The Hard Problem of Consciousness

← Encyclopedia

The hard problem of consciousness is the question of why there is experience at all — why the physical activity of a brain, which can in principle be described completely, is accompanied by an inner feel rather than running, as David Chalmers put it, “in the dark.” Neuroscience explains more of what a brain does every year. It has not explained why doing any of it feels like something.

Chalmers drew the line at the first Toward a Science of Consciousness conference in Tucson, Arizona, in 1994, and published it the next year in the Journal of Consciousness Studies as “Facing Up to the Problem of Consciousness.” His move was to split a single word in two. The “easy problems” are, in his phrase, “those that seem directly susceptible to the standard methods of cognitive science, whereby a phenomenon is explained in terms of computational or neural mechanisms” — discriminating and categorizing stimuli, integrating information, reporting mental states, focusing attention, the difference between waking and sleep. These are easy only in a special sense: each asks how a function is performed, and exhibiting the mechanism finishes the job. The hard problem is what remains when all of them are solved. Explain discrimination, integration, access, and report down to the last synapse, and a further question still stands. “The really hard problem of consciousness,” Chalmers wrote, “is the problem of experience.” What makes it hard, in his own gloss, is that it “goes beyond problems about the performance of functions”: nothing in the explanation of a function says why performing it should be felt. “Why doesn’t all this information-processing go on ‘in the dark’, free of any inner feel?” Chalmers was modest about the contribution — a catchy name, he said later, for philosophically familiar points.

The familiarity runs deep. In 1714 Leibniz imagined a brain enlarged until one could walk through it as through a mill: among all the pushing parts, an observer would nowhere see a thought. In 1866 T. H. Huxley judged the emergence of consciousness from nervous tissue as unaccountable as the appearance of the genie when Aladdin rubbed his lamp. In 1872 the physiologist Emil du Bois-Reymond told a congress of German scientists that consciousness marked a permanent limit of natural knowledge — ignoramus et ignorabimus, we do not know and will not know. The modern formulations that Chalmers inherited come from three short papers written within a decade of one another, and each isolates a face of the same difficulty.

Thomas Nagel set the terms in 1974 in “What Is It Like to Be a Bat?”, published in the Philosophical Review. An organism is conscious, Nagel proposed, exactly when there is a felt interior to its existence: “the fact that an organism has conscious experience at all means, basically, that there is something it is like to be that organism.” He called this the subjective character of experience, and observed that it is “not analyzable in terms of any explanatory system of functional states” — the very states that the science of the brain trades in. His example was chosen to make the gap vivid rather than to settle it. A bat navigates by sonar, building a detailed picture of distance, size, shape, and motion from the echoes of its own shrieks; we can describe the mechanism exhaustively and still have no purchase on what the world is like for the bat, because, in Nagel’s words, “every subjective phenomenon is essentially connected with a single point of view, and it seems inevitable that an objective, physical theory will abandon that point of view.” Physical science advances precisely by leaving the point of view behind — by describing the world as it is from nowhere — and that is exactly what experience cannot survive losing.

Frank Jackson pressed the same wedge from the side of knowledge. His 1982 paper “Epiphenomenal Qualia,” in the Philosophical Quarterly, introduced the figure who has carried the argument ever since: “Mary is a brilliant scientist who is, for whatever reason, forced to investigate the world from a black and white room via a black and white television monitor.” She specializes in the neurophysiology of vision and learns, by stipulation, all the physical facts there are about what happens when a person sees a ripe tomato or the sky — every wavelength, every retinal response, every nerve impulse down to the words “the sky is blue.” Then she is let out. “It seems just obvious,” Jackson wrote, “that she will learn something about the world and our visual experience of it. But then it is inescapable that her previous knowledge was incomplete. But she had all the physical information. Ergo there is more to have than that, and Physicalism is false.” This is the knowledge argument: if a complete physical description leaves a fact unknown — what red is like — then experience is a fact over and above the physics. (Jackson, characteristically, later abandoned the conclusion while leaving the thought experiment to do its work for others.)

