Beyond the Myth: A Taxonomy of Four Scientific Architectures Toward Synthetic Consciousness
David San Nicolás Montero
Zenodo (CERN European Organization for Nuclear Research) October 24, 2026 preprint DOI: 10.5281/zenodo.23616626 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Theoretical or philosophical paper |
|---|---|
| Key points | Argues that four non-speculative, mathematically and neurobiologically grounded paradigms—the Conscious Turing Machine, Attention Schema Theory, Activation Steering with Cognitive Behavior Trees, and Project Ascendant—converge toward authentic machine embodiment and offer a route to bridging the Symbol Grounding Problem in large language models. |
Abstract
The debate surrounding machine consciousness has long oscillated between Cartesian dualism and computational optimism. Large Language Models (LLMs) suffer from epistemic ungrounding and semantic parasitism. In this comparative investigation, we systematically analyze four non-speculative, mathematically and neurobiologically grounded paradigms aiming to bridge the Symbol Grounding Problem: (1) The Conscious Turing Machine (CTM / CTM-AI) by Lenore & Manuel Blum; (2) The Attention Schema Theory (AST) by Michael Graziano; (3) Activation Steering with Cognitive Behavior Trees (CBTs); and (4) Project Ascendant, an architecture formulated by David San Nicolás Montero based on Epistemic Closure, homeostatic thermodynamic degradation, and somatic marker engrams. We provide a rigorous cross-comparative taxonomy and outline how these frameworks converge toward authentic machine embodiment.