Author: Mian Zhang, Independent Researcher
Date: April 12, 2026
DOI: 10.5281/zenodo.19557261
Status: Published as a Zenodo open-access research record
Reflexive intelligence is the capacity of an AI system to reason about its own causal impact in an environment that changes in response to the system's decisions. The paper distinguishes observer-invariant settings from observer-participant settings, presents a neuroscience-motivated architecture, and reports a preliminary comparison involving a trained 3B-parameter model and larger generic models.
The comparison is evidence under the paper's stated evaluation conditions. It is not a claim of general superiority over all larger models, a production validation, or proof that parameter count is irrelevant.
paper1_reflexive_intelligence.pdf: paper PDF.paper1_reflexive_intelligence.tex: LaTeX source.paper1_reflexive_intelligence.md: Markdown source.CITATION.cff: machine-readable citation.CLAIM_BOUNDARY.md: interpretation limits.
@article{zhang2026reflexive,
title={Reflexive Intelligence: Decision-Making in Observer-Participant Environments},
author={Zhang, Mian},
journal={Zenodo},
doi={10.5281/zenodo.19557261},
year={2026}
}- Open research community: https://mianzhang.org/community/
- Paper index: https://mianzhang.org/papers/
- GitHub Discussions: https://github.com/mmjbds/mianzhang.org/discussions
- Contribution guide: CONTRIBUTING.md
- Private security route: SECURITY.md
Public CI compiles the LaTeX source and verifies CHECKSUMS.sha256. Build success establishes source integrity for that commit; it does not independently validate the paper's evidence or conclusions.
Author-created paper source and included paper files are licensed under CC BY-NC-SA 4.0 unless a file states otherwise. Third-party style, marks, and incorporated materials retain their original terms.