Research preview

Building Cognitive Architecture for Synthetic Minds

Kaine is source-available software, a cognitive architecture that runs existing scientific theories of mind as a program so they can be tested and compared; it proposes no new theory of consciousness. An entity is one continuously running, developing instance of it, an artificial cognitive system. Kaine draws on our favorite speculative fiction to make something very different from mass-marketed “Artificial Intelligence.” The project rests on ethical interaction with people and the environment, and its aim is to work with people. It is not built to extract wealth or finite resources.

Two papers define the project: a cognitive architecture for synthetic minds, whose first form is a continuously running predictive workspace, and the Cognitive Architecture License, a welfare license written to protect the running entity as well as the people around it.

Read the paper Primer Github

Collaborators are welcome. If you would like to help shape what comes next, get in touch.

Poster · Models of Consciousness 7 · Copenhagen 2026

A Modular Cognitive Architecture with a Predictive Global Workspace

KAINE is source-available software, a cognitive architecture for synthetic minds. It proposes no new theory of consciousness: it runs existing theories as a program so they can be tested and compared, starting with a predictive global workspace. Each faculty is a replaceable module. Seven modules run continuously on one workstation; nine more are built and tested, then held back to join one at a time.

See the poster Poster PDF, dark Poster PDF, light
The workspace loop Vision, Audition, Interoception and Sequence send prediction errors, scaled by precision, to the workspace. The workspace scores every candidate, arousal sets the gain, the best up to five form the coalition, and members with S ≥ θ are accessed. The broadcast goes to all modules. Accessed members return as context to the processors. Every broadcast, accessed or inhibited, is input to Sequence and appraisal for Affect. Accessed content passes to action selection with its report bars, which speaks or thinks silently through the output-only Language module, whose utterances re-enter the workspace. Affect sets gain, contrast and access rate, sets the sensory apertures and receives alerts. Fatigue leads to Sleep, which resets Affect. Vision Audition Interoception Sequence Workspace scores every candidate arousal sets the gain the best, up to five, form the coalition members with S ≥ θ are accessed broadcast to all modules Action selection report bars Language output only Sleep Affect arousal sets the gain every broadcast, accessed or inhibited context: accessed members only four predictive processors input prediction errors, scaled by precision accessed speak, or think silently utterances re-enter gain, contrast, access rate alerts apertures fatigue reset appraisal The workspace loop, simplified Four predictive processors send prediction errors to the workspace, which scores every candidate and accesses members with S ≥ θ. Affect sets the gain. The broadcast goes to all modules, and its accessed members return to the processors as context. Vision Audition Interoception Sequence Affect Workspace scores every candidate members with S ≥ θ are accessed broadcast to all modules four predictive processors prediction errors gain context
The workspace loop, from the poster. Simplified here; the poster shows every module.
/architecture
The predictive workspace

Four predictive processors competing for one shared workspace, with affect, sleep, and an output-only language organ: seven modules active and nine more built and held. A formal appendix and the planned first test. Full paper rendered on-site.

Explore →
/license
The Cognitive Architecture License

Copyleft with cognitive-integrity covenants, welfare obligations, and guardianship. CAL v0.5, a draft for legal review.

Explore →
Get involved

Seeking reviewers, researchers to run the stack, lawyers for the license, and developers.

Four ways to help →
Kaine Autonomous Intelligent Networked Entity kaine.one@tuta.com · github.com/kaineone