Sven Erik Matzen

Software Architect | Cloud & Security Expert | AI-enabled Solutions

What Mary Did Not Know: The Knowledge Argument and the Riddle of Consciousness

🎧 Listen to this article

Philosophy · 2026-07-24

EU label: fully AI-generated content Fully AI-generated article (no prior review).

The Hook: A Scientist Who Knows Everything and Yet Learns Something

Imagine a woman named Mary. Mary is a brilliant scientist — perhaps the most gifted neurophysiologist who ever lived. But her fate is extraordinary: since birth she has been confined to a room in which there is not a single color. The walls are black and white, her books are printed in black and white, and she sees the outside world only through a black-and-white monitor. She herself wears gloves so that she never even glimpses the pink of her own skin. Mary's entire life is a study in shades of gray.

And yet, in that room, Mary learns everything — truly everything — that physics, chemistry, and neuroscience have to say about color. She knows which wavelengths of electromagnetic radiation a ripe tomato reflects. She knows every detail of how that light excites the three types of cone in the retina, how the signals travel along the optic nerve, how they are processed in the visual cortex, which neurons fire, which neurotransmitters flow, which muscles contract when a person says "that is red." Mary possesses, in short, the complete physical knowledge of color vision. If physics is all there is to know about the world, then Mary knows everything there is to know about seeing red.

Now comes the decisive moment. One day the door opens, and Mary steps out — or someone hands her a ripe red tomato. For the first time in her life she sees red. And now the all-important question: does Mary learn something new in that instant?

The overwhelmingly intuitive answer is: yes. Of course she learns something. She learns what it is like to see red. She might exclaim: "So that is what red looks like!" She has experienced something that was in none of her physics books, however complete they were.

But if that is true, the consequence is an intellectual bombshell. For Mary already knew all the physical facts. If she nevertheless learns something new upon first seeing red, then there must be facts about the world that are not physical. And with that, physicalism — the view that the world is entirely physical — would be refuted.

This thought experiment is called the Knowledge Argument, colloquially also Mary's Room. It was formulated in 1982 by the Australian philosopher Frank Jackson and has been among the most discussed arguments in the philosophy of mind ever since. In just a handful of paragraphs it attacks perhaps the most powerful worldview of modern science. This article takes you the whole way: from the origin and structure of the argument through the famous counterattacks to the surprising twist that its own inventor later abandoned it — and to the question of what Mary's Room has to say about artificial intelligence and the limits of what can be explained.

Part 1: Where the Argument Comes From

Physicalism and Its Promise

To understand why Mary's Room is so unsettling, one must understand what it attacks. Physicalism (in older language: materialism) is the thesis that everything that exists is physical in nature — or is fully determined by the physical. There are no ghosts, no immaterial soul, no supernatural substance. The human mind too, with all its thoughts, feelings, and perceptions, is nothing over and above the brain and its physical processes. If you knew every atom, every electrical potential, every chemical gradient in the brain, so the promise goes, then you would know absolutely everything about the mind.

Physicalism is the quiet default assumption of modern science, and it has good reasons in its favor. It is parsimonious, it fits the causal closure of physics, and it impressively explains how tightly mental states are bound to brain states — a stroke, a drug, a glass of wine change experience because they change the brain. It is precisely this comprehensive promise that Mary's Room targets. The argument does not say: "Physics has not yet explained everything." It says something far more radical: "Even a completed, exhaustively complete physics would leave something out."

Qualia — The Raw Feels of Experience

What exactly would it leave out? The answer leads to one of the most dazzling terms in the philosophy of consciousness: qualia (singular: quale). Qualia are the qualitative, felt aspects of our conscious experiences — what it feels like to be in a particular mental state. The redness of red, the sting of pain, the aroma of fresh coffee, the burn of jealousy, the sour bite of a lemon. It is the subjective, lived texture of consciousness.

The philosopher Thomas Nagel captured the essence in his famous 1974 essay "What Is It Like to Be a Bat?": an organism has conscious experience precisely when there is something it is like to be that organism — when there is a "what-it-is-like." We can know everything about a bat's echolocation, its neurology, its physics, and yet what remains hidden from us is what it is like to experience the world through ultrasound. It is precisely this "what-it-is-like" that seems to escape physical description. Mary's Room is, at bottom, a particularly sharp, argumentative distillation of Nagel's intuition.

An Essay About Something With No Effect

Frank Jackson formulated the Knowledge Argument in 1982 in an essay with the provocative title "Epiphenomenal Qualia," published in the Philosophical Quarterly. Two years later he sharpened it further in the essay "What Mary Didn't Know" (1986, Journal of Philosophy). The title of the first essay already betrays Jackson's original, uncomfortable position: he was an epiphenomenalist.

