Thalamic contributions to the state and contents of consciousness

Abstract

Consciousness can be conceptualized as varying along at least two dimensions: the global state of consciousness and the content of conscious experience. Here, we highlight the cellular and systems-level contributions of the thalamus to conscious state and then argue for thalamic contributions to conscious content, including the integrated, segregated, and continuous nature of our experience. We underscore vital, yet distinct roles for core- and matrix-type thalamic neurons. Through reciprocal interactions with deep-layer cortical neurons, matrix neurons support wakefulness and determine perceptual thresholds, whereas the cortical interactions of core neurons maintain content and enable perceptual constancy. We further propose that conscious integration, segregation, and continuity depend on the convergent nature of corticothalamic projections enabling dimensionality reduction, a thalamic reticular nucleus-mediated divisive normalization-like process, and sustained coherent activity in thalamocortical loops, respectively. Overall, we conclude that the thalamus plays a central topological role in brain structures controlling conscious experience.

Keywords: conscious state; core; matrix; perception; thalamocortical.

Figure 1.. The micro-scale, meso-scale, and macro-scale organization of the thalamocortical system.

A) At the micro-scale, distinct excitatory cell types in the thalamus (core [olive] and matrix [plum]) project to distinct layers (numbered 1–6) of the cerebral cortex, wherein they make contact with specific cell-type populations that have different impacts on cortical computations. Core cells send targeted projections to stellate cells (L4, light orange) in the middle (granular) layers, which then innervate superficial (L2/3, tangerine) and deep (L6A, teal) pyramidal neurons, which either project back to the thalamus (L6A) or within the cerebral cortex (L2/3). In contrast, matrix cells innervate superficial (L1) and deep (L5) layers, with particular targeting of large, thick-tufted pyramidal tract (PT)-type pyramidal neurons (ttL5B, blue) that are the sole cortical output beyond the midbrain. Both ttL5B and matrix cells are also strongly connected with orexin-sensitive layer 6 pyramidal neurons (L6B, teal). There is also extensive inhibitory control over thalamic spiking: e.g., the reticular nucleus of the thalamus (RTn, dark blue), which is excited by thalamic projection neurons and cortical inputs to the thalamus (not shown), and the globus pallidus internus (GPi), which is gated by ttL5B inputs to the globus pallidus externus (GPe, medium gray); NB - this is an idealized diagram intended to convey a general overview of the main circuits, and hence not all known connections are shown. B) The different nuclei of the thalamus can be characterized according to: left - First-Order (light orange) and Higher-Order (brown); whether driver inputs come from the subcortex or cerebral cortex; right - the expression of parvalbumin (in core cells) and calbindin (in matrix cells) – note that individual nuclei contain a blend of both cell-types (denoted approximately by relative color intensity; distribution within individual nuclei not depicted). C) At the meso-scale, the thalamus is proposed to control the local (core) and long-range (matrix) excitability of distributed regions of the cerebral cortex through projections that augment local or distributed resonance in corticothalamic loops across the cortical mantle. D) At the macroscale, the thalamus is deeply interconnected with the entire cortical mantle, with axonal connections originating in both the cerebral cortex and thalamus that together form a distributed network that shapes conscious experience. Of particular importance is the presence of both convergent (multiple cortical areas projecting to one thalamic area) and divergent (single thalamic areas projecting to multiple cortical areas) architectures within the same system. The hypothalamus (dot-dashed line) projects to multiple hubs within the ascending arousal system, wherein specialized cells release a variety of neuromodulatory neurochemicals (dopamine, light blue; acetylcholine, green; noradrenaline, dark red; serotonin, purple). Key: CL - central lateral, dLGN - dorsal lateral geniculate nucleus, MD - mediodorsal, Pulv - pulvinar, RTN - reticular thalamic nucleus, VA - ventral anterior, and VL - ventral lateral nucleus.

Figure 2.. Matrix thalamic inputs interface with the cerebral cortex at different scales to alter conscious state.

