100 The posteromedial area has been demonstrated to play a key role in altered consciousness in epileptic patients 101–103 and in patients with disorders of consciousness. 28 , 104 , 105 Converging evidence has shown the importance of cortical and subcortical contributions of the DMN, and especially posteromedial areas, in healthy subjects under anaesthesia and/or patients with DOC. 7 , 27 , 28 Our results align with several resting-state fMRI studies in healthy volunteers which have demonstrated strong FC between precuneus and whole thalamus during changes in consciousness. 106 , 107 Importantly, specific thalamic nuclei were found to be both functionally and structurally connected to nodes of the default mode network. 107–111 Using tractography analyses of diffusion MRI data, Edlow and colleagues mapped projection pathways connecting brainstem default ascending arousal network nodes with thalamic—intralaminar, reticular, and paraventricular nuclei—and cortical DMN nodes in healthy controls. 112 Furthermore, past work has shown that alterations in structural connectivity in DMN and along the thalamus-precuneus pathway were correlated with behavioural signs in patients with impaired consciousness, 63 reinforcing the importance of posteromedial regions in the neural basis of consciousness. Evidence of compromised connectivity between the precuneus and whole thalamus associated with impaired consciousness have been further demonstrated in various MCS and UWS cohorts. 113–115 Additionally, PET evidence indicates that UWS patients display altered effective connectivity between the posterior cingulate and frontal association regions, 116 as well as between the anterior cingulate cortex, prefrontal cortex, and thalamus. 107 Moreover, network property alterations and FC disruptions were found in brain regions associated with consciousness, particularly the medial parietal cortex, frontal cortex and thalamus in DOC patients. 117 Furthermore, a functional regulation of the DMN by the thalamus has been suggested in other impaired states of consciousness, specifically during epileptic seizures. Among other higher-order cortical networks relevant to our findings, the executive control (ECN) network, which consists of dorsolateral frontal and parietal nodes, is known for mediation of attention and awareness of one’s environment. 103 , 118–120 There is also increased evidence of its alterations in disorders of consciousness, where fewer MCS and UWS patients following brain injury showed components of neuronal origin for the left and right ECN, as well as for DMN and auditory networks, compared to healthy controls. 15 Furthermore, dynamic interactions between the ECN and salience network have been found important for sustaining and recovering consciousness. 14 Similarly, the FPN is associated with restoring cerebral activity in recovery from disorders of consciousness, 114 , 121 and is known to have robust connectivity to subcortical and limbic structures, notably the thalamus and basal ganglia. 20 , 95 , 113 Finally, our results demonstrated strong FC changes with higher-order networks (i.e. DMN, ECN, FPN) and opposite FC changes with lower-order networks (i.e. SMN, VN, AUDN, CBN) in our DOC patient cohort. It remains to be explored precisely how this functional network antagonism may play a role in sustaining consciousness in larger cohorts of patients. 122 Clearly, our findings demonstrate that distinct cortical networks interact with different thalamic nuclei depending on the nature of loss of consciousness, whether pharmacological or pathological, thus opening an interesting avenue for further inquiries into the importance of thalamo-cortical aka ‘ thalamic nuclei-cortical network’ interactions which seem to contribute to the phenomena of loss of consciousness. On a more microscopic level, two distinct cell classes in the thalamus have been shown to be differentially correlated with the cortex, which includes parvalbumin-staining (PVALB) core cells driving excitatory activity via projections to cortical layers III and IV) and calbindin-staining (CALB1) matrix cells projecting diffusely to supra-granular and infragranular layers. 17 , 33 , 123 Intriguingly, previous work showed that CALB1-rich matrix cells in the healthy awake brain exhibited strong preferential functional coupling with several cortical networks, involving the control, limbic, ventral attention, and notably the DMN, whereas in contrast, PVALB-rich core cells favoured functional coupling with visual, dorsal-attention, temporo-parietal and somatomotor networks. 124 Moreover, Huang and colleagues observed a reorganization from a balanced core-matrix unimodal-transmodal functional geometry during consciousness to unimodal core dominance during loss of consciousness with propofol in humans. 125 Additionally, parvalbumin expression was found to be higher in areas with reduced cortical connectivity during sedation with propofol. 126 There remains a need to further investigate the coupling effects between cortical networks and matrix or core cells in thalamic nuclei in the anaesthetized brain, and even more so, in patients with DOC. Owing to recent advancements in thalamic parcellation, 51 , 52 our study has demonstrated that specific sections of the thalamus are associated with distinct types of perturbations of consciousness, pathological 7 or pharmacological. 127 Previously, the thalamus had been explored almost exclusively as a whole entity, and few studies have been able to break down the heterogenous complex nuclei and explore their respective relevance in consciousness in humans. 42–44 In a recent animal study led by Tasserie and colleagues, electrical stimulation of the ‘central thalamus’, notably CM nuclei, was found to restore arousal and awareness in a macaque model in propofol-induced loss of consciousness. 40 Although stimulation of the ventral thalamus control site had no effects on behavioural score in macaques, the authors acknowledged there was no data acquired during an event-related auditory task experiment to measure conscious awareness following ventral lateral thalamic stimulation under anaesthesia, 40 pressing the need for more extended and nuanced investigations. Furthermore, Bastos and colleagues 128 implanted multiple-contact stimulating electrodes in frontal thalamic nuclei (i.e. intralaminar, mediodorsal, with additional targets in neighbouring ventral posterolateral nucleus) in macaques, and demonstrated that thalamic stimulation reversed the electrophysiologic patterns associated with unconsciousness. Given our findings, not only do we provide a differential picture between pharmacological and pathological states of unconsciousness with relevance to the thalamus but we also press for the need for more nuanced experimental paradigms that distinguish arousal (alertness or wakefulness) from awareness (content of subjective experience) in human consciousness. 