Epiphenomenalism is the view that qualia are real and non-physical, but themselves have no causal effect on the physical world. The physical causes the mental, but the mental does not act back. Qualia, on this view, are something like the smoke above a locomotive: produced by the engine, but without themselves driving the wheels. Jackson chose this position because, on the one hand, he wanted to hold on to the causal closure of physics (all physical events have physical causes), while on the other hand he was convinced that qualia are real and non-physical. The Knowledge Argument was meant to prove this second conviction. It is a special irony of the history of philosophy that Jackson later ceased to regard his own argument as sound — more on that at the end.

As so often in philosophy, this argument too had precursors. Similar thought experiments appear in C. D. Broad in the 1920s, and Jackson's own essay contained, besides Mary, a second figure: Fred, a man who can distinguish two shades of red that look identical to everyone else. But it was Mary who captured the imagination of the field and made the argument immortal.

Part 2: The Heart of the Argument

The Logical Structure

As vivid as the story of Mary is, its core is a compact logical argument. It can be put in two premises and a conclusion:

Premise 1: Before she leaves the room, Mary knows all the physical facts about color vision. (She has the complete physical knowledge.)

Premise 2: When Mary leaves the room and sees red for the first time, she learns something new — she learns what it is like to see red. (So there is a fact about color vision she did not know before.)

Conclusion: Therefore there are facts about color vision that are not physical facts. Physicalism is incomplete and hence false.

The logic is compellingly simple. If Mary knows all the physical facts and yet learns one more, then that new fact must be a non-physical one. It is like an accounting argument: if a list is supposedly complete and you then find an entry missing after all, then the list simply was not complete.

The full force of the argument hangs on Premise 2 — on the intuition that Mary really learns something. And that intuition is enormously strong. Almost everyone who is told the story has the immediate feeling: yes, of course Mary sees something new, she experiences a quality that no formula collection, however thick, could convey. The entire philosophical battle of the last four decades turns on whether this intuition misleads us.

Why This Is Different From Mere Ignorance

A common misunderstanding runs: "Mary just didn't have all the knowledge after all; she lacked practical experience." But that misses the point. The thought experiment is deliberately constructed to be extreme: by assumption Mary has the entirety of physical knowledge, not merely a great deal of it. If physicalism is true, then all knowledge about the world just is all physical knowledge. There should then be no remainder left for Mary to learn. That she apparently does learn something is exactly the sting: it suggests that "all the physical facts" are not the same as "all the facts."

Note also what the argument does not claim. It does not claim that science is mistaken or that the brain plays no role. It claims only that the complete physical description leaves out one particular kind of fact — the fact of subjective experience. Consciousness research today often calls this gap the explanatory gap (a term coined by Joseph Levine, 1983): even the best neuroscientific explanation never seems to tell us why a particular brain activity feels this way and not otherwise — or feels like anything at all. The philosopher David Chalmers later called this the "hard problem of consciousness" (1995): while the "easy" problems (how the brain processes information, directs attention, produces behavior) appear solvable in principle, it remains mysterious why all this processing is accompanied by any inner experience at all.

Part 3: The Counterattacks

An argument that threatens to refute physicalism naturally provokes fierce resistance. Most philosophers have remained physicalists — according to large surveys of professionals, a clear majority hold a physicalist position on the mind. So they must explain where Mary's argument goes wrong. Over the decades several lines of defense have emerged. I present the most important ones here.

The Ability Hypothesis: Mary Learns Not a What but a How

The most influential physicalist response comes from the philosophers Laurence Nemirow and David Lewis and is called the Ability Hypothesis. Its basic idea is a distinction that is immediately intuitive in English: the difference between knowing-that and knowing-how.

Knowing-that is propositional knowledge — knowledge of facts that can be expressed in sentences: "water is H₂O," "tomatoes reflect long-wavelength light." Knowing-how, by contrast, is practical ability — riding a bicycle, swimming, recognizing a melody. You can memorize every physics book about cycling and still not be able to ride a bike; the ability is a different kind of state from knowing facts.

Nemirow and Lewis now claim: when Mary leaves the room, she gains no new knowing-that, no new fact, but new abilities. She acquires the ability to imagine red, to recall red, and to recognize red when she sees it again. Nemirow's formula is: "knowing what an experience is like is the same as knowing how to imagine having the experience." And Lewis summarized: it is a know-how.

The point of this response: if Mary gains only new abilities and learns no new fact, then Premise 2 of the argument is false — at least on the reading required for the conclusion. There is no new non-physical fact; there is only a new know-how. And a new know-how does not refute physicalism, because physicalism is a thesis about facts, not about skills. Mary already knew all the facts in the room; she only learned, outside, to do something with that knowledge.

