Li (2026), Frontiers in Computational Neuroscience 20:1778902 (2026), article type “Hypothesis and Theory”; open access, CC BY 4.0 — DOI-10.3389/fncom.2026.1778902
Standing in the record
Filed on 2026-10-01 at the owner’s request.
published: is the first appearance (ADR-002).
It was filed Deferred, unread. NOTE-321 is the close reading of 2026-10-01, and it placed the work: Rejected — not worth a reader’s time as a source for either of its claims. The complexity and topology premises are misstated (Savitch is read backwards, Urysohn is overstated). The scaling “theorem” is unproved and, by the paper’s own statement, AI-assisted. Every neural mapping is asserted. What survives is three loosely specified predictions (§6.2) that would need a model to make them discriminating.
The author’s follow-up, read 2026-10-01. The owner offered Li’s “Structural Decoupling” (LIT-374, NOTE-322) as work that might change this entry’s standing. It does not: it cites this paper but repairs none of NOTE-321’s grounds. Neither the Savitch nor the Urysohn argument recurs, nothing corresponds to Theorem 2, and it has no data. Its own new results are assumed, self-contradictory or false as stated, and it too is filed Rejected. NOTE-321’s remark that the paper’s metabolic maintenance cost runs against THEORY-030 was wrong and is corrected there; it was never a ground for rejection.
Re-placed 2026-10-01, on different grounds. At the owner’s request both of Li’s papers were re-read together, as one research programme, and charitably: a technical slip counts only if a major takeaway depends on it and its sensible repair does not save it. The mathematical slips the first reading rejected on (the Urysohn “if and only if” and linear readout, Savitch read backwards, the Definition 2 clash) are repairable and are no longer grounds. It stays Rejected for what remains after repair: a restatement of amortised inference and complementary learning systems, neural predictions that do not discriminate it from standard consolidation accounts, and an unargued Theorem 2. The programme is better read in LIT-374 and its companion. See the curation entry of 2026-10-01.