Artificial Conscious Access: A Falsifiable Causal and Formal Framework for Global Information Availability in Neural Systems

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This repository contains the code, data, formal proofs, and manuscript for Artificial Conscious Access: A Falsifiable Causal and Formal Framework for Global Information Availability in Neural Systems. ACA is a falsification framework, not a detector of consciousness. It operationalizes a narrow functional criterion involving cross-task information availability, local representation, control-task preservation, causal intervention, and activation restoration.The release includes an engineered positive-control replication, frozen behavioral Gate v2 evaluations, independent model-selection results, interface controls, reproducible analysis scripts, Lean 4 formalizations, theorem-audit outputs, and the submission manuscript. The experiments validate the instrumentation and provide negative behavioral/model-selection evidence. No evaluated trained model reached the mechanistic-intervention stage, and no claim about phenomenal consciousness is made.The archive is deterministic and secret-scanned. It includes per-file SHA-256 hashes and excludes model weights, runtime environments, and API credentials. Author: Gustavo Venegas, Argorix (gustavo@argorix.com). Version 0.1.2, 2026.