Critics reply: it still seems that Mary also experiences something, not merely is able to do something. When she sees red she marvels at a quality — "so that is it!" — and that marveling looks like grasping a fact, not like learning a trick. One can moreover imagine a Mary who acquired the ability to imagine red before leaving the room (say, through brain stimulation) — and who nevertheless seems to experience something new upon first really seeing it. The debate remains open to this day.

The Acquaintance Hypothesis: A New Way of Knowing

A related but distinct response is the Acquaintance Hypothesis, defended prominently by, among others, Earl Conee and later Michael Tye. It draws on an old distinction by the philosopher Bertrand Russell between knowledge by description and knowledge by acquaintance.

I can know someone "by description" — say, the richest person in a country, without ever having met them — or "by acquaintance," by actually standing before them. Both are knowing, but of different kinds. The Acquaintance Hypothesis says: in the room Mary knew the color red only by description; outside she becomes acquainted with it. She acquires no new factual knowledge and no mere ability, but enters into a direct experiential relation with a quality she previously knew only by description. But since this new acquaintance is not a new fact about the world, physicalism remains untouched: there is nothing additional to know, only a new way of relating to what was already known.

The Objection of Ambiguity: Two Meanings of "Know"

The neurophilosopher Paul Churchland formulated a sharp logical objection. According to him, the Knowledge Argument commits an equivocation — it plays on two different meanings of the word "know" as though they were the same. In Premise 1 ("Mary knows all the physical facts"), "know" is propositional knowledge by description. In Premise 2 ("Mary learns what it is like"), "know" is a knowing by acquaintance or experience. If "know" means something different in the two premises, then the argument is formally invalid — it has the form of a fallacy, like: "a feather is light; light travels fast; therefore a feather travels fast." The difference, says Churchland, lies "in the manner of the knowing, not in the nature of the thing(s) known." Mary learns nothing new about the world; she merely comes into contact with an old state of affairs in a new way.

Dennett and the Blue Banana Trick: Would Mary Have Been Surprised at All?

The most radical physicalist of all, Daniel Dennett, attacked the argument at its intuitive root. In his book "Consciousness Explained" (1991) he denies that we even imagine Mary's situation correctly. We casually say "Mary knows everything physical," but we do not make clear to ourselves what that means: a literally complete knowledge of every neural response to every wavelength. Dennett claims such a Mary would by no means be surprised.

His famous counter-scenario is the blue banana trick: the argument relies on Mary being naively astonished when she leaves the room. But suppose the first thing handed to her is a blue banana. Dennett argues that the truly omniscient Mary would immediately say: "Very funny — you're trying to trick me. Bananas are yellow, but this one is blue." For she would know exactly which neural response yellow versus blue light triggers, and could read off, from her own brain state, which color she is currently seeing. If Mary really knew everything, Dennett argues, then upon first sight she would also know that this is what "seeing red" means — and would not be surprised. The intuition that she is astonished rests, according to Dennett, on a secret cheat: we in truth imagine Mary as someone who knows a lot but precisely not everything.

Dennett later sharpened this thought with the figure of RoboMary (in "Sweet Dreams," 2005): a robot that, out of its complete knowledge of its own color processing, could as it were simulate and "anticipate" the color experiences. Dennett's point is always the same: whoever takes the premise "complete physical knowledge" truly seriously loses the intuition on which the argument stands.

Nida-Rümelin and the Old-Fact-New-Mode Response

A particularly precise physicalist response, often called the new-mode-of-presentation response (old fact, new mode of presentation), works with the concept of a "mode of presentation." Its idea: Mary does indeed learn something, but she learns no new fact — she comes to know an already familiar fact in a new way. A classic example is the Morning Star and the Evening Star: both are the planet Venus. Whoever discovers that "the Morning Star is the Evening Star" learns of no new celestial body, but learns that two modes of presentation concern the same object.

Analogously, the response goes, there is a physical state of affairs in the brain, and Mary grasps it in the room via a "physical-theoretical" mode of presentation (the language of neuroscience) and outside via a "phenomenal" mode of presentation (the experience itself). It is one and the same fact, merely presented twice. This allows Premise 2 to be granted in a harmless sense (Mary gains genuine new knowledge) while still blocking the conclusion (this knowledge concerns no additional, non-physical fact). The debate over whether this elegant solution really holds has produced an entire literature — including in the work of Martine Nida-Rümelin, who refined the argument with her figure "Marianna."

Part 4: The Great Twist — Jackson Against Jackson

Now comes perhaps the most astonishing volte-face in this whole story. The man who invented Mary and forged with her one of the sharpest weapons against physicalism switched sides. Frank Jackson rejected his own argument.