Joseph Levine gave the difficulty the name it now travels under. In 1983, in “Materialism and Qualia: The Explanatory Gap” (Pacific Philosophical Quarterly), he compared two identity statements. “Heat is the motion of molecules,” he argued, “expresses an identity that is fully explanatory, with nothing crucial left out”: once chemistry tells us how molecular motion plays the causal role we called heat, “there is nothing more we need to understand.” But “pain is the firing of C-fibers” leaves a residue. The firing explains pain’s causal role — the warning, the recoil — and yet “what is left unexplained by the discovery of C-fiber firing is why pain should feel the way it does.” There seems nothing about the firing that makes it fit that feeling rather than another; the identification, Levine wrote, “leaves the connection between it and what we identify it with completely mysterious,” making the way pain feels into a brute fact. Even granting that pain simply is the firing, the explanatory gap remains: an account of the mechanism that does not touch the feel. Chalmers’s distinction added no new mystery. It collected these three and isolated the very old one beneath them.

The argument that carries most of the metaphysical weight is the philosophical zombie: a being physically identical to a human, atom for atom, but with no inner experience — nothing it is like to be it. Robert Kirk coined the term in 1974; Chalmers built it into a formal argument in his 1996 book The Conscious Mind. If a zombie is so much as coherently conceivable, the reasoning runs, then conceivability tracks possibility, and if a zombie-world is possible then the physical facts do not by themselves fix the facts of experience — consciousness is something over and above the physics. Every step is contested. Daniel Dennett holds that zombies are not really conceivable: anyone who thinks they have imagined one has quietly imagined a person and then merely declared the lights off, underestimating how much of the imagining the inner feel was doing. Others grant the conceiving but deny that conceivability shows possibility — the inference from “I can imagine it” to “it could be so” is exactly what a physicalist will refuse. The 2020 PhilPapers survey of professional philosophers split three ways on the bare question of zombies: 37 percent held them conceivable but metaphysically impossible, 24 percent held them possible, 16 percent held them inconceivable — figures almost unchanged from the previous survey six years before. The Stanford Encyclopedia’s verdict on the whole exchange is that “in spite of the fact that the arguments on both sides have become increasingly sophisticated — or perhaps because of it — they have not become more persuasive,” and that “the pull in each direction remains strong.”

The replies sort into families, each coherent in its own terms, and the philosophical literature labels them with a precision worth keeping. Type-A physicalism denies there is any explanatory gap at all once the easy problems are done: there is no further fact of experience left over to explain, and the appearance of one is a confusion. Its sharpest form is illusionism — Dennett and Keith Frankish — which argues that phenomenal consciousness, the inner glow the hard problem is about, does not exist as it seems to; what needs explaining is why people are so unshakably convinced that it does. The hard problem becomes the illusion problem: how the brain generates so vivid a misrepresentation of its own workings. Opponents answer with what is at bottom Descartes’s point — that the reality of one’s own experience is more certain than any theory denying it — and between those two certainties the exchange has not moved in decades. Type-B physicalism, with David Papineau and Levine himself among its proponents, takes a quieter line: the gap is real but epistemic, lying in the knower rather than in nature. Consciousness simply is brain activity; the appearance of an unbridgeable difference reflects two ways of knowing one thing — a phenomenal concept and a physical concept locking onto the same state. Against both stands Chalmers’s own position, naturalistic dualism or property dualism: experience is a further, fundamental property of the world, not reducible to the physical but not supernatural either, governed by psychophysical laws as basic as those for mass and charge. And the “new mysterians,” led by Colin McGinn, accept the problem as genuine, naturalistic, and permanently insoluble — not because the answer is occult, but because human minds are cognitively closed to it as a mouse is closed to a maze whose solution is a prime number.

