You wake to rain against a window. There is a sound, the dim light of the room, a remembered appointment and the feeling of being the one who is hearing and remembering. Neuroscience can measure brain activity during this episode, and psychology can ask what you noticed or recall. Neither measurement is identical to the episode as lived. The difficulty of relating first-person experience to third-person evidence is part of why consciousness remains a central, unsettled question.
Consciousness often means that there is something it is like for an organism to be in a state. In empirical work the word is also used for a person’s overall state, such as wakefulness compared with certain forms of sleep or anesthesia, and for specific contents, such as seeing a face rather than not seeing it. These uses are related but not interchangeable. A person can be awake without noticing a particular object; a vivid dream can occur while the person is asleep. A useful account must name which aspect of consciousness is under discussion before drawing conclusions from a test.
Experience, state and content
The broadest question concerns subjective experience: the felt character of seeing red, hearing a melody, feeling pain or thinking about tomorrow. Researchers cannot directly observe another person’s experience in the way they can observe a button press. They can ask for a report, study behavior and measure physiology. Those are indispensable sources of evidence, but each is a bridge to the target rather than the target itself.
A second question concerns a conscious state. Is the person awake, dreaming, deeply anesthetized or in a clinically altered state? Researchers examine responsiveness, brain activity and the capacity to sustain differentiated interactions. A third question concerns conscious contents: what, if anything, is experienced within a state? An awake participant may report seeing one image rather than another. A state measure and a content measure need not move together. The ability to classify wakefulness does not reveal every detail of what was experienced.
Seth and Bayne’s review of theories emphasizes that consciousness science includes multiple explanatory projects. Some theories focus on the availability of information for report and flexible control; others on recurrent sensory processing, higher-order representations, or the organization and integration of information. These are not merely different names for one agreed mechanism. Their predictions, levels of explanation and philosophical commitments differ. The field gains clarity when a theory specifies which phenomenon it seeks to explain and which observation would count against it.
Consciousness is not the same as attention
Attention selects or prioritizes information for a task. Consciousness concerns experience. The two often interact: attended information is frequently easier to report, and an experienced event may capture attention. Yet defining one by the other creates problems. A person may attend to a location without clearly experiencing a faint stimulus there, or notice an unexpected sound that was not the object of deliberate attention. Experimental designs that vary attention while holding other factors as steady as possible help study the relationship, but the separation is technically difficult.
Report is a further complication. A participant who presses a button to indicate seeing a stimulus must perceive, remember, decide and act. Neural activity linked to the button press may reflect those later operations rather than the experience itself. Koch and colleagues describe the challenge of separating neural correlates of consciousness from prerequisites and consequences. Finding activity that accompanies a report is informative, but it does not automatically establish the minimal process sufficient for experience.
The word awareness is also used in different ways. It may mean noticing a particular stimulus, knowing something about one’s own mental state or simply being awake. In this Learning Hub, the neighboring awareness entry can address those uses in their own right. This consciousness entry focuses on the broader problem of subjective experience and on how state, content and report are studied. It does not assume that a questionnaire about noticing is a complete measurement of consciousness.
How can a first-person phenomenon be studied?
A common laboratory design presents a stimulus near a threshold and compares trials in which participants report seeing it with trials in which they do not. Researchers may hold the physical stimulus similar while examining changes in performance and brain activity. This strategy helps identify correlates of conscious content. Its limits are equally important: attention, expectation, memory, decision criteria and the act of reporting can differ between the two groups of trials. A correlation does not alone identify a cause.
Another strategy changes the global state through sleep, anesthesia or brain injury and examines what physiological patterns accompany loss and return of responsiveness. These conditions are not identical. Some sleep includes vivid dreaming; anesthetic drugs differ in mechanisms; a person with severe motor impairment may be conscious while unable to give an ordinary response. A single behavioral scale or electrophysiological index must therefore be interpreted within the clinical and experimental context for which it was validated.
Consciousness research also uses naturalistic experience, introspective report and philosophical analysis. The last is not a substitute for experiments; it can clarify what a proposed measure is meant to explain. When researchers disagree about whether a study tested subjective experience or access for report, the dispute is partly conceptual and partly methodological. Better experiments require both clear questions and carefully matched comparisons.
Neural correlates are not the whole explanation
A neural correlate of consciousness is a pattern of brain activity reliably associated with an aspect of experience under specified conditions. Locating such patterns is a major scientific aim. But “correlate” is a disciplined word. Some activity may enable a conscious episode, some may constitute part of the process, and some may follow from attending to or reporting it. Koch and colleagues review progress and unresolved problems in separating those roles. A colored brain image is not a direct photograph of experience.