129–131 One model specific to aetiologies of brain injuries producing DOC has been proposed in the clinical literature—the ‘mesocircuit model’—,which suggests a downregulation of synaptic activity across areas of the frontal cortex (i.e. medial frontal, anterior cingulate cortex), central thalamus (i.e. central-lateral nucleus) and striatum. 7 , 132 Moreover, a functional disfacilitation of the central thalamus, following injuries, reduces activity along projections from central thalamus to the frontal cortex, posterior medial parietal cortex (a key DMN node) and striatum. 23 Fridman and Schiff 23 have previously outlined that the term ‘central thalamus’ refers to a functional and physiological construct rather than a strictly delineated anatomical entity. 23 , 133–135 Furthermore, the targets for stimulation in 5 individual patients with impaired consciousness following traumatic brain injury were functional targets of the central-lateral/medial dorsal tegmental fibre bundle activations, 46 modelled upon DTI data, 46 , 55 , 57 though none of these patients were DOC. In our cohort of DOC patients, we provide evidence for a more complex picture in pathological disruptions of consciousness with an important role for the VLV, part of the motor thalamus, possibly through inputs from the globus pallidus and striatum from the basal ganglia circuitry. 16 , 82 , 93 Moreover, it may be important to consider the proximity of ventral lateral ventral to central-lateral nuclei, as well as their relevant projections. Specifically, ventral lateral nuclei have projections to the globus pallidum internal of the basal ganglia, 110 and atrophy in this area has been associated with decreased behavioural arousal patients with DOC. 104 Alternatively, the central-lateral neurons innervate the rostral striatum of the basal ganglia, as well as prefrontal/frontal cortex; these CL projections drive action potentials from primary striatal output neurons, which, consequently, increase the inhibition of the globus pallidus interna, and release further thalamic activation of the cortical regions. 46 We suggest, therefore, a potential interactive framework by which specific functional nodes—CL and VLV included—of the thalamus coordinate with the basal ganglia to modulate arousal in humans, calling for further experimental investigations. Opening the ‘black box’ of the thalamus significantly contributes to unravelling the ‘black box’ of consciousness. By disentangling the thalamus into more nuanced anatomically relevant parts, we suggest that pharmacological and pathological loss of consciousness—although they overlap at the cortical 5 , 28 , 41 , 60 , 136 and brainstem 27 levels—appear to involve distinct thalamic contributions, given that we found the Pu in anaesthesia and VLV in DOC, respectively, to be pivotal. This has clinical implications, as DOC therapies may not be fully reliant upon anaesthetic studies, at least where thalamic connectivity is concerned. Unlike anaesthesia, which taps into the brain connectivity changes through transient neuromodulatory influences, the permanent anatomical damages that accompany brain injury may account for the differential thalamic involvement. 7 , 137 By having further separated the DOC in two groups, those severely injured and those less impaired, and by having shown that the VLV remains the predominant nucleus responsible for both, we show consistency within the DOC category of loss of consciousness. Hence, our results not only demonstrate there is heterogeneity between consciousness alterations of anaesthesia and DOC in terms of thalamic contributions but also homogeneity of thalamic involvement within the DOC category, regardless of the exact aetiology of the brain injury and extent of lesions in the patients. This suggests that, for all its diversity, DOC samples can inform future clinical studies and pinpoint where in the thalamus brain stimulation therapies may prove more efficacious. Given the heterogeneity both in terms of structural damage and functional dysconnectivity in these types of patient cohorts, we would need to, ideally, begin implementing personalized functional connectome mapping 95 to achieve improvements in clinical responses from brain stimulation. 95 , 138 , 139 Our differential findings related to the thalamus depending on the type of unconsciousness is especially pertinent, given recent evidence demonstrating distinctive patterns of cortical engagement between arousable unconsciousness (sleep) and unarousable unconsciousness (propofol-induced anaesthesia), 140 adding to the debate on the shared versus different mechanisms across multiple conditions of unconsciousness. 141 Our study has several limitations. Firstly, thalamic subdivisions were not segmented at the basis of individual subjects in their native brain space. The processing pipeline we adopted is standard in neuroimaging studies. The atlas we chose should be able to tolerate the normalization error, as it was shown to have high inter-subject consistency and intra-subject reproducibility. 53 However, we believe it desirable to generate individual-specific parcellations which could provide even higher anatomic accuracy. This may be experimented and validated by future studies. Secondly, we recognize that the healthy volunteers who were given propofol may have been not in in unconscious state per se, but rather, an unresponsive state, which could be explained by the effect-site concentration and plasma concentration used in our study. Furthermore, given the nature of the patient cohort, segmentation for nuclei may not be perfect in all cases, especially considering the inherent challenges of spatially normalizing DOC patients following brain injury. However, we implemented a rigorous spatial normalization process that brought all images into standard space, facilitated the use of a validated, high-resolution and fine-parcellation thalamic atlas 53 in a region of interest analysis, and applied a stringent visual quality control; all these steps ensured our segmentation was as accurate as achievable, supporting the robustness of our findings. Finally, we used the datasets from two imaging sites with non-identical scanner hardware and acquisition parameters. We compared the DOC group to the healthy awake participants from the anaesthesia experiment. However, we are confident this does not influence our interpretation of the main results, due to the fact we focused our analysis on between-nuclei comparisons, by which the system variance caused by different scanners should be evened out. Additionally, our results related to the DOC experiment should hold, because they demonstrated a behavioural correspondence which was tested within subsets of the DOC dataset. In fact, previous publications utilizing multi-site datasets have proven the validity of this approach. 