David J. Chalmers (1995), in Thomas Metzinger (ed.), Conscious Experience, Thorverton: Imprint Academic, co-published in Paderborn (by Schöningh according to Chalmers and Open Library, by mentis according to Metzinger). Author’s text: consc.net/papers/qualia.html. Reprinted in T. O’Connor & D. Robb (eds), Philosophy of Mind: Contemporary Readings (Routledge, 2003), DOI 10.4324/9780203987698-28. The argument is also chapter 7 of The Conscious Mind (LIT-441).
The brief cited the venue as Imprint Academic. That is right but incomplete; see the history note.
Key takeaways
- The principle (§1). “Given any system that has conscious experiences, then any system that has the same functional organization at a fine enough grain will have qualitatively identical conscious experiences.” A functional organisation is a set of abstract components, their possible states, and the dependency relations among them and the inputs and outputs. “Fine enough” means fine enough to fix the system’s behavioural dispositions, and the two systems must be in corresponding states.
- The target is nomological possibility (§2). Absent and inverted qualia may be logically possible. The question is whether they can happen in this world. Block’s China population and Searle’s water-pipes give intuitions, but “it seems equally counterintuitive that a mass of 10^{11} appropriately organized neurons should give rise to consciousness, and yet it happens”.
- Fading qualia (§3). If a silicon isomorph (“Robot”) lacked experience, then along a one-neuron-at-a-time replacement series either experience vanishes suddenly, which would need a brute discontinuity in the laws, or it fades. A half-way “Joe” would then sincerely report vivid red while experiencing faded pink: a rational, unimpaired system systematically wrong about its own experience. Searle’s variant, with true but impotent beliefs, is ruled out because there is “no room” in an unchanged organisation for new beliefs.
- Explicitly extended to populations (§3). Neurons can be replaced one by one with tiny homunculi, or with normal-sized external ones linked by radio, ending in “a network of homunculi that is essentially equivalent to the population controlling a robot”. It can go further, to a single homunculus who does all the work from written records, as in Searle’s Chinese room. In that case “we should not expect the homunculus itself to have the experiences; it is merely acting as a sort of causal facilitator”.
- Dancing qualia (§4). If an isomorph had inverted (or absent) qualia, some two systems ten percent apart would differ noticeably in experience. Install the second system’s circuit beside the first’s, with a switch, and flip it. Experience changes “before my eyes” with no change in functioning, so it cannot be noticed. Taking that seriously would let our qualia dance all the time, since our low-level physiology changes constantly. Loopholes are listed: speed, history, very mild inversions, unattended qualia.
- Nonreductive functionalism (§5). The arguments show at most that organisation determines experience with natural necessity. They cannot show that it does so with logical necessity, because their premise, that some brains are conscious, is empirical. The view is “just as compatible with certain forms of property dualism … as with certain forms of physicalism”. It leaves open “just what sort of organization gives rise to experience”.
Standing in the record
Filed on 2026-10-02 at the owner’s request for David Chalmers in this record, and read against the society-of-mind bridge (THEORY-043) and Block’s absent-qualia argument (LIT-407, read in NOTE-366).
Against Block. This paper is the main published reply to the nation-of-China form of the Absent Qualia Argument, and it attacks it on Block’s own ground of nomological possibility. Against the claims as NOTE-366 numbers them:
- C3 (the doubt about a homunculi-head’s qualia has a rational basis): contested. Chalmers treats the doubt as an intuition, and says intuition is “unreliable as a guide to empirical possibility”. Block’s stated grounds are dissimilarity to our psychology and neurophysiology, and design to mimic. Chalmers does not answer them by name. His argument answers them only for a system isomorphic to us at a fine grain, where the psychological dissimilarity Block relies on is absent.
- C5 (the doubt does not rest on the homunculi’s mentality): agreed in effect. Chalmers’s single-homunculus case makes the homunculus a mere “causal facilitator”, and his argument nowhere turns on whether the parts are minded.
- Block’s n. 17 case, a billion people each simulating one of your neurons, is where Block says intuition “seems to founder”, and NOTE-366 leaves it open. It is exactly the case this paper’s §3 population extension reaches by argument rather than intuition, and the paper’s verdict is that the system is conscious.
- A characterisation to correct. The paper says Block “argues that it is bizarre to suppose that this would somehow give rise to a group mind”. By NOTE-366, Block’s doubt is explicitly not about group minds or nesting (n. 19). Chalmers also cites Block as “1980” in the text and as “1981” (the Harvard Readings reprint) in the references. The original is 1978.
