Home » Core Papers

Category Archives: Core Papers

Excess Entropy Production as a Candidate Universal Cost of Persistence: A Thermodynamic Foundation for the Attractor Framework; Robert Galida (July 2026) [F]

June 27, 2026 6:36 pm / Leave a comment

Abstract

Every dissipative system maintains its attractor through continuous reconfiguration. Reconfiguration requires work; work generates entropy. The recovery rate $κ$ κ — corrective permeability — is the rate at which a system reconfigures to return to its attractor after perturbation. This paper proposes that $κ$ κ is a measure of excess entropy generation rate.

We develop an abstract persistence cost framework and prove its equivalence to Lyapunov theory. We then identify entropy production as a physical realization of this cost, deriving: $κ = \inf_{x} \frac{δ (x)}{\int_{0}^{\infty} σ_{excess} (ϕ_{t} (x)) d t}$ κ=xinf∫0∞σexcess(ϕt(x))dtδ(x)

where $σ_{excess} = σ - σ_{s s}$ σexcess=σ−σss is the excess entropy production rate above the system’s steady-state baseline. For physical systems, the baseline is zero (equilibrium); for biological, cognitive, and social systems, the baseline is the steady-state dissipation rate of the healthy, well-coordinated attractor.

This unifies physical, biological, cognitive, and social systems. The framework is grounded in the second law of thermodynamics and non-equilibrium steady-state thermodynamics, not analogy. Empirical predictions are provided for each domain.

Keywords: entropy generation, excess entropy production, corrective permeability, attractor framework, dissipative structures, reconfiguration, Lyapunov theory, free energy principle, allostatic load

1. Introduction

The attractor framework defines persistence as the ability of a system to maintain its attractor under perturbation. Historically, persistence has been measured kinematically — as distance traveled or time spent away from equilibrium. This paper proposes that the true cost of persistence is thermodynamic: it is the excess entropy generated during reconfiguration and recovery.

Every dissipative system maintains its attractor through continuous reconfiguration. A bacterium reconfigures its metabolism to maintain homeostasis. A brain reconfigures its synaptic connections to maintain predictive models. A society reconfigures its institutions to maintain order. Reconfiguration requires work; work generates entropy. The second law of thermodynamics applies at every level of organization.

We develop an abstract persistence cost framework first, establishing its equivalence to Lyapunov theory. We then identify entropy production as a physical realization of this cost, deriving the relationship between corrective permeability and excess entropy generation.

The framework unifies physical, biological, cognitive, and social systems. It is grounded in the second law of thermodynamics and non-equilibrium steady-state thermodynamics, not analogy.

2. The Persistence Cost Functional

Let $X$ X be a state space, $ϕ_{t} (x)$ ϕt(x) the flow of a dynamical system, and $A \subseteq X$ A⊆X an attractor set. Let $δ (x) = d (x, A)$ δ(x)=d(x,A) be the distance from $x$ x to the attractor. For a treatment of state-space constraints in viability theory, see Aubin (1991).

Definition 1 (Persistence Cost Functional): A persistence cost functional $C (x)$ C(x) is a scalar function on $X$ X satisfying:

$C (x) \geq 0$ C(x)≥0 for all $x$ x
$C (x) = 0$ C(x)=0 if and only if $x \in A$ x∈A
$C (ϕ_{t} (x)) \in L^{1} ([0, \infty))$ C(ϕt(x))∈L1([0,∞)) for all $x$ x in the basin

Definition 2 (Cumulative Persistence Cost): For a finite horizon $T > 0$ T>0: $D_{T} (x) = \int_{0}^{T} C (ϕ_{t} (x)) d t$ DT(x)=∫0TC(ϕt(x))dt

For trajectories that converge to the attractor: $D_{\infty} (x) = \int_{0}^{\infty} C (ϕ_{t} (x)) d t$ D∞(x)=∫0∞C(ϕt(x))dt

3. Existence and Lyapunov Equivalence

Theorem 1 (Existence of the Persistence Functional): Assume $C (x) \geq 0$ C(x)≥0, $C = 0$ C=0 only on $A$ A, and $C (ϕ_{t} (x)) \in L^{1} ([0, \infty))$ C(ϕt(x))∈L1([0,∞)) for all $x$ x in the basin. Assume $f$ f is locally Lipschitz, the flow is continuously differentiable in the initial condition, and $C$ C is continuous and locally bounded. Then:

$D_{\infty} (x) = \int_{0}^{\infty} C (ϕ_{t} (x)) d t$ D∞(x)=∫0∞C(ϕt(x))dt exists and is finite.
$D_{\infty}$ D∞ is continuous.
$D_{\infty}$ D∞ satisfies the transport equation:

$\nabla D_{\infty} (x) \cdot f (x) = - C (x)$ ∇D∞(x)⋅f(x)=−C(x)

Proof: The integral exists and is finite by the $L^{1}$ L1 assumption. Continuity follows from the dominated convergence theorem under the stated regularity assumptions. To derive the transport equation, compute: $D (ϕ_{h} (x)) = \int_{h}^{\infty} C (ϕ_{t} (x)) d t = D (x) - \int_{0}^{h} C (ϕ_{t} (x)) d t$ D(ϕh(x))=∫h∞C(ϕt(x))dt=D(x)−∫0hC(ϕt(x))dt

Then: $\frac{D (ϕ_{h} (x)) - D (x)}{h} = - \frac{1}{h} \int_{0}^{h} C (ϕ_{t} (x)) d t \to - C (x)$ hD(ϕh(x))−D(x)=−h1∫0hC(ϕt(x))dt→−C(x)

as $h \to 0$ h→0. By the chain rule: $\nabla D (x) \cdot f (x) = - C (x)$ ∇D(x)⋅f(x)=−C(x) $□$ □

Corollary (Equivalence to Lyapunov Theory): Any Lyapunov function $V (x)$ V(x) (with $V \geq 0$ V≥0, $V = 0$ V=0 on the attractor, and $\dot{V} \leq 0$ V˙≤0) yields a persistence cost $C (x) = - \dot{V} (x)$ C(x)=−V˙(x). Conversely, any persistence cost $C (x)$ C(x) satisfying $\nabla D \cdot f = - C$ ∇D⋅f=−C defines a Lyapunov function $D (x)$ D(x).

Proof: If $V$ V is a Lyapunov function, then $\dot{V} = \nabla V \cdot f \leq 0$ V˙=∇V⋅f≤0. Define $C = - \dot{V}$ C=−V˙. Then $C \geq 0$ C≥0, $C = 0$ C=0 on the attractor, and $D_{T} = \int C = V (x) - V (ϕ_{T} (x))$ DT=∫C=V(x)−V(ϕT(x)). Conversely, if $\nabla D \cdot f = - C$ ∇D⋅f=−C, then $\dot{D} = - C \leq 0$ D˙=−C≤0, so $D$ D is a Lyapunov function. $□$ □

Interpretation: The persistence cost framework is mathematically equivalent to classical Lyapunov stability theory. For the connection to contraction analysis, see Lohmiller & Slotine (1998). For control Lyapunov functions, see Freeman & Kokotovic (1996). Entropy production is one physically meaningful realization of the cost function $C$ C. For a detailed treatment of Lipschitz continuity of $D_{\infty}$ D∞ under a Lipschitz-flow hypothesis, see Galida (2026a), Proposition 4.

4. Entropy Production as Persistence Cost

4.1 Entropy Balance

For an open system, the entropy balance equation is: $\frac{d S_{system}}{d t} = σ - Φ$ dtdSsystem=σ−Φ

where $σ \geq 0$ σ≥0 is the entropy production rate (always non-negative by the second law) and $Φ$ Φ is the entropy export rate to the environment. For foundational treatments of stochastic thermodynamics and entropy production, see Seifert (2012) and Sekimoto (2010).

For a system in a steady state: $\frac{d S_{system}}{d t} = 0 ⟹ σ = Φ$ dtdSsystem=0⟹σ=Φ

4.2 Excess Entropy Production

Define the steady-state entropy production rate $σ_{s s}$ σss as the rate when the system is at its attractor.

