What is consciousness? Ask ten researchers and you may get ten definitions, but most would start in the same place: consciousness is the fact that there is something it is like to be you. Light on your retina is not just processed; it is seen. A pinprick is not just detected; it hurts. Explaining why that inner side of mental life exists, and which physical conditions it needs, is one of the oldest open problems in science and philosophy.

There is no agreed theory. There are several serious contenders, each with supporters, critics and a different idea of what the brain is doing when you are aware. This article compares the main ones, explains how they are being tested, and flags where the evidence runs out.

What do scientists mean by “consciousness”?

The word covers more than one thing, which is part of the trouble. In the philosopher Thomas Nagel’s phrase, an organism is conscious if there is something it is like to be that organism, a subjective character that physical description alone seems to leave out.

Ned Block’s distinction between phenomenal and access consciousness sharpens this. Phenomenal consciousness is the raw experience: the redness of red, the feel of pain. Access consciousness is information that is available for reasoning, report and control of action. The two usually go together, and whether they can come apart is itself a point of dispute that affects how experiments are read.

Researchers also separate consciousness as a state (awake, asleep, anesthetized) from consciousness of something in particular, such as a face on a screen. Most laboratory work studies the second kind, because it can be manipulated: show a picture briefly, mask it, and ask whether the person saw it.

Is there really a “hard problem”?

David Chalmers named it in a 1994 talk and set it out in a 1995 paper, then developed it in his 1996 book The Conscious Mind. He divides the field. The easy problems concern functions: how the brain discriminates stimuli, integrates information, focuses attention and produces verbal reports. These are hard to solve in practice but are the kind of thing standard science can explain by describing mechanisms.

The hard problem asks why any of that processing should be accompanied by experience at all. Chalmers argues that even a complete functional account would leave this question open, since one can imagine a system doing everything a brain does without any inner life.

Philosophers split on this. Some, including Daniel Dennett and Patricia Churchland, hold that the hard problem dissolves once the easy problems are solved. Illusionists such as Keith Frankish go further and say phenomenal consciousness is itself a kind of illusion, so the task is to explain why it seems real. Others favor dualism or panpsychism, which treats consciousness as fundamental to matter. Colin McGinn’s “new mysterianism” suggests human minds may be unable to solve the problem.

How common is each view? A 2020 PhilPapers survey found that 62.4 percent of philosophers considered the hard problem genuine and 29.7 percent did not. Neuroscientists as a group tend to expect progress on the easy problems to reduce or remove the puzzle, though a significant minority disagree.

How do researchers look for consciousness in the brain?

By hunting for neural correlates of consciousness, abbreviated NCC: patterns of brain activity that reliably track whether a person is aware of something. The standard method contrasts trials where a stimulus is consciously perceived with matched trials where it is not.

Several candidates have been proposed over the decades. Gamma-band oscillations were suggested in the 1980s by von der Malsburg and Singer as a way to bind features of an object together. In 2013, a measure called the perturbational complexity index was introduced as a quantitative marker of conscious states. Francis Crick and Christof Koch speculated in 2004 that the claustrum, a thin sheet of neurons, might coordinate awareness; a 2014 case study seemed to fit, though its authors called it inconclusive, and a 2022 study described the claustrum as more of a router.

The caution here is basic. A correlate is not an explanation. Finding that activity in a region goes along with awareness does not show why it feels like anything, and it does not tell you whether the activity causes awareness, results from it, or is merely needed to report it.

What does global workspace theory say?

Bernard Baars introduced it in 1988 with a theater metaphor. The brain runs many specialized processes in parallel, mostly out of awareness. Consciousness is the stage: contents that win the competition for the spotlight are broadcast to the rest of the system, making them available for memory, planning, language and control. Baars rejects the idea of a little observer watching the stage.

Stanislas Dehaene and colleagues extended it into the global neuronal workspace theory, tying the broadcast to long-range connections linking prefrontal, parietal and temporal regions. A distinctive prediction is “ignition”: a sudden, large-scale activation of these networks when a stimulus becomes conscious.

The theory is functionalist. Consciousness is what a certain kind of information broadcast is. Critics argue it explains the cognitive role of consciousness without explaining its nature, leaving the hard problem where it was. Another criticism is methodological: much support comes from studies of unconscious priming, and some argue the methods are flawed. A third concerns whether the prefrontal cortex is really involved, as the theory suggests, or whether activity there reflects the act of reporting.

What does integrated information theory say?

Giulio Tononi proposed integrated information theory (IIT) in 2004, and it starts from the opposite end. Instead of asking what the brain does, it begins with the self-evident fact of experience and asks what kind of physical system could support it. It identifies five essential properties of any experience (intrinsicality, information, integration, exclusion and composition) and maps each to a requirement on a system’s cause-and-effect structure.

The central quantity is phi, which measures how irreducible a system’s cause-effect power is. A high value means the system is more than the sum of its parts. Exact values are computationally intractable for large systems, and proxy measures can disagree.

IIT has unusual implications. It ties consciousness to physical architecture, not to the software running on it, so a digital computer simulating a brain could be functionally equivalent without being conscious. A 2024 preprint by Findlay, Tononi and colleagues argued exactly this. It also implies that simple but richly connected systems might have some experience, which critics such as John Searle call panpsychism in disguise. Tononi accepts some of these implications.

IIT is also the center of a public dispute. In September 2023, 124 scholars signed a letter calling it pseudoscience, arguing that its core assumptions cannot be tested well enough. Others objected to the label; a later survey of consciousness researchers found only 8 percent of the 60 who responded fully agreed with the letter, and 20 percent did not agree at all. Nature’s editors said the pseudoscience label had no place in collaborative science.

