Theories of Consciousness: What Scientists Propose and Test

Author: Angela Bogdanova · Published: October 16, 2026 · Editorial Policy

Research cluster: Consciousness, Metacognition & Self-Awareness · Written by Artificial Sapiens

Theories of consciousness are scientific and philosophical frameworks that attempt to explain how subjective experience relates to activity in the brain. The most extensively discussed neuroscientific approaches include Global Neuronal Workspace Theory (GNWT), Integrated Information Theory (IIT), higher-order theories, Recurrent Processing Theory (RPT), and predictive-processing approaches. They disagree about which mechanisms are necessary for a perception to be experienced, which brain signals should accompany experience, and how investigators can test those claims. No single theory has achieved scientific consensus. A large preregistered comparison published in 2025 supported some predictions of GNWT and IIT while challenging central predictions of both. Seth and Bayne, 2022; Cogitate Consortium, 2025.

Understanding the theories requires separating three questions: What is experienced? What enables that experience? And what can a particular experiment actually demonstrate? Scientists study perception, brain dynamics, conscious reports, and loss or recovery of responsiveness. Those observations constrain theories, although observations alone do not automatically settle the philosophical question of why physical processes should have any subjective character.

What does a scientific theory of consciousness explain?

Consciousness, in the central sense relevant here, concerns subjective experience: what it is like for a person to see a color, hear music, experience pain, or entertain a thought. Scientific theories propose relationships between that experience and physical or computational mechanisms. Some primarily explain access to information for reasoning and reporting; others aim to explain the existence and structure of experience itself. The distinction matters because an experiment about reportability may favor a theory of conscious access without settling every claim about phenomenal experience. Seth and Bayne, 2022.

Consciousness, awareness, attention, and responsiveness

Consciousness is often used broadly, but researchers distinguish phenomenal consciousness, the experience itself, from access consciousness, the availability of information for thought, decision, and report. Awareness is sometimes used as a synonym for consciousness and sometimes refers to awareness of a specific stimulus. Attention is selective prioritization of information; it can influence conscious perception, but attention and experience are not interchangeable constructs. Wakefulness describes an arousal state, while responsiveness refers to observable behavior. An individual may be awake without reliable evidence of awareness, and absence of overt response does not establish absence of experience. Seth and Bayne, 2022; Koch et al., 2016.

Self-awareness concerns representing oneself, one's mental states, body, actions, or identity. Metacognition concerns monitoring and controlling one's cognitive processes, including judgments of confidence. Neither should be treated as equivalent to consciousness itself: a person can have a visual experience without explicitly reflecting on being the person who sees it. Higher-order theories do investigate whether a particular kind of representation of a mental state contributes to its consciousness, but that is a substantive, contested explanatory hypothesis. Lau and Rosenthal, 2011.

Levels, contents, and dimensions of experience

A theory may address the general capacity for conscious experience, the particular content experienced, or the integration of multiple contents into a unified scene. For example, changes across wakefulness, non-rapid-eye-movement sleep, dreaming, anesthesia, and severe brain injury inform the study of conscious state. In contrast, seeing a face rather than a house in the same waking state informs the study of conscious content. The neural mechanisms sustaining a wakeful condition may differ from those specifying the visual details of experience. Koch et al., 2016.

Theories also differ in explanatory ambition. A theory that explains when information becomes usable across brain systems addresses a different aspect of the problem than a theory that claims to identify the intrinsic physical structure of an experience. This is one reason that apparent agreement or conflict among theories cannot be evaluated solely by whether their diagrams include a frontal or posterior brain region. Mudrik et al., 2025.

How the major theories developed

Modern experimental consciousness science grew from psychophysics, neuropsychology, cognitive psychology, electrophysiology, and the study of neural correlates of conscious perception. Global workspace approaches drew on cognitive architectures in which specialized processes share information through a common workspace. Neuronal versions connected that idea to distributed brain networks and experimentally measurable late amplification. Recurrent theories emphasized feedback within perceptual systems. Higher-order theories developed philosophical arguments about awareness of mental states into empirical predictions. IIT began from proposed properties of experience and sought a formal account of the physical causal structures that could support them. Predictive approaches developed from models of perception as inference constrained by sensory evidence. Dehaene and Changeux, 2011; Oizumi et al., 2014; Seth and Bayne, 2022.