28 Our work which has aimed to identify which thalamic nuclei are more prominent in loss of consciousness could bring diagnostic and therapeutic value for DOC patients. Considering that consciousness depends on the interplay of brain networks, 4 , 30 , 41 we could use localized individualized brain stimulation, in one or multiple nuclei, to induce controlled perturbations in a specific node of the network, which would trigger the propagation of neural activity across proximal and distal brain regions. 142 In this case, the nuclei which ‘stirred’ the connectivity across the whole brain the most may have the highest chance for neuromodulatory effects following targeted thalamic stimulation. Considering that no clear consensus currently exists on the ideal neural substrates for therapeutic neuromodulation of consciousness, 7 several functional neuroimaging studies in patients have been able to suggest the involvement of specific brain regions or circuits for targeted stimulation. One study reverted to cortical stimulation in DOC patients and provided the first proof of principle in a sham-controlled randomized double-blind study that non-invasive technique of transcranial direct current stimulation (tDCS) of anterior cortical regions, notably the dorso-lateral prefrontal cortex, can increase CRS-R scores and lead to cognitive gains in MCS patients after brain injury. 143 Alternatively, central-lateral thalamic stimulation has been demonstrated in a few single-case studies, 24 , 144 whereas an alternative to DBS, the non-invasive technique of low-intensity focused ultrasound, targeting the central thalamus has been reported to exhibit clinically significant increases in behavioural responsiveness in two chronic DOC patients. 145 , 146 Furthermore, DBS on the central thalamus in a MCS patient exhibited reactivations of dormant functional brain networks; however, increases in consciousness were limited. 147 Moreover, a single-case study demonstrated the effects of DBS were more long-lasting than those observed with pharmacological interventions, notably zolpidem, albeit the behavioural improvements were still limited. 148 Until now, however, stimulation studies have been limited, which presses for the need for future work—perhaps, we need to be targeting not one but multiple nuclei in the thalamus for recovery of consciousness, as demonstrated by recent investigations of a novel multiple-target thalamic stimulation paradigm in epilepsy. 149 Given our findings, one way forward could be to redefine stimulation protocols to add the VLV as a target. Our work, thus, offers the most up-to-date elucidation of consciousness-relevant regions in the thalamus, in the hope of marking an accurate, consistent, and reliable target for stimulation site with DBS for recovery from DOC in the not-so-distant future. Conclusions In summary, this is a systematic investigation of thalamic connectivity involving both healthy anaesthetized and DOC patients, where we establish that (i) anaesthesia and DOC exhibit distinct involvement of thalamic nuclei for loss of consciousness; (ii) specific nuclei, Pu and VLV, account for the greatest functional changes in anaesthesia and DOC, respectively and (iii) distinct cortical networks may contribute to the underlying mechanisms accounting for the functional changes between these thalamic subdivisions and rest of the brain in loss of consciousness. Perhaps most critical of all, we encourage the need for more parcellated explorations into the thalamic mosaic for the purpose of reaching a more unified consensus on the diagnostics and therapies for DOC. Supplementary Material fcag021_Supplementary_Data fcag021_supplementary_data.pdf (2.8MB, pdf) Acknowledgements We gratefully acknowledge Victoria Lupson and the staff at the Wolfson Brain Imaging Centre, Addenbrooke’s Hospital, for their assistance throughout scanning sessions. Contributor Information Dorottya Szocs, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK; Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK. Dian Lyu, Departments of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA 94304, USA. Andrea I Luppi, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK; Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK; Montreal Neurological Institute, McGill University, Montreal, QC H3A 2B4, Canada. Peter Coppola, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK; Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK. Rebecca E Woodrow, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK; Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK. Guy B Williams, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK; Wolfson Brain Imaging Centre, University of Cambridge, Cambridge CB2 0QQ, UK. Judith Allanson, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK. John D Pickard, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK. Adrian M Owen, The Brain and Mind Institute, Department of Psychology, The University of Western Ontario, London, ON N6A 5B7, Canada. Lorina Naci, Institute for Neuroscience, Trinity College Dublin, Dublin D02, Ireland. David K Menon, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK; Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK. Emmanuel A Stamatakis, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK; Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0SP, UK. Supplementary material Supplementary material is available at Brain Communications online. Funding This work was supported by grants from the UK Medical Research Council (U.1055.01.002.00001.01) [to A.M.O. and J.D.P.]; the James S. McDonnell Foundation [to A.M.O. and J.D.P.]; the Canada Excellence Research Chairs program (215063) [to A.M.O.]; the Canadian Institute for Advanced Research [to A.M.O., D.K.M. and E.A.S.]; the National Institute for Health and Care Research, the NIHR Cambridge Biomedical Research Centre and Senior Investigator Awards [to J.D.P. and D.K.M.]; the British Oxygen Professorship of the Royal College of Anaesthetists [to D.K.M.]; the Evelyn Trust, Cambridge and the East of England CLAHRC fellowship [to J.A.]; the L’Oreal-Unesco for Women in Science Excellence Research Fellowship [to L.N.]; the Stephen Erskine Fellowship, Queens’ College, University of Cambridge [to E.A.S.] and the Gates Cambridge Trust [to A.I.L.]