Skip to content
Robotic Brain Universe

The thalamus and consciousness

How the thalamus relays and gates signals to the cortex, how consciousness is lost and measured, and what science says about conscious machines.

Advanced · about 9 min · updated 2026-10-02 · awaiting clinical review

See it in 3D:

Illustrative simulation excitatory inhibitory

First-order and higher-order thalamic relays, drivers and modulators, burst and tonic firing and sleep oscillations; the thalamus in arousal and attention; sleep, anaesthesia and disorders of consciousness; thalamic stroke and deep brain stimulation for tremor and the minimally conscious state; Shannon entropy, integrated information and the perturbational complexity index; global workspace and integrated information theories, their 2025 adversarial test, and indicator properties for AI consciousness.

Contents
  1. The gateway and the switch
  2. What the thalamus is
  3. Why route signals through the thalamus
  4. How the thalamus gates the cortex
  5. When consciousness fades
  6. When the thalamus fails
  7. Measuring consciousness
  8. Technology: stimulating and reading consciousness
  9. Milestones
  10. Frontiers: testing theories, and the question of AI
  11. Check yourself

The gateway and the switch

Signals from the eyes and other senses pass through the thalamus on their way to the cortex, and the thalamus also helps decide whether the cortex is awake at all. When researchers stimulated a small part of it, the central lateral thalamus, in anaesthetised monkeys, they restored arousal and wake-like brain activity.[1,2]

This matters at the bedside. Bilateral deep brain stimulation of the central thalamus improved behavioural responsiveness in a patient who had been minimally conscious for six years after a traumatic brain injury, and brain imaging revealed that a patient diagnosed as vegetative could follow instructions to imagine playing tennis, showing preserved awareness.[3,4]

The thalamus sits at the centre of the hardest question in neuroscience: how brain activity becomes experience. This reading covers what the thalamus does, the leading theories of consciousness and the first head-to-head test between them, how consciousness can be measured, and what the science says about whether AI could be conscious.[5,6]

What the thalamus is

The thalamus is a paired nuclear complex made up of many nuclei; the segmentation in this atlas labels it as a whole without distinguishing them. Its blood comes from four main arterial territories: tuberothalamic, paramedian, inferolateral and posterior choroidal.[7,8,9]

Sherman and Guillery distinguish two kinds of relay. First-order relays, such as the lateral geniculate nucleus for vision, carry information from below to the cortex for the first time. Higher-order relays, such as much of the pulvinar, carry information from layer 5 of one cortical area to another, so that a great deal of communication between cortical areas may run through the thalamus.[1]

Key numbers

Share of input to visual relay cells that comes from the retina (the driving input)
5–10%[1]
Thalami with complete or major tremor relief after ventral intermediate nucleus stimulation
38 of 43 (88%)[10]
Years in a minimally conscious state before central thalamic stimulation improved responsiveness
6[3]
Participants in the 2025 adversarial test of consciousness theories
256[5]

Why route signals through the thalamus

Most of a relay cell's input does not come from the sense organ it relays. In the lateral geniculate nucleus only 5–10% of the input to relay cells comes from the retina, the 'driving' input; the rest is modulatory, from local inhibitory cells, from layer 6 of visual cortex and from the brainstem, and they control how the signal is passed on.[1]

The thalamus also coordinates the cortex. Halassa and Kastner review evidence that, besides relaying information, thalamic circuits shift and sustain functional interactions within and between cortical areas, assembling the networks a task needs, which may be critical for attention and cognitive flexibility.[11]

How the thalamus gates the cortex

Two modes. Thalamic relay cells fire either in bursts or in a steady tonic mode, depending on their membrane potential, which modulatory inputs control; the mode relates to the attentional demands of the moment.[1]

Sleep rhythms. In sleep, billions of coupled neurons in thalamocortical systems fire in synchronised, slow oscillations. On waking, neuromodulatory transmitter systems block these slow rhythms, induce fast ones and restore the brain's full responsiveness.[12]

Deep layers and consciousness. Recording simultaneously from the central lateral thalamus and every layer of frontoparietal cortex in macaques, Redinbaugh and colleagues found that thalamic and deep-layer cortical neurons were the most sensitive to changes in the level of consciousness, across anaesthetics and sleep, and that deep-layer activity was sustained by its interaction with the thalamus.[2]