(A) Thalamocortical circuits implicated in conscious state. Matrix thalamic cells (purple) are important targets of the reticular activating system. They reciprocally connect to ttL5B (dark blue) neurons, and are ideally positioned to modulate their excitability. Orexin-gated L6B neurons further modulate these thalamocortical interactions. (B) During wakefulness, matrix thalamic inputs excite L5 pyramidal neurons via metabotropic glutamatergic inputs to the oblique dendrites, allowing calcium spikes in the apical dendrites to influence somatic activity. Under general anesthesia, in the absence of thalamic excitation, L5 pyramidal neurons fail to propagate signals from the dendrites through to the soma, contributing to functional dissociation of brain areas. Figure adapted with permission from . (C) Minimally sufficient mechanism for CL thalamic DBS to restore or disrupt consciousness. Top) Thalamocortical (TC), feedforward (FF), and feedback (FB) connections most altered by thalamic stimulations that successfully reversed general anesthesia (propofol and isoflurane) in macaques. Higher frequency (50Hz) thalamic DBS simultaneously via 16 contacts of a linear array (200um spacing) centered in CL mimics wake-state firing (yellow) from thalamocortical efferents. This increases firing in deep cortical layers (yellow) which reinstates intra-columnar, feedforward, and feedback communication at alpha and gamma frequencies, despite inhibitory pressure from anesthesia. Adapted with permission from. Bottom) Low (10Hz) and very high (200Hz) frequency DBS centered on CL in awake macaques instead produces a vacant state similar to the symptoms of absence epilepsy. Low activity in thalamus and deep cortical layers (blue) leads to perturbed connectivity in parietal intra-columnar circuits and disrupted feedforward communication (light shading). Based on results from. (D) A large-scale corticothalamic neural mass model (imbued with matrix and core thalamic cells) that replicated recovery of consciousness with thalamic stimulation following propofol anesthesia in macaques. Simulated stimulation of matrix, but not core cells, increased cortical excitability, allowing interareal information transfer and more complex, less stereotyped, dynamics. At the macroscale this was reflected in a flattening of the whole-brain energy landscape that is typical of the waking state. Figure reproduced with permission from.

Figure 3.. Thalamic contributions to conscious contents.

(A) Thalamocortical circuit hypothesized to support the contents of consciousness. Matrix thalamic cells (purple) project to the apical dendrites of ttL5B (dark blue) and L2/3 (orange) pyramidal cells, and the oblique dendrites of ttL5B cells in L5A. The projections to the oblique dendrites depolarise the trunk of the cell through the action of metabotropic glutamatergic (and cholinergic) receptors which has been hypothesized and shown to facilitate coupling between the apical and somatic compartments allowing sodium spikes to backpropagate from the soma to the apical dendrites contributing to the generation of calcium spikes and vice versa. Matrix and core (light green) thalamic cells both receive driving inputs from ttL5B cells. Core cells send driving projections to L4 IT type cells and L2/3 pyramidal cells (not shown). L4 IT type cells then form a closed excitatory loop with core cells via L6A neurons (cyan). Similarly, excitatory output from layer 4 cells target the basal dendrites of ttL5B via L2/3. (B) Illustration of the threshold detection paradigm across both studies by Takashi and colleagues ^,. (C) Psychometric function showing response probability as a function of (scaled) stimulus intensity across mice and recording sessions. (D) Example raster plots for the soma of ttL5B cells in hit and miss trials. Sparse bursts (green) occur in both conditions but are time-locked to stimulus onset on hit trials. (E-G) Detection probability in the threshold detection paradigm as a function of stimulus intensity comparing control to (E) striatal inhibition, (F) POm inhibition, or (G) superior colliculus inhibition from. (H) Generalized flash suppression paradigm used by. Monkeys fixated a central location for 1.5s at which point the target stimulus (red disk) appeared in the visual periphery. 2s following the onset of the target stimulus, a surround pattern appeared rendering the target stimulus invisible with a probability proportional to the density of the pattern. Monkeys were trained to report the disappearance of the stimulus with a lever press. (I) Average firing rate of neurons in the dorsal and ventral pulvinar that were suppressed by the disappearance of the stimulus (red) or were activated by the disappearance of the stimulus (blue). Dashed lines show pulvinar responses for the physical removal of the stimulus, whilst solid lines show the pulvinar responses to the perceived disappearance. Panels B-I recreated with permission from^, and.

Figure 4.. Thalamic contributions to the character of conscious contents.

(A-C) The constraints placed on ongoing dynamics by the organization of the thalamus may help to explain key features of conscious information processing, such as: integration (A), which we argue arises as a product of the dimensionality reduction imposed by the converging inputs to the smaller number of cells in the thalamus (lower row), relative to the cerebral cortex (upper row); segregation (B), which is a natural byproduct of the divisive normalization-like process that arises from activity-dependent recruitment of the inhibitory thalamic reticular nucleus (dark blue); and continuity (C), wherein the combination of driver and modulatory roles in the thalamus imposes a ‘Matthew effect’ on cellular populations that provides an activity boost to neurons that are connected to the currently dominant thalamocortical ensemble. The colors of cells in Panel A-C match those in Figures 1–3. (D) A spiking neural model was created to mimic a key feature of tt5LB cells – namely, that they shift from a regular-spiking (green; input to basal dendrites only) to a burst-firing (orange; simultaneous activation of apical and basal dendrites) mode when gated by matrix thalamic inputs. After fitting the model to electrophysiological data from anesthetized/awake humans and naturally-sleeping macaques, the authors showed that the location in model parameter space that best fit to sleep was associated with predominantly regular spiking activity (green dots), whereas in the awake regime (following the gray dotted square), model tt5LB neurons (embedded on a N = 100×100 cortical sheet) were capable of entering into a burst-firing mode that, in turn, triggered burst-firing in synaptically connected neurons leading to sustained activity that could propagate coherently across the cortical sheet, a process labeled as “connected bursting cascades”. Panel D adapted from with permission. Note that these mechanisms are simply intended to convey plausible, non-exhaustive means for enacting each feature.