- Grain. Block’s main China system implements a machine table, a line per person. The paper’s argument needs a series of intermediate systems that are fine-grained isomorphs. It runs neuron by neuron, and the step to a coarser, machine-table realisation is covered only by “the arguments generalize” (§1). So the paper reaches Block’s n. 17 case directly and his main case only by that assertion.
Does organizational invariance license Schwitzgebel’s upward move (LIT-159)? No. It licenses only the replacement cases. The answer follows from the principle’s form, and the reasoning is mine:
- What it licenses. The principle quantifies over systems with the same fine-grained organisation as a system already conscious. Its supporting arguments work only where an organisation-preserving replacement series can be built. That covers silicon brains, and, by §3, a population of people realising a human’s neural organisation, external and radio-linked if need be. So, if true, it settles one thing the bridge treats as open. A whole whose parts are minded people can be conscious, with the experiences of the person it is isomorphic to, while the parts keep their own. That is the composition claim of NOTE-131’s C3, established for isomorphs.
- What it does not license. The United States is not a functional isomorph of any being known to be conscious. It shares no human’s or rabbit’s behavioural dispositions, and no organisation-preserving series runs from a brain to it. Schwitzgebel’s step is by similarity of kind and degree: the US has the materialist marks “in a greater degree” than a rabbit (NOTE-131, C7). Invariance is silent on that. The paper says that which organisations give rise to experience, and which experience, is left open (§5). The upward move needs exactly the principle it leaves out.
- What it does for Schwitzgebel anyway. It closes three of the exits his argument has to close. Substrate: “the only place to draw a principled line is at the functional level” (§4), which is against neurochauvinism, his §7. Spatial contiguity: normal-sized external homunculi with radios are allowed, which is his C2. Absolute anti-nesting: a population isomorph is conscious, which is his C4 and §2. It also bears on Kammerer’s escape from Chalmers’s objection (NOTE-362, C4). Kammerer says workspace modules are not conscious, so his principle spares humans. Invariance makes the “conscious or not” status of any module a fact about its organisation, so the principle stays statable in functional terms.
For THEORY-043. If invariance is true, the THEORY’s first way to refuse the upward move is unavailable in its absolute form. Putnam’s stipulation and IIT’s exclusion postulate both deny consciousness to a whole whose parts are conscious. A neural-level population isomorph is such a whole, and invariance makes it conscious. Under exclusion, experience goes to the homunculi or to the whole, whichever has the larger Φ, never to both. Invariance requires the whole to have it, and does not take it from the parts. The two theories also part more widely. IIT holds that a functional equivalent can differ in Φ (NOTE-365, C7), and invariance denies that this can make a difference to experience. Kammerer’s functional principle survives, because a homunculus filling a neuron’s role is one a chip could fill. Everything left is the THEORY’s second route, architecture, and invariance cannot decide it, since a nation and a modular mind are not isomorphs of each other or of us. So, on invariance, the THEORY’s dispute becomes entirely a dispute about which organisations suffice. That is the question this paper sets aside. The inference is mine.
On the hard problem. The paper is the long form of the second principle in LIT-406. It is compatible with that paper’s naturalistic dualism by design.
On AI consciousness (THEORY-023). The paper says its arguments “support some of the ambitions of artificial intelligence”. The support is narrower than it sounds. An exact fine-grained isomorph of a human brain would be conscious. No current AI system is such an isomorph, and THEORY-023 is about systems that are not. Seth (LIT-135) doubts the substrate-flexibility half of functionalism, and this paper argues for it for isomorphs only. Chalmers’s later work on implementation and on AI, which tries to carry the principle to computations, is being filed in this record by another reader at the same time.
On the extended mind (LIT-097). No direct bearing. The paper’s principle is about the organisation that fixes experience. Whether that organisation’s components can lie outside the head is the question Chalmers’s later paper on extended consciousness takes up, and another reader is filing it now.
It carries no instruction for machine-learning practice, and it does not belong in the anthology.
Its opposite number. Dennett’s “Quining Qualia” (LIT-433) starts from the same premise, that a change in experience the subject cannot notice is incoherent or impossible, and concludes there is no such property to fix or lose. This paper concludes the property is fixed by organisation. The pairing is the reader’s, not either author’s. The paper’s verdict on the population-of-people case also answers the case Block (LIT-407) says intuition “seems to founder” on.