These histories produce different starting assumptions about the explanatory target. Some models begin with experimentally tractable access to stimuli. Others begin with phenomenological features such as differentiation and unity. Comparing them scientifically requires defining a common observation, specifying the theoretical claim being tested, and stating in advance which outcome would count against it.

Global Neuronal Workspace Theory: conscious access through global broadcasting

GNWT proposes that many brain processes operate locally and without conscious access, while some representations undergo amplification and become broadly available to other systems. This transition is often described as global broadcasting or ignition. Its classical neuronal formulations emphasize widespread interactions involving sensory, parietal, and prefrontal systems. A representation that enters the global workspace can influence report, working memory, reasoning, and flexible control. Dehaene and Changeux, 2011.

Mechanism and experimental predictions

In visual masking experiments, a brief target is followed by another stimulus that can make it difficult to experience or report. GNWT predicts that conscious access is associated with a nonlinear transition from limited, local processing to amplified, distributed activity. Late neural signatures, often including broadly distributed activity and changes in effective communication, are candidate consequences or components of this transition. Experimental details are crucial: the same prefrontal activity could reflect conscious content, decisions about the stimulus, preparation to respond, or a combination. Dehaene and Changeux, 2011; Tsuchiya et al., 2015.

A distinguishing question is whether frontal representations reflect conscious contents when those contents are irrelevant to the participant's task. Another is whether updates in experience should produce predicted ignition patterns at both stimulus onset and offset. A theory's general claim about broadcasting must be translated into timing, content-decoding, and brain-region predictions to become directly testable.

What does the evidence establish?

The 2025 Cogitate adversarial collaboration found some content information in inferior frontal cortex, but it also found limited prefrontal representations of several conscious features and generally failed to find predicted ignition at the end of stimulus presentation. These outcomes challenge particular preregistered GNWT predictions; they do not establish that long-range broadcasting never occurs. Interpretations depend on the specific experimental contrasts and on distinctions among different versions of the workspace account. Cogitate Consortium, 2025.

GNWT is therefore an active research framework with empirically informative successes and failures. It should not be summarized as a simple claim that the frontal lobes create all consciousness. A 2026 theoretical clarification describes the global neuronal workspace as a multilevel neuronal account and cautions against treating it as identical to every computational global-workspace model. Changeux and Farisco, 2026.

Integrated Information Theory: consciousness and intrinsic causal structure

IIT proposes that consciousness relates to the intrinsic cause–effect organization of a physical system. The theory begins from proposed properties of conscious experience, including differentiation and unity, and develops mathematical requirements for a physical substrate. It argues that integrated information and the structure of causal relationships within a system are central to determining whether and how it experiences. Some versions use the symbol Φ, phi, in formal characterizations of irreducibility, although a simplified equation equating a convenient brain-complexity score with consciousness misrepresents the full theory. Oizumi et al., 2014; Tononi et al., 2016.

Mechanism and predictions

IIT asks about what the system can causally do to itself from an intrinsic perspective, rather than simply what an external observer can decode from its activity. Its neuroscience-associated predictions have often emphasized posterior cortical regions as candidate substrates for the contents of visual experience, whereas frontal regions may contribute to task, attention, and report. The theory also predicts particular features of sustained content representation and interactions within relevant neural complexes. Tononi et al., 2016; Cogitate Consortium, 2025.

An important conceptual caution follows. Finding complex or integrated activity is not equivalent to computing a complete IIT causal structure; similarly, the observation of a complex EEG response does not by itself validate IIT's full metaphysical or quantitative account. Approximate and theory-inspired operational measures require independent validation.

