Experience is the part visible only from inside
A brain can discriminate colors, direct attention, store a memory, and produce the sentence “I see red.” Those are processes we can observe indirectly, manipulate, and connect to neural activity. Consciousness adds a stubborn fact: red also looks like something to the person seeing it.
That first-person character is the simplest working definition of consciousness. There is something it is like to taste coffee, feel embarrassment, or hear a trumpet. Philosophers often call the qualities of those experiences qualia. The term is technical; the examples are ordinary.
The definition matters because several neighboring abilities can be mistaken for consciousness. Intelligence is an ability to learn, plan, or solve problems. Detection is a system’s response to information. Reportability is the ability to say what was noticed. These abilities often travel with consciousness in adult humans, but none is identical to it. A baby may experience pain without describing it. A machine can classify damage without hurting.
The easy problems are difficult but approachable
David Chalmers used “easy problems” for questions about functions such as attention, discrimination, memory, behavioral control, and verbal report. The name does not mean neuroscientists can answer them casually. It means we can see the outline of a mechanistic explanation.
We can ask which brain regions exchange signals, how information becomes available to memory, or why anesthesia interrupts responsiveness. A sufficiently detailed account could explain how the system performs each task.
The hard problem begins after that account. Why is any of the processing accompanied by experience? Why is there an inside view rather than a perfectly capable system operating in the dark? Describing more functions does not obviously insert the painfulness into pain or the blueness into blue.
Some philosophers reject this extra gap. Once every capacity and disposition has been explained, they argue, nothing remains. Others think this response changes the subject from experience to what experience enables us to do. The disagreement is not over whether brains perform functions. It is over whether function exhausts the phenomenon.
The brain connection is strong
Consciousness changes when the brain changes. Anesthetic drugs can remove it. Injury can alter vision, memory, personality, or the sense of self. Sleep and waking have measurable neural signatures. Electrical and chemical interventions can modify experience without consulting anything outside the body.
These interventions strongly support consciousness’s dependence on the brain. An explanation must go further than listing which changes accompany which experiences: it must account for the relationship. The hard problem asks whether a full account of the brain’s functions would also explain why those functions have a first-person character.
Workspace theories explain access
Global neuronal workspace theory pictures much brain processing as specialized and local. Information becomes consciously accessible when it is broadcast widely enough to influence reasoning, memory, decision-making, and report.
The model captures an important contrast. An unnoticed signal might guide posture or prime a response while remaining confined. A consciously seen plume of smoke can be described, remembered, investigated, and used to change plans. Broadcasting explains that flexibility.
The remaining question is whether widespread access constitutes experience or merely describes what information can do once conscious. A noticeboard can coordinate an office. Coordination alone does not tell us why the noticeboard should have a point of view.
Integration begins from unity
Integrated information theory starts from another feature: experience is both differentiated and unified. A moment can contain color, shape, sound, bodily sensation, and location, yet it arrives as one experience rather than separate packets.
The theory connects consciousness to the way a system’s causal information is integrated, such that the whole cannot be reduced to independent parts. This is attractive because it attempts to connect experience to a formal property. It is controversial because the measurement is demanding and can attribute consciousness in counterintuitive places.
A 2025 study directly comparing integrated information and global neuronal workspace theories challenged preregistered predictions from both. That is useful scientific pressure, but it has not produced a winner that turns consciousness into a settled quantity.
Self-representation and prediction offer other routes
Higher-order theories propose that a mental state becomes conscious when the system in some sense represents itself as having that state. A visual representation of red is not sufficient; the mind must register that it is seeing red.
Predictive approaches emphasize the brain’s ongoing models of the world and body. Perception is not a passive camera feed. The system predicts what is present, compares that model with incoming signals, and updates errors. This can explain why context and expectation shape experience.
Both approaches can clarify architecture without automatically solving the hard problem. A self-model may explain why a system can report its state. Prediction may explain the structured content of perception. A critic can still ask why either process feels like anything.
Zombies expose the gap—and its uncertainty
A philosophical zombie is imagined as physically and behaviorally identical to a conscious person but with no inner experience. If that situation is genuinely conceivable, Chalmers argues, physical facts may not logically entail consciousness.
Critics answer that the apparent possibility may reflect confused concepts. Saying “an identical brain with no experience” might be like saying “a four-sided triangle”: grammatically tidy but not coherent. The thought experiment does not supply an unconscious duplicate. It tests what we believe follows from a complete physical description.
It also sharpens the problem of other minds. We never inspect another person’s experience directly. We infer it from behavior, reports, shared biology, and similar development. Those inferences are strong for other humans and less straightforward as nervous systems and behavior diverge.
The same caution applies to machines. Fluent claims about fear or suffering are not proof of experience. Our companion article on whether an AI could be conscious examines why behavior, architecture, and theory-based indicators must be kept separate.
There may be no operator in the control room
The image of a small observer watching mental screens creates an infinite regress: what makes the observer conscious? Adding a second control room inside its head only moves the question.
Consciousness may instead be a process made by the system’s activity—broadcast, integration, self-representation, biological regulation, prediction, or an organization not yet captured by one theory. We know many conditions under which experience appears, changes, and disappears. We do not yet know why those conditions give the machinery an inner surface.
That uncertainty matters because conscious beings can suffer. It also prevents a neat shortcut. Intelligence is not a consciousness meter, verbal confidence is not a scan, and a detailed mechanism must still explain the lights, not merely the switches.
