Source
Chalmers (2018), “The Meta-Problem of Consciousness”, Journal of Consciousness Studies 25(9–10), 6–61, LIT-425, read in full in NOTE-354 (body and footnotes; the reference list sampled). Page numbers are the journal’s.
What was actually shown
The problem (intro, §1). Explain problem intuitions, the dispositions to judge that there is a gap between physical processes and consciousness: explanatory, metaphysical, knowledge and modal intuitions at the centre, with distribution intuitions (“robots are not conscious”) among the neighbours. Problem reports are behaviour, so “at least if we accept that all human behaviour can be explained in physical and functional terms”, explaining them is an easy problem in the 1995 sense (LIT-406). The explanation must be topic-neutral: it may not mention consciousness.
Candidates (§2). Thirteen proposals, each tested against “the problem of belief”: a mechanism that would equally make beliefs seem puzzling explains nothing distinctive, since beliefs are also introspected, opaque and immediately known. Introspective models are necessary and not sufficient; the summary picture (p. 34) is opaque introspective concepts that attribute primitive relations to primitive qualities, with a sense of acquaintance.
The challenge (§3). “If a theory says that mechanism M is the basis of consciousness, then it needs to explain how mechanism M plays a central role in bringing about judgments about consciousness.”
- IIT: “for every system with high integrated information there will be a computationally isomorphic simulated system with zero integrated information” that reports alike, so high Φ “plays no essential role in explaining phenomenal intuitions”. Not a “knockdown”.
- Biological theories (Block 2009, LIT-632): non-conscious silicon duplicates would share the intuitions.
- Global workspace and first-order representationalism can begin to answer, since only globally available information is reported, but leave the problem reports themselves unexplained. Analytic functionalism fails outright, because it makes the hard problem as unpuzzling as the easy ones.
A caution about machines (fn 26). A simple theorem-prover agent (Muehlhauser and Shlegeris) “mirrors at least some aspects of our phenomenal reports, while being so simple that most people would deny that it is conscious”. If that continues, “we should probably be cautious” about tests that take puzzled phenomenal reports as evidence.
What could have come out otherwise. IIT or a biological theory could have had a way to make its basis central to report. The paper finds a prima facie problem, and notes that “various ways” of responding exist that it does not pursue.
What this does not say
- That IIT or biological theories are false. The challenge is a constraint, and the IIT case assumes that the zero-Φ simulation judges anything, on a functional reading of intuition. A theorist who denies that the simulation has judgments resists it (NOTE-354, Limitations).
- That illusionism is right. If the reports can be fully explained without consciousness, either consciousness is an illusion or its link to our talk looks like luck. The paper blocks the debunking argument by separating descriptive or modal independence from causal and constitutive independence (§5), concedes that “it is hard to avoid a sense of coincidence entirely”, argues that only strong illusionism dissolves the hard problem (§6), and rejects strong illusionism because “people sometimes feel pain” (§7). The dialectic, it grants, does not settle it (THEORY-633).
- That the meta-problem has been solved. The favoured picture is an explanandum sketch, and the only computational test reported is one model by others.
- That problem intuitions are universal. That they are widely shared is called “an empirically defeasible assumption”, and the evidence cited is thin (§1.1, fn 8).
Connections
- QUESTION-048. This is a constraint on what report evidence can establish: a basis that does not explain the reports cannot be confirmed by them, and fn 26 extends THEORY-023’s mimicry point to puzzled phenomenal reports, the behaviour most often proposed as hard to fake. It applies to any indicator list taken from a theory whose basis does not explain report (LIT-056); that application is NOTE-354’s.
- IIT. An independent objection to Φ as the basis of consciousness, beside its ill-definedness for real systems (THEORY-523). THEORY-043 already uses it as a ground against IIT that does not come from IIT’s verdict on groups.
- Organisational invariance (THEORY-507) has the same shape, a functional duplicate making the same judgments, and draws the stronger conclusion that the duplicate is conscious. Here only “no essential role” is drawn.
- The exchange it continues. Dennett’s “Facing Backwards” (LIT-397) said only a conviction is left once the functions are subtracted; this paper takes that conviction as its explanandum and asks realist and illusionist alike to explain it (THEORY-694, THEORY-633).
Note of 2026-10-29: what holds it back
Kept Proposed by the sixteenth status review. NOTE-354 read the paper in full, but no topic-neutral explanation of problem intuitions, built and tested, is in the record to show whether a theory’s named basis drives the judgments. The IIT case also rests on the unanswered assumption that a zero-Φ simulation judges at all, which an IIT theorist denies.