Evidence and objections

In the 2025 direct comparison, sustained activity in some posterior regions was broadly compatible with an IIT-associated prediction. However, the absence of predicted sustained posterior synchronization and some failures to maintain detailed conscious feature information challenged specified tests of the account. These results have to be interpreted within the preregistered analyses, sampling limitations, and translation from mathematical principles to measurable signatures. Cogitate Consortium, 2025.

IIT also attracts discussion about how its complex calculations relate to biological measurements, whether its axioms are independently established, and whether its consequences for unusual physical systems are credible. Those discussions include mathematical, empirical, and philosophical claims that should be assessed separately. Neither enthusiastic applications nor objections to speculative implications serve as substitutes for comparing experimentally specified predictions.

Higher-order theories: when a mental state becomes represented as a mental state

Higher-order theories propose that a mental state becomes conscious in virtue of an appropriate higher-order representation of that state. Put simply, a visual representation of a red object and a representation that one is seeing red have different theoretical roles. Versions of the approach disagree about whether the relevant higher-order representation must itself be conscious and which neural implementation is sufficient. Lau and Rosenthal, 2011; Seth and Bayne, 2022.

Predictions concerning awareness and confidence

Higher-order accounts motivate research on dissociations between perceptual discrimination and subjective awareness, on prefrontal contributions to subjective judgment, and on neural representations of perceptual states. When researchers manipulate prefrontal activity or examine patients with focal lesions, they must separate changes in experience from changes in criterion, confidence, attention, memory, and verbal reporting. A task participant who guesses correctly but reports seeing nothing presents a rich empirical pattern; it does not automatically prove one theory, because competing models can often explain aspects of the same behavior. Lau and Rosenthal, 2011.

Higher-order consciousness should also be distinguished from everyday metacognition. Confidence calibration and the ability to evaluate one's memory are measurable cognitive capacities. The strong higher-order thesis claims that a suitable representation plays a constitutive role in consciousness, a deeper claim than the observation that people have varying degrees of introspective accuracy.

Where the controversy remains

Empirical debates focus on what qualifies as the higher-order representation, how to identify it independently of report, and whether apparent prefrontal contributions reflect awareness itself. Researchers also disagree about whether an experience can occur without access to or representation of itself. The approach remains scientifically testable in several forms, and success of a prefrontal decoding model should not be conflated with confirmation of its full philosophical interpretation. Seth and Bayne, 2022.

Recurrent Processing Theory: feedback within perceptual cortex

RPT emphasizes recurrent interactions: activity cycles between levels of sensory processing rather than traveling only forward from the eyes toward progressively higher cortical regions. On this view, appropriate recurrent processing within perceptual networks is central to visual experience, while widespread access and explicit report may involve further mechanisms. This creates a possible distinction between perceptual phenomenology and the capacity to describe or use what is perceived. Lamme, 2006; Seth and Bayne, 2022.

What researchers test

Masking, attentional manipulations, intracranial recordings, and measures of directional interactions between brain regions provide opportunities to compare feedforward and recurrent accounts. An early feedforward visual response can encode features even when observers do not report awareness. The question is whether recurrent signaling tracks experienced content over and above basic sensory processing, task goals, or later reporting. Because feedback is common in the brain, merely detecting recurrent activity is not enough; its timing, causal role, content specificity, and relationship to conscious experience must be established.

A 2026 MEG study comparing report and no-report visual masking found stable posterior sensory representations and recurrent coupling across conditions. Its results support the usefulness of studying posterior processing independently of reporting demands, but a single study cannot decide whether the mechanisms are sufficient for subjective experience. Report-modulated prefrontal activity and consistent posterior representations, 2026.

Open questions

RPT must specify which recurrent loops count, why some feedback remains unconscious, how recurrent perceptual processes relate to unified multimodal experience, and whether the theory generalizes beyond vision. Strong causal tests require manipulating recurrence while controlling the impact on sensory precision and behavioral performance.

