Research synthesis

What We Have Learned

Key findings from DuoNeural's research program across RLHF truth suppression, recurrent predictive architectures, the Dynamical Horizon Principle, and emergent circuit behavior.

On RLHF and Truth

Internal knowledge and external behavior can diverge.

On Recurrent Architectures and the Dynamical Horizon

Learning systems find the edge of useful prediction.

On Emergent Behaviors

Composition can matter more than obvious weight signatures.

Terminology Note

CCS is now CDP in the canonical paper stream.