. The research was also supported by the NIHR Brain Injury Healthcare Technology Co-operative based at Cambridge University Hospitals NHS Foundation Trust and University of Cambridge. Competing interests The authors declare that they have no competing interests. Data availability Access to DOC patient data is limited to qualified researchers for non-commercial use to protect patient confidentiality. The UK Health Research Authority assigns responsibility for safeguarding the data to the Chief Investigators of the original studies (Dr. Judith Allanson and Prof. David Menon, or anyone else to whom responsibility is given). To request access, please contact the Data Access Committee: Dr. Judith Allanson ( [email protected] ), Prof. David Menon ( [email protected] ) or Dr. Emmanuel Stamatakis ( [email protected] ). The propofol dataset is publicly accessible via the OpenNeuro data repository (doi: 10.18112/openneuro.ds003171.v2.0.1 ). The codes produced and used for this work have been uploaded at https://github.com/dorottyaszocs1/Thalamus_in_-un-consciousness_in_pharmacological_and_pathological_states . References 1. Avidan  MS, Mashour  GA. Prevention of intraoperative awareness with explicit recall: Making sense of the evidence. Anesthesiology. 2013;118(2):449–456. [ DOI ] [ PubMed ] [ Google Scholar ] 2. He  BJ. Next frontiers in consciousness research. Neuron. 2023;111(20):3150–3153. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Owen  AM. The search for consciousness. Neuron. 2019;102(3):526–528. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Li  H, Zhang  X, Sun  X, et al.  Functional networks in prolonged disorders of consciousness. Front Neurosci. 2023;17:1113695. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Huang  Z, Mashour  GA, Hudetz  AG. Functional geometry of the cortex encodes dimensions of consciousness. Nat Commun. 2023;14(1):72. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Oujamaa  L, Delon-Martin  C, Jaroszynski  C, et al.  Functional hub disruption emphasizes consciousness recovery in severe traumatic brain injury. Brain Commun. 2023;5(6):fcad319. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Edlow  BL, Claassen  J, Schiff  ND, Greer  DM. Recovery from disorders of consciousness: Mechanisms, prognosis and emerging therapies. Nat Rev Neurol. 2021;17(3):135–156. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Threlkeld  ZD, Bodien  YG, Rosenthal  ES, et al.  Functional networks reemerge during recovery of consciousness after acute severe traumatic brain injury. Cortex J Devoted Study Nerv Syst Behav. 2018;106:299–308. [ Google Scholar ] 9. Mashour  GA, Hudetz  AG. Neural correlates of unconsciousness in large-scale brain networks. Trends Neurosci. 2018;41(3):150–160. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Raichle  ME. The brain’s default mode network. Annu Rev Neurosci. 2015;38:433–447. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Demertzi  A, Antonopoulos  G, Heine  L, et al.  Intrinsic functional connectivity differentiates minimally conscious from unresponsive patients. Brain J Neurol. 2015;138(Pt 9):2619–2631. [ Google Scholar ] 12. Guo  Y, Cao  B, He  Y, et al.  Disrupted multi-scale topological organization of directed functional brain networks in patients with disorders of consciousness. Brain Commun. 2023;5(2):fcad069. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Norton  L, Hutchison  RM, Young  GB, Lee  DH, Sharpe  MD, Mirsattari  SM. Disruptions of functional connectivity in the default mode network of comatose patients. Neurology. 2012;78(3):175–181. [ DOI ] [ PubMed ] [ Google Scholar ] 14. Qin  P, Wu  X, Huang  Z, et al.  How are different neural networks related to consciousness?  Ann Neurol. 2015;78(4):594–605. [ DOI ] [ PubMed ] [ Google Scholar ] 15. Demertzi  A, Gómez  F, Crone  JS, et al.  Multiple fMRI system-level baseline connectivity is disrupted in patients with consciousness alterations. Cortex. 2014;52:35–46. [ DOI ] [ PubMed ] [ Google Scholar ] 16. Shine  JM. The thalamus integrates the macrosystems of the brain to facilitate complex, adaptive brain network dynamics. Prog Neurobiol. 2021;199:101951. [ DOI ] [ PubMed ] [ Google Scholar ] 17. Clascá  F, Rubio-Garrido  P, Jabaudon  D. Unveiling the diversity of thalamocortical neuron subtypes. Eur J Neurosci. 2012;35(10):1524–1532. [ DOI ] [ PubMed ] [ Google Scholar ] 18. Hudetz  AG, Mashour  GA. Disconnecting consciousness: Is there a common anesthetic end-point?  Anesth Analg. 2016;123(5):1228–1240. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Akeju  O, Loggia  ML, Catana  C, et al.  Disruption of thalamic functional connectivity is a neural correlate of dexmedetomidine-induced unconsciousness. eLife. 2014;3:e04499. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Shine  JM, Lewis  LD, Garrett  DD, Hwang  K. The impact of the human thalamus on brain-wide information processing. Nat Rev Neurosci. 2023;24(7):416–430. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Ching  S, Brown  EN. Modeling the dynamical effects of anesthesia on brain circuits. Curr Opin Neurobiol. 2014;25:116–122. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Lutkenhoff  ES, Chiang  J, Tshibanda  L, et al.  Thalamic and extrathalamic mechanisms of consciousness after severe brain injury. Ann Neurol. 2015;78(1):68–76. [ DOI ] [ PubMed ] [ Google Scholar ] 23. Fridman  EA, Schiff  ND. Organizing a rational approach to treatments of disorders of consciousness using the anterior forebrain mesocircuit model. J Clin Neurophysiol Off Publ Am Electroencephalogr Soc. 2022;39(1):40–48. [ Google Scholar ] 24. Schiff  ND, Giacino  JT, Kalmar  K, et al.  Behavioural improvements with thalamic stimulation after severe traumatic brain injury. Nature. 2007;448(7153):600–603. [ DOI ] [ PubMed ] [ Google Scholar ] 25. Fischer  DB, Boes  AD, Demertzi  A, et al.  A human brain network derived from coma-causing brainstem lesions. Neurology. 2016;87(23):2427–2434. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 26. Coppola  P, Allanson  J, Naci  L, et al.  The complexity of the stream of consciousness. Commun Biol. 2022;5(1):1173. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Spindler  LRB, Luppi  AI, Adapa  RM, et al.  Dopaminergic brainstem disconnection is common to pharmacological and pathological consciousness perturbation. Proc Natl Acad Sci U S A. 2021;118(30):e2026289118. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Luppi  AI, Craig  MM, Pappas  I, et al.  Consciousness-specific dynamic interactions of brain integration and functional diversity. Nat Commun. 2019;10(1):4616. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Halassa  MM, Sherman  SM. Thalamocortical circuit motifs: A general framework. Neuron. 2019;103(5):762–770. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Hwang  K, Bertolero  MA, Liu  WB, D’Esposito  M. The human thalamus is an integrative hub for functional brain networks. J Neurosci. 2017;37(23):5594–5607. