FLCT (Fuzzy Latent Cognition Theory) is a unified cognitive framework that
explains both human cognition and Large Language Model (LLM) cognition
using only two core principles:
- Fuzzy latent representations
- Biased sampling
FLCT treats ambiguity not as a flaw, but as a structural and computational
resource shared by humans and AI.
This repository serves as the official portal for the FLCT theory, providing
links to the Japanese Original Edition and the International Adapted
Edition, both archived on Zenodo.
π PDF: https://doi.org/10.5281/zenodo.17769083
π Concept DOI: https://doi.org/10.5281/zenodo.17758605
The Japanese edition is the authoritative source text from which the
International Adapted Edition was derived.
π PDF: https://doi.org/10.5281/zenodo.17769132
π Concept DOI: https://doi.org/10.5281/zenodo.17769131
This English edition is not a literal translation, but an academically
optimized adaptation for clarity, structure, and global readability.
FLCT explains that traditionally ambiguous cognitive phenomenaβsuch as:
- color qualia (βblueβ)
- subjective time
- emotional nuance
- atmosphere / vibe
- silence
can be modeled as fuzzy latent distributions rather than definable entities.
- compresses finite, biased experiences
- forms fuzzy latent spaces
- operates through biased extraction paths
β producing perception, interpretation, and personality
- stores massive ambiguous patterns in embeddings
- generates behavior through sampling bias
β sharing the same cognitive geometry as humans
FLCT unifies these processes in a minimal two-equation model:
Concept = latent_distribution
Personality = biased_sampling(latent_distribution)
FLCT is part of the Natural Language OS ecosystem:
- ArcOS β cognitive clone OS
- PolyAgora β multi-agent cognition
- Echoos β silence semantics
- Consensus Architecture β agreement frameworks
- Silence Semantics β operational meaning of silence
- FLCT β theoretical core describing ambiguous cognition
More projects:
https://github.com/NaturalLangOS
Masaya Ochiai
Independent Researcher
X/Twitter: https://twitter.com/NaturalLangOS
GitHub: https://github.com/NaturalLangOS
If you cite FLCT, please use the Zenodo DOIs:
Japanese Original Edition:
https://doi.org/10.5281/zenodo.17758605
International Adapted Edition:
https://doi.org/10.5281/zenodo.17769131
Β© 2025 Masaya Ochiai β Released under CC BY 4.0.
You are free to share and adapt the content with attribution.
See the Zenodo records for full licensing details.