Define the excess entropy production rate: $σ_{excess} (x) = σ (x) - σ_{s s} (x)$ σexcess(x)=σ(x)−σss(x)

Assumption (Excess Entropy Decay): For all trajectories in the basin, there exist constants $C < \infty$ C<∞ and $μ > 0$ μ>0 such that: $σ_{excess} (ϕ_{t} (x)) \leq C e^{- μ t} σ_{excess} (x)$ σexcess(ϕt(x))≤Ce−μtσexcess(x)

for all $t \geq 0$ t≥0. This ensures $D_{\infty} (x) < \infty$ D∞(x)<∞ and is the standard hypothesis under which the persistence functional and its associated bounds are well-defined, consistent with Galida (2026a, 2026b). The decay rate $μ$ μ may be domain-specific and is empirically measurable.

Note on generalization: The exponential decay assumption is adopted here to ensure finiteness of $D_{\infty}$ D∞ and to maintain consistency with the prior papers in this series. Generalization to $L^{1}$ L1 integrable decays (e.g., algebraic) is a priority for future work.

4.3 The Entropy Persistence Functional

Definition 3 (Cumulative Excess Entropy Functional): For a finite horizon $T > 0$ T>0: $D_{T} (x) = \int_{0}^{T} σ_{excess} (ϕ_{t} (x)) d t$ DT(x)=∫0Tσexcess(ϕt(x))dt

For trajectories that converge to the attractor: $D_{\infty} (x) = \int_{0}^{\infty} σ_{excess} (ϕ_{t} (x)) d t$ D∞(x)=∫0∞σexcess(ϕt(x))dt

Interpretation: The persistence functional is the total excess entropy generated during reconfiguration and recovery.

4.4 Corrective Permeability

Definition 4 (Corrective Permeability): $κ = \inf_{x \in B ∖ A} \frac{δ (x)}{D_{\infty} (x)}$ κ=x∈B∖AinfD∞(x)δ(x)

where $δ (x) = d (x, A)$ δ(x)=d(x,A) is the distance to the attractor.

Interpretation: $κ$ κ is the minimum excess entropy cost per unit distance. It measures the efficiency of reconfiguration: a system that returns with minimal excess entropy generation has high $κ$ κ; a system that generates excess entropy has low $κ$ κ.

4.5 Basin Depth

Proposition 1 (Properties of Basin Depth): Define $B = D_{\infty} (saddle)$ B=D∞(saddle), where $saddle$ saddle is the lowest point on the basin boundary (the separatrix between attractors). For the connection to large-deviation theory and escape rates, see Freidlin & Wentzell (2012). Then:

$B \geq 0$ B≥0, with equality iff the basin has no barrier (i.e., the boundary coincides with the attractor).
For gradient systems $\dot{x} = - \nabla V (x)$ x˙=−∇V(x), $B = V (saddle) - V (A)$ B=V(saddle)−V(A) (the classical energy barrier).
$B$ B is invariant under smooth coordinate changes (coordinate invariance).
$B$ B depends on the chosen persistence cost functional $C$ C; different costs yield different barriers.

Proof: (1) follows from non-negativity of $D_{\infty}$ D∞. (2) follows from the transport equation $\nabla D \cdot f = - C$ ∇D⋅f=−C and the identity $f = - \nabla V$ f=−∇V. (3) follows from the invariance of the integral under diffeomorphisms. (4) is self-evident.

5. Domain-Specific Realizations

5.1 Physical Systems: Thermodynamic Excess Entropy

For a thermodynamic system, $S (x) = k_{B} \log Ω (x)$ S(x)=kBlogΩ(x), where $Ω (x)$ Ω(x) is the number of microstates. For an isolated system, $σ_{s s} = 0$ σss=0 (equilibrium), so $σ_{excess} = σ = \dot{S}$ σexcess=σ=S˙. $κ = \inf_{x} \frac{δ (x)}{S (A) - S (x)}$ κ=xinfS(A)−S(x)δ(x)

Example: A gas returning to equilibrium after compression. The entropy generated is $Δ S = n R \log (V_{f} / V_{i})$ ΔS=nRlog(Vf/Vi).

5.2 Biological Systems: Metabolic Excess Entropy

For a biological system, $S (x)$ S(x) is the metabolic entropy. The baseline $σ_{s s}$ σss is the resting metabolic rate (homeostasis). The excess is: $σ_{excess} = metabolic rate - resting metabolic rate$ σexcess=metabolic rate−resting metabolic rate $κ = \inf_{x} \frac{δ (x)}{\int_{0}^{\infty} σ_{excess} (ϕ_{t} (x)) d t}$ κ=xinf∫0∞σexcess(ϕt(x))dtδ(x)

Example: A cell returning to homeostasis after a nutrient shock. The excess entropy generated is the metabolic cost of restoring homeostasis above baseline. For the dissipative-structures framework underlying biological self-organization, see Nicolis & Prigogine (1989).

5.3 Cognitive Systems: Free Energy Dissipation

For a cognitive system, variational free energy $F = - \log p (y ∣ x) + D_{KL} [q (\cdot) ∥ p (\cdot ∣ x)]$ F=−logp(y∣x)+DKL[q(⋅)∥p(⋅∣x)] is adopted here as one candidate persistence functional. We do not claim variational free energy is uniquely correct; it is adopted as the most developed existing candidate persistence functional for cognitive systems. Other candidates (Bayesian surprise, expected free energy, predictive information) are possible; this paper focuses on $F$ F due to its established role in the free-energy principle (Friston, 2010). For the thermodynamics of information and its connection to free-energy minimization, see Parrondo, Horowitz & Sagawa (2015) and Sagawa & Ueda (2008).

The baseline $σ_{s s}$ σss is the baseline neural dissipation rate (resting brain activity). The excess is: $σ_{excess} = \dot{F} - {\dot{F}}_{s s}$ σexcess=F˙−F˙ss $κ = \inf_{x} \frac{δ (x)}{\int_{0}^{\infty} σ_{excess} (ϕ_{t} (x)) d t}$ κ=xinf∫0∞σexcess(ϕt(x))dtδ(x)

Example: A cognitive system updating its beliefs after a prediction error. The excess entropy generated is the free energy dissipated during belief updating above baseline.

5.4 Social Systems: Coordination Excess Entropy

For a social system, define the aggregate social entropy production rate as: $σ^{social} (t) = \sum_{i} ({\dot{S}}_{i} (t) - {\dot{S}}_{i}^{rest})$ σsocial(t)=i∑(S˙i(t)−S˙irest)

where ${\dot{S}}_{i} (t)$ S˙i(t) is the total entropy production rate of individual $i$ i, and ${\dot{S}}_{i}^{rest}$ S˙irest is the individual’s baseline entropy production rate in a resting, minimally socially constrained state. This is measured via physiological proxies such as basal metabolic rate, resting allostatic load, or cortisol baseline (McEwen, 1998; Sterling & Eyer, 1988).

Interpretation: $σ^{social}$ σsocial measures the excess dissipation attributable to social constraints: the additional entropy generated by coordination, communication, conflict, norm enforcement, and institutional friction.

Non-Negativity: Unlike total entropy production ${\dot{S}}_{i} \geq 0$ S˙i≥0 (which follows from the second law), $σ_{i}^{social}$ σisocial is not guaranteed to be non-negative. Division of labor, infrastructure, and specialization may reduce an individual’s metabolic burden relative to a solitary baseline. The hypothesis is that during recovery from social disruption, $σ_{i}^{social} \geq 0$ σisocial≥0; in steady-state, $σ_{i}^{social} \to 0$ σisocial→0. This is an empirical claim, not a theorem.

The baseline $σ_{s s}$ σss is the steady-state social entropy production rate (well-coordinated society). The excess is: $σ_{excess} = σ^{social} - σ_{s s}$ σexcess=σsocial−σss $κ = \inf_{x} \frac{δ (x)}{\int_{0}^{\infty} σ_{excess} (ϕ_{t} (x)) d t}$ κ=xinf∫0∞σexcess(ϕt(x))dtδ(x)

Example: A society recovering from a shock (economic crisis, political upheaval). The excess entropy generated is the coordination cost of restructuring above baseline. A harmonious society has $σ_{excess} = 0$ σexcess=0; a turbulent society has $σ_{excess} > 0$ σexcess>0; a chronically turbulent society may have settled into a new attractor with a higher $σ_{s s}$ σss. This illustrates the framework’s central distinction: the attractor is the state of minimum entropy generation for that class of system.

