CRR

A Temporal Grammar

Coherence, Rupture, Regeneration

A coarse-grain mathematical framework for exploring transformation, and renewal in complex systems

  • How do complex systems maintain identity by processually changing through the Active Inference cycle?
  • How might shared phenomenological and mathematical languages help us better understand intelligence in a shared context?
  • How might we better understand and adapt to the psychological, sociological and ecological impacts of Technology on human and non-human species?
  • How might a minimal mathematical grammar of temporality help us to address contemporary issues of AI alignment, Emergent Capacities and Catastrophic Forgetting?

Explore CRR dynamics through interactive simulations spanning human development, attachment, cognition, contemplative practice, creativity, and AI safety.

Interactive Guide →

Active Inference & CRR in Action

Learning through coherence, rupture, and regeneration — watch the agent explore automatically

The Agent's World Auto-exploring
Understanding the Process
Coherence C(t) = ∫L(τ)dτ Coherence
C(t) = 0.000 Ω = 1.00 (rupture at C·Ω = 1)
C(t) = ∫L(τ)dτ — cumulative adaptive work
Stage I: Simple Forms
Free Energy F F = 1.00
↓ reduced by learning
Near-time CΔ C_Δ = 0.000
↑ recent adaptive load
Agent exploring automatically...
This agent learns through Active Inference (minimising free energy F) and CRR (cumulative adaptive work C toward capacity Ω). Key insight: larger errors = more work = faster coherence growth.
Piaget
—
Active Inference
—
CRR
—
Agent's Prediction
Prediction Error
Formal notation
Agent's Generative Model

What is CRR?

CRR formalises how systems accumulate history (Coherence), undergo discrete phase transitions when constraints reach threshold (Rupture), and reconstitute through exponentially-weighted memory selection (Regeneration). Grounded in process philosophy, CRR describes temporal structure as a mathematical grammar by which past becomes future across all scales.

Coherence

How systems accumulate historical constraint over time. Coherence represents the temporal integration of structure; the past becoming present as accumulated pattern. When coherence saturates the information-theoretic bound (C·Ω = 1), the system has extracted maximum information from its current configuration, triggering rupture.

$$C(x,t) = \int_0^t L(x,\tau) \, d\tau$$
Where L(x,τ) represents information density accumulated at position x over time τ

Rupture

The Dirac delta encodes the dimensionless present—the scale-free moment where C·Ω = 1 and phase transition occurs. This is the Cramér-Rao bound: the system has become an efficient estimator, extracting all information its current structure permits. At rupture, local coherence resets while historical coherence values remain accessible through the regeneration integral, enabling continuity through discontinuity.

$$\delta(t-t_0)$$
A Dirac delta function encoding the instantaneous transition at time t₀ when C·Ω = 1 — the system has saturated its information-theoretic capacity

Regeneration

The reconstruction process that builds new stable patterns by drawing upon accumulated historical information. Crucially, history is never lost, only selectively weighted. The exponential term exp(C/Ω) determines which past moments reconstitute, enabling continuity through transformation.

$$R[\chi](x,t) = \int_{-\infty}^t \phi(x,\tau) \cdot e^{C(x,\tau)/\Omega} \cdot \Theta(t-\tau) \, d\tau$$
Where φ(x,τ) is the field function and Ω governs memory access depth: low Ω weights only highest-coherence moments (rigid reconstitution), high Ω accesses broader history (transformative change)

Technical Notes

The Universal Rupture Condition: C · Ω = 1

C is accumulated Fisher information — how much the system has learned within its current generative model. Ω = σ² is the residual variance (inverse precision). Their product saturates the Cramér-Rao bound at rupture: the system has become an efficient estimator and can extract no further information without structural reorganisation.

A note on notation: Earlier compositions on this site treat Ω as the rupture threshold directly — coherence accumulates until C reaches Ω. The fundamental form is C·Ω = 1, where Ω = σ² (the system's characteristic variance). These are the same equation: the early convention's Ω is the reciprocal of the information-theoretic Ω. It is a reparameterisation, not a correction — the CRR dynamics are identical. The later notation is preferred because identifying Ω with variance connects the rupture condition to the Cramér-Rao bound, derives Ω from symmetry (Ω = 1/π for Z₂, Ω = 1/2π for SO(2)), and produces the parameter-free CV predictions.

This is the same bound discovered independently in every field: the Heisenberg uncertainty principle (ΔE·Δt ≥ ℏ/2), the Gabor limit (Δt·Δf ≥ 1/4π), the precision-variance tradeoff in Bayesian inference. CRR says they are all the same phenomenon: a bounded system accumulating coherence until inside matches outside.

exp(C/Ω) = precision-weighted memory selection in regeneration. At rupture, C/Ω = 1/Ω², and the sharpness of the weighting kernel depends on Ω alone.