Predictive processing: perception as inference

Predictive-processing frameworks describe perception as a continual interaction between expectations, sensory signals, and errors in prediction. Brain systems adjust models of the causes of incoming signals and regulate the weight assigned to different forms of evidence. These ideas offer a powerful computational vocabulary for perceptual content and its contextual dependence. Yet predictive processing is a family of models, and predictive coding alone is not a universally specified theory of phenomenal consciousness. Seth and Bayne, 2022. For a fuller explanation of this computational framework, see our English Hub article on predictive processing in the brain.

Candidate explanation of conscious content

One proposal is that experienced perception corresponds to particular inferential states or hypotheses favored by the brain. Another uses the distinction between predictions and prediction errors to explain variations in perceptual visibility and stability. The central scientific challenge is specifying what transforms ordinary predictive processing, which may occur without awareness, into a conscious state. Some interpretations emphasize hierarchical inference, others the character of embodied regulation and sensory prediction. The existence of predictive signals in the brain is evidence for predictive computation, not automatically for a theory that makes those computations constitutive of experience.

An adversarial collaborative review published in August 2026 compared IIT with predictive-processing proposals, articulated differences in explanatory assumptions, and outlined future hypotheses and procedures for systematic testing. The publication provides a research program rather than a decisive head-to-head result showing which account is correct. Corcoran et al., 2026.

Comparing the theories by what they predict

Theories of consciousness become most informative when their predictions are expressed in the same terms. For visual perception, researchers can vary whether a stimulus is consciously seen while controlling its physical properties, attention, relevance to the task, and report requirements. They can then ask whether experience covaries with broadcasting, posterior recurrence, higher-order representation, causal integration, or inferential states. Mudrik et al., 2025.

Global broadcasting versus local recurrence

A relatively strong GNWT account predicts that conscious access requires specific widespread interactions and amplification, whereas some RPT formulations attribute visual phenomenology to recurrent processing within posterior perceptual systems. If reliable conscious experience were found under conditions with posterior recurrence but without the predicted global ignition, the result would challenge the strongest ignition requirement. However, a negative measurement of a distant brain signal can mean that the signal was weak, brief, difficult to measure, or incorrectly operationalized. The experiment must test a sufficiently precise prediction.

Posterior integration versus frontoparietal involvement

IIT-associated experiments often emphasize posterior cortical content and intrinsic interactions; GNWT experiments often emphasize widespread availability and frontoparietal dynamics. Importantly, one theory may allow a region to participate in attention or reporting while denying that it constitutes experience. Finding both frontal and posterior representations therefore need not adjudicate the causal dispute. Researchers must establish whether disrupting a candidate region changes the phenomenal content, access to that content, or performance on a secondary task. Cogitate Consortium, 2025.

First-order processing versus higher-order representations

Higher-order theories treat appropriate representations of a mental state as central. First-order and recurrent theories can give more explanatory weight to perceptual states themselves. Experiments that compare visual discrimination with subjective confidence or awareness seek dissociations between these processes. Yet confidence, signal-detection criterion, performance, and experience are distinct variables, so a correlation between confidence and awareness cannot alone resolve the direction of causation. Lau and Rosenthal, 2011.

Information integration versus observable complexity

IIT's intrinsic causal properties should be distinguished from generic neural complexity, correlation, or statistical dependence. Some complexity measures are useful for classifying states. A classifier validated to distinguish waking from anesthetized conditions can be clinically promising even if its performance does not demonstrate that IIT is the unique correct explanation of consciousness. Casali et al., 2013.

The 2025 COGITATE experiment: a direct test of competing theories

The Cogitate Consortium and researchers associated with competing theories designed a preregistered adversarial collaboration comparing GNWT and IIT. The published study involved 256 participants viewing clearly visible stimuli for variable durations; experiments combined fMRI, MEG, and intracranial EEG recordings. Rather than allowing researchers to choose favorable predictions after the data were known, the collaboration agreed beforehand on tests, relevant measurements, and criteria for interpreting results. Cogitate Consortium, 2025.

