The question
Every kind of evidence the record holds about experience reaches it through access: through report, through a capacity that is lost, or through an architecture that a theory says matters. The question is what each kind can establish about experience rather than access. It also asks whether any of it carries over to systems unlike the human brain: a bird’s nuclear pallium, a machine, an organization.
Why the record needs it
Two claims on the organizational-affect line depend on the answer. CLAIM-039 needs functional affect to come apart from experience. CLAIM-003 needs experience, not only access, to be compartmented. Neither can be tested until it is clear which evidence bears on experience at all. The same limits decide what the distributed-agency line could ever learn about whether a combined form is a subject (CLAIM-165).
What the record holds
- THEORY-091 (deficits): when consciousness and a capacity are lost together, this does not show the capacity is experience’s, because “access is lost too”. Limit: one source; the test case, superblindsight, has not been found.
- THEORY-094 (report and no-report): neither isolates experience, because “post-perceptual cognition that tracks the percept persists without report”. Only removing differential cognition does. Limit: shown for rivalry only.
- THEORY-100 (comparative): a correlate of report “does not require laminar cortex”. So report evidence can be had without cortex, but it “cannot separate access from phenomenal consciousness”. Limit: second-hand, one laboratory.
- THEORY-023 (architecture and behaviour): mimicry undercuts behavioural evidence, and architectural indicators “presuppose the disputed computational functionalism”. So the dispute is “over what counts as evidence as well as the answer”. Limit: about AI. The architectural half applies to any non-biological system; the mimicry half needs training on human output.
- COGITATE (LIT-973, NOTE-812; THEORY-245, THEORY-296) is task-based no-report evidence. Percept and stimulus never came apart, and the authors concede that “some might argue our approach tracks stimulus processing rather than consciousness”, so nothing in it separates experience from processing, which is THEORY-094’s limit. Its own view, that theory tests reach implementations and not cores, supports THEORY-023’s evidential point.
- LIT-965 (NOTE-797), on emergence from anaesthesia, is deficit-and-recovery evidence of the kind this question asks about, and it confirms THEORY-091’s limit: its marker of returning consciousness, response to command, is a marker of access. Limit: it uses human emergence data only. Birds and early mammals are reached by homology of anatomy, not by anaesthesia evidence from them.
- THEORY-298 (LIT-982, NOTE-849): report-based evidence cannot test whether posterior cortex suffices for experience, because report itself needs anterior activity, and no marker that bypasses report has a validation resting neither on introspection nor on a contested theory. Its Box 3 extends the problem to infants, patients, animals and organoids. Limit: an argument in a review, not a result, and it says the methods are lacking “at present”; it offers no evidence that carries over to systems without cortex.
What would answer it
For each kind of evidence, a case where access and experience come apart and the evidence follows one of them. Superblindsight, inconspicuous rivalry recorded intracranially, or a marker both sides accept in advance would each do this. More correlates found with report would not.
Note of 2026-11-01: Metzinger on reports of minimal experience
Metzinger’s later work, read on 2026-11-01, is the largest body of report evidence the record holds for one kind of experience: “pure awareness”, or minimal phenomenal experience (MPE). It states the limits of report evidence itself, and they belong in “What the record holds”.
- Retrospective report cannot show that a state was selfless. Metzinger 2024 (LIT-1943, NOTE-1499, p. 19): a report of having been selfless is “highly dubious from a methodological perspective”, since autobiographical memory represents events as a self’s. The state may be “stored and later retrieved in an autobiographical format, as a post hoc mnemonic misrepresentation”. He leaves the problem open. The 2020 paper (LIT-1946, NOTE-1498, p. 15) goes further: the content of a full-absorption episode “cannot be reported”, and only entry and exit are remembered. The book’s summary calls such states “empirically established” all the same (p. 457). THEORY-1295 states the gap. Limit: a conceptual argument, which the book does not test, for instance with reports taken at the end of a session, as it recommends (p. 18).
- No phenomenal contrast for the minimal case. If MPE is consciousness as such, “the method of phenomenal contrast … cannot be applied” (2020, p. 13), because its contrast class is unconsciousness. So no-report designs that contrast seen with unseen cannot isolate it.
- Reports about what an episode was. The E-fallacy and the C-fallacy, with its mirror image: “Just because something is reported as feeling not like an experience at all, this doesn’t license the conclusion that it isn’t” (2024, p. 417).
- A function without experience, by Metzinger’s reading. Tonic alertness without wakefulness, in unresponsive wakefulness (2020, p. 33). Limit: the absence of experience there is inferred from behaviour, which is THEORY-091’s limit.
- The architectural route he proposes. The Triple Triangulation Project looks for the “minimally sufficient neural correlates” common to full absorption entered from waking, from NREM sleep and from lucid dreams (2024, pp. 220–221). Limit: every episode would still be picked out by report, so it inherits THEORY-298’s problem.
No status, title or warrant changes.