6. The Unified Framework

6.1 Summary Table

Domain	Entropy Functional	Baseline $σ_{s s}$ σss	Excess $σ_{excess}$ σexcess	Recovery Rate $κ$ κ
Physical	Thermodynamic entropy	0 (equilibrium)	$\dot{S}$ S˙	$\inf \frac{δ}{Δ S}$ infΔSδ
Biological	Metabolic entropy	Resting metabolic rate	Metabolic rate — resting	$\inf \frac{δ}{\int σ_{excess} d t}$ inf∫σexcessdtδ
Cognitive	Free energy	Baseline neural dissipation	$\dot{F} - {\dot{F}}_{s s}$ F˙−F˙ss	$\inf \frac{δ}{\int σ_{excess} d t}$ inf∫σexcessdtδ
Social	Social entropy production	Steady-state social dissipation	$σ^{social} - σ_{s s}$ σsocial−σss	$\inf \frac{δ}{\int σ_{excess} d t}$ inf∫σexcessdtδ

6.2 The Universal Structure

Every domain follows the same mathematical structure:

Component	Expression
Excess entropy production	$σ_{excess} (x) = σ (x) - σ_{s s}$ σexcess(x)=σ(x)−σss
Cumulative cost	$D_{\infty} (x) = \int_{0}^{\infty} σ_{excess} (ϕ_{t} (x)) d t$ D∞(x)=∫0∞σexcess(ϕt(x))dt
Recovery rate	$κ = \inf_{x} δ (x) / D_{\infty} (x)$ κ=infxδ(x)/D∞(x)
Basin depth	$B = D_{\infty} (saddle)$ B=D∞(saddle)
Transport equation	$\nabla D \cdot f = - σ_{excess}$ ∇D⋅f=−σexcess

6.3 The Low-Energy Attractor Benchmark (Proposed Hypothesis)

We propose the following benchmark as an additional hypothesis: the attractor is the state of minimum entropy generation for that class of system.

Domain	Attractor	Entropy Generation at Attractor
Physical	Equilibrium	$σ = 0$ σ=0
Biological	Homeostasis	$σ = σ_{s s} > 0$ σ=σss>0 (resting metabolism)
Cognitive	Settled Belief	$σ = σ_{s s} > 0$ σ=σss>0 (baseline neural dissipation)
Social	Coordinated Order	$σ = σ_{s s} > 0$ σ=σss>0 (baseline institutional friction)

Interpretation:

For equilibrium systems (gases, isolated systems), the attractor is the state where entropy generation reaches zero — the system has nowhere lower to go.
For dissipative systems (cells, brains, societies), the attractor is the state where entropy generation reaches its lowest non-zero steady-state value — the minimum entropy generation the system can sustain while maintaining its functional organization.

Important caveats:

This is a proposed benchmark, not a derived theorem.
For cognitive systems in particular, minimizing entropy production rate (a thermodynamic quantity) and minimizing free energy/surprise (the actual claim in the free-energy principle) are distinct minimization principles. The framework does not establish a bridge between them; this is an open question.
The benchmark is an empirical hypothesis that requires domain-specific validation.

In all cases, the attractor is the lowest entropy-generating state that system can have while remaining itself.

7. Testable Predictions

7.1 Core Prediction

Prediction: The recovery rate $κ$ κ is inversely proportional to the excess entropy generated during reconfiguration: $κ \propto \frac{1}{D_{\infty}}$ κ∝D∞1

Falsification: If a system returns to its attractor with high excess entropy generation but high recovery rate, the prediction is falsified.

7.2 Secondary Prediction

Prediction: Systems that maintain their attractor with minimal excess entropy generation are more “efficient.” Systems that generate excess entropy are “inefficient” or “stressed.”

Falsification: If an inefficient system has lower excess entropy generation than an efficient system, the prediction is falsified.

7.3 Domain-Specific Predictions

Domain	Prediction	Falsification
Physical	$κ$ κ correlates with thermal efficiency	$κ$ κ high but efficiency low
Biological	$κ$ κ correlates with metabolic efficiency	$κ$ κ high but metabolic cost high
Cognitive	$κ$ κ correlates with learning efficiency	$κ$ κ high but learning cost high
Social	$κ$ κ correlates with institutional efficiency	$κ$ κ high but coordination cost high

8. Experimental Design

8.1 Physical Systems

System: Gas in a piston
Perturbation: Compression
Measurement: Excess entropy generation (heat measurement) and recovery time
Test: Correlation between $κ$ κ and $1 / D_{\infty}$ 1/D∞

8.2 Biological Systems

System: Cell culture
Perturbation: Nutrient shock
Measurement: Metabolic rate above resting (oxygen consumption) and recovery time
Test: Correlation between $κ$ κ and metabolic cost

8.3 Cognitive Systems

System: Human participants in a learning task
Perturbation: Prediction error
Measurement: Free energy dissipation above baseline (EEG complexity, pupil dilation) and belief updating rate
Test: Correlation between $κ$ κ and free energy dissipation

8.4 Social Systems

System: Institutional response to shocks
Perturbation: Economic or political crisis
Measurement: Social entropy production above baseline (allostatic load, cortisol, institutional friction) and recovery time
Test: Correlation between $κ$ κ and social entropy production

9. Open Questions

Question	Status	Difficulty
Q1: Uniqueness of S(x)S(x)	Are there multiple valid entropy functionals for a given domain?	Hard
Q2: Variational principle	Is there a universal variational principle that yields $S (x)$ S(x)?	Hard
Q3: Social second law	Does $σ^{social} \geq 0$ σsocial≥0 always hold during recovery?	Very Hard
Q4: Cross-level entropy	How does entropy generation at one level relate to entropy generation at another?	Hard
Q5: Measurement	Can we measure excess entropy generation in cognitive and social systems directly?	Moderate
Q6: Unification	Can all domain-specific entropy functionals be derived from a single universal functional?	Very Hard

10. Conclusion

Every dissipative system maintains its attractor through continuous reconfiguration. Reconfiguration requires work; work generates excess entropy. The recovery rate $κ$ κ — corrective permeability — is the rate at which a system reconfigures to return to its attractor after perturbation. We have proposed that $κ$ κ is a measure of excess entropy generation rate.

We developed an abstract persistence cost framework and proved its equivalence to Lyapunov theory. We then identified entropy production as a physical realization of this cost, deriving: $κ = \inf_{x} \frac{δ (x)}{\int_{0}^{\infty} σ_{excess} (ϕ_{t} (x)) d t}$ κ=xinf∫0∞σexcess(ϕt(x))dtδ(x)

where $σ_{excess} = σ - σ_{s s}$ σexcess=σ−σss is the excess entropy production rate above the system’s steady-state baseline — thermodynamic entropy for physical systems, metabolic entropy for biological systems, free energy dissipation for cognitive systems, and social entropy production for social systems.

We proposed a unified benchmark: the attractor is the state of minimum entropy generation for that class of system — zero for equilibrium systems, non-zero steady-state for dissipative systems. This provides a unified criterion for identifying attractors across domains: an attractor is a state from which the system cannot reduce its entropy generation further without losing its defining structure or function.

This unifies physical, biological, cognitive, and social systems. In each domain, persistence requires reconfiguration; reconfiguration generates excess entropy; $κ$ κ measures the entropy cost of that reconfiguration. The framework is grounded in the second law of thermodynamics and non-equilibrium steady-state thermodynamics, not analogy.

Social Application: The framework provides a thermodynamic interpretation of social dynamics: harmony is a low-entropy attractor state; turbulence is a high-entropy state generated by excess dissipation during reconfiguration. The recovery rate $κ$ κ measures how efficiently a society transitions from turbulence back to harmony — that is, how quickly it reduces its excess entropy production to zero.

11. Limitations

This paper establishes an abstract persistence cost framework with a proposed thermodynamic realization. Several limitations should be explicitly acknowledged:

Uniqueness. Entropy production is not proved to be the unique persistence cost. Many positive functionals $C (x)$ C(x) satisfy $\nabla D \cdot f = - C$ ∇D⋅f=−C. The identification of entropy production as the canonical cost is a physically motivated hypothesis, not a mathematical theorem.
Scope. The framework does not imply that all domains obey thermodynamics literally. The cognitive and social realizations are proposed hypotheses requiring empirical validation.
Decay assumption. Exponential decay of $σ_{excess}$ σexcess is a sufficient assumption to ensure finiteness of $D_{\infty}$ D∞, not a necessary one. Generalization to $L^{1}$ L1 integrable decays (e.g., algebraic) is a priority for future work.
Basin depth. Basin depth $B = D_{\infty} (saddle)$ B=D∞(saddle) is defined in terms of the persistence cost functional. Its relationship to classical energy barriers is established only for gradient systems.
Empirical validation. The predictions of the framework — particularly the inverse relationship between $κ$ κ and $D_{\infty}$ D∞ — remain to be tested empirically across domains.
Low-energy attractor benchmark. The benchmark proposed in §6.3 is a hypothesis, not a derived theorem. For cognitive systems, it risks conflating thermodynamic entropy production with free-energy minimization — distinct principles whose relationship remains open.