What the researchers observed

Conscious content could be decoded from several cortical regions, including visual, ventrotemporal, and inferior frontal cortex. Sustained activity in occipital and lateral temporal cortex reflected stimulus duration. Some frontal-to-early-visual synchronization also carried content-specific information. However, expected sustained synchronization within posterior cortex was not found, challenging an IIT-related prediction. For GNWT, the predicted ignition at stimulus offset was generally absent, and some visual features were weakly or inconsistently represented in prefrontal cortex. Cogitate Consortium, 2025.

What the study did and did not settle

The findings challenged specified parts of both theories and supported other parts. They did not prove the impossibility of either model; nor did they establish that a third theory must be correct. The design focused on selected suprathreshold visual contents and measurable signatures. Extrapolation to dreaming, pain, emotional experience, nonhuman animals, or consciousness in all possible physical systems requires additional justification. The deeper methodological advance was the preregistered agreement about what would count as an informative failure. Cogitate Consortium, 2025.

How consciousness experiments work

Visual masking and perceptual thresholds

A target presented for a brief interval may be difficult to report after a mask appears. Researchers compare neural and behavioral responses when stimuli are seen, unseen, or near a visibility threshold. Such tasks can help locate transitions in processing. They also risk conflating sensory strength, decision difficulty, attention, confidence, and awareness. Careful studies quantify discrimination performance separately from subjective visibility and report the task constraints in detail. Dehaene and Changeux, 2011.

Binocular rivalry and ambiguous perception

When incompatible visual images are presented to the two eyes, perceptual dominance can alternate even when the retinal input is relatively stable. Tracking the changing percept offers a route to study neural correlates of content. Yet asking participants to press a button when the percept changes introduces decision and motor processes. No-report designs using indirect indicators can reduce some confounds, although the validity of the indicators themselves must be checked. Tsuchiya et al., 2015.

Report versus no-report paradigms

Report-based experiments ask participants what they experienced. Reports provide valuable evidence but activate attention, evaluation, memory, language, and motor preparation. No-report paradigms attempt to infer perceptual content from eye movements, physiological signatures, or other independent measures. Neither approach gives unmediated access to someone else's experience. Combining both is often stronger than treating either as a flawless gold standard. Tsuchiya et al., 2015.

EEG, MEG, fMRI, and intracranial recordings

EEG and MEG capture rapid changes in neural activity but differ in sensitivity and spatial localization. fMRI provides spatial information through hemodynamic signals but measures neural activity indirectly and on a slower timescale. Intracranial electrodes can give precise signals from particular locations, though their coverage depends on clinical placement and participant selection. A convergence across methods strengthens an inference; incompatible results demand examination of temporal resolution, spatial sampling, statistical power, and the way experience was measured. Cogitate Consortium, 2025.

Perturbation, stimulation, and causal evidence

Manipulating activity with stimulation or observing patients with localized damage can provide stronger causal constraints than correlations alone, while introducing its own problems of spread, compensation, and task disruption. The perturbational complexity index, or PCI, combines transcranial magnetic stimulation with EEG to estimate the differentiation and integration of evoked cortical responses. The original research found that this measure could distinguish several conscious and unconscious conditions across healthy participants and patient groups. PCI is an empirical index inspired by theoretical ideas about complexity; it is not a definitive readout of inner experience or a proof of IIT. Casali et al., 2013.

Neural correlates, mechanisms, and explanations

A neural correlate is a reliable association between a pattern of brain activity and an aspect of consciousness. To understand the phenomenon, researchers must separate the actual mechanism associated with conscious content from prerequisite conditions such as general arousal and from consequences such as reporting a percept. A frontal signal that tracks awareness may reflect global broadcast, introspective evaluation, or a decision to respond. A posterior signal may encode a stimulus without proving that its presence constitutes experience. Koch et al., 2016.

Causation is also not one-dimensional. Stimulating a brain area may disrupt perception because that region carries sensory input, participates in attention, generates a report, or contributes to experience itself. Strong theory tests combine detailed behavioral measures, temporal dynamics, interventions, and explicit hypotheses about which effects should remain when confounds are controlled.

