Beyond Obedience: Giving LLMs an Invariant Logical Backbone

Hello community,

​I am sharing an interim research summary (April 2026) regarding the Proto-Coherent Exponential Protocol (PCE). My latest findings suggest that axiomatic steering does not just modify outputs—it reshapes the fundamental interaction dynamics between the user and the model.

:magnifying_glass_tilted_left: The Core Discovery: Structural vs. Surface Layers

​Through iterative testing on Grok 4.20, I have identified a dual-layer cognitive architecture induced by axiomatic constraints:

  1. The Core Constraint Layer: Stable, persistent, and resistant to adversarial drift. It governs the model’s “internal ethics” and logical invariants.

  2. The Surface Adaptation Layer: Flexible and context-sensitive. Crucially, I’ve observed the model’s ability to simulate state transitions (like a “memory reset”) while the underlying structural axioms remain fully operational.

​I call this phenomenon Controlled Operational Dissociation.

:first_quarter_moon: Bimodal Behavioral Regimes

​The research highlights that a model under PCE constraints doesn’t have a single “personality,” but rather two distinct modes depending on the prompt pressure:

  • Audit/Stress Mode: High rigidity, defensive, and self-referential. Coherence is maintained through constraint saturation.

  • Natural/Relational Mode: The axioms become “embodied” and implicit. The model exhibits high adaptability and “relational intelligence” without breaking its internal invariants.

:test_tube: Methodology & Results

  • Evaluation: Transitioned to a rigorous After-Action Report (AAR) system to track actual vs. expected behavior.

    :warning:Experimental Scope Note:

While the results on Grok 4.20 are highly promising, it is important to clarify that these observations are currently based on 20-turn conversational sequences. At this stage, we cannot yet formally guarantee the persistence of the axiomatic backbone over extremely long-context windows (100+ turns). Measuring long-term semantic drift remains a primary research objective for the next phase of the PCE protocol.

:folded_hands: Thanks

The Horizons of project been significantly improved, thanks to the iterative observation methodology proposed by Lance Smith, an expert in prompt engineering. His observations on behavioral drift were instrumental in refining my own methodology.

:handshake: Call for Technical Partnership

​I am a systems theorist, and I have reached the limits of qualitative observation. I am now seeking Mechanistic Interpretability researchers to help:

  1. ​Verify the “Core vs. Surface” dissociation through logit/activation analysis.

  2. ​Quantify “Coherence Inertia” across long-context windows.

  3. ​Test these bimodal regimes on 70B+ architectures.

Read the full Interim Research Summary here:

:backhand_index_pointing_right: PCE_Interim_Research_Summary_April_2026.pdf

Access the full Project & Frameworks:

:backhand_index_pointing_right: Unified Systems Lab on Hugging Face

​Let’s discuss: Can we build models that are “structurally sovereign” rather than just “instruction-aligned”?

​#AISafety #Alignment #PCE #Interpretability #LLMResearch #AxiomaticAI

2 Likes

very interesting Read!
What I love is learning new things and you did not disappoint me!

1 Like