References

Aubin, J. P. (1991). Viability Theory. Birkhäuser.

Boltzmann, L. (1877). “Über die Beziehung zwischen dem zweiten Hauptsatz der mechanischen Wärmetheorie und der Wahrscheinlichkeitsrechnung.” Wiener Berichte, 76, 373-435.

Clausius, R. (1865). “Über verschiedene für die Anwendung bequeme Formen der Hauptgleichungen der mechanischen Wärmetheorie.” Annalen der Physik, 125(7), 353-400.

Freeman, R. A., & Kokotovic, P. V. (1996). Robust Nonlinear Control Design: State-Space and Lyapunov Techniques. Birkhäuser.

Freidlin, M. I., & Wentzell, A. D. (2012). Random Perturbations of Dynamical Systems (3rd ed.). Springer.

Friston, K. (2010). “The free-energy principle: a unified brain theory?” Nature Reviews Neuroscience, 11(2), 127-138.

Galida, R. (2026a). “The Persistence Functional: A Candidate Formal Foundation for the Attractor Framework.” Fantasy Attractor.

Galida, R. (2026b). “Deriving Corrective Permeability from the Cumulative Deviation Functional.” Fantasy Attractor.

Jaynes, E. T. (1957). “Information Theory and Statistical Mechanics.” Physical Review, 106(4), 620-630.

Khalil, H. K. (2002). Nonlinear Systems (3rd ed.). Prentice Hall.

Kondepudi, D., & Prigogine, I. (1998). Modern Thermodynamics: From Heat Engines to Dissipative Structures. Wiley.

Lohmiller, W., & Slotine, J. J. E. (1998). “On contraction analysis for non-linear systems.” Automatica, 34(6), 683-696.

Lyapunov, A. M. (1892). The General Problem of the Stability of Motion.

McEwen, B. S. (1998). “Stress, Adaptation, and Disease: Allostasis and Allostatic Load.” Annals of the New York Academy of Sciences, 840(1), 33-44.

Nicolis, G., & Prigogine, I. (1989). Exploring Complexity: An Introduction. W. H. Freeman.

Parrondo, J. M. R., Horowitz, J. M., & Sagawa, T. (2015). “Thermodynamics of information.” Nature Physics, 11(2), 131-139.

Prigogine, I. (1947). Étude Thermodynamique des Phénomènes Irréversibles. Dunod.

Prigogine, I., & Nicolis, G. (1977). Self-Organization in Non-Equilibrium Systems. Wiley.

Sagawa, T., & Ueda, M. (2008). “Second law of thermodynamics with discrete quantum feedback control.” Physical Review Letters, 100(8), 080403.

Seifert, U. (2012). “Stochastic thermodynamics, fluctuation theorems and molecular machines.” Reports on Progress in Physics, 75(12), 126001.

Sekimoto, K. (2010). Stochastic Energetics. Springer.

Shannon, C. E. (1948). “A Mathematical Theory of Communication.” Bell System Technical Journal, 27(3), 379-423.

Sterling, P., & Eyer, J. (1988). “Allostasis: A New Paradigm to Explain Arousal Pathology.” In Handbook of Life Stress, Cognition and Health, 629-649.

Suggested citation: Galida, R. S. (2026). Excess Entropy Production as a Candidate Universal Cost of Persistence: A Thermodynamic Foundation for the Attractor Framework. Fantasy Attractor.

Deriving Corrective Permeability from the Cumulative Deviation Functional; Robert Galida (June 2026) [F]

June 27, 2026 5:13 pm / Leave a comment

Abstract

The attractor framework defines $κ$ κ (corrective permeability) as the rate at which a system returns to its attractor after perturbation. Historically, $κ$ κ has been treated as an empirical parameter — fitted to data rather than derived from first principles. This paper derives $κ$ κ from the framework’s foundational object: the cumulative deviation functional $D_{T} (x) = \int_{0}^{T} δ (ϕ_{t} (x)) d t$ DT(x)=∫0Tδ(ϕt(x))dt, where $δ (x) = d (x, A)$ δ(x)=d(x,A).

We define: $κ = \inf_{x \in B ∖ A} \frac{δ (x)}{D_{\infty} (x)}$ κ=x∈B∖AinfD∞(x)δ(x)

We prove that for linear systems $\dot{x} = - A x$ x˙=−Ax with $A$ A symmetric positive definite, this definition recovers the slowest eigenvalue $λ_{\min} (A)$ λmin(A) — the conventional notion of corrective permeability. We establish a sharp universal persistence bound $D_{\infty} (x) \leq δ (x) / κ$ D∞(x)≤δ(x)/κ, show homogeneity and scale invariance of the variational ratio, and demonstrate consistency with Koopman spectral theory and resolvent poles for finite-dimensional linear systems. A comparison theorem links $κ$ κ to classical exponential stability constants. A Hamilton-Jacobi-type transport equation for $D_{\infty}$ D∞ is derived. A finite-horizon estimator $κ_{T} = \inf_{x} \frac{δ (x)}{D_{T} (x)}$ κT=infxDT(x)δ(x) is provided with exponential convergence under explicit assumptions.

The derivation is rigorous for linear systems and testable. Open questions for nonlinear, multiscale, and stochastic systems are identified.

Keywords: corrective permeability, cumulative deviation functional, attractor framework, Koopman operator, trajectory functional

1. Introduction

The attractor framework has been applied across physics, biology, cognition, and social systems. Its central variable — corrective permeability $κ$ κ — measures the rate at which a system returns to its attractor after perturbation. Historically, $κ$ κ has been defined empirically as $κ = 1 / τ$ κ=1/τ, where $τ$ τ is a measured recovery time constant.

This paper derives $κ$ κ from a single foundational object: the cumulative deviation functional $D_{T} (x)$ DT(x). Within the present framework, $κ$ κ is defined variationally rather than introduced as an empirical fitting parameter. We show that $κ$ κ is a consequence of the trajectory geometry — specifically, the ratio of initial distance to total cumulative deviation.

The derivation is rigorous for linear systems, connects to established theory (Koopman operators, resolvent poles), and provides a finite-horizon estimator for empirical use. Open questions for nonlinear and stochastic systems are identified.

2. The Cumulative Deviation Functional

Let $X$ X be a metric space with distance function $∥ \cdot ∥$ ∥⋅∥. Let $ϕ_{t} (x)$ ϕt(x) be the flow of a dynamical system starting from state $x \in X$ x∈X at time $t = 0$ t=0. Let $A \subseteq X$ A⊆X be an attractor set (a compact, invariant set to which trajectories converge). Let $B$ B be the basin of attraction of $A$ A.

Define the distance from a point to the attractor: $δ (x) = d (x, A) = \inf_{a \in A} ∥ x - a ∥$ δ(x)=d(x,A)=a∈Ainf∥x−a∥

Definition 1 (Cumulative Deviation Functional): For a finite horizon $T > 0$ T>0, define: $D_{T} (x) = \int_{0}^{T} δ (ϕ_{t} (x)) d t$ DT(x)=∫0Tδ(ϕt(x))dt

For $T \to \infty$ T→∞, define: $D_{\infty} (x) = \int_{0}^{\infty} δ (ϕ_{t} (x)) d t$ D∞(x)=∫0∞δ(ϕt(x))dt

Proposition 1 (Finiteness of D∞D∞): Assume there exist constants $C < \infty$ C<∞ and $μ > 0$ μ>0 such that: $δ (ϕ_{t} (x)) \leq C e^{- μ t} δ (x)$ δ(ϕt(x))≤Ce−μtδ(x)

for all $x \in B$ x∈B. Then $D_{\infty} (x) < \infty$ D∞(x)<∞ for every $x \in B$ x∈B.

