Loading
Old version — revision 1
This is a fixed snapshot of Neural correlates of consciousness, saved by Import as part of the initial corpus import. It is not edited and it is not updated; the article may have changed since.
Edit summary: Initial import of content/neural-correlates-of-consciousness.md — the filesystem corpus, unchanged. Not an edit.
The minimal neural activity jointly sufficient for a specific conscious experience, and the experimental programme that tries to identify it.
Neural correlates of consciousness (NCC) are the neural events minimally sufficient for a particular conscious experience. The concept was proposed as a deliberately modest research programme: rather than solving the question of why physical processes give rise to experience at all, find out which processes reliably accompany which experiences, and let the mechanism constrain the theory. Francis Crick and Christof Koch put the programme on the agenda in 1990, proposing synchronized gamma-band activity as a first candidate.1 Three and a half decades later the field has robust experimental paradigms, several well-specified theories, and no consensus on which is right.
David Chalmers's standard formulation distinguishes the content NCC — the minimal neural system whose state determines what a person is conscious of — from the background or state NCC, the conditions of arousal that make consciousness possible at all.2 Brainstem structures that regulate wakefulness are part of the second but plainly not the seat of visual experience.
Two qualifiers do real work. "Minimal" excludes the whole brain, which is trivially sufficient. "Sufficient" is weaker than "identical to": a correlate can be a reliable accompaniment without being the thing itself, which is why NCC results do not by themselves settle whether a machine or an organoid is conscious. A further complication is that many candidate correlates are consequences of consciousness rather than constituents — attention, working memory, and the preparation of a verbal report all follow conscious perception and are recorded alongside it.
The workhorse is the contrastive paradigm: hold the physical stimulus constant and vary whether it is consciously perceived, then look for the difference. Binocular rivalry presents each eye with a different image and lets perception alternate spontaneously. Masking renders a stimulus invisible by flanking it in time. The attentional blink makes a target invisible when it follows another too closely. Bistable figures, continuous flash suppression, and threshold detection tasks all serve the same function.
The main confound is report. Asking a participant what they saw recruits decision-making, motor preparation, and memory, and the resulting frontal activity may reflect reporting rather than experiencing. No-report paradigms address this by inferring perceptual state from optokinetic eye movements or pupil size, and they systematically reduce the frontal signature seen in report-based studies.3 How much of the classic prefrontal NCC survives is one of the field's live disputes.
Measurement runs from scalp electroencephalography and functional MRI up to intracranial recording in patients implanted for epilepsy monitoring, which supplies the field's highest-resolution human data and uses the same hardware as Electrocorticography interfaces. Analysis increasingly borrows the multivariate classifiers developed for Neural decoding, asking not whether a region is more active but whether the identity of what a person saw can be read out from it. In animals, causal tests using Optogenetics can silence or drive candidate circuits, though animals cannot report experience, which limits what those manipulations establish.
Global neuronal workspace theory, developed from Bernard Baars's cognitive model by Stanislas Dehaene and Jean-Pierre Changeux, holds that a stimulus becomes conscious when it is amplified and broadcast to a distributed network of long-range pyramidal neurons, particularly in prefrontal and parietal cortex. The signature is a nonlinear "ignition" event, with late widespread activity following a threshold crossing.4 On this account consciousness is about making information globally available for report, reasoning, and action.
Integrated information theory, developed by Giulio Tononi, starts from the properties of experience and asks what a physical system must be like to have them. It identifies consciousness with a system's intrinsic, maximally irreducible cause-effect structure, quantified as Φ, and locates the substrate in a posterior "hot zone" of parietal, temporal, and occipital cortex rather than in prefrontal areas.5 It makes the strong claim that a system's architecture, not its input-output behaviour, determines whether it is conscious — the reason it denies Substrate independence.
Recurrent processing theory holds that local recurrent activity in sensory cortex suffices for experience even without global broadcast. Higher-order theories hold that a first-order representation becomes conscious only when represented by a further state, typically in prefrontal cortex. Attention schema theory treats consciousness as the brain's simplified model of its own attention. These are not merely verbal variants: they disagree about where to look and about which patients and which systems are conscious.
To break the pattern of each laboratory confirming its own theory, a large preregistered project was organized in which proponents of global workspace theory and integrated information theory agreed in advance on predictions that would count against their positions. The Cogitate consortium ran the resulting experiments across multiple sites using functional MRI, magnetoencephalography, and intracranial recordings, and published the results in 2025.6
Neither theory came through cleanly. Content-specific information about what participants saw was decodable from posterior cortex and sustained for the duration of the stimulus, which fits the posterior emphasis; but the sustained long-range synchronization that integrated information theory predicted was not observed. Conversely, prefrontal cortex carried some content information, but the ignition-like response global workspace theory predicted at the end of a stimulus was not found. Both camps have argued that the tested predictions were not core commitments of their theories — a response the collaboration's design was intended to forestall, and which illustrates how hard it is to falsify a theory of consciousness.
