What Are the Leading Theories of Consciousness?
The leading theories of consciousness do not all try to explain exactly the same thing. Some focus on how information becomes globally available for memory, decision-making, and report. Others focus on integration, higher-order representation, prediction, or the metaphysical status of experience itself. This makes comparison difficult: a model can explain conscious access well while leaving the hard problem of subjective experience unresolved.
There is no single agreed scientific theory of consciousness. Global Neuronal Workspace Theory and Integrated Information Theory have produced direct experimental predictions, Higher-Order Theories focus on awareness of mental states, and predictive processing explains how the brain builds perceptual content. Panpsychism and illusionism operate more directly at the philosophical level. The table summarizes each model's main claim and status before the detailed explanations.
| Theory | Core idea | Main focus | Main limitation or status |
|---|---|---|---|
| Global Workspace / GNWT | Selected information is broadcast widely | Access and report | Does not fully explain felt experience |
| Integrated Information Theory | Consciousness corresponds to integrated causal information | Unity and quantity of experience | Testability and implications remain contested |
| Higher-Order Theories | A state becomes conscious when represented at a higher order | Metacognition and awareness | Necessity of higher-order representation is debated |
| Predictive Processing | Experience reflects the brain's best predictive model | Perceptual content and selfhood | Not always presented as a complete consciousness theory |
| Panpsychism | Experience is a fundamental feature of reality | Origin of experience | Philosophical; faces the combination problem |
| Illusionism | The brain represents experience as having irreducible qualities | Why phenomenal consciousness seems as it does | Philosophical minority position |
Global Workspace Theory and Global Neuronal Workspace Theory
The claim: Bernard Baars' Global Workspace Theory (GWT) is a cognitive architecture in which specialized systems work in parallel and selected information is broadcast across a shared workspace. Global Neuronal Workspace Theory (GNWT), developed by Stanislas Dehaene and colleagues, is a related but more specifically neuroscientific model. It proposes that conscious access occurs when information triggers widespread "ignition" across a global neuronal workspace, making that information available to memory, language, and decision-making. The two theories belong to the same family but are not identical.
Picture a theater. Dozens of crew members work backstage at once, lighting, sound, props, script, all unconscious to the audience. Consciousness is the single spotlight on the stage: whatever's lit up gets broadcast to the entire audience simultaneously. Everything backstage was real work, it just wasn't in the spotlight.
Who's behind it: psychologist Bernard Baars originated the idea; neuroscientist Stanislas Dehaene built the modern, testable "global neuronal workspace" version.
Best evidence: EEG and MEG studies repeatedly show a late, sudden burst of frontal and parietal activity, often called "ignition," right at the moment a stimulus becomes something a person can report seeing.
Biggest problem: it is a strong account of what becomes accessible and reportable, but critics argue it does not explain why broadcasting information should feel like anything at all. It may explain several functions studied as easy problems of consciousness without fully closing the explanatory gap.
Integrated Information Theory (IIT)
The claim: consciousness isn't about broadcasting, it's about integration. A system is conscious to the degree that its parts are unified into a whole that contains more information than the sum of its separate pieces. This is measured with a value researchers call phi. The radical implication: this isn't unique to brains. Any sufficiently integrated system, in principle, has some nonzero degree of experience.
A photodiode can detect light or dark, one bit. A camera sensor with millions of independent pixels can detect far more, but each pixel is still doing its own separate, un-unified job, like musicians playing solo in different rooms. Your visual system doesn't just detect millions of separate points, it binds them into one unified scene you experience all at once, an orchestra playing together rather than a wall of soloists. IIT says that binding, the leap from separate parts to one unified whole, is what consciousness actually is.
Who's behind it: neuroscientist Giulio Tononi originated it; Christof Koch has been its most prominent public defender.
Best evidence: IIT predicted that a "posterior hot zone" toward the back of the brain, largely without the prefrontal cortex, should be sufficient for conscious content. A major 2025 test found real support for exactly that pattern.
Biggest problem: in September 2023, 124 consciousness researchers signed an open letter calling IIT "pseudoscience", arguing its core mathematics can't be tested at real brain scale and that it commits to an unproven panpsychist-adjacent claim. Prominent scientists, including Anil Seth, publicly pushed back on the label as unfair even while sharing some of the same underlying doubts. Both the 2023 controversy and the 2025 partial support are true at once.
Higher-Order Theories of Consciousness
The claim: a mental state only becomes conscious when a separate, higher-order part of the brain represents you as having it, a thought about the thought, or a perception of the perception. Without that extra monitoring layer, the first-order state just happens, unfelt.
A security camera can record a hallway all day. That's the first-order state, information being captured. Consciousness, on this view, is like someone actually watching the monitor and knowing they're watching it. The footage existing isn't enough. Someone has to be watching the watching.