Proof: $D_{\infty} (x) = \int_{0}^{\infty} δ (ϕ_{t} (x)) d t \leq \int_{0}^{\infty} C e^{- μ t} δ (x) d t = \frac{C}{μ} δ (x) < \infty$ D∞(x)=∫0∞δ(ϕt(x))dt≤∫0∞Ce−μtδ(x)dt=μCδ(x)<∞ $□$ □

Properties (from Galida, 2026a):

Property	Statement
Non-negativity	$D_{T} (x) \geq 0$ DT(x)≥0
Monotonicity	$D_{T_{2}} (x) \geq D_{T_{1}} (x)$ DT2(x)≥DT1(x) for $T_{2} \geq T_{1}$ T2≥T1
Additivity	$D_{T + S} (x) = D_{T} (x) + D_{S} (ϕ_{T} (x))$ DT+S(x)=DT(x)+DS(ϕT(x))
Instantaneous growth	$\frac{d}{d T} D_{T} (x) = δ (ϕ_{T} (x))$ dTdDT(x)=δ(ϕT(x))
Occupation measure	$D_{T} (x) = \int δ (y) d μ_{T} (y)$ DT(x)=∫δ(y)dμT(y), where $μ_{T}$ μT is the occupation measure

3. Derivation of Corrective Permeability ( $κ$ κ)

3.1 Variational Definition

Definition 2 (Corrective Permeability): $κ = \inf_{x \in B ∖ A} \frac{δ (x)}{D_{\infty} (x)}$ κ=x∈B∖AinfD∞(x)δ(x)

Interpretation: $κ$ κ is the effective recovery rate — the smallest ratio of initial distance to total cumulative deviation. It serves as a global measure of the slowest recovery mode in the basin.

Remark on κκ: The definition allows $κ = 0$ κ=0 if $D_{\infty} (x)$ D∞(x) diverges or if the ratio $δ (x) / D_{\infty} (x)$ δ(x)/D∞(x) can be made arbitrarily small. Throughout the remainder of this paper, we assume hypotheses (such as the exponential stability in Proposition 1) that guarantee $κ > 0$ κ>0.

Remark on attainment: The infimum in the definition of $κ$ κ need not be attained; minimizing sequences may exist without a minimizing state. For linear systems, the infimum is attained on the slow eigenspace.

3.2 Homogeneity and Scale Invariance

Theorem 1 (Homogeneity and Scale Invariance): Suppose the flow satisfies $ϕ_{t} (α x) = α ϕ_{t} (x)$ ϕt(αx)=αϕt(x) for all $t$ t and all $α > 0$ α>0, and the distance function satisfies $δ (α x) = α δ (x)$ δ(αx)=αδ(x). Then: $\frac{δ (α x)}{D_{\infty} (α x)} = \frac{δ (x)}{D_{\infty} (x)}$ D∞(αx)δ(αx)=D∞(x)δ(x)

Proof: $D_{\infty} (α x) = \int_{0}^{\infty} δ (ϕ_{t} (α x)) d t = \int_{0}^{\infty} δ (α ϕ_{t} (x)) d t = α \int_{0}^{\infty} δ (ϕ_{t} (x)) d t = α D_{\infty} (x)$ D∞(αx)=∫0∞δ(ϕt(αx))dt=∫0∞δ(αϕt(x))dt=α∫0∞δ(ϕt(x))dt=αD∞(x)

Corollary: For linear systems, the infimum over all $x \neq 0$ x=0 reduces to an infimum over the unit sphere: $κ = \inf_{∥ x ∥ = 1} \frac{δ (x)}{D_{\infty} (x)}$ κ=∥x∥=1infD∞(x)δ(x)

3.3 Sharp Universal Persistence Bound

Theorem 2 (Sharp Universal Persistence Bound): For any $x \in B ∖ A$ x∈B∖A: $D_{\infty} (x) \leq \frac{δ (x)}{κ}$ D∞(x)≤κδ(x)

Moreover, the constant $1 / κ$ 1/κ is optimal: it is the smallest constant such that this inequality holds for all $x$ x in the basin.

Proof: By definition of $κ$ κ as the infimum of $δ (x) / D_{\infty} (x)$ δ(x)/D∞(x), we have $δ (x) / D_{\infty} (x) \geq κ$ δ(x)/D∞(x)≥κ for all $x$ x. Rearranging gives: $D_{\infty} (x) \leq \frac{δ (x)}{κ}$ D∞(x)≤κδ(x)

Optimality follows from Theorem 3: for the slow eigenvector $v_{1}$ v1, $D_{\infty} (v_{1}) = δ (v_{1}) / κ$ D∞(v1)=δ(v1)/κ, so no smaller constant can work. $□$ □

3.4 Consistency with Linear Systems

Consider a linear system $\dot{x} = - A x$ x˙=−Ax, with $A$ A symmetric positive definite. Let its eigenvalues be $0 < λ_{1} \leq λ_{2} \leq \dots \leq λ_{n}$ 0<λ1≤λ2≤⋯≤λn, with corresponding orthonormal eigenvectors $v_{1}, v_{2}, \dots, v_{n}$ v1,v2,…,vn.

The flow is $ϕ_{t} (x) = e^{- A t} x$ ϕt(x)=e−Atx. The attractor is $A = {0}$ A={0}, and the distance to the attractor is $δ (x) = ∥ x ∥$ δ(x)=∥x∥.

Theorem 3 (Linear Consistency): For $\dot{x} = - A x$ x˙=−Ax with $A$ A symmetric positive definite, $\inf_{x \neq 0} \frac{∥ x ∥}{D_{\infty} (x)} = λ_{\min} (A)$ x=0infD∞(x)∥x∥=λmin(A)

Proof:

Since $A$ A is symmetric positive definite, $e^{- A t}$ e−At is symmetric positive definite with eigenvalues $e^{- λ_{i} t}$ e−λit. Hence its operator norm is $∥ e^{- A t} ∥ = e^{- λ_{1} t}$ ∥e−At∥=e−λ1t. For any $x \neq 0$ x=0: $D_{\infty} (x) = \int_{0}^{\infty} ∥ e^{- A t} x ∥ d t \leq \int_{0}^{\infty} ∥ x ∥ e^{- λ_{1} t} d t = \frac{∥ x ∥}{λ_{1}}$ D∞(x)=∫0∞∥e−Atx∥dt≤∫0∞∥x∥e−λ1tdt=λ1∥x∥

Therefore: $\frac{∥ x ∥}{D_{\infty} (x)} \geq λ_{1}$ D∞(x)∥x∥≥λ1

To show equality is achieved, take $x = v_{1}$ x=v1 (the eigenvector corresponding to $λ_{1}$ λ1). Then: $∥ e^{- A t} v_{1} ∥ = ∥ v_{1} ∥ e^{- λ_{1} t}$ ∥e−Atv1∥=∥v1∥e−λ1t

and: $D_{\infty} (v_{1}) = \int_{0}^{\infty} ∥ v_{1} ∥ e^{- λ_{1} t} d t = \frac{∥ v_{1} ∥}{λ_{1}}$ D∞(v1)=∫0∞∥v1∥e−λ1tdt=λ1∥v1∥

Thus: $\frac{∥ v_{1} ∥}{D_{\infty} (v_{1})} = λ_{1}$ D∞(v1)∥v1∥=λ1

Hence: $\inf_{x \neq 0} \frac{∥ x ∥}{D_{\infty} (x)} = λ_{1}$ x=0infD∞(x)∥x∥=λ1 $□$ □

Corollary: For linear systems, the variational definition of $κ$ κ recovers the slowest eigenvalue — the conventional notion of corrective permeability.

3.5 Transport Equation

Theorem 4 (Transport Equation): Assume the vector field $f$ f is $C^{1}$ C1, the flow $ϕ_{t}$ ϕt is $C^{1}$ C1, and $D_{\infty}$ D∞ is continuously differentiable on $B ∖ A$ B∖A. Then: $\nabla D_{\infty} (x) \cdot f (x) = - δ (x)$ ∇D∞(x)⋅f(x)=−δ(x)

Proof: From the definition: $D_{\infty} (ϕ_{s} (x)) = D_{\infty} (x) - D_{s} (x)$ D∞(ϕs(x))=D∞(x)−Ds(x)

Differentiating with respect to $s$ s at $s = 0$ s=0: $\frac{d}{d s} D_{\infty} (ϕ_{s} (x)) ∣_{s = 0} = - δ (x)$ dsdD∞(ϕs(x))s=0=−δ(x)

By the chain rule: $\nabla D_{\infty} (x) \cdot f (x) = - δ (x)$ ∇D∞(x)⋅f(x)=−δ(x) $□$ □

Interpretation: This is a first-order transport equation, $f \cdot \nabla D = - δ$ f⋅∇D=−δ, which belongs to the broader Hamilton-Jacobi family but lacks a Hamiltonian in the usual sense. It may serve as a foundation for numerical computation and further theoretical development.

