Source
Graziano (2024), eNeuro 11(10), LIT-1358, read in full in NOTE-1071: “From Self Model to Speech” (p. 4), Hypotheses 1–6 (pp. 4–12) and Figures 2–4.
What was actually shown
An argument and a proposal. No new data.
- The encoding premise. If Bob says he has conscious experience, “whatever conscious experience is or is not, Bob must have a neural embedding that represents it” (p. 4).
- Arrow B. A feeling produced by neurons explains the report only if it in turn causes that embedding. That needs neurons that respond to “the emergent feeling itself” (p. 5). Without it, “the conscious feeling that emerges from the brain cannot be the consciousness that we say we have”. With it, what is reported “is entirely dependent on the information embedding” (p. 5).
- Subtle illusionism. Brain models are schematic, so “Subtle illusionism is necessarily true of all models in the brain and must be true of consciousness as well” (p. 7). Hard illusionism, a model of nothing, is set aside as evolutionarily improbable.
- The target. An Item 1 should affect neurons, resemble the reported consciousness in outline, and covary with report (p. 9). Attention is offered as meeting all three, on a century of cited covariation, with laboratory dissociations called “difficult, though possible” (p. 10). The attention schema depicts attention as “an amorphous, mental essence, a mind vividly possessing something in the moment” (p. 10).
- Deep illusionism. Attention, the workspace and integrated information are “different descriptions of the same underlying object—the deep, selective processing of information in the cerebral cortex” (p. 11).
What could have come out otherwise is the covariation premise. Reports of awareness could have followed something other than attention. The paper reports that they mostly do not.
What this does not say
- That nothing real is going on. Item 1 is a real process, and the paper’s word is “caricature” (p. 2). But what it keeps real is the process, not the phenomenal property: “The subjective, phenomenal experience that Bob describes is a distorted construct” (p. 7).
- That every schematic model is false. The argument’s premise, that “nothing is exactly as the brain represents it”, is true of any model and does not by itself show that a depicted property is absent from its target. That step is made on causal-closure grounds, not from the premise.
- That the attention schema has been found. Its location is “not important to the present argument” (p. 4), and the paper reports no manipulation of the model apart from attention.
- That attention, workspace and integration are one thing. This is asserted from overlap. The integration cited, Treisman’s, is not IIT’s.
- That machines are or are not conscious. The framework “could have applied to artificial as well as biological consciousness” (p. 12), and the topic is set aside.
Connections
- THEORY-633. Frankish’s strong illusionism, with a mechanism for the representation and a named target. Graziano’s “small” illusionism is not Frankish’s weak illusionism, since no phenomenal property survives. His “hard” illusionism, a model of nothing, goes beyond THEORY-633.
- THEORY-577. Arrow B is Chalmers’s meta-problem challenge from the other side: Chalmers asks a theory to explain the reports, and Graziano says a theory that cannot is magic. Graziano’s illusionist IIT answers the challenge Chalmers puts to IIT by making integration the target of a model.
- THEORY-064. The workspace model “depicts me as a magical essence inhabiting my body” (p. 8), a self-model the system cannot see through, extended from the self to experience.
- CLAIM-355. Same structure, a real organization misdescribed by a model whose bearer is named, but extended from the owner to phenomenality, which CLAIM-355 leaves open.
- CLAIM-412 and TERM-072. The rival to the constitutive claim’s bridge. On this account what is called experience is a represented perspective (sense 2) of a real organization (sense 1), and sense 3 is the representation’s content.
- CLAIM-411. A candidate for the further condition, being modelled by the attention schema. That condition is represented, not organizational, and the paper states it only through report.
Note of 2026-10-30: what holds it back
No manipulation of the attention schema with attention held fixed, or of attention with the model held, is in the record. The identification of the modelled process with attention rests on covariation with report cited from others, and the paper reports no such manipulation (NOTE-1071).