First-order and higher-order thalamic relaysRetinadriving inputLateral geniculate (firstorder)relays subcortical inputVisual cortexlayer 5 output; layer 6 feedbackPulvinar (higher order)relays cortex to cortexHigher cortical areareceives the re-entered signalBrainstemmodulatory input, arousaldriverlayer 5 drivermodulator
First-order and higher-order thalamic relays. Sherman and Guillery's scheme: the lateral geniculate nucleus relays retinal input to cortex for the first time, while higher-order relays such as the pulvinar pass layer-5 output from one cortical area to another. Every relay also receives modulatory input from cortical layer 6 and from the brainstem.[1]
Text version of the diagram
  1. Retina: driving input. Leads to Lateral geniculate (first order) (driver).
  2. Lateral geniculate (first order): relays subcortical input. Leads to Visual cortex.
  3. Visual cortex: layer 5 output; layer 6 feedback. Leads to Pulvinar (higher order) (layer 5 driver).
  4. Pulvinar (higher order): relays cortex to cortex. Leads to Higher cortical area.
  5. Higher cortical area: receives the re-entered signal.
  6. Brainstem: modulatory input, arousal. Leads to Lateral geniculate (first order) (modulator).

When consciousness fades

Sleep. Sleep is characterised by synchronised activity in thalamocortical systems, whose slow oscillations waking abolishes.[12]

Anaesthesia. Anaesthesia reliably produces unresponsiveness and amnesia, but whether it always removes consciousness is harder to establish. Alkire, Hudetz and Tononi argue that unconsciousness follows when anaesthetics disconnect a posterior parietal complex, interrupting cortical communication, or drive the brain into stereotyped responses: in both cases, a loss of the ability to integrate information.[13]

After brain injury. People can move from coma to a vegetative state or a minimally conscious state, which is defined by inconsistent but clearly discernible behavioural evidence of consciousness and can be transient or permanent.[14]

When the thalamus fails

Thalamic stroke. Infarcts produce syndromes that depend on the arterial territory involved, including sensory loss, memory and behavioural change, and impaired arousal.[9]

Disorders of consciousness. Bedside behaviour can underestimate awareness. A patient who met the criteria for the vegetative state, asked to imagine playing tennis or walking around her home, activated the same brain areas as healthy volunteers.[4]

Tremor. The thalamus is also a treatment target. High-frequency stimulation of the ventral intermediate nucleus relieved disabling tremor from Parkinson's disease or essential tremor completely on 27 of 43 treated sides and substantially on 11 more, with reversible side-effects.[10]

Measuring consciousness

Two influential approaches turn the idea that consciousness needs a brain that is both integrated and differentiated into numbers.[15,16]

Shannon entropy[15]
H=−∑ipilog⁡2piH = -\sum_{i} p_i \log_2 p_i

The average information, in bits, of a source whose states occur with probabilities pip_i. Theories of consciousness build on it: a conscious brain is thought to have a very large repertoire of states (high information) that are nevertheless unified (integrated).

Symbols in Shannon entropy
SymbolMeaningUnit
pip_iprobability of state i—
HHentropybits
Integrated information[15]
Φ(S)=EI(A↔B)∣{A,B}=MIB(S)\Phi(S) = \mathrm{EI}\big(A \leftrightarrow B\big)\Big|_{\{A, B\} = \mathrm{MIB}(S)}

In Tononi's integrated information theory, the amount of consciousness of a system SS is the effective information that can be integrated across its weakest link: the bipartition into AA and BB that loses least information (the minimum information bipartition). A system with Φ>0\Phi > 0 that is not part of a subset with higher Φ\Phi is a 'complex'.

Symbols in Integrated information
SymbolMeaningUnit
Φ\Phiintegrated informationbits
EIEIeffective information exchanged between the two parts when each is perturbed—
MIBMIBminimum information bipartition—
Perturbational complexity index[16]
PCI=cL log⁡2LL H(L)\mathrm{PCI} = \frac{c_L \,\log_2 L}{L\, H(L)}

Stimulate the cortex with transcranial magnetic stimulation, record the spread of activity with EEG, convert it to a binary space-time pattern of length LL and measure how compressible it is (cLc_L is its Lempel–Ziv complexity), normalised by its source entropy H(L)H(L). Rich, widespread but differentiated responses score high; PCI distinguished wakefulness from sleep and anaesthesia, and minimally conscious from vegetative patients.

Symbols in Perturbational complexity index
SymbolMeaningUnit
cLc_LLempel–Ziv complexity of the binary response pattern—
LLnumber of entries in the pattern (channels × time samples)—
H(L)H(L)entropy of the pattern's source—

Technology: stimulating and reading consciousness

Deep brain stimulation. In 1991 Benabid and colleagues reported that chronic stimulation of the ventral intermediate thalamus relieved tremor; because stimulation is reversible and adjustable, they considered it preferable to destroying tissue (thalamotomy), especially when both sides needed treatment.[10]

Stimulating awareness. In a six-month double-blind crossover study, central thalamic stimulation increased specific cognitively mediated behaviours, limb control and oral feeding in a minimally conscious patient, compared with periods with the stimulator off.[3]

Reading awareness. Functional MRI command-following and the perturbational complexity index, computed from TMS and EEG, can reveal consciousness in people who cannot show it through behaviour.[4,16]