Consider a participant who reports a faint visual image. Activity in sensory cortex could reflect stimulus processing, activity in wider networks could reflect access and decision-making, and motor regions could reflect the response. A strong theory must say which pattern it predicts and why. If several processes covary, the experiment must try to dissociate them. This is why no-report and altered-report paradigms attract interest, although removing an overt report introduces its own inference problems.
Neural measurements also differ in spatial and temporal resolution. Brain imaging can help locate distributed patterns; electrophysiology can follow fast changes; stimulation can test whether perturbing a system affects behavior or state. Each method constrains a claim without making every competing interpretation disappear. The Institute’s practical emphasis on careful inference should not be mistaken for a new theory of consciousness. It is a way to read evidence without leaping from one measurement to a universal answer.
What competing theories propose
The global neuronal workspace family proposes that some information becomes available to many brain systems through large-scale, recurrent broadcasting. Mashour and colleagues review its account of conscious processing and the neural architecture it invokes. The account has motivated experiments about the timing and distribution of brain activity when a stimulus is reported. It should be described as a hypothesis family with testable predictions, not as a settled description of every conscious episode.
Integrated information theory begins from claims about the structure of experience and proposes that certain causal properties of a physical system are relevant to consciousness. Other approaches emphasize recurrent processing in sensory systems, higher-order representation or predictive processes. Seth and Bayne compare prominent theories and identify the work needed for sharper tests. A theory can offer a useful formalism or explanatory vocabulary while still facing hard questions about measurement, scope and the connection between its mathematical terms and lived experience.
The theories sometimes address different targets. An account of information becoming available for flexible report may be excellent at explaining access while leaving critics unconvinced that it explains why experience feels like anything. A theory focused on subjective structure may face a different challenge: making distinct, feasible predictions about neural data. Comparing theories fairly requires asking whether they predict different outcomes for the same task, not simply counting which theory can describe a result after it has occurred.
What an adversarial test can and cannot settle
In 2025, the Cogitate Consortium published an adversarial collaboration testing pre-specified predictions associated with global neuronal workspace and integrated information theories. Researchers used multiple brain-recording methods and defined analyses intended to give each theory a fair challenge. The study reported results that challenged aspects of both accounts. Its value lies partly in requiring explicit predictions, collecting common data and making disagreements more testable.
It would be a mistake to turn that result into a sporting verdict that one complete theory won and the other lost. Each theory contains multiple claims; an experimental implementation tests a subset. Stimulus choices, timing, brain coverage, analysis decisions and definitions of conscious content matter. A study can disconfirm a particular prediction while leaving a broader family of ideas open to revision. That is normal scientific progress, not a reason to declare consciousness unscientific.
Nor does theory competition establish a direct clinical test. A result about healthy participants viewing stimuli cannot, by itself, determine the condition of a person who cannot speak or move. Clinical assessment requires methods tested for the relevant population, repeated observations, specialist judgment and attention to sources of uncertainty. The ethical stakes are too high for a single theoretical slogan or one brain signal to carry the whole conclusion.
Levels of consciousness and clinical caution
Casali and colleagues introduced a perturbational complexity index, or PCI, based on how complex the brain’s response is after a controlled magnetic stimulation pulse. They tested it in people under different conditions, including wakefulness, sleep, anesthesia and brain injury. The work is important because it sought a measure less dependent on ordinary sensory input and behavioral response. It does not make the index a universal or self-interpreting meter of experience.
A physiological index is trained and evaluated against particular groups and reference judgments. Its threshold, uncertainty and applicability must be checked in a new setting. Brain injury can alter pathways needed to generate or record a signal, while medication and other conditions may affect measurements. A false inference in this area can have serious consequences. The article therefore describes PCI as a research and clinical-research contribution, not as a tool a reader should apply to an individual person.
The distinction between responsiveness and consciousness is vital. A person may not respond because motor output is impaired, sedation is present or a task was misunderstood. Conversely, a reflexive response does not necessarily establish reflective awareness of the stimulus. Clinical teams combine examination with appropriate supplementary tests and revisit judgments over time. This article cannot substitute for that process and does not label any individual state from a story or a scan.
The hard problem and the explanatory gap
David Chalmers named the “hard problem” to distinguish explaining functions and behaviors from explaining why and how physical processes are accompanied by subjective experience. We can investigate how a person discriminates a sound, reports a color or integrates information while still asking why those mechanisms have a felt aspect. The hard problem is a philosophical challenge about explanation. It is not an empirical finding that neuroscience has failed, and it is not proof that any one metaphysical answer is correct.
