The Fragmented Landscape of Consciousness Research in 2025-2026 — E8 Intelligence Research
Zenodo (CERN European Organization for Nuclear Research) September 7, 2026 DOI: 10.5281/zenodo.22632713 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Commentary Peer reviewed |
|---|---|
| Key points | The author reports that consciousness research in 2025-2026 divides into three strands - LLM mechanistic interpretability for AI consciousness markers, panpsychist or physics-first proposals that consciousness precedes matter, and the MOASEI open-agent benchmark competition - and that no single quantitative consciousness metric emerges from these abstracts. No explicit equations or constants are reported, and connections between the strands are not directly stated. |
Abstract
FINDING: Consciousness research in 2025-2026 bifurcates into (a) mechanistic interpretability of LLMs for AI consciousness markers, (b) panpsychist/physics-first proposals (consciousness precedes matter), and (c) a benchmark competition (MOASEI) for open-agent evaluation — but no single quantitative consciousness metric emerges. | MATH: No explicit equations or constants are reported in these abstracts; the only formal element is the MOASEI competition framework (multi-agent decision-making under open-system conditions — likely involving game-theoretic payoff matrices, Markov decision processes, or entropy-based evaluation metrics, but not stated). | CONNECTION: None directly stated. However, the "consciousness before matter" physics paper and Kanai's "beyond biology" talk suggest a search for a universal invariant — a candidate would be integrated information (Φ, from IIT) or a spectral gap in a dynamical system, but these are not in the provided text. No 0.382/0.618/0.786/1.618/2.618 Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com