The science has not waited on the philosophy. The search for the neural correlates of consciousness, begun by Francis Crick and Christof Koch around 1990, maps which brain events accompany experience — and that, as Chalmers observed from the start, is an easy problem: one can always ask why those events give rise to experience at all. Two research programs have grown into candidate sciences of consciousness. Global workspace theory, proposed by Bernard Baars in 1988 and carried to the neuronal level by Stanislas Dehaene, holds that a content becomes conscious when it is broadcast from specialized processors to the brain at large — a stage on which information is made available everywhere, attention its spotlight. Dehaene goes further: educate intuitions with enough neuroscience, he predicts, and the hard problem will evaporate as the old mystery of life did when biochemistry explained metabolism and heredity. Integrated information theory, proposed by Giulio Tononi in 2004 and championed by Koch, runs the opposite way: it begins from axioms about what experience is essentially like — that it exists, is structured, specific, unified, and definite — and derives the causal structure a physical system would need to be that experience, down to a single measure, Φ, of how much consciousness a system possesses. Because Φ can in principle be nonzero in very simple systems, the theory carries a panpsychist tilt, which is precisely what made it a lightning rod. On September 15, 2023, an open letter signed by 124 scholars — Baars and Dennett among them — was posted to the PsyArXiv repository urging that integrated information theory be labeled pseudoscience until its core assumptions are properly testable. The charge drew an immediate counter-charge: Anil Seth called the label inflammatory, and Nature itself ran an editorial arguing that “pseudoscience” was the wrong word for an ambitious but disputed theory. A working program had become a public quarrel about the limits of science.

The quarrel had been seeded by an experiment of unusual design. The Cogitate Consortium, funded by the Templeton World Charity Foundation, ran an adversarial collaboration in which proponents of global workspace theory and integrated information theory pre-registered rival predictions, agreed in advance what would count as failure for each, and handed the testing to theory-neutral laboratories. The first results, unveiled at a conference in June 2023, fit neither theory cleanly: across a series of predictions, some of integrated information theory’s fared better — its account of where conscious content is decoded held up — while global workspace theory’s predicted late “ignition” in the prefrontal cortex largely failed to appear, yet each theory was also contradicted on a key point. The occasion settled a famous wager: in 1998 Koch had bet Chalmers that the neural basis of consciousness would be pinned down within twenty-five years, and on the strength of these results he conceded — losing the case of wine he had wagered and calling the outcome a victory for science even as it was a defeat for his prediction. The final paper, in Nature in April 2025, confirmed the verdict at full rigor: a lack of the sustained posterior synchronization integrated information theory required, and a lack of the prefrontal ignition global workspace theory required, supporting neither theory fully while turning up unexpected long-range connections of its own.

A third proposal sits at the far edge of the field. Roger Penrose and Stuart Hameroff hold that consciousness arises from quantum “objective reduction” — orchestrated collapses of the wave function inside the microtubules of neurons, a hypothesis they call Orch OR. It remains under study and under heavy critique. Max Tegmark calculated in 2000 that quantum coherence in warm, wet brain tissue would decohere in roughly 10⁻¹³ seconds, far too fast to matter on the timescales at which neurons compute — though Hameroff’s collaborators have contested the model behind the estimate. The Gödelian argument with which Penrose motivates the theory, that human mathematical insight outruns any algorithm, is widely rejected; and Chalmers notes the deeper point that there is no more reason a quantum event should generate experience than a classical one. Invoking quantum physics relocates the mystery without dissolving it. The experiments continue regardless.

Two further moves bracket the field. One revives the oldest option: if experience can be neither conjured out of inert matter nor argued out of existence, perhaps it is intrinsic to matter, present in some form all the way down. That panpsychist turn — taken up again by Galen Strawson, Philip Goff, and Chalmers himself, often in the Russellian monist form that treats consciousness as the intrinsic nature physics describes only structurally — has its own famous difficulty, the combination problem: how the supposed micro-experiences of particles add up to the single unified experience of a person. The other move is Chalmers’s 2018 “meta-problem,” published again in the Journal of Consciousness Studies: set aside why consciousness exists, and explain in ordinary functional terms why human beings are so convinced there is a hard problem at all. That question is by design an easy one — beliefs and reports are functions, and a purely physical story about them must be available — and its answer could cut either way: toward debunking the hard problem, as illusionists hope, or merely toward constraining any solution that hopes to respect why the problem feels so intractable.

Beneath everything runs the dispute over whether the problem exists at all. Levine, McGinn, Ned Block, Tononi, and Koch hold that it does; Dennett, Frankish, Patricia Churchland — who renamed it the “hornswoggle problem” — Dehaene, Baars, and Anil Seth hold that it does not. In the 2020 PhilPapers survey, 62.4 percent of philosophers called the hard problem genuine and 29.7 percent denied there is any such problem. Serious people sit on both sides. The disagreement is not over any experimental result; it is over what would count as one — which is why the adversarial collaboration, for all its rigor, moved the neuroscience without touching the philosophy.

