Why does red feel like red?

256 people looked at images for varying lengths of time while three measurement methods recorded simultaneously what was happening in their heads: functional magnetic resonance imaging, magnetoencephalography and electrodes placed directly in the brain. The design was unusual. Two competing theories of consciousness had publicly committed themselves in advance to which findings would count for them and which against; an uninvolved consortium ran the study.

The result, published in 2025, was uncomfortable for both sides. Some of the predictions held, while central assumptions of both theories were seriously called into question. And even if one of them had won cleanly, one question would have remained untouched: why is all this processing accompanied by experience at all? A scanner can show which networks become active when a person sees red. It can track attention, report and decision. What red looks like for that person is not visible in the data.

Chalmers did not stop at the diagnosis. In the same paper he proposes not to reduce experience to physical processes but to posit it as a fundamental feature of the world — alongside mass, charge and spacetime — and to connect it to physics through psychophysical laws of its own. He sketches three such principles: a structural coherence between experience and cognitive access; an organisational invariance, on which systems of the same functional organisation have the same experience; and a double-aspect theory of information, on which information always has a physical and a phenomenal side. Chalmers calls this an outline himself, not a theory. The price is plainly visible: a quantity that cannot be measured is admitted into the basic furniture of the world.

Easy problems and one hard one

David J. Chalmers named this gap the “hard problem” of consciousness in 1995. The “easy problems” are by no means easy — the label is Chalmers’s own irony. They ask how the brain discriminates stimuli, integrates information, controls behaviour or makes content available for report. Difficult, but formulable in functional terms: one can say which mechanism would have to deliver the performance, and then go and look.

The hard problem asks in addition why there is any subjective experience involved — the hurt of pain, the timbre of a sound, the taste of coffee. Chalmers had a precursor: in 1974 Thomas Nagel had asked what it is like to be a bat, and used the question to show that the subjective character of experience escapes objective description. In 1983 Joseph Levine gave the distance the soberer name “explanatory gap”.

Nagel’s example was deliberately an alien animal. Bats orient themselves by echolocation, a sense resembling none of our own and for which our language holds no experiential terms. We can imagine what it would be like for us to behave as a bat does; what we cannot reach is what it is like for the bat to be one. That, for Nagel, is the heart of it: experience has a standpoint, and objective description works methodically away from every standpoint.

What the theories explain

Global workspace theory describes consciousness as global availability: content is “broadcast” through a far-reaching network and can influence memory, speech and action planning. Anil K. Seth and Tim Bayne sort the field into four families. Alongside the global workspace stand higher-order theories, for which a state becomes conscious when a further state represents it; approaches that build on recurrent processing and prediction; and integrated information theory.

Such models generate testable predictions about access and processing. That is exactly what made the 2025 study possible, in which global workspace and integrated information were pitted directly against one another. Integrated information was contradicted by the absence of the expected sustained synchronisation in the posterior cortex; the global workspace was contradicted by the largely missing ignition expected when the stimulus disappeared. That is progress on the easy problems. It is not a step on the hard one.

The objections to the hard problem

The dispute begins where interpretation sets in — and it is aimed not only at possible answers but at the question. Patricia S. Churchland regards the division into easy and hard problems as premature. Nobody, she argues, can know in advance which problem will turn out to be the fundamentally insoluble one; historically, questions held to be closed in principle have dissolved once the mechanism was found. She calls this, ironically, the “hornswoggle problem”: a label that elevates a difficulty into a special class before the work has been done.

Since Consciousness Explained, Daniel C. Dennett has held the deflationary counter-position: what has to be explained are the capacities and the reports, and once those are fully explained, nothing is left over still awaiting explanation. Keith Frankish has shaped this into illusionism. His proposal shifts the task: what needs explaining is not phenomenal experience but why it seems so undeniably to us that we have it.

The other side has its own tools, and they are thought experiments, not laboratory findings. In 1996 Chalmers himself introduced the philosophical zombie: a being that behaves exactly like a human, gives every answer, passes every test — and experiences nothing. If such a being is so much as conceivable, the argument goes, then describing the behaviour does not describe everything. The objection to this is just as old: conceivability is a state of our imagination, not a finding about the world.

In 1982 Frank Jackson devised the scientist who knows all the physical facts about colour and has nevertheless never seen red: does she learn something new at first sight? If so, the physical knowledge was not complete.

The most remarkable turn in this debate comes from Jackson himself. A good twenty years later he abandoned his own argument and declared that the appearance of an additional fact is deceptive. Anyone who wants to know how seriously this dispute is conducted will find the proof here: the author of the strongest objection has withdrawn it, and the objection continues to be used anyway — because it does not depend on its author.

Why the question stays practical

Consciousness research needs reports, behaviour and brain data, and every route has its limits. Reports are subjective, behaviour can arise without conscious access, and a neural correlation is not yet an explanation. Progress comes about the way 2025 demonstrated: when theories say in advance what they expect and which finding would refute them.

The consequences are not academic. Whether animals, infants, people without behavioural responses or future machines possess experience cannot be settled by asking them. Researchers therefore look for bundles of indicators rather than a single marker of consciousness. The less direct the access, the more important ethical caution becomes: missing behaviour must not be equated too quickly with missing experience. At the same time, a theory has to do more than reconcile every possible result with consciousness after the fact.

What remains

What holds: the easy problems are in motion. There are competing theories, they make different predictions, and those predictions can be tested against one another — the 2025 study showed that a field can bear this without either side winning. What remains contested is the status of the question that is left over. For one side it is the real one; for the other a conceptual error that dissolves as soon as the mechanics are completely described. This disagreement is old, it cannot be settled by an experiment, and it has not narrowed in thirty years.

How to recognise the debate in everyday life: by a silent change of subject. Where someone explains which networks become active during a sensation and concludes that the sensation has thereby been explained, the description of a mechanism has been passed off as the answer to a different question. That may be correct — it is the position of several serious philosophers. It is then a thesis, though, and not a measurement.