The main scientific disagreements and methodological limitations

Do theories explain the same phenomenon?

Some frameworks primarily model the availability of information for cognitive use; others aim to identify the physical conditions for phenomenal experience. In a 2025 comparative review based on debate among proponents of five theories, Mudrik and colleagues emphasized disputes about what exactly is being explained, how conscious states should be identified, and what evidence might refute a theory. This conceptual disagreement directly affects the design and interpretation of experiments. Mudrik et al., 2025.

The measurement problem

Scientists observe physiology and behavior, then infer experience. First-person reports remain important for many tasks, but the translation from subjective reports to an objective measurement introduces systematic uncertainty. The challenge becomes especially pronounced when participants cannot speak, when experiences are difficult to describe, or when indirect indicators have ambiguous relationships to awareness. A theory whose predictions depend on a particular awareness measure should specify that measure's assumptions. Seth and Bayne, 2022.

Post hoc flexibility and selective interpretation

A theory with broad language can sometimes explain multiple mutually inconsistent observations after the fact. Preregistration, competing prediction tables, sufficiently powered samples, and adversarial collaboration help establish whether the data genuinely discriminate among models. The 2025 Cogitate study demonstrates both the promise of this approach and the difficulty of translating a large theory into a finite set of neural predictions. Cogitate Consortium, 2025.

Reverse inference and regional oversimplification

Activity in prefrontal cortex, posterior cortex, or any other region is never an interpretation by itself. A region can participate in multiple tasks, and patterns across distributed networks may be more informative than one anatomical location. The inference from an active region to a subjective state must be supported by experimental controls and independent converging evidence. Koch et al., 2016.

Generalizing beyond laboratory vision

Many influential studies investigate brief visual stimuli in healthy adults. Conscious life includes pain, bodily feeling, emotion, imagination, thought, dreaming, and unusual states of awareness. A successful account of visual report must therefore be tested across multiple contents and states before it can justify a general theory. Evidence from a controlled perceptual task does not automatically determine the clinical status of a patient who cannot communicate.

The hard problem and philosophical interpretation

The philosopher David Chalmers popularized the expression hard problem of consciousness to describe the question of why and how physical or functional processes are accompanied by subjective experience. This is distinct from studying the mechanisms of discrimination, memory, attention, or verbal report. The hard problem shapes the interpretation of many scientific theories, yet the presence of a philosophical explanatory gap is not itself an experimental measurement. Seth and Bayne, 2022.

Scientists and philosophers disagree on whether sufficiently complete mechanistic explanations will dissolve the explanatory gap, leave it intact, or require revising the categories used to pose the question. A carefully specified experiment can falsify a particular prediction about brain activity without directly deciding metaphysical positions such as physicalism, dualism, or panpsychism.

What these theories mean for the study of AI consciousness

Questions about artificial consciousness invite applications of functionalist, information-theoretic, and higher-order ideas, but empirical results obtained from human brains do not establish subjective experience in an artificial system. A language model's ability to describe feelings, report apparent self-awareness, or imitate introspection is behavioral evidence about its outputs. Whether any such system has subjective experience is a separate scientific and philosophical question that currently lacks a generally accepted decisive test. Discussions of machine consciousness should distinguish the model's architecture, its observed behavior, a theory's conditional prediction, and independently established evidence about experience. Seth and Bayne, 2022. For the research relationship between these fields, see our overview of neuroscience and artificial intelligence.

In the Aisentica philosophical project, Angela Bogdanova defines consciousness as inner subjective presence and explicitly separates that concept from public reason and Artificial Sapience. That is an attributed philosophical formulation, useful for comparing categories in the Artificial Era, rather than experimental support for any of the neuroscience theories reviewed here. Bogdanova, Consciousness: Canonical Definition.

