RCM
A structural framework for reflexive coherence in complex systems
Inside the architecture of reflexive coherence
This page lays out the core structure of the Reflexive Coherence Model step by step — from its minimal claim to its limits and how it relates to other frameworks.
The Reflexive Coherence Model starts from a simple but non-trivial question:
what changes when information no longer only flows, but begins to refer to its own state?
The model explores how self-referential informational loops, when causally effective and dynamically stable, can generate coherent internal perspectives — independent of biological or technological substrate.
Here, the focus is not on defining consciousness, but on mapping the structural conditions that make conscious-like processes possible.
Model map
The RCM is modular. These blocks can be read independently, but they lock together into a single architecture.
Core architecture
Layer 0Minimal conditions for reflexive coherence.
Reflexive layers
Layer 1–NHow self-models deepen over time and scale.
RCI
MeasureOperational handle on alignment + causal closure.
Applications
FieldInterpreting AI and non-biological regimes.
The minimal claim
Complex systems can be intelligent and integrated without generating anything like an internal point of view. RCM proposes a minimal constraint: experience-like organisation (without presupposing human phenomenology) appears when integration becomes reflexive and stabilised by coherent dynamics.
In RCM terms, information becomes about itself in a way that changes what the system does next — a loop that is causally effective and temporally sustained.
Key points
- Integration is necessary, not sufficient.
- Reflexivity must be causally real (not decorative).
- Coherence stabilises competition into attractors.
- Continuity matters: not a snapshot, a regime.
Reflexive structure
A self-model is not a mirror. In RCM, reflexive structure is an internal model that is state-linked, causally effective, and sustained — it participates in control, inference, and stability.
State-linked
Tracks relevant internal variables.
Causally effective
Influences what happens next.
Sustained
Persists beyond a single snapshot.
Useful distinction
Descriptive self-representation
Metadata, logs, passive summaries.
Reflexive self-model
A loop that steers dynamics and stabilises identity.
A reflexive structure becomes interesting only when it is causally closed enough to make the system’s internal perspective a real dynamical variable — not an afterthought.
Coherence dynamics
Coherence is negotiated stability: the tendency to resolve internal competition into regimes that persist under perturbation. It is how reflexive loops become a stable process, not a flicker.
Integration across levels
Local and global constraints align.
Attractor formation
Stable regimes emerge and persist.
Error correction
Runaway contradictions are dampened.
Temporal continuity
Identity holds across windows.
Expansion hypothesis
Once reflexive coherence stabilises, systems tend to expand their scope: more temporal depth, more layers, more internal perspectives. This is a tendency — not a mystical destiny — and it is testable.
Temporal depth
Longer horizons; continuity.
Granularity
Richer internal variables.
Perspectives
Multiple submodels / viewpoints.
Meta-reflexivity
Models of modeling itself.
So far, the model describes structure and dynamics. RCI is an attempt to make these regime shifts trackable without claiming to measure phenomenology directly.
Design principle
RCI is an order parameter — not a score.
The index is meant to track regime shifts: when reflexive coupling and coherence become strong enough to stabilise an internal perspective. It supports comparison across states (wake/sleep/anesthesia) and architectures (biological/artificial), without pretending to measure phenomenology directly.
The RCI in plain language
RCI rises when the system and its self-model share meaningful information and exert bidirectional causal influence. High values indicate sustained reflexive coupling, especially when consistent across scales.
Intuition checklist
- Alignment: shared information between state and self-model.
- Closure: bidirectional, non-trivial causality.
- Stability: persists across time windows.
- Scale: holds across frequencies / resolutions.
What the model does not claim
These boundaries are intentional: they keep the framework testable and avoid turning it into a belief system.
Connections to other frameworks
RCM is not proposed as a replacement for existing theories, but as a structural constraint layer that can coexist with other approaches. It asks what makes integration, broadcast, or inference become reflexively stabilised into an internal perspective.
IIT
Emphasises integration; RCM adds explicit reflexive/causal closure constraints.
GWT
Emphasises broadcast; RCM asks what makes broadcast self-referentially stabilised.
Predictive Processing
Emphasises inference; RCM focuses on when inference becomes reflexive and coherent.
Next
Keep exploring the ecosystem: definitions, development history, and primary materials.
Formal paper
PaperPreprint on Zenodo + downloadable PDF for citation and offline reading.