What happened: The Continuous/Discrete framework proposes two representational formats — System C, a distributed parallel sensory-motor engine over continuous high-dimensional manifolds, and System D, a centralised serial engine of discrete scale-invariant symbols decoupled from that substrate — and models conscious access as the structure-preserving translation between them, with an inverse projection grounding symbols back in sensorimotor dynamics. It derives representational-geometry predictions (a measurable collapse from graded similarity to low-dimensional categorical equivalence classes) and contrasts them with global workspace, integrated information, predictive processing and higher-order accounts. §5.2 characterises language models as “a very large System C whose training signal has been the output of human System D operations, filtered through language”, predicting their successes where distributional patterns of symbolic behaviour are rich, and their failures where those patterns diverge from the discrete structure beneath — out-of-distribution arithmetic, compositional generalisation, holding a stable discrete state across reasoning steps. The framing is stated as a deliberate bracket: the account clarifies “what conscious states functionally accomplish without requiring resolution of the hard problem of phenomenology”, and “our aim throughout is not to declare the Hard Problem solved, dissolved, or irrelevant”.
Tracks: ARG-02 v1.7, through its declined-criterion clause rather than the trilemma. No criterion for experiencing is offered here, so none runs onto the three horns — the bracket is what keeps it off them, and it holds as long as the bracket does. That is the datum: the trilemma is avoided at the price of not answering the question, which is what a declined-criterion position inherits. Also FWK-03 v1.5 — the paper is a clean specimen of the pragmatic bracket as a move: redefine the question as computational access, gain tractability, and leave the ontological question standing where it stood.
Where the bracket strains, stated as the project’s observation and not the author’s: the term retained is conscious access, and §5.2 concludes that current systems reproduce the outputs of symbolic cognition “without implementing the underlying architecture”. No consciousness verdict about language models appears anywhere in the paper — but the sentence will be read as one, and the geometric signatures offered are not specific to conscious systems, since categorisation and dimensional compression occur in much that nobody treats as conscious. The symmetry: Not acknowledged tag records that the symmetry requirement is not engaged; it is not a criticism here, because no exclusion claim is made that would need testing.
Not novel against: no prior entry. The log holds no other entry on a positive theory of conscious access, and none on the bracketing move itself.
Read in full 2026-08-28. A theoretical proposal with predictions, not a result. Logged as a specimen of the bracketing move, not as evidence about experiencing. The paper has no limitations section; the non-specificity observation below is the project's, not the author's — an earlier version of this note wrongly attributed it to him.