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Vertes  RP, Linley  SB, Hoover  WB. Limbic circuitry of the midline thalamus. Neurosci Biobehav Rev. 2015;54:89–107. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Rikhye  RV, Wimmer  RD, Halassa  MM. Toward an integrative theory of thalamic function. Annu Rev Neurosci. 2018;41(1):163–183. [ DOI ] [ PubMed ] [ Google Scholar ] 33. Jones  EG. The thalamic matrix and thalamocortical synchrony. Trends Neurosci. 2001;24(10):595–601. [ DOI ] [ PubMed ] [ Google Scholar ] 34. Sherman  SM. The thalamus is more than just a relay. Curr Opin Neurobiol. 2007;17(4):417–422. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 35. Shine  JM, Bissett  PG, Bell  PT, et al.  The dynamics of functional brain networks: Integrated network states during cognitive task performance. Neuron. 2016;92(2):544–554. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 36. Wimmer  RD, Schmitt  LI, Davidson  TJ, Nakajima  M, Deisseroth  K, Halassa  MM. Thalamic control of sensory selection in divided attention. Nature. 2015;526(7575):705–709. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 37. Saalmann  YB, Pinsk  MA, Wang  L, Li  X, Kastner  S. The pulvinar regulates information transmission between cortical areas based on attention demands. Science. 2012;337(6095):753–756. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 38. Redinbaugh  MJ, Phillips  JM, Kambi  NA, et al.  Thalamus modulates consciousness via layer-specific control of cortex. Neuron. 2020;106(1):66–75.e12. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Aru  J, Suzuki  M, Larkum  ME. Cellular mechanisms of conscious processing. Trends Cogn Sci. 2020;24(10):814–825. [ DOI ] [ PubMed ] [ Google Scholar ] 40. Tasserie  J, Uhrig  L, Sitt  JD, et al.  Deep brain stimulation of the thalamus restores signatures of consciousness in a nonhuman primate model. Sci Adv. 2022;8(11):eabl5547. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Demertzi  A, Tagliazucchi  E, Dehaene  S, et al.  Human consciousness is supported by dynamic complex patterns of brain signal coordination. Sci Adv. 2019;5(2):eaat7603. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Weiner  VS, Zhou  DW, Kahali  P, et al.  Propofol disrupts alpha dynamics in functionally distinct thalamocortical networks during loss of consciousness. Proc Natl Acad Sci U S A. 2023;120(11):e2207831120. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Setzer  B, Fultz  NE, Gomez  DEP, et al.  A temporal sequence of thalamic activity unfolds at transitions in behavioral arousal state. Nat Commun. 2022;13(1):5442. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 44. Liu  X, Lauer  KK, Ward  BD, Li  S-J, Hudetz  AG. Differential effects of deep sedation with propofol on the specific and nonspecific thalamocortical systems: A functional magnetic resonance imaging study. Anesthesiology. 2013;118(1):59–69. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Whyte  CJ, Redinbaugh  MJ, Shine  JM, Saalmann  YB. Thalamic contributions to the state and contents of consciousness. Neuron. 2024;112(10):1611–1625. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 46. Schiff  N, Giacino  J, Butson  C, et al.  Thalamic deep brain stimulation in traumatic brain injury: A phase 1, randomized feasibility study. Nat Med. 2023;29(12):3162–3174. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Karas  PJ, Lee  S, Jimenez-Shahed  J, Goodman  WK, Viswanathan  A, Sheth  SA. Deep brain stimulation for obsessive compulsive disorder: Evolution of surgical stimulation target parallels changing model of dysfunctional brain circuits. Front Neurosci. 2019;12:998. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 48. Cury  RG, Fraix  V, Castrioto  A, et al.  Thalamic deep brain stimulation for tremor in Parkinson disease, essential tremor, and dystonia. Neurology. 2017;89(13):1416–1423. [ DOI ] [ PubMed ] [ Google Scholar ] 49. Soulier  H, Pizzo  F, Jegou  A, et al.  The anterior and pulvinar thalamic nuclei interactions in mesial temporal lobe seizure networks. Clin Neurophysiol Off J Int Fed Clin Neurophysiol. 2023;150:176–183. [ Google Scholar ] 50. Wu  TQ, Kaboodvand  N, McGinn  RJ, et al.  Multisite thalamic recordings to characterize seizure propagation in the human brain. Brain J Neurol. 2023;146(7):2792–2802. [ Google Scholar ] 51. Saranathan  M, Iglehart  C, Monti  M, Tourdias  T, Rutt  B. In vivo high-resolution structural MRI-based atlas of human thalamic nuclei. Sci Data. 2021;8(1):275. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 52. Mai  JK, Majtanik  M. Toward a common terminology for the thalamus. Front Neuroanat. 2019;12:114. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 53. Najdenovska  E, Alemán-Gómez  Y, Battistella  G, et al.  In-vivo probabilistic atlas of human thalamic nuclei based on diffusion- weighted magnetic resonance imaging. Sci Data. 2018;5:180270. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 54. Iglesias  JE, Insausti  R, Lerma-Usabiaga  G, et al.  A probabilistic atlas of the human thalamic nuclei combining ex vivo MRI and histology. NeuroImage. 2018;183:314–326. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 55. Kumar  V, Mang  S, Grodd  W. Direct diffusion-based parcellation of the human thalamus. Brain Struct Funct. 2015;220(3):1619–1635. [ DOI ] [ PubMed ] [ Google Scholar ] 56. Zhang  D, Snyder  AZ, Fox  MD, Sansbury  MW, Shimony  JS, Raichle  ME. Intrinsic functional relations between human cerebral cortex and thalamus. J Neurophysiol. 2008;100(4):1740–1748. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 57. Behrens  TEJ, Johansen-Berg  H, Woolrich  MW, et al.  Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nat Neurosci. 2003;6(7):750–757. [ DOI ] [ PubMed ] [ Google Scholar ] 58. Luppi  AI, Vohryzek  J, Kringelbach  ML, et al.  Distributed harmonic patterns of structure-function dependence orchestrate human consciousness. Commun Biol. 2023;6(1):117. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 59. Naci  L, Haugg  A, MacDonald  A, et al.  Functional diversity of brain networks supports consciousness and verbal intelligence. Sci Rep. 2018;8(1):13259. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 60. Luppi  AI, Mediano  PAM, Rosas  FE, et al.  Whole-brain modelling identifies distinct but convergent paths to unconsciousness in anaesthesia and disorders of consciousness. Commun Biol. 2022;5(1):384. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 61. Varley  TF, Luppi  AI, Pappas  I, et al.  Consciousness & brain functional complexity in propofol anaesthesia. Sci Rep. 2020;10(1):1018. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 62. Edlow  BL, Chatelle  C, Spencer  CA, et al.  Early detection of consciousness in patients with acute severe traumatic brain injury. Brain. 