Two leading theories[5,15,17]
AspectGlobal neuronal workspaceIntegrated information theory
Core claimConscious access is information made globally available through long-range prefronto-parietal networksConsciousness is the capacity of a system to integrate information (Φ)
SignatureLate 'ignition' of a large-scale prefronto-parietal networkSustained, integrated activity, especially in posterior cortex
2025 testChallenged by a general lack of ignition at stimulus offsetChallenged by a lack of sustained synchronisation within posterior cortex

Milestones

The thalamus and consciousness

  1. 1991Chronic thalamic stimulation suppresses tremor.[10]
  2. 1993Thalamocortical oscillations in sleep and waking are reviewed.[12]
  3. 2002The minimally conscious state is defined; first-order and higher-order thalamic relays are distinguished.[1,14]
  4. 2004Integrated information theory proposes Φ as a measure of consciousness.[15]
  5. 2006fMRI reveals awareness in a patient diagnosed as vegetative.[4]
  6. 2007Central thalamic stimulation improves a minimally conscious patient.[3]
  7. 2008Anaesthesia is linked to loss of information integration.[13]
  8. 2011The global neuronal workspace model of conscious access is reviewed.[17]
  9. 2013The perturbational complexity index measures consciousness with TMS and EEG.[16]
  10. 2017The thalamus is recast as a coordinator of cortical networks.[11]
  11. 2020Stimulating the central lateral thalamus wakes anaesthetised monkeys.[2]
  12. 2023Indicator properties are proposed for assessing consciousness in AI systems.[6]
  13. 2025An adversarial collaboration tests two theories of consciousness head to head.[5]

Frontiers: testing theories, and the question of AI

In an adversarial collaboration, proponents of integrated information theory and global neuronal workspace theory agreed predictions in advance; 256 participants were then studied with fMRI, MEG and intracranial EEG. Information about conscious content was found in visual, ventrotemporal and inferior frontal cortex. The results matched some predictions of each theory but substantially challenged key tenets of both.[5]

Could an AI be conscious? Butlin and colleagues derived 'indicator properties' from neuroscientific theories, including recurrent processing, global workspace, higher-order and predictive processing theories, and assessed recent AI systems against them. They concluded that no current AI system is conscious, but that there are no obvious technical barriers to building systems that satisfy the indicators.[6]

Check yourself

Check yourself

  1. What is the difference between first-order and higher-order thalamic relays?
    Show answer

    First-order relays pass subcortical input (such as from the retina) to cortex; higher-order relays pass layer-5 output from one cortical area to another.

  2. What proportion of input to visual relay cells comes from the retina?
    Show answer

    Only about 5–10%; the rest is modulatory.

  3. What happened when the central lateral thalamus was stimulated in anaesthetised macaques?
    Show answer

    Arousal and wake-like neural processing were restored.

  4. How was awareness detected in a patient diagnosed as vegetative?
    Show answer

    With fMRI, while she imagined playing tennis or moving around her home, her brain activated like that of healthy volunteers.

  5. What does the perturbational complexity index measure?
    Show answer

    How complex (incompressible) the brain's EEG response to a TMS pulse is, reflecting both integration and differentiation.

  6. What did the 2025 adversarial collaboration conclude?
    Show answer

    Results matched some predictions of both integrated information theory and global workspace theory but challenged key tenets of each.

  7. Are current AI systems conscious, according to the indicator-property assessment?
    Show answer

    No current system satisfied the indicators, though the authors saw no obvious technical barrier to building one that does.

Glossary[1,10,14,15,17]

Thalamus
A paired complex of nuclei that relays and modulates signals to the cortex.
Relay nucleus
A thalamic nucleus that passes driving input on to the cortex.
Pulvinar
A large posterior thalamic region that is largely a higher-order relay.
Driver and modulator
Inputs that carry the main message and inputs that adjust how it is transmitted.
Vegetative state
Wakefulness without behavioural signs of awareness.
Minimally conscious state
Inconsistent but clearly discernible behavioural evidence of consciousness.
Integrated information (Φ)
In integrated information theory, the information a system integrates across its weakest link.
Global workspace
A proposed brain-wide network that makes information globally available, producing conscious access.
Ignition
Sudden, sustained activation of a large prefronto-parietal network during conscious access.
Deep brain stimulation
Electrical stimulation through electrodes implanted in deep brain structures.