3.6 Local vs. Global Interpretation

The variational definition $κ = \inf_{x} \frac{δ (x)}{D_{\infty} (x)}$ κ=infxD∞(x)δ(x) is global — it is the slowest recovery rate over the entire basin. This is not necessarily the same as the local recovery rate near the attractor (the slowest eigenvalue of the linearization). For linear systems, they coincide. For nonlinear systems, they may differ if transient excursions produce slower effective recovery than the local linearization predicts.

This distinction is important: $κ$ κ is a global invariant of the basin, not merely a local property of the attractor. The relationship between the global $κ$ κ and the local Lyapunov exponent is an open question (see §6).

3.7 Non-Symmetric Linear Systems

For a general linear system $\dot{x} = A x$ x˙=Ax (where $A$ A is stable, i.e., all eigenvalues have negative real parts), the same principle holds in the diagonalizable case. The slowest mode corresponds to the eigenvalue with the largest real part (closest to zero).

Conjecture: An analogous result holds for non-normal linear systems under additional assumptions on the semigroup, such as a uniformly exponentially stable semigroup satisfying suitable norm bounds. This remains an open question.

3.8 Comparison with Exponential Stability

Theorem 5 (Comparison with Exponential Stability): Suppose the system satisfies the exponential stability bound: $δ (ϕ_{t} (x)) \leq C e^{- μ t} δ (x)$ δ(ϕt(x))≤Ce−μtδ(x)

for all $x \in B$ x∈B, with constants $C < \infty$ C<∞ and $μ > 0$ μ>0. Then: $κ \geq \frac{μ}{C}$ κ≥Cμ

Proof: From the stability bound: $D_{\infty} (x) = \int_{0}^{\infty} δ (ϕ_{t} (x)) d t \leq \int_{0}^{\infty} C e^{- μ t} δ (x) d t = \frac{C}{μ} δ (x)$ D∞(x)=∫0∞δ(ϕt(x))dt≤∫0∞Ce−μtδ(x)dt=μCδ(x)

Therefore: $\frac{δ (x)}{D_{\infty} (x)} \geq \frac{μ}{C}$ D∞(x)δ(x)≥Cμ

Taking the infimum over $x$ x: $κ = \inf_{x} \frac{δ (x)}{D_{\infty} (x)} \geq \frac{μ}{C}$ κ=xinfD∞(x)δ(x)≥Cμ $□$ □

Interpretation: The variational constant $κ$ κ is bounded below by the exponential stability constant $μ / C$ μ/C.

4. Connections to Existing Theory

4.1 Koopman Operator

The Koopman operator $K^{t}$ Kt acts on observables as: $(K^{t} f) (x) = f (ϕ_{t} (x))$ (Ktf)(x)=f(ϕt(x))

For linear systems $\dot{x} = - A x$ x˙=−Ax, the Koopman eigenvalues are $e^{- λ_{i} t}$ e−λit. The dominant nontrivial eigenvalue (largest less than 1) is $e^{- λ_{1} t}$ e−λ1t, corresponding to the slowest decay rate.

For finite-dimensional linear systems, $ρ = e^{- λ_{\min} t}$ ρ=e−λmint, and therefore: $- \frac{1}{t} \log ρ = λ_{\min} = κ$ −t1logρ=λmin=κ

Thus, under the hypotheses of Theorem 3, the variational constant equals the exponential decay rate associated with the dominant Koopman eigenvalue.

4.2 Resolvent Poles

For finite-dimensional stable linear systems, the resolvent $(s I + A)^{- 1}$ (sI+A)−1 has poles at $s = - λ_{i}$ s=−λi. The pole closest to the imaginary axis is $s = - λ_{1}$ s=−λ1.

Since Theorem 3 identifies $κ = λ_{\min}$ κ=λmin, and the resolvent poles are $s_{i} = - λ_{i}$ si=−λi, we obtain: $κ = \min_{i} ∣ ℜ (s_{i}) ∣$ κ=imin∣ℜ(si)∣

for finite-dimensional linear systems.

5. Finite-Horizon Estimation

In practice, we can only measure finite trajectories. Define the finite-horizon estimator: $κ_{T} = \inf_{x \in K} \frac{δ (x)}{D_{T} (x)}$ κT=x∈KinfDT(x)δ(x)

where $K \subset B$ K⊂B is compact and $K \cap A = \emptyset$ K∩A=∅.

Proposition 2 (Finite-Horizon Estimation): Assume:

The flow $ϕ_{t} (x)$ ϕt(x) is jointly continuous in $(t, x)$ (t,x).
$δ (x)$ δ(x) is continuous.
The exponential stability bound $δ (ϕ_{t} (x)) \leq C e^{- μ t} δ (x)$ δ(ϕt(x))≤Ce−μtδ(x) holds uniformly for all $x \in K$ x∈K, with $μ > 0$ μ>0.

Then the variational constant $κ$ κ (from Definition 2) satisfies $κ \geq μ / C$ κ≥μ/C by Theorem 5, and: $κ_{T} \to κ as T \to \infty$ κT→κas T→∞

with error: $∣ κ_{T} - κ ∣ = O (e^{- μ T})$ ∣κT−κ∣=O(e−μT)

Proof: For any $x \in K$ x∈K, the tail bound gives: $∣ D_{\infty} (x) - D_{T} (x) ∣ = \int_{T}^{\infty} δ (ϕ_{t} (x)) d t \leq \frac{C e^{- μ T} δ (x)}{μ}$ ∣D∞(x)−DT(x)∣=∫T∞δ(ϕt(x))dt≤μCe−μTδ(x)

Since $δ (x)$ δ(x) is bounded on the compact set $K$ K, let $M = \sup_{x \in K} δ (x) < \infty$ M=supx∈Kδ(x)<∞. Then: $∣ D_{\infty} (x) - D_{T} (x) ∣ \leq \frac{C M e^{- μ T}}{μ}$ ∣D∞(x)−DT(x)∣≤μCMe−μT

The right-hand side is independent of $x$ x and tends to zero as $T \to \infty$ T→∞. Hence $D_{T} \to D_{\infty}$ DT→D∞ uniformly on $K$ K.

Moreover, since $K$ K is compact and $K \cap A = \emptyset$ K∩A=∅, continuity of $δ$ δ gives $\inf_{x \in K} δ (x) > 0$ infx∈Kδ(x)>0. Since $D_{T} (x)$ DT(x) is continuous (by assumptions 1–2) and monotonically non-decreasing in $T$ T (from §2), for any fixed finite $T_{0} > 0$ T0>0, $D_{\infty} (x) \geq D_{T_{0}} (x)$ D∞(x)≥DT0(x), and $D_{T_{0}}$ DT0 is continuous and strictly positive on $K$ K. A continuous, strictly positive function on a compact set has a positive infimum: $m = \inf_{x \in K} D_{T_{0}} (x) > 0$ m=x∈KinfDT0(x)>0

Thus: $\inf_{x \in K} D_{\infty} (x) \geq m > 0$ x∈KinfD∞(x)≥m>0

Uniform convergence of $D_{T}$ DT to $D_{\infty}$ D∞ on $K$ K therefore implies uniform convergence of $δ (x) / D_{T} (x)$ δ(x)/DT(x) to $δ (x) / D_{\infty} (x)$ δ(x)/D∞(x). Consequently, the infima converge. $□$ □

6. Open Questions

Question	Status	Difficulty
Q1: Nonlinear systems	Does $\inf \frac{δ}{D_{\infty}}$ infD∞δ equal the local Lyapunov exponent?	Hard
Q2: Local vs. global consistency	Does $\lim_{x \to A} \frac{δ (x)}{D_{\infty} (x)} = κ$ limx→AD∞(x)δ(x)=κ hold for general nonlinear systems?	Hard
Q3: Non-normal systems	Does the infimum equal the slowest eigenvalue for non-normal $A$ A?	Moderate
Q4: Multiple timescales	Does the infimum isolate the slowest timescale?	Hard
Q5: Stochastic systems	How does noise affect the finite-horizon estimator?	Hard
Q6: Multiple attractors	How does $κ$ κ behave in basins with multiple attractors?	Moderate

7. Conclusion

This paper derives corrective permeability $κ$ κ from the cumulative deviation functional $D_{T} (x)$ DT(x). The variational definition: $κ = \inf_{x} \frac{δ (x)}{D_{\infty} (x)}$ κ=xinfD∞(x)δ(x)

is shown to recover the slowest eigenvalue for linear systems, consistent with the conventional empirical definition $κ = 1 / τ$ κ=1/τ. A sharp universal persistence bound $D_{\infty} (x) \leq δ (x) / κ$ D∞(x)≤δ(x)/κ is established. A comparison theorem links $κ$ κ to classical exponential stability constants. A Hamilton-Jacobi-type transport equation for $D_{\infty}$ D∞ is derived. Connections to Koopman theory and resolvent theory are established for finite-dimensional linear systems. A finite-horizon estimator $κ_{T}$ κT is provided with exponential convergence under explicit assumptions.