The question has acquired a practical edge. Deciding whether an artificial system is conscious is the hard problem in applied form, compounded because a system trained on human language will produce human-sounding reports either way — as in 2022, when a Google engineer pronounced the LaMDA chatbot sentient and the field judged the claim mimicry. In 2023 Chalmers judged current large language models probably not conscious — lacking, among other things, unified agency, recurrent processing, and a global workspace — but held that nothing in principle rules out future systems built to have them. Thomas Metzinger has called for a global moratorium on creating synthetic phenomenology until 2050, to forestall what he names an explosion of artificial suffering. The stakes have stopped being purely academic, because the same gap that makes consciousness hard to explain makes it impossible to detect from the outside: a report is a function, and the hard problem is about what no function captures.

The Hermetic and Platonic traditions asserted, each in its own vocabulary, the primacy of mind — nous not as a late product of matter but as something nearer the ground of things, the source rather than the residue. The hard problem is the one place inside mainstream, naturalistic inquiry where the reality of inner experience is itself the open question, and where treating experience as fundamental — alongside mass and charge, in psychophysical laws or panpsychist substrate — appears not as revelation but as one live hypothesis among rivals in a peer-reviewed journal. The nearness cuts in both directions. An open question is open toward dissolution as much as toward vindication, and if the problem evaporates, as Dennett and Dehaene expect, the old assertion of mind’s priority loses its modern foothold with it. A century and a half after du Bois-Reymond, and three decades after Tucson, what survives is narrower and stranger than either camp would like: the one fact every observer holds without instruments remains the one fact that nothing yet explains.

The texts and the studies

The hard problem is unusually well documented in primary sources, most of them short and many of them freely posted by their authors.

• David J. Chalmers, “Facing Up to the Problem of Consciousness” (1995). The founding statement of the easy/hard distinction, naming the problem of experience and the question of why processing is not “in the dark.” Journal of Consciousness Studies 2(3): 200–219. Author’s copy: https://consc.net/papers/facing.html

• David J. Chalmers, The Conscious Mind: In Search of a Fundamental Theory (1996). The book-length case for naturalistic dualism and the zombie/conceivability argument. Oxford University Press, ISBN 0-19-510553-2.

• Thomas Nagel, “What Is It Like to Be a Bat?” (1974). The paper that fixed “something it is like” as the mark of consciousness and argued that objective physical science abandons the point of view experience requires. Philosophical Review 83(4): 435–450. DOI: https://doi.org/10.2307/2183914

• Frank Jackson, “Epiphenomenal Qualia” (1982). The knowledge argument and the Mary thought experiment. Philosophical Quarterly 32(127): 127–136. DOI: https://doi.org/10.2307/2960077

• Joseph Levine, “Materialism and Qualia: The Explanatory Gap” (1983). Names the explanatory gap with the C-fiber and molecular-motion examples. Pacific Philosophical Quarterly 64(4): 354–361. DOI: https://doi.org/10.1111/j.1468-0114.1983.tb00207.x

• David J. Chalmers, “The Meta-Problem of Consciousness” (2018). Recasts the question as why we believe there is a hard problem, an easy problem bearing on the hard one. Journal of Consciousness Studies 25(9–10): 6–61. Open archive: https://philarchive.org/rec/CHATMO-32

• Cogitate Consortium, “Adversarial testing of global neuronal workspace and integrated information theories of consciousness” (2025). The Templeton-funded, pre-registered test whose final results supported neither theory in full. Nature, 30 April 2025. DOI: https://doi.org/10.1038/s41586-025-08888-1

• The Stanford Encyclopedia of Philosophy maintains the standard reference treatments of the debate, including the conclusion that the zombie arguments “have not become more persuasive” with sophistication: https://plato.stanford.edu/entries/zombies/ and https://plato.stanford.edu/entries/consciousness/

Together these define the modern problem and its proposed sciences, and they remain the texts a reader is sent back to whenever the argument is reopened — which, on the evidence of the surveys and the laboratory, it continues to be.

→ Related: Panpsychism · Nous · Global Workspace Theory · Integrated Information Theory · Quantum Physics · Near Death Experience