Researchers differ on whether the gap reflects an enduring limit, an incomplete science, a mistaken intuition about explanation or a problem that will change as theories improve. Seth and Bayne’s theory review shows how philosophical assumptions shape empirical programs. A useful article can take the question seriously without presenting it as solved or beyond research. It should also avoid implying that finding a neural correlate alone settles why the correlate accompanies experience.
At the same time, unanswered philosophical questions do not erase empirical progress. Researchers can distinguish some brain states, identify conditions under which reports change, perturb systems and test competing predictions. Those achievements matter for science and potentially for care. Their limits matter just as much. Clear language keeps the two claims together: we can learn a great deal about mechanisms and markers without claiming a final explanation of subjective experience.
What consciousness is not
Consciousness is not identical to intelligence. Solving a difficult problem, producing coherent language or acting flexibly does not alone settle whether there is felt experience. Consciousness is not identical to self-reflection: a person can have a simple sensory experience without explicitly thinking about themselves. Nor is it the same as moral worth. Ethical regard for a patient, animal or other entity cannot be mechanically derived from one contested consciousness measure.
Likewise, consciousness should not be collapsed into a single score. A measure of state, a report of content, a task requiring attention and a questionnaire about awareness each target different aspects. Combining them can be informative if their meanings are kept clear. If they disagree, the disagreement may reveal a measurement problem, a theory problem or a genuine dissociation. It should not be hidden by averaging everything into a generic “consciousness level.”
These distinctions also protect public discussion of artificial intelligence. Fluent text from a system is evidence of a particular capability, not direct evidence of subjective experience. The theories reviewed here do not provide a consensus test that can settle consciousness in every biological or artificial system. Claims about AI experience require more than analogy to human report or confidence in one favored model.
How to read a consciousness claim
First identify the target: overall state, specific content, access for report or subjective experience as such. Then ask what was measured: report, task performance, neural recording, stimulation response or a combination. Ask whether the study compared plausible alternatives, controlled for attention and response demands, and specified where its participants and methods can generalize. A theory paper should state a prediction that could be wrong; an empirical paper should say which interpretation of the result remains open.
Next examine the language of certainty. “Associated with” is different from “causes,” and a validated discriminator within a sample is different from a universal diagnostic test. A study using healthy participants viewing images cannot automatically answer questions about dreaming, anesthesia, infant experience or severe brain injury. These are related topics but not interchangeable populations and tasks. The boundary is part of the finding.
Finally ask whether the article separates scientific and philosophical questions. Brain recordings can constrain theories; philosophical analysis can expose hidden assumptions; neither should be used to pretend the other does not matter. The Institute’s proposed reading approach is a synthesis, not a new empirical scale. It encourages precise questions so that genuine advances are visible without being inflated.
What the evidence does not show
The evidence does not establish one agreed definition that resolves every use of “consciousness,” nor one accepted neural theory. The 2025 adversarial test challenged predictions from prominent accounts but did not settle the full debate. A neural correlate is not automatically a cause or a complete explanation. A physiological index is not a direct, universally valid window into another person’s experience.
Report and non-report methods each have limitations. A person may fail to report because of attention, memory, language or motor constraints; a brain pattern can be hard to interpret without a reliable reference. Neither apparent unresponsiveness nor one positive neural signal should be treated as a stand-alone diagnosis. This article cannot judge an individual’s clinical state and should not be used for care decisions.
The evidence also does not show that attention, awareness, intelligence, self-reflection and consciousness are synonyms. They overlap in some tasks and diverge in others. The hard problem remains a live dispute about explanation, not a scientific result that closes investigation. The useful stance is precise curiosity: define the target, compare methods, test theories and keep the first-person phenomenon in view without claiming more than the evidence can bear.
Research references
- Seth, A. K., & Bayne, T. (2022). Theories of consciousness. Nature Reviews Neuroscience, 23, 439–452. View source
- Koch, C., Massimini, M., Boly, M., & Tononi, G. (2016). Neural correlates of consciousness: Progress and problems. Nature Reviews Neuroscience, 17, 307–321. View source
- Mashour, G. A., Roelfsema, P., Changeux, J.-P., & Dehaene, S. (2020). Conscious processing and the global neuronal workspace hypothesis. Neuron, 105(5), 776–798. View source
- Cogitate Consortium et al. (2025). Adversarial testing of global neuronal workspace and integrated information theories of consciousness. Nature, 642, 133–142. View source
- Casali, A. G., Gosseries, O., Rosanova, M., et al. (2013). A theoretically based index of consciousness independent of sensory processing and behavior. Science Translational Medicine, 5(198), 198ra105. View source
- Chalmers, D. J. (1995). Facing up to the problem of consciousness. Journal of Consciousness Studies, 2(3), 200–219. View source