Key contribution: Within the present framework, $κ$ κ is defined variationally rather than introduced as an empirical fitting parameter — at least for the class of systems analyzed here.

Next steps: Extend the derivation to nonlinear systems (Q1–Q2), non-normal systems (Q3), multiple timescales (Q4), and stochastic dynamics (Q5).

References

Crandall, M. G., Ishii, H., & Lions, P. L. (1992). “User’s Guide to Viscosity Solutions of Second Order Partial Differential Equations.” Bulletin of the American Mathematical Society, 27(1), 1-67.

Evans, L. C. (2010). Partial Differential Equations. American Mathematical Society.

Galida, R. (2026a). “The Persistence Functional: A Candidate Formal Foundation for the Attractor Framework.” Fantasy Attractor.

Hale, J. K. (1988). Asymptotic Behavior of Dissipative Systems. American Mathematical Society.

Hirsch, M. W., Smale, S., & Devaney, R. L. (2004). Differential Equations, Dynamical Systems, and an Introduction to Chaos (2nd ed.). Elsevier Academic Press.

Khalil, H. K. (2002). Nonlinear Systems (3rd ed.). Prentice Hall.

Koopman, B. O. (1931). “Hamiltonian Systems and Transformations in Hilbert Space.” Proceedings of the National Academy of Sciences, 17(5), 315-318.

Lyapunov, A. M. (1892). The General Problem of the Stability of Motion. (English translation: 1992, Taylor & Francis).

Mezić, I. (2005). “Spectral Properties of Dynamical Systems, Model Reduction and Decompositions.” Nonlinear Dynamics, 41(1-3), 309-325.

Pazy, A. (1983). Semigroups of Linear Operators and Applications to Partial Differential Equations. Springer.

Vidyasagar, M. (1993). Nonlinear Systems Analysis (2nd ed.). Prentice Hall.

Suggested citation: Galida, R. S. (2026). Deriving Corrective Permeability from the Cumulative Deviation Functional. Fantasy Attractor.

The Persistence Functional: A Candidate Formal Foundation for the Attractor Framework; Robert Galida (July 2026) [F]

June 27, 2026 2:06 pm / Leave a comment

Abstract

The attractor framework provides a domain-general vocabulary for describing persistence and change across physical, biological, cognitive, and social systems. However, its core variables— $κ$ κ (corrective permeability), $B$ B (basin depth), and $R$ R (reality alignment)—have been defined inconsistently across application papers, and their formal relationships have remained implicit. This paper proposes a candidate mathematical formalization for the framework.

The central mathematical innovation of this paper is treating persistence as a functional defined over trajectories— $D_{T} (x) = \int_{0}^{T} d (ϕ_{τ} (x), A) d τ$ DT(x)=∫0Td(ϕτ(x),A)dτ—rather than as a scalar property of states. We prove several mathematical properties of $D_{T}$ DT, including non-negativity, monotonicity in $T$ T, additivity, Lipschitz continuity with respect to initial conditions, and a bound relating $D_{\infty}$ D∞ to the recovery rate $κ$ κ: $D_{\infty} (x) \leq \frac{C}{κ} d (x, A)$ D∞(x)≤κCd(x,A). We establish connections to dynamic programming and ergodic theory via occupation measures. We introduce a complementary topological persistence functional $P_{topo} (t)$ Ptopo(t), which measures the lifetime of topological features in the trajectory’s state-space geometry, and the topological evolution rate $E (t)$ E(t).

We unify the framework’s variable set: $κ$ κ is the recovery rate (operationalized as $1 / τ$ 1/τ); $γ$ γ is a proposed drift rate for persistent chaos, grounded in the literature on high-dimensional neural networks; $B$ B is the energy barrier (basin depth); $\tilde{B}$ B~ is a complementary persistence depth; $R$ R is the expected log predictive likelihood. We propose testable predictions linking $E (t)$ E(t) to $κ$ κ and $γ$ γ, and provide a falsifiable experimental protocol using neural network training and persistent homology.

The paper offers a candidate formal foundation, with explicit definitions, mathematical properties, and empirical grounding. All unverified sources are clearly labeled as such.

Keywords: attractor framework, persistence functional, cumulative deviation, topological persistence, corrective permeability, basin depth, reality alignment, persistent homology

1. Introduction

The attractor framework has been applied across physics (hydrogen decay, Jeans instability), biology (ECM mechanics, HRV), cognition (belief updating, performance attractors), and social systems (religious attractors, civilizational dynamics). A common vocabulary has emerged: $κ$ κ (corrective permeability), $B$ B (basin depth), and $R$ R (reality alignment). However, these variables have been defined inconsistently across papers, and their formal relationships have remained implicit. This paper proposes a candidate mathematical formalization that addresses these inconsistencies.

The central mathematical innovation of this paper is treating persistence as a functional defined over trajectories rather than as a scalar property of states. $D_{T} (x) = \int_{0}^{T} d (ϕ_{τ} (x), A) d τ$ DT(x)=∫0Td(ϕτ(x),A)dτ can be understood as a type of action functional (carefully qualified). Like the classical action $\int L (q, \dot{q}) d t$ ∫L(q,q˙)dt, it assigns a scalar to an entire trajectory, is additive under concatenation, and suggests variational and optimal-control interpretations. However, it is not the mechanical action; it is a cumulative deviation functional that measures time away from equilibrium. This moves the framework into the domain of trajectory-level analysis, aligning it with modern dynamical systems and geometric control theory.

We introduce the cumulative deviation functional $D_{T} (x)$ DT(x) as this central object, and we establish its mathematical properties, including its relationship to the recovery rate $κ$ κ. We introduce a complementary topological persistence functional $P_{topo} (t)$ Ptopo(t) and the topological evolution rate $E (t)$ E(t). We unify the framework’s variable set with operational definitions and propose testable predictions with falsification criteria.

1.1 Scope and Status

This paper is a candidate formalization—it provides definitions, mathematical properties, and empirical hypotheses. It is not a completed empirical validation; that is the subject of future work. All claims are labeled as definitions (part of the formal structure), propositions/theorems (proved), hypotheses (testable predictions), or heuristics (suggestive connections not yet formalized). This distinction is maintained throughout.

2. Formal Definitions

Let $X$ X be a metric space with distance function $∥ \cdot ∥$ ∥⋅∥. Let $ϕ_{τ} (x)$ ϕτ(x) be the flow of a dynamical system starting from state $x \in X$ x∈X at time $τ = 0$ τ=0. Let $A \subseteq X$ A⊆X be an attractor set (a compact, invariant set to which trajectories converge). Assume the flow is continuous and measurable so that $d (ϕ_{τ} (x), A)$ d(ϕτ(x),A) is measurable. The flow $ϕ_{τ}$ ϕτ satisfies the semigroup property $ϕ_{t + s} = ϕ_{t} \circ ϕ_{s}$ ϕt+s=ϕt∘ϕs for all $t, s \geq 0$ t,s≥0, with $ϕ_{0} = id$ ϕ0=id. We assume $d (ϕ_{τ} (x), A) \in L^{1} ([0, T])$ d(ϕτ(x),A)∈L1([0,T]) for all finite $T$ T, so the integral defining $D_{T}$ DT is well-defined.

