Cognitive runtime architecture
Defines the externally observable surface for runtime cognition: state, attractors, safety signals, and replay.
Research
Research themes.
The project studies AI cognition as runtime structure: architecture, bounded dynamics, and concept formation that can be inspected outside hidden model weights.
Determinism and bounded dynamics
Studies whether runtime cognition can remain separable, measurable, and reproducible.
Operational concept formation
Explores how new runtime structures may form, specialize, and shape later behavior.
Publication Track
Public manuscripts will be linked as they mature.
The site will eventually host public abstracts, preprints, and supplementary material. For now, this page presents the research themes in a form suitable for external readers.
Architecture
What would it mean for cognition to live in a runtime?
This theme defines the observable system surface: field state, attractor structure, safety pressure, memory, and replay.
Dynamics
Can runtime cognition remain measurable and bounded?
This theme studies whether cognitive state can be separated from language generation and revisited across controlled runs.
Concepts
Can new cognitive structures form outside hidden weights?
This theme asks whether concepts can be treated as inspectable runtime structures rather than only implicit model behavior.
Release
Preprints and supplements coming later
External links will be added only when manuscripts and artifacts are ready for public review.