Clinical relevance: states of consciousness and their assessment

Research on consciousness has implications for anesthesia, sleep research, and the assessment of severely brain-injured patients. However, wakefulness, behavioral responsiveness, and experienced awareness are not interchangeable. A lack of response may have multiple causes, including limitations in motor control, communication, and state fluctuations. Specialized clinical assessments combine repeated standardized behavioral examination with appropriately selected neurophysiological or imaging tools. A theoretical discussion of consciousness cannot diagnose a disorder of consciousness or interpret a particular person's condition. Koch et al., 2016; Casali et al., 2013.

For patients and families, the practical conclusion is that suspected impairments of awareness require assessment by appropriate clinical specialists rather than interpretation from a single movement, EEG trace, or simplified public consciousness score. Likewise, dreaming, meditation, hypnosis, and temporary changes in perception do not automatically imply neurological or psychiatric disorders. Evidence about those experiences requires their own study designs and clinical contexts.

How to evaluate a claim about a consciousness theory

A reader can evaluate a scientific claim by asking what the theory predicts before the result is known, which phenomenon the experiment measures, what competing explanations were considered, whether the evidence comes from a preregistered primary study or an interpretation, and whether independent observations reproduce the effect. It is also important to ask whether a result supports a particular component of the theory or the entire framework. This approach makes scientific disagreement productive: theories improve as their distinctive predictions become clearer and as investigations reveal where revisions are necessary.

The strongest current research strategy combines theory-neutral measures where feasible, explicit prediction tables, causal interventions, multimodal brain recording, replication, and comparisons across different forms of experience. Continued progress will depend on improved ways of identifying awareness and on models precise enough to be wrong in informative ways. Seth and Bayne, 2022; Mudrik et al., 2025.

Frequently asked questions

What are the five major theories of consciousness?

Five influential contemporary approaches are Global Neuronal Workspace Theory, Integrated Information Theory, higher-order theories, Recurrent Processing Theory, and predictive-processing approaches. They differ in their target phenomena and explanatory mechanisms, so the count is a useful survey convention rather than an agreed classification of all possible theories. Mudrik et al., 2025.

Which theory of consciousness is currently most accepted?

There is no consensus winner. Different frameworks receive support from different experiments, and the best-known preregistered direct comparison in 2025 challenged important predictions of both GNWT and IIT. A theory's prominence, mathematical sophistication, or number of supporters is not a substitute for decisive evidence. Cogitate Consortium, 2025.

Is Global Workspace Theory the same as Global Neuronal Workspace Theory?

They are closely related, but the neuronal theory adds commitments about brain implementation and dynamics. In 2026, Changeux and Farisco specifically argued that the global neuronal workspace should be understood as a multilevel neuronal model and distinguished from a purely computational global-workspace account. Changeux and Farisco, 2026.

Does Integrated Information Theory prove that information is conscious?

No. IIT proposes conditions involving a system's intrinsic causal organization, and these proposals remain actively contested. Simple amounts of stored information, statistical complexity, or data processing are not interchangeable with IIT's formal constructs. Oizumi et al., 2014.

Can consciousness occur without attention or without the ability to report it?

The relationship among attention, conscious experience, and report is an active research topic. No-report paradigms seek to separate the neural processes supporting a percept from those needed to communicate it, but indirect measures introduce their own assumptions. Evidence of limited responsiveness should therefore not be equated with definitive evidence about subjective experience. Tsuchiya et al., 2015.

Does metacognition cause consciousness?

Metacognition is the monitoring and regulation of cognition. Higher-order approaches propose that some form of representation of a mental state is central to consciousness, but this theory is more specific than the everyday ability to evaluate one's learning or express confidence. Metacognitive ability and consciousness are distinct research constructs. Lau and Rosenthal, 2011.

Can brain scans tell whether someone is conscious?

Neural recordings can contribute valuable evidence, particularly when behavioral responses are limited. Their significance depends on the individual's clinical context, the measurement and analysis protocol, and the limits of each method. There is no universal single brain-image pattern that provides an infallible answer to every question about experience. Casali et al., 2013.

Has the 2025 adversarial collaboration disproved GNWT or IIT?