References

  1. Sherman SM, Guillery RW. The role of the thalamus in the flow of information to the cortex. Philosophical Transactions of the Royal Society B 2002;357(1428):1695-1708. doi:10.1098/rstb.2002.1161
  2. Redinbaugh MJ, Phillips JM, Kambi NA, Mohanta S, Andryk S, Dooley GL, et al.. Thalamus modulates consciousness via layer-specific control of cortex. Neuron 2020;106(1):66-75.e12. doi:10.1016/j.neuron.2020.01.005
  3. Schiff ND, Giacino JT, Kalmar K, Victor JD, Baker K, Gerber M, et al.. Behavioural improvements with thalamic stimulation after severe traumatic brain injury. Nature 2007;448(7153):600-603. doi:10.1038/nature06041
  4. Owen AM, Coleman MR, Boly M, Davis MH, Laureys S, Pickard JD. Detecting awareness in the vegetative state. Science 2006;313(5792):1402. doi:10.1126/science.1130197
  5. Cogitate Consortium, Ferrante O, Gorska-Klimowska U, et al.. Adversarial testing of global neuronal workspace and integrated information theories of consciousness. Nature 2025;642(8066):133-142. doi:10.1038/s41586-025-08888-1
  6. Butlin P, Long R, Elmoznino E, Bengio Y, Birch J, Constant A, et al.. Consciousness in artificial intelligence: insights from the science of consciousness. arXiv 2023. doi:10.48550/arXiv.2308.08708
  7. Morel A, Magnin M, Jeanmonod D. Multiarchitectonic and stereotactic atlas of the human thalamus. The Journal of Comparative Neurology 1997;387(4):588-630. doi:10.1002/(SICI)1096-9861(19971103)387:4<588::AID-CNE8>3.0.CO;2-Z
  8. Fischl B, Salat DH, Busa E, Albert M, Dieterich M, Haselgrove C, et al.. Whole Brain Segmentation. Neuron 2002;33(3):341-355. doi:10.1016/S0896-6273(02)00569-X
  9. Schmahmann JD. Vascular Syndromes of the Thalamus. Stroke 2003;34(9):2264-2278. doi:10.1161/01.STR.0000087786.38997.9E
  10. Benabid AL, Pollak P, Hoffmann D, Gervason C, Hommel M, Perret JE, et al.. Long-term suppression of tremor by chronic stimulation of the ventral intermediate thalamic nucleus. The Lancet 1991;337(8738):403-406. doi:10.1016/0140-6736(91)91175-T
  11. Halassa MM, Kastner S. Thalamic functions in distributed cognitive control. Nature Neuroscience 2017;20(12):1669-1679. doi:10.1038/s41593-017-0020-1
  12. Steriade M, McCormick DA, Sejnowski TJ. Thalamocortical oscillations in the sleeping and aroused brain. Science 1993;262(5134):679-685. doi:10.1126/science.8235588
  13. Alkire MT, Hudetz AG, Tononi G. Consciousness and anesthesia. Science 2008;322(5903):876-880. doi:10.1126/science.1149213
  14. Giacino JT, Ashwal S, Childs N, Cranford R, Jennett B, Katz DI, et al.. The minimally conscious state. Neurology 2002;58(3):349-353. doi:10.1212/WNL.58.3.349
  15. Tononi G. An information integration theory of consciousness. BMC Neuroscience 2004;5:42. doi:10.1186/1471-2202-5-42
  16. Casali AG, Gosseries O, Rosanova M, Boly M, Sarasso S, Casali KR, et al.. A theoretically based index of consciousness independent of sensory processing and behavior. Science Translational Medicine 2013;5(198):198ra105. doi:10.1126/scitranslmed.3006294
  17. Dehaene S, Changeux JP. Experimental and theoretical approaches to conscious processing. Neuron 2011;70(2):200-227. doi:10.1016/j.neuron.2011.03.018

Related readings

  • Vision and the occipital lobe

    How spikes from the retina become edges, faces and scenes, the equations of receptive fields, and the implants and AI models built on them.

    Intermediate

  • The cerebral cortex and its maps

    What the folded outer layer of the brain is made of, why it folds, and how scientists map it from Brodmann's microscope to petabyte connectomes.

    Introductory

  • Neural networks, biological and artificial

    How brains and machines learn from experience, the maths both share, and why AI and neuroscience keep borrowing from each other.

    Intermediate

  • Blood supply and stroke

    The arteries of the brain and their territories, how flow is regulated, what a stroke destroys minute by minute, and the trials that changed treatment.

    Intermediate

  • The limbic system, cingulate and insula

    How the brain's inner rim and the hidden insula sense the body, shape feelings, register pain and conflict, and how stimulators and decoders now target mood.

    Intermediate

  • Hearing and the temporal lobe

    How the cochlea analyses sound, how auditory cortex maps frequency, speech and music, what goes wrong, and the implants and gene therapies that restore hearing.

    Intermediate

  • The brainstem and cranial nerves

    How the brainstem keeps us breathing and awake, what the twelve cranial nerves do, how they are tested, and what happens when they fail.

    Intermediate

Template anatomy for education. Not patient-specific. Not for clinical decision-making.