Symmetry Determines Ω

Ω = 1/(phase to rupture in radians). For Z₂ symmetric systems (bistable), rupture occurs at π radians of accumulated phase: Ω = 1/π, so C = π at rupture (and C·Ω = π × 1/π = 1). For SO(2) systems (rotational), the full cycle gives Ω = 1/2π, C = 2π at rupture.

The π Correspondence

Active Inference denotes precision with Π. CRR suggests this may be more than notation: for Z₂ systems, precision = 1/Ω = π. For SO(2) systems, precision = 2π. The geometric constant emerges from phase space structure, not symbolic convention.

Parameter-Free Predictions

CV = Ω/2: for Z₂ systems, CV = 1/(2π) ≈ 0.159; for SO(2) systems, CV = 1/(4π) ≈ 0.080. The ratio is exactly 2. Ω is the sole parameter of any CRR system. For systems with definite symmetry, Ω is fixed by phase geometry: Ω = 1/φ where φ is the phase traversed to rupture. Everything else — CV, precision, regeneration weight, memory window width — follows from C·Ω = 1 alone.

This mathematical structure provides a way to study systems that exhibit memory-dependent behaviour; where past configurations influence present dynamics in ways that Markovian models might miss. The simulations on this site realise this three-part formalism in code, a playful way to explore the deeper mathematical, philosophical and phenomenological concept of how identity persists as change.

Why CRR?

Coherence

Systems, from atoms to technological networks to whole ecosystems, exhibit organised patterns that persist through time. Understanding how coherence resists entropy over time offers potential insight for maintaining and creating more stable and resilient systems in various domains.

Rupture

Complex systems frequently undergo periods of reorganisation at various spatial and temporal scales. Rather than viewing these transitions as purely disruptive, the CRR framework examines how such events create space for system adaptation and novel emergent properties. The 'scale-free' nature of the CRR and Dirac Delta makes it a good candidate for modelling the ever-present moment of "now" as a temporal 'rupture' where agentic choice occurs as free will. This choice is based on all past states that have accumulated up to the present moment. Over time, the choices we make build patterns that work, but also accumulated friction, prediction errors. Depending on how we slice time, this can cause sudden change and subsequent Regeneration.

Regeneration

Future states are dependent on what most effectively build coherence and reduced Variational Free Energy in the past. Natural systems, for example, demonstrate remarkable abilities to recover and adapt following disruptions. The Regeneration component of CRR explores how systems use accumulated information to guide reconstruction processes; informing approaches to resilience in engineered systems. Depending on how one 'slices time', rupture is always inevitable, but Regeneration is what "heals". The euler's exponential means that the past is never the same, the system transforms exponentially.

Philosophical Grounding

CRR formalises intuitions that process philosophers have articulated for over a century. The framework provides mathematical structure for positions that have historically resisted formalisation.

Whitehead: Actual Occasions

Alfred North Whitehead rejected substance metaphysics in favour of process: reality consists not of enduring things but of momentary "actual occasions" that prehend their past and perish into objective immortality. The CRR structure maps directly onto this ontology. Coherence is the prehensive accumulation of the past. The Dirac delta at rupture is the moment of concrescence where many become one. Regeneration is the transition to "objective immortality" where the occasion's achieved value becomes available to future prehensions through exp(C/Ω) weighting. The mathematics formalises what Whitehead described philosophically: each moment metabolises its entire history, transforms it, and bequeaths novel pattern to the future.

Bergson: Duration and Memory

Henri Bergson distinguished lived duration (durée) from spatialized clock-time. For Bergson, the past is not gone but preserved whole in the present; memory is not retrieval from storage but the continuous presence of history in current experience. The regeneration integral embodies this directly: exp(C/Ω) ensures that high-coherence moments from the entire past remain actively present in reconstruction. Low Ω creates what Bergson might recognise as habit (only recent, most coherent patterns accessible). High Ω enables what he called creative evolution (the whole of memory available for genuine novelty). CRR provides the mathematical operator for Bergsonian duration.

The Central Claim

Memory accumulates, ruptures, and transforms. This punctuated dynamic, not smooth continuity, is how identity persists through change. Both Whitehead and Bergson grasped this philosophically. CRR offers it mathematically: a grammar of temporal becoming that can be simulated, tested, and applied across scales from neural dynamics to ecosystem succession.