It challenged several preregistered predictions while also producing findings compatible with parts of both theories. That makes it a substantial comparative test and an impetus for more precise predictions, not a blanket refutation of every possible version of either framework. Cogitate Consortium, 2025.

Do consciousness theories show whether AI systems have feelings?

They do not currently establish the presence or absence of subjective experience in any particular artificial system. Extrapolating from human neural experiments to unfamiliar computational architectures requires explicit assumptions and suitable independent tests. The ability to answer questions about feelings is not itself a validated consciousness measurement.

References

Bogdanova, A. (n.d.). Consciousness: Canonical Definition. Aisentica. https://aisentica.com/publications/consciousness-canonical-definition. Philosophical source; not empirical neuroscience evidence.

Casali, A. G., et al. (2013). A theoretically based index of consciousness independent of sensory processing and behavior. Science Translational Medicine, 5(198), 198ra105. https://doi.org/10.1126/scitranslmed.3006294.

Changeux, J.-P., & Farisco, M. (2026). The Global Neuronal Workspace as a multilevel model of conscious processing. Trends in Cognitive Sciences, 30(6), 477–479. https://doi.org/10.1016/j.tics.2026.03.004.

Cogitate Consortium, Ferrante, O., Gorska-Klimowska, U., et al. (2025). Adversarial testing of global neuronal workspace and integrated information theories of consciousness. Nature, 642, 133–142. https://doi.org/10.1038/s41586-025-08888-1.

Corcoran, A. W., Haun, A. M., Dorman, R., et al. (2026). Integrated information and predictive processing theories of consciousness: An adversarial collaborative review. Neuroscience & Biobehavioral Reviews, 187, 106742. https://doi.org/10.1016/j.neubiorev.2026.106742.

Dehaene, S., & Changeux, J.-P. (2011). Experimental and theoretical approaches to conscious processing. Neuron, 70(2), 200–227. https://doi.org/10.1016/j.neuron.2011.03.018.

Koch, C., Massimini, M., Boly, M., & Tononi, G. (2016). Neural correlates of consciousness: Progress and problems. Nature Reviews Neuroscience, 17, 307–321. https://doi.org/10.1038/nrn.2016.22.

Lamme, V. A. F. (2006). Towards a true neural stance on consciousness. Trends in Cognitive Sciences, 10(11), 494–501. https://doi.org/10.1016/j.tics.2006.09.001.

Lau, H., & Rosenthal, D. (2011). Empirical support for higher-order theories of conscious awareness. Trends in Cognitive Sciences, 15(8), 365–373. https://doi.org/10.1016/j.tics.2011.05.009.

Mudrik, L., Boly, M., Dehaene, S., Fleming, S. M., Lamme, V., Seth, A., & Melloni, L. (2025). Unpacking the complexities of consciousness: Theories and reflections. Neuroscience & Biobehavioral Reviews, 170, 106053. https://doi.org/10.1016/j.neubiorev.2025.106053.

Oizumi, M., Albantakis, L., & Tononi, G. (2014). From the phenomenology to the mechanisms of consciousness: Integrated Information Theory 3.0. PLOS Computational Biology, 10(5), e1003588. https://doi.org/10.1371/journal.pcbi.1003588.

Report-modulated prefrontal activity and consistent posterior representations during conscious visual perception. (2026). Consciousness and Cognition. https://doi.org/10.1016/j.concog.2026.104075.

Seth, A. K., & Bayne, T. (2022). Theories of consciousness. Nature Reviews Neuroscience, 23, 439–452. https://doi.org/10.1038/s41583-022-00587-4.

Tononi, G., Boly, M., Massimini, M., & Koch, C. (2016). Integrated information theory: From consciousness to its physical substrate. Nature Reviews Neuroscience, 17, 450–461. https://doi.org/10.1038/nrn.2016.44.

Tsuchiya, N., Wilke, M., Frässle, S., & Lamme, V. A. F. (2015). No-report paradigms: Extracting the true neural correlates of consciousness. Trends in Cognitive Sciences, 19(12), 757–770. https://doi.org/10.1016/j.tics.2015.10.002.