Computational Neural Processing: Scarcity, allocation, and total reliance under a shared rate budget
Year: 2026
Type: article
Abstract: Description Sīdus E1 RC4 tests scarcity, value-sensitive allocation, causal total reliance and allocation prediction in a finite synthetic system with three source-local monitoring channels under a shared KL-rate constraint. The registered panel contains 501 paired training seeds across six conditions; all 3,006 conditions completed. Under the frozen all-seed analysis, invalid components contribute zero. T1, T2, U and V were confirmed, yielding the recorded PASS at the declared fixture scope. A retrospective AI-assisted audit reproduced the original adjudication byte-for-byte and a separately written, source-aware arithmetic calculation matched 481 comparisons using the same SciPy library. The attached selected analytic evidence supports arithmetic reproduction from assembled data; it is not the full execution archive. This is a public working paper, not a peer-reviewed or independently replicated result. A retained record reports an unplanned environment-distribution verification refusal and subsequent restart; the cause and remediation remain unresolved. Current verification does not establish historical continuity. Primary U is evaluated at high capacity; low-capacity replacement results are descriptive diagnostics. No claim of strict metarepresentation, architecture-general law, biological cognition or demonstrated metarepresentational ecology is made. AI assistance is disclosed in the paper. Companion: https://sidus.quest/research/e1Interactive explanation: https://sidus.quest/#/results/e1Website release: https://github.com/RavenSeldon/Sidus/commit/593ed2f55dcb74030ce4ef309ca1f8850e6f7bbe
Author: Amuwo, Benjamin Maitreya
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225450131