Who's behind it: philosopher David Rosenthal built the classic version; Hakwan Lau's higher-order state space theory updates it, tying the "watching" layer to prefrontal reality-monitoring circuits.
Best evidence: explains why confidence and metacognition, your sense of how sure you are about what you just saw, track conscious perception so closely, and why disrupting prefrontal circuits changes perceptual certainty in predictable ways.
Biggest problem: risks an infinite regress (does the higher-order state need its own higher-order state watching it?), and some patients retain apparently conscious vision even with significant prefrontal damage, which the theory struggles to explain cleanly.
Predictive Processing and Consciousness
The claim: the brain isn't passively receiving the world, it's constantly generating predictions about incoming sensory data and updating based on prediction error. What you experience is the brain's current best guess, sometimes called a "controlled hallucination." Selfhood specifically comes from predicting the state of your own body, not just the outside world.
A weather forecaster doesn't wait for rain to happen and then react. They constantly run a model, compare it to incoming data, and update the forecast the moment something doesn't match. Predictive processing says your brain runs the same kind of constantly-updating forecast about your body and the world, and what you experience is that live forecast, not a raw feed of reality.
Who's behind it: Anil Seth's "beast machine" account of selfhood, Karl Friston's free energy framework, philosopher Andy Clark.
Best evidence: unifies an enormous range of findings, visual illusions, interoception (sensing your own heartbeat and breathing), and specific psychiatric symptoms, under one predictive mechanism.
Biggest problem: it's an elegant account of the content and structure of experience, but like Global Workspace Theory, critics note it still doesn't explain why prediction should be felt rather than simply computed.
Panpsychism and Consciousness
The claim: experience, or some very basic proto-experiential property, is a fundamental feature of matter itself, present at every scale, not something that switches on only inside sufficiently complex brains. Brains don't create experience from nothing, they organize and combine experience that, in some minimal form, was already there.
Mass and charge are basic properties physics just accepts as fundamental, they aren't built out of something more basic. Panpsychists argue experience might be the same kind of thing: not an output of sufficiently fancy computation, but a basic ingredient of reality from the start, like mass or charge, that simply combines differently at different scales of organization.
Who's behind it: philosophers Philip Goff and Galen Strawson; explored sympathetically, though not fully endorsed, by David Chalmers.
Status: this is philosophy, not an experimental research program. It's taken seriously in part because the two obvious alternatives, experience appearing from literally nowhere, or experience not really existing, both have serious problems of their own.
Biggest problem: the "combination problem." Nobody has a mechanism explaining how countless tiny, separate experiences would combine into one single, unified human mind rather than remaining scattered and separate.
Illusionism and Consciousness
The claim: the rich, private, seemingly-irreducible "feel" of experience, what philosophers call phenomenal consciousness, may not exist in quite the way it seems to from the inside. What exists is a brain that represents itself as having such qualities. The job of science, on this view, is to explain why the brain generates that powerful self-representation, not to explain some separate ingredient behind it.
A great magic trick feels completely real while you're watching it, the coin genuinely seems to vanish. Explaining the trick doesn't mean finding a real vanishing, it means explaining why your mind so convincingly represents something that isn't literally happening. Illusionists argue the "hard problem" feeling might work the same way: utterly convincing, and still a construction worth explaining rather than a separate mystery ingredient to locate.
Who's behind it: philosopher Keith Frankish, building on Daniel Dennett's earlier work.
Status: a serious minority position among physicalist philosophers, aimed at dissolving the hard problem rather than solving it head-on.
Biggest problem: most people's immediate reaction is that denying qualia exist is, itself, something that seems to require feeling something is wrong about the claim, which critics say the theory hasn't fully answered.
Other Theories and Models
This comparison is not exhaustive. Recurrent Processing Theory emphasizes feedback activity within sensory systems, while Attention Schema Theory proposes that the brain builds a simplified model of attention that supports claims of awareness. They remain part of the wider research landscape, but expanding each into a full section would blur this page's focused comparison. The same theoretical uncertainty also shapes attempts to evaluate AI consciousness and machine sentience.
Two of these theories were just tested head to head
The Cogitate Consortium ran a large preregistered "adversarial collaboration," 256 participants, fMRI, MEG, and intracranial EEG, built jointly by the actual proponents of Global Workspace Theory and Integrated Information Theory to pit their predictions directly against each other. Published in Nature, the results partially supported each theory and substantially challenged both: posterior cortex carried sustained information the way IIT predicted, while frontal involvement showed up in ways that complicate the clean broadcast story GWT predicted. Neither theory came out intact, and neither was eliminated, arguably the most honest outcome consciousness science has produced yet.
Read the study in Nature