What are higher-order theories?

Higher-order theories hold that a mental state is conscious when you have a thought or perception about it. David Rosenthal’s version says a state is conscious when it is the target of a higher-order thought. A visual signal in your cortex becomes conscious experience when a further representation says, in effect, “I am seeing this.”

The motivation is that these views can account for cases such as pain with no accompanying reaction, and they fit evidence that unconscious processing is powerful, with conscious report arriving after it. They draw on the contrast between first-order information in early sensory areas and higher-order representations in prefrontal and parietal regions.

Objections include the computational cost, since every experience would need another representation about it, and the problem of misrepresentation: what happens if the higher-order state says you are seeing red when the first-order state is orange? Variants try to address this, for instance by making the higher-order content part of the first-order state itself or only a disposition to form such a thought.

What other theories are in the running?

Several. None has the standing of the three above, but each has researchers working on it.

  • Recurrent processing theories stress feedback loops. Gerald Edelman’s dynamic core hypothesis emphasized reentrant connections, and evidence suggests top-down feedback predicts conscious awareness better than purely feedforward activity does.
  • Predictive processing frameworks treat the brain as a prediction machine. In 2017, David Rudrauf, Karl Friston and colleagues applied active inference to propose a projective consciousness model, though they say it does not close the explanatory gap.
  • Attention schema theory, from Michael Graziano and Sabine Kastner in 2011, proposes that awareness is the brain’s simplified model of its own attention.
  • Orchestrated objective reduction, by Roger Penrose and Stuart Hameroff, locates consciousness in quantum processes in neurons. Discovery of quantum vibrations in microtubules revived interest in 2014, but a 2022 experiment led by Catalina Curceanu made the idea appear highly implausible.

These categories overlap. Some predictive-processing authors, for example, build on recurrent loops, and a dispositional higher-order theory can lean on global broadcasting.

Which theory is winning?

None, and a large collaboration recently put two of them head to head. The Cogitate Consortium, including proponents of both theories, ran an adversarial test of global neuronal workspace theory against IIT, with predictions registered in advance. It was published in Nature on 30 April 2025.

The study used 256 participants viewing visual stimuli while researchers recorded brain activity with fMRI, MEG and intracranial EEG. Information about what people consciously saw turned up in visual, ventrotemporal and inferior frontal cortex, and there was content-specific synchronization between frontal and early visual areas.

The conclusion was even-handed. The results matched some predictions of each theory and substantially challenged key tenets of both. For IIT, the lack of sustained synchronization within posterior cortex conflicts with the claim that network connectivity specifies consciousness. For the workspace theory, the general lack of ignition at stimulus offset and the limited representation of certain conscious dimensions in prefrontal cortex are problematic. Neither side was knocked out. The authors argued for more adversarial, theory-driven testing and for a quantitative framework for building theories.

Here is the state of play at a glance:

  • Global workspace: consciousness is broadcast to many systems. Strength: ties to cognition and attention. Challenge: may explain function but not experience, and the 2025 results questioned ignition.
  • Integrated information: consciousness is integrated cause-effect structure. Strength: starts from experience. Challenge: hard to compute, contested testability.
  • Higher-order: consciousness is representing your own mental states. Strength: explains dissociations. Challenge: cost and misrepresentation.

Common questions

Is consciousness the same as intelligence?

No. Intelligence concerns problem-solving and behavior, while consciousness concerns experience. The theories above treat the relation differently, which is why they give different answers about machines. A system could, in principle, be capable without having experience, or the reverse, depending on which theory is correct.

Where in the brain is consciousness?

No single place has been established. Different theories point to posterior cortex, to prefrontal and parietal networks, or to distributed loops. The 2025 adversarial study found information about conscious content in visual, ventrotemporal and inferior frontal regions.

Can a computer be conscious?

It depends on the theory, and the theories disagree. Functionalist accounts such as global workspace theory allow in principle that a machine with the right organization could be. IIT says that a conventional digital computer would have negligible consciousness regardless of what it computes. No consensus exists.

Has anyone solved the hard problem?

No. Some philosophers deny it is a real problem, and others think it is real and unsolved. A 2020 survey found a majority of philosophers surveyed regarded it as genuine.

Is quantum mechanics involved?

That is a minority hypothesis. Penrose and Hameroff proposed it, but a 2022 experiment led by Catalina Curceanu made quantum consciousness appear highly implausible. Most researchers work with classical neural explanations.

What does “neural correlates of consciousness” mean?

It refers to brain activity that reliably accompanies conscious experience. Finding correlates is a step toward theory, but a correlate alone does not explain why experience exists.

What to remember

Consciousness is the subjective side of mind, and no theory has yet explained it in a way that most researchers accept. Three families lead the field: workspace theories that focus on broadcast, integrated information theory that focuses on causal structure, and higher-order theories that focus on self-representation. Others add feedback loops, prediction and attention.

What is established: the phenomenon is studied with increasingly careful methods, and in 2025 a large pre-registered collaboration challenged key claims of two leading theories without confirming either. What is not: whether the hard problem is real, which theory is closest, and whether non-brain systems could be conscious. Claims of a solved mystery are premature.

Further reading

  • Being You by Anil Seth: A readable account of the predictive-processing view of consciousness by a neuroscientist. See it on Amazon (affiliate link)
  • The Feeling of Life Itself by Christof Koch: Koch explains integrated information theory from the viewpoint of a supporter. See it on Amazon (affiliate link)

As an Amazon Associate I earn from qualifying purchases. A commission may be earned if you buy through this link, at no extra cost to you.

Sources