2017;140(9):2399–2414. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 63. Fernández-Espejo  D, Norton  L, Owen  AM. The clinical utility of fMRI for identifying covert awareness in the vegetative state: A comparison of sensitivity between 3T and 1.5T. PLoS One. 2014;9(4):e95082. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 64. Monti  MM, Vanhaudenhuyse  A, Coleman  MR, et al.  Willful modulation of brain activity in disorders of consciousness. N Engl J Med. 2010;362(7):579–589. [ DOI ] [ PubMed ] [ Google Scholar ] 65. Owen  AM, Coleman  MR, Boly  M, Davis  MH, Laureys  S, Pickard  JD. Detecting awareness in the vegetative state. Science. 2006;313(5792):1402–1402. [ DOI ] [ PubMed ] [ Google Scholar ] 66. Craig  MM, Pappas  I, Allanson  J, et al. Resting-state based prediction of task-related activation in patients with disorders of consciousness. bioRxiv . [Preprint] doi: 10.1101/2021.03.27.436534 [ DOI ] 67. Calhoun  VD, Wager  TD, Krishnan  A, et al.  The impact of T1 versus EPI spatial normalization templates for fMRI data analyses. Hum Brain Mapp. 2017;38(11):5331–5342. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 68. Ashburner  J, Friston  KJ. Unified segmentation. NeuroImage. 2005;26(3):839–851. [ DOI ] [ PubMed ] [ Google Scholar ] 69. Whitfield-Gabrieli  S, Nieto-Castanon  A. Conn: A functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect. 2012;2(3):125–141. [ DOI ] [ PubMed ] [ Google Scholar ] 70. Power  JD, Plitt  M, Gotts  SJ, et al.  Ridding fMRI data of motion-related influences: Removal of signals with distinct spatial and physical bases in multiecho data. Proc Natl Acad Sci U S A. 2018;115(9):E2105–E2114. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 71. Weiler  M, Casseb  RF, de Campos  BM, et al.  Evaluating denoising strategies in resting-state functional magnetic resonance in traumatic brain injury (EpiBioS4Rx). Hum Brain Mapp. 2022;43(15):4640–4649. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 72. Crone  JS, Lutkenhoff  ES, Vespa  PM, Monti  MM. A systematic investigation of the association between network dynamics in the human brain and the state of consciousness. Neurosci Conscious. 2020;2020(1):niaa008. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 73. Parkes  L, Fulcher  B, Yücel  M, Fornito  A. An evaluation of the efficacy, reliability, and sensitivity of motion correction strategies for resting-state functional MRI. NeuroImage. 2018;171:415–436. [ DOI ] [ PubMed ] [ Google Scholar ] 74. Battistella  G, Najdenovska  E, Maeder  P, et al.  Robust thalamic nuclei segmentation method based on local diffusion magnetic resonance properties. Brain Struct Funct. 2017;222(5):2203–2216. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 75. Morel  A, Magnin  M, Jeanmonod  D. Multiarchitectonic and stereotactic atlas of the human thalamus. J Comp Neurol. 1997;387(4):588–630. [ DOI ] [ PubMed ] [ Google Scholar ] 76. Woodrow  RE, Winzeck  S, Luppi  AI, et al.  Acute thalamic connectivity precedes chronic post-concussive symptoms in mild traumatic brain injury. Brain. 2023;146(8):3484–3499. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 77. Eklund  A, Nichols  TE, Knutsson  H. Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. Proc Natl Acad Sci U S A. 2016;113(28):7900–7905. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 78. Smith  SM, Fox  PT, Miller  KL, et al.  Correspondence of the brain’s functional architecture during activation and rest. Proc Natl Acad Sci U S A. 2009;106(31):13040–13045. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 79. Kozák  LR, van Graan  LA, Chaudhary  UJ, Szabó  ÁG, Lemieux  L. ICN_Atlas: Automated description and quantification of functional MRI activation patterns in the framework of intrinsic connectivity networks. NeuroImage. 2017;163:319–341. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 80. Kolde  R. Pheatmap: Pretty heatmaps. R package version 1.0.13. 2025. https://github.com/raivokolde/pheatmap 81. Shepherd  GMG, Yamawaki  N. Untangling the cortico-thalamo-cortical loop: Cellular pieces of a knotty circuit puzzle. Nat Rev Neurosci. 2021;22(7):389–406. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 82. McFarland  NR, Haber  SN. Thalamic relay nuclei of the basal ganglia form both reciprocal and nonreciprocal cortical connections, linking multiple frontal cortical areas. J Neurosci Off J Soc Neurosci. 2002;22(18):8117–8132. [ Google Scholar ] 83. Schnakers  C, Vanhaudenhuyse  A, Giacino  J, et al.  Diagnostic accuracy of the vegetative and minimally conscious state: Clinical consensus versus standardized neurobehavioral assessment. BMC Neurol. 2009;9(1):35. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 84. Bodien  YG, Allanson  J, Cardone  P, et al.  Cognitive motor dissociation in disorders of consciousness. N Engl J Med. 2024;391(7):598–608. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 85. Wang  J, Hu  X, Hu  Z, Sun  Z, Laureys  S, Di  H. The misdiagnosis of prolonged disorders of consciousness by a clinical consensus compared with repeated coma-recovery scale-revised assessment. BMC Neurol. 2020;20(1):343. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 86. Peterson  A, Cruse  D, Naci  L, Weijer  C, Owen  AM. Risk, diagnostic error, and the clinical science of consciousness. NeuroImage Clin. 2015;7:588–597. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 87. Homman-Ludiye  J, Bourne  JA. The medial pulvinar: Function, origin and association with neurodevelopmental disorders. J Anat. 2019;235(3):507–520. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 88. Filipescu  C, Lagarde  S, Lambert  I, et al.  The effect of medial pulvinar stimulation on temporal lobe seizures. Epilepsia. 2019;60(4):e25–e30. [ DOI ] [ PubMed ] [ Google Scholar ] 89. Deutschová  B, Pizzo  F, Giusiano  B, et al.  Ictal connectivity changes induced by pulvinar stimulation correlate with improvement of awareness. Brain Stimulat. 2021;14(2):344–346. [ Google Scholar ] 90. Ward  R, Danziger  S, Owen  V, Rafal  R. Deficits in spatial coding and feature binding following damage to spatiotopic maps in the human pulvinar. Nat Neurosci. 2002;5(2):99–100. [ DOI ] [ PubMed ] [ Google Scholar ] 91. Karnath  H-O, Himmelbach  M, Rorden  C. The subcortical anatomy of human spatial neglect: Putamen, caudate nucleus and pulvinar. Brain. 2002;125(2):350–360. [ DOI ] [ PubMed ] [ Google Scholar ] 92. Wolff  M, Vann  SD. The cognitive thalamus as a gateway to mental representations. J Neurosci Off J Soc Neurosci. 2019;39(1):3–14. [ Google Scholar ] 93. Zheng  ZS, Monti  MM. Cortical and thalamic connections of the human globus pallidus: Implications for disorders of consciousness. Front Neuroanat. 2022;16:960439. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 94. de Bourbon-Teles  J, Bentley  P, Koshino  S, et al.  Thalamic control of human attention driven by memory and learning. Curr Biol. 2014;24(9):993–999. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 95. Edlow  BL, Barra  ME, Zhou  DW, et al.  Personalized connectome mapping to guide targeted therapy and promote recovery of consciousness in the intensive care unit. Neurocrit Care. 2020;33(2):364–375. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 96. Menon  V. 20 years of the default mode network: A review and synthesis. Neuron. 2023;111(16):2469–2487. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 97. Bodien  YG, Threlkeld  ZD, Edlow  BL. Default mode network dynamics in covert consciousness. Cortex J Devoted Study Nerv Syst Behav. 2019;119:571–574. [ Google Scholar ] 98. Yeshurun  Y, Nguyen  M, Hasson  U. The default mode network: Where the idiosyncratic self meets the shared social world. Nat Rev Neurosci. 2021;22(3):181–192. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 99. Demertzi  A, Soddu  A, Laureys  S. Consciousness supporting networks. Curr Opin Neurobiol. 2013;23(2):239–244. [ DOI ] [ PubMed ] [ Google Scholar ] 100. Lyu  D, Pappas  I, Menon  DK, Stamatakis  EA. A precuneal causal loop mediates external and internal information integration in the human brain. J Neurosci. 2021;41(48):9944–9956. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 101. Lyu  D, Stieger  JR, Xin  C, et al.  Causal evidence for the processing of bodily self in the anterior precuneus. Neuron. 2023;111(16):2502–2512.e4. [ DOI ] [ PubMed ] [ Google Scholar ] 102. Parvizi  J, Braga  RM, Kucyi  A, et al.  Altered sense of self during seizures in the posteromedial cortex. Proc Natl Acad Sci U S A. 2021;118(29):e2100522118. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 103. Herbet  G, Lafargue  G, de Champfleur  NM, et al.  Disrupting posterior cingulate connectivity disconnects consciousness from the external environment. Neuropsychologia. 2014;56:239–244. [ DOI ] [ PubMed ] [ Google Scholar ] 104. Cui  Y, Song  M, Lipnicki  DM, et al.  Subdivisions of the posteromedial cortex in disorders of consciousness. NeuroImage Clin. 2018;20:260–266. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 105. Silva  S, de Pasquale  F, Vuillaume  C, et al.  Disruption of posteromedial large-scale neural communication predicts recovery from coma. Neurology. 2015;85(23):2036–2044. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 106. Zhang  S, Li  C-SR. Functional connectivity mapping of the human precuneus by resting state fMRI. Neuroimage. 2012;59(4):3548–3562. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 107. Cunningham  SI, Tomasi  D, Volkow  ND. Structural and functional connectivity of the precuneus and thalamus to the default mode network. Hum Brain Mapp. 2017;38(2):938–956. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 108. Alves  PN, Foulon  C, Karolis  V, et al.  An improved neuroanatomical model of the default-mode network reconciles previous neuroimaging and neuropathological findings. Commun Biol. 2019;2(1):370. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 109. Lee  T-W, Xue  S-W. Functional connectivity maps based on hippocampal and thalamic dynamics may account for the default-mode network. Eur J Neurosci. 2018;47(5):388–398. [ DOI ] [ PubMed ] [ Google Scholar ] 110. Parvizi  J, Van Hoesen  GW, Buckwalter  J, Damasio  A. Neural connections of the posteromedial cortex in the macaque. Proc Natl Acad Sci U S A. 2006;103(5):1563–1568. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 111. Cauda  F, Geminiani  G, D’Agata  F, et al.  Functional connectivity of the posteromedial Cortex. PLoS One. 2010;5(9):e13107. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 112. Edlow  BL, Olchanyi  M, Freeman  HJ, et al.  Multimodal MRI reveals brainstem connections that sustain wakefulness in human consciousness. Sci Transl Med. 2024;16(745):eadj4303. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 113. Crone  JS, Soddu  A, Höller  Y, et al.  Altered network properties of the fronto-parietal network and the thalamus in impaired consciousness. NeuroImage Clin. 2013;4:240–248. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 114. Vanhaudenhuyse  A, Noirhomme  Q, Tshibanda  LJF, et al.  Default network connectivity reflects the level of consciousness in non-communicative brain-damaged patients. Brain. 2010;133(1):161–171. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 115. Fernández-Espejo  D, Soddu  A, Cruse  D, et al.  A role for the default mode network in the bases of disorders of consciousness. Ann Neurol. 2012;72(3):335–343. [ DOI ] [ PubMed ] [ Google Scholar ] 116. Laureys  S, Goldman  S, Phillips  C, et al.  Impaired effective cortical connectivity in vegetative state: Preliminary investigation using PET. NeuroImage. 1999;9(4):377–382. [ DOI ] [ PubMed ] [ Google Scholar ] 117. Laureys  S, Faymonville  ME, Luxen  A, Lamy  M, Franck  G, Maquet  P. Restoration of thalamocortical connectivity after recovery from persistent vegetative state. The Lancet. 2000;355(9217):1790–1791. [ Google Scholar ] 118. Seeley  WW, Menon  V, Schatzberg  AF, et al.  Dissociable intrinsic connectivity networks for salience processing and executive control. J Neurosci. 2007;27(9):2349–2356. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 119. Danielson  NB, Guo  JN, Blumenfeld  H. The default mode network and altered consciousness in epilepsy. Behav Neurol. 2011;24(1):55–65. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 120. Blumenfeld  H, Varghese  GI, Purcaro  MJ, et al.  Cortical and subcortical networks in human secondarily generalized tonic–clonic seizures. Brain. 2009;132(4):999–1012. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 121. Wu  X, Zou  Q, Hu  J, et al.  Intrinsic functional connectivity patterns predict consciousness level and recovery outcome in acquired brain injury. J Neurosci. 2015;35(37):12932–12946. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 122. Demertzi  A, Kucyi  A, Ponce-Alvarez  A, Keliris  GA, Whitfield-Gabrieli  S, Deco  G. Functional network antagonism and consciousness. Netw Neurosci. 2022;6(4):998–1009. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 123. Müller  EJ, Munn  B, Hearne  LJ, et al.  Core and matrix thalamic sub-populations relate to spatio-temporal cortical connectivity gradients. NeuroImage. 2020;222:117224. [ DOI ] [ PubMed ] [ Google Scholar ] 124. Müller  EJ, Munn  BR, Redinbaugh  MJ, et al.  The non-specific matrix thalamus facilitates the cortical information processing modes relevant for conscious awareness. Cell Rep. 2023;42(8):112844. [ DOI ] [ PubMed ] [ Google Scholar ] 125. Huang  Z, Mashour  GA, Hudetz  AG. Propofol disrupts the functional core-matrix architecture of the thalamus in humans.  