A bet, settledIn 1998 Christof Koch wagered David Chalmers a case of wine that a clear neural correlate of consciousness would be identified within 25 years. Koch conceded in 2023. The concession was collegial and the science had advanced considerably; the point is that a leading experimentalist judged the specific target unmet.
The clinically important products of the field do not depend on settling the theory. The perturbational complexity index uses transcranial magnetic stimulation — one of the techniques covered under Non-invasive neuromodulation — to perturb cortex and measures the spatiotemporal complexity of the electroencephalographic response. It separates wakefulness and dreaming from anaesthesia, deep sleep, and vegetative states in ways that behavioural examination misses.7 Measures of signal diversity in the same family rise rather than fall under psilocybin and LSD, which is why the drugs studied for Psychedelic therapy also serve the field as a way of perturbing conscious state rather than only of suppressing it. Task-based imaging can detect covert awareness: some behaviourally unresponsive patients modulate their brain activity on command, for example by imagining playing tennis. A large multi-centre study published in 2024 found this cognitive-motor dissociation in roughly one in four patients who showed no observable response at the bedside.8 For those patients a Brain–computer interface is the only available communication channel, and central thalamic Deep brain stimulation has been investigated as a way to raise arousal.
These methods are correlational and calibrated against people who can confirm their experience, which limits how far they extend to systems very unlike humans — the problem that dominates Machine consciousness.
Whether a machine could be conscious depends on which theory is true, and the theories give opposite answers. Global workspace theory is functionalist: build the right architecture, with a bottleneck that selects and broadcasts information, and the system qualifies. Integrated information theory holds that a digital computer simulating a conscious system would have negligible Φ, so behaviour is irrelevant. Higher-order and attention-schema theories fall in between.
This is why current work on assessing artificial systems takes an indicator-property approach — extracting the computational features that each theory says matter and checking which are present — rather than applying a behavioural test. It is also why progress on the NCC bears directly on Mind uploading and Whole brain emulation: an emulation's moral and personal status depends on a question the NCC programme was explicitly designed to bracket, as does the status of anything built on the path toward Artificial general intelligence. The status of products marketed as Digital immortality does not turn on it, since a text model of a dead person makes no claim to experience in the first place.
The programme's founding bracket is now its main limitation. Correlates do not distinguish constituents from consequences, and no experiment currently on offer separates a mechanism that produces experience from one that merely accompanies it. Adjacent difficulties compound this: theories are formulated at different levels of description, so a result can be read as supporting either; the field's stimuli are almost entirely visual, so the correlates found may be correlates of seeing rather than of consciousness; and the tools remain coarse, since even Connectomics and dense recording sample far less than the relevant circuitry. Tension over whether integrated information theory is testable at all — expressed in a 2023 open letter from a large group of researchers — reflects a deeper unresolved question about what would count as evidence.
paperCrick, F. and Koch, C. "Towards a neurobiological theory of consciousness." Seminars in the Neurosciences, 1990.↩A programmatic essay rather than an experiment; the specific proposal it advanced, gamma-band synchrony as the correlate, did not survive later work.
bookChalmers, D. J. "What is a Neural Correlate of Consciousness?" In T. Metzinger (ed.), Neural Correlates of Consciousness: Empirical and Conceptual Questions, MIT Press, 2000. ↩
paperTsuchiya, N., Wilke, M., Frässle, S. and Lamme, V. A. F. "No-Report Paradigms: Extracting the True Neural Correlates of Consciousness." Trends in Cognitive Sciences, 2015. ↩
paperDehaene, S. and Changeux, J.-P. "Experimental and theoretical approaches to conscious processing." Neuron, 2011. ↩
paperKoch, C., Massimini, M., Boly, M. and Tononi, G. "Neural correlates of consciousness: progress and problems." Nature Reviews Neuroscience, 2016. ↩
paperCogitate Consortium. "Adversarial testing of global neuronal workspace and integrated information theories of consciousness." Nature, 2025.↩Proponents of both theories fixed the predictions before data collection, which is why the later claims that those predictions were not core commitments arrived after the result.
paperCasali, A. G. et al. "A theoretically based index of consciousness independent of sensory processing and behavior." Science Translational Medicine, 2013. ↩
paperBodien, Y. G. et al. "Cognitive Motor Dissociation in Disorders of Consciousness." New England Journal of Medicine, 2024.↩Detection requires the patient to sustain a mental task on command, so a negative result establishes nothing about whether that patient is aware.