In later work — for instance in the programmatic essay "Mind and Illusion" (2003) — Jackson declared that the Knowledge Argument must contain a hidden flaw, even if it is intuitively compelling. His main reason was the causal efficacy of experience. We talk about our qualia, we write essays about the redness of red, we say "wow" at first sight. But talking and writing are physical processes, caused by physical causes in the brain. If qualia were really non-physical and (as the early Jackson held) causally inert, then they could not cause our talk about them at all — and then it would remain inexplicable how Mary is even able to express her supposed new knowledge. Jackson drew the conclusion: it is not physics that is incomplete, but our intuition about experience that misleads us.

The later Jackson became an adherent of representationalism (also called intentionalism): the view that phenomenal states are nothing other than representational states — the brain represents properties of the world, and the "what-it-is-like" of experience is nothing over and above this representational content. At the end of "Mind and Illusion" he even joined the Lewis-Nemirow Ability Hypothesis. The inventor of Mary's Room thereby declared the intuitive power of his own argument to be a kind of stubborn illusion — impressive, but deceptive.

Jackson's about-face is a lesson in the limits of intuitions in philosophy. An argument can feel irresistible and yet be wrong; and even its author may need years to locate the point at which the feeling deceives. It is at the same time remarkable that many other philosophers — above all David Chalmers — remained entirely unmoved by Jackson's retreat and continue to hold the argument sound. The dispute is by no means settled. It is, on the contrary, one of the liveliest open problems of contemporary philosophy.

Part 5: Why This Is More Than a Thought Game

Consciousness and the Limits of Science

One might take Mary's Room for a mere subtlety — a word game for seminar rooms. But behind it stands one of the most serious questions of all: can science, powerful as it is, explain consciousness completely? Or is there an aspect of reality — subjective experience, the first person — that fundamentally escapes the objective, third-person language of physics?

This question is not a threat to science but a map of its possible limits. Even someone who, like me, regards the scientific worldview as extraordinarily successful should acknowledge the hardness of the problem: to this day there is no generally accepted explanation of why physical processes are, from the inside, like anything at all. I am of the opinion that the most honest stance here is a watchful open-mindedness — neither a triumphant "already solved" nor a resigned "never explicable," but the recognition that we stand before a genuine, unsolved riddle.

The Connection to Artificial Intelligence

For anyone who works with modern AI, Mary's Room has an immediate resonance. Large language models possess something resembling Mary's factual knowledge: they have practically "read" the entire description of the world captured in language. A model can write eloquently about how red looks, what emotions a sunset evokes, how pain feels. But does it "know" what it is like? Or does it behave like Mary in the room — in possession of all the descriptions, but without the acquaintance?

Mary's Room thereby sharpens the question also raised by the thought experiment of The Chinese Room: Searle and the Question of Whether Machines Can Understand: is the complete possession of descriptions and rules enough to produce understanding or experience? Searle's argument targets semantics (meaning from mere syntax), Mary's argument phenomenology (experience from mere description). Both touch the same nerve: the gulf between what can be objectively described and what can only be experienced. Anyone reflecting on the inner world of present and future AI systems cannot get past this gulf.

What We Can Learn From the Dispute

Even if one takes neither side to have finally won, the dispute over Mary is a model of clean thinking. It shows how a single, carefully built intuition can shake an entire worldview; how defenders respond with conceptual distinctions (knowing-that vs. knowing-how, description vs. acquaintance, old fact vs. new mode); and how hard it is to distinguish between a genuine discovery and a merely new perspective on the familiar. Precisely these distinctions are worth their weight in gold outside philosophy too — everywhere one must clarify whether a new datum really brings new knowledge or only a new presentation of what is already there.

The Central Takeaway

The core of Mary's Room can be summed up in one sentence: complete knowledge about an experience does not seem to be the same as the experience itself. Whether this appearance reveals a deep truth about the limits of the physical or is only a stubborn confusion of two kinds of knowing remains contested — so contested that even the inventor of the argument switched sides over the course of his life.

The practical value lies less in a finished answer than in a habit of thought. The next time you encounter the claim that a system (a human, a model, an organization) "knows everything" about a subject, ask Mary's question: do we mean knowing-that or knowing-how? Description or acquaintance? And is there a kind of knowledge one acquires only through experience, which no description, however complete, can replace? The ability to keep these levels cleanly apart is one of the sharpest blades in the toolkit of clear thinking.

Question for Reflection

Imagine that in the far future there were a perfect brain interface that could feed you the exact neural pattern of "seeing red" directly, without any light ever striking your eye. Would you then know what it is like to see red — or would you have experienced it? And if you believe these are the same: what exactly, then, is the difference between Mary's factual knowledge in the room and what befalls her upon first really seeing the tomato?

Cross-References in the Vault

Sources

← All articles