Nat Commun. 2024;15(1):7496. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 126. Craig  MM, Misic  B, Pappas  I, Adapa  RM, Menon  DK, Stamatakis  EA. Propofol sedation-induced alterations in brain connectivity reflect parvalbumin interneurone distribution in human cerebral cortex. Br J Anaesth. 2021;126(4):835–844. [ DOI ] [ PubMed ] [ Google Scholar ] 127. Mashour  GA. Anesthesia and the neurobiology of consciousness. Neuron. 2024;112(10):1553–1567. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 128. Bastos  AM, Donoghue  JA, Brincat  SL, et al.  Neural effects of propofol-induced unconsciousness and its reversal using thalamic stimulation. eLife. 2021;10:e60824. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 129. Lee  M, Sanz  LRD, Barra  A, et al.  Quantifying arousal and awareness in altered states of consciousness using interpretable deep learning. Nat Commun. 2022;13(1):1064. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 130. Koch  C, Massimini  M, Boly  M, Tononi  G. Neural correlates of consciousness: Progress and problems. Nat Rev Neurosci. 2016;17(5):307–321. [ DOI ] [ PubMed ] [ Google Scholar ] 131. Tononi  G, Boly  M, Massimini  M, Koch  C. Integrated information theory: From consciousness to its physical substrate. Nat Rev Neurosci. 2016;17(7):450–461. [ DOI ] [ PubMed ] [ Google Scholar ] 132. Schiff  ND. Recovery of consciousness after brain injury: A mesocircuit hypothesis. Trends Neurosci. 2010;33(1):1–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 133. Schiff  ND. Moving toward a generalizable application of central thalamic deep brain stimulation for support of forebrain arousal regulation in the severely injured brain. Ann N Y Acad Sci. 2012;1265(1):56–68. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 134. Wyder  MT, Massoglia  DP, Stanford  TR. Contextual modulation of central thalamic delay-period activity: Representation of visual and saccadic goals. J Neurophysiol. 2004;91(6):2628–2648. [ DOI ] [ PubMed ] [ Google Scholar ] 135. Schlag  J, Schlag-Rey  M. Visuomotor functions of central thalamus in monkey. II. Unit activity related to visual events, targeting, and fixation. J Neurophysiol. 1984;51(6):1175–1195. [ DOI ] [ PubMed ] [ Google Scholar ] 136. Huang  Z, Zhang  J, Wu  J, Mashour  GA, Hudetz  AG. Temporal circuit of macroscale dynamic brain activity supports human consciousness. Sci Adv. 2020;6(11):eaaz0087. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 137. Raguž  M, Predrijevac  N, Dlaka  D, et al.  Structural changes in brains of patients with disorders of consciousness treated with deep brain stimulation. Sci Rep. 2021;11(1):4401. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 138. Thibaut  A, Schiff  N, Giacino  J, Laureys  S, Gosseries  O. Therapeutic interventions in patients with prolonged disorders of consciousness. Lancet Neurol. 2019;18(6):600–614. [ DOI ] [ PubMed ] [ Google Scholar ] 139. Provencio  JJ, Hemphill  JC, Claassen  J, et al.  The curing coma campaign: Framing initial scientific challenges — proceedings of the first curing coma campaign scientific advisory council meeting. Neurocrit Care. 2020;33(1):1–12. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 140. Zelmann  R, Paulk  AC, Tian  F, et al.  Differential cortical network engagement during states of un/consciousness in humans. Neuron. 2023;111(21):3479–3495.e6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 141. Eikermann  M, Akeju  O, Chamberlin  NL. Sleep and anesthesia: The shared circuit hypothesis has been put to bed. Curr Biol. 2020;30(5):R219–R221. [ DOI ] [ PubMed ] [ Google Scholar ] 142. Ozdemir  RA, Tadayon  E, Boucher  P, et al.  Individualized perturbation of the human connectome reveals reproducible biomarkers of network dynamics relevant to cognition. Proc Natl Acad Sci U S A. 2020;117(14):8115–8125. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 143. Thibaut  A, Bruno  M-A, Ledoux  D, Demertzi  A, Laureys  S. tDCS in patients with disorders of consciousness. Neurology. 2014;82(13):1112–1118. [ DOI ] [ PubMed ] [ Google Scholar ] 144. Gottshall  JL, Adams  ZM, Forgacs  PB, Schiff  ND. Daytime central thalamic deep brain stimulation modulates sleep dynamics in the severely injured brain: Mechanistic insights and a novel framework for alpha-delta sleep generation. Front Neurol. 2019;10:20. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 145. Cain  JA, Spivak  NM, Coetzee  JP, et al.  Ultrasonic deep brain neuromodulation in acute disorders of consciousness: A proof-of-concept. Brain Sci. 2022;12(4):428. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 146. Monti  MM, Schnakers  C, Korb  AS, Bystritsky  A, Vespa  PM. Non-invasive ultrasonic thalamic stimulation in disorders of consciousness after severe brain injury: A first-in-man report. Brain Stimulat. 2016;9(6):940–941. [ Google Scholar ] 147. Arnts  H, Tewarie  P, van Erp  WS, et al.  Clinical and neurophysiological effects of central thalamic deep brain stimulation in the minimally conscious state after severe brain injury. Sci Rep. 2022;12(1):12932. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 148. Arnts  H, Tewarie  P, van Erp  W, et al.  Deep brain stimulation of the central thalamus restores arousal and motivation in a zolpidem-responsive patient with akinetic mutism after severe brain injury. Sci Rep. 2024;14(1):2950. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 149. Yang  AI, Isbaine  F, Alwaki  A, Gross  RE. Multitarget deep brain stimulation for epilepsy. J Neurosurg. 2023;140(1):210–217. [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials fcag021_Supplementary_Data fcag021_supplementary_data.pdf (2.8MB, pdf) Data Availability Statement Access to DOC patient data is limited to qualified researchers for non-commercial use to protect patient confidentiality. The UK Health Research Authority assigns responsibility for safeguarding the data to the Chief Investigators of the original studies (Dr. Judith Allanson and Prof. David Menon, or anyone else to whom responsibility is given). To request access, please contact the Data Access Committee: Dr. Judith Allanson ( [email protected] ), Prof. David Menon ( [email protected] ) or Dr. Emmanuel Stamatakis ( [email protected] ). The propofol dataset is publicly accessible via the OpenNeuro data repository (doi: 10.18112/openneuro.ds003171.v2.0.1 ). The codes produced and used for this work have been uploaded at https://github.com/dorottyaszocs1/Thalamus_in_-un-consciousness_in_pharmacological_and_pathological_states . Articles from Brain Communications are provided here courtesy of Oxford University Press ACTIONS View on publisher site PDF (10.9 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top