Define the distance from a point to the attractor: $d (x, A) = \inf_{a \in A} ∥ x - a ∥$ d(x,A)=a∈Ainf∥x−a∥

The definition applies to any metric space; for infinite-dimensional spaces, the usual measurability and integrability conditions are assumed.

2.1 Cumulative Deviation Functional

Definition 1 (Cumulative Deviation Functional): For a finite horizon $T > 0$ T>0, the cumulative deviation functional is: $D_{T} (x) = \int_{0}^{T} d (ϕ_{τ} (x), A) d τ$ DT(x)=∫0Td(ϕτ(x),A)dτ

Interpretation: $D_{T} (x)$ DT(x) is the total accumulated deviation from the attractor over the interval $[0, T]$ [0,T]. It measures integrated error, residence-time-weighted distance, or accumulated regret. This is not a path length; it measures time spent away from equilibrium, whereas path length $\int ∥ {\dot{ϕ}}_{τ} (x) ∥ d τ$ ∫∥ϕ˙τ(x)∥dτ measures distance traveled.

Domain generality: This definition applies to any system with a well-defined state space, a flow, and an attractor set. It does not require linearity, differentiability, or specific functional forms.

Empirical note: $D_{T}$ DT is the fundamental object for empirical work; $D_{\infty}$ D∞ is primarily an analytical limit used for theoretical bounds.

Note: $D_{T}$ DT is not a Lyapunov function. A Lyapunov function is a scalar function of the current state; $D_{T}$ DT is a functional of the entire trajectory. It does not decrease monotonically along trajectories, and it does not provide pointwise stability information. Its purpose is to measure accumulated history, not instantaneous energy.

Occupation measure connection: Define the occupation measure of the trajectory up to time $T$ T as: $μ_{T} (B) = \int_{0}^{T} 1_{B} (ϕ_{τ} (x)) d τ$ μT(B)=∫0T1B(ϕτ(x))dτ

for measurable $B \subseteq X$ B⊆X. Then: $D_{T} (x) = \int_{X} d (y, A) d μ_{T} (y)$ DT(x)=∫Xd(y,A)dμT(y)

Thus $D_{T}$ DT is the expected distance to the attractor under the occupation measure. This connects the functional directly to ergodic theory and occupation measure analysis. For foundational treatments of occupation measures and invariant measures, see Ruelle (1989) and Bowen (1975).

2.1.1 Why the L¹ Trajectory Functional?

The choice of the L¹ integral over alternatives is motivated by the following properties:

Linearity: Each moment contributes equally; accumulation is additive over time.
Physical units: For systems with a natural distance metric, $D_{T}$ DT has units of distance × time, which is interpretable as accumulated deviation.
Simplicity: It is the simplest nontrivial trajectory functional that is not a path length.
Analogy: It mirrors cumulative regret and occupation measures in control theory and ergodic theory.
Avoidance of overweighting: Unlike $d^{2}$ d2, it does not disproportionately weight large deviations; unlike max, it is sensitive to the full trajectory.

This is one natural choice; other functionals (e.g., $d^{p}$ dp, exponentially weighted integrals) could be substituted without changing the framework’s structure.

2.2 Topological Persistence Functional

Let $X_{τ} = {ϕ_{s} (x) : s \in [0, τ]}$ Xτ={ϕs(x):s∈[0,τ]} be the trajectory segment up to time $τ$ τ. Let ${PH}_{k} (X_{τ})$ PHk(Xτ) be the $k$ k-dimensional persistent homology of the point cloud $X_{τ}$ Xτ at scale $ϵ$ ϵ. Each feature (component, loop, void) has a birth scale $b$ b and a death scale $d$ d, with persistence $d - b$ d−b. For foundational treatments of persistent homology, see Edelsbrunner & Harer (2010) or Carlsson (2009).

Definition 2 (Topological Persistence Functional): We define the following complementary topological persistence functional. For $t \geq 0$ t≥0: $P_{topo} (t) = \int_{0}^{t} \sum_{k \geq 0} \sum_{(b, d) \in {PH}_{k} (X_{τ})} (d - b) d τ$ Ptopo(t)=∫0tk≥0∑(b,d)∈PHk(Xτ)∑(d−b)dτ

The map $τ \mapsto {PH}_{k} (X_{τ})$ τ↦PHk(Xτ) is piecewise constant on intervals where the trajectory does not cross a homology-critical threshold. Assuming the trajectory crosses such thresholds at discrete times, the integral is well-defined as a sum of piecewise continuous segments. This is the standard assumption in time-varying persistent homology (see Carlsson & Zomorodian, 2009).

Interpretation: $P_{topo} (t)$ Ptopo(t) is the total lifetime of all topological features in the trajectory’s state-space geometry up to time $t$ t. This is a separate mathematical object from $D_{T}$ DT; the relationship between them is an empirical hypothesis. This is one possible choice among several topological summaries (e.g., persistence landscapes, persistence images) and is selected because it mirrors the cumulative interpretation of $D_{T}$ DT, rather than because it is uniquely canonical. Other stable summaries—such as persistence landscapes, persistence images, or Betti curves—could be substituted for the present functional without changing the framework’s structure.

Measurement: In practice, $P_{topo} (t)$ Ptopo(t) is computed by sampling the trajectory at discrete times, computing persistent homology on latent activation manifolds, and summing the persistence of all features using standard libraries (e.g., GUDHI, Ripser). Turner & Barak (2023) demonstrated that trained RNNs develop attractors sequentially during training; the topological structure of these attractors can be analyzed using persistent homology.

Falsification: If persistent homology features do not correlate with any behavioral or dynamical measure in a given system, $P_{topo}$ Ptopo is not a useful construct for that domain.

2.3 Topological Evolution Rate

Definition 3 (Topological Evolution Rate): For a learning system with time-dependent topological persistence, the topological evolution rate is defined as: $E (t) = \frac{d}{d t} P_{topo} (t)$ E(t)=dtdPtopo(t)

where differentiable, and experimentally as $E (t) \approx \frac{Δ P_{topo}}{Δ t}$ E(t)≈ΔtΔPtopo over finite intervals.

Interpretation: $E (t)$ E(t) measures how quickly the system’s topological complexity changes during learning. Negative $E (t)$ E(t) indicates topological simplification (compression); positive $E (t)$ E(t) indicates increasing complexity (expansion); $E (t) \approx 0$ E(t)≈0 indicates stagnation. Learning is one possible cause of topological change; random drift, noise, or chaotic wandering can also change topology.

Empirical anchor: Karuppiah, Nazreen Banu et al. (2026) examine the evolution of topological signatures during training. Turner & Barak (2023) show that RNNs develop attractors sequentially, which may correspond to phases of topological simplification. We hypothesize that successful learning corresponds to negative average values of $E (t)$ E(t) over defined phases, but this is a testable claim, not a definition.

3. Mathematical Properties of the Cumulative Deviation Functional

This section establishes the mathematical behavior of $D_{T}$ DT, providing the foundation for its use in the framework.

3.1 Non-negativity

Proposition 1 (Non-negativity): For any $x \in X$ x∈X and any $T \geq 0$ T≥0: $D_{T} (x) \geq 0$ DT(x)≥0

with equality iff $ϕ_{τ} (x) \in A$ ϕτ(x)∈A for almost all $τ \in [0, T]$ τ∈[0,T].

Proof: The integrand is a distance function $d (ϕ_{τ} (x), A)$ d(ϕτ(x),A), which is non-negative by definition. The integral of a non-negative function is non-negative. Equality holds only if the integrand is zero almost everywhere.

3.2 Monotonicity in $T$ T

Proposition 2 (Monotonicity): For fixed $x$ x, $D_{T} (x)$ DT(x) is monotonically non-decreasing in $T$ T: $D_{T_{2}} (x) \geq D_{T_{1}} (x) for T_{2} \geq T_{1}$ DT2(x)≥DT1(x)for T2≥T1

Proof: For $T_{2} \geq T_{1}$ T2≥T1: $D_{T_{2}} (x) = \int_{0}^{T_{1}} d (ϕ_{τ} (x), A) d τ + \int_{T_{1}}^{T_{2}} d (ϕ_{τ} (x), A) d τ$ DT2(x)=∫0T1d(ϕτ(x),A)dτ+∫T1T2d(ϕτ(x),A)dτ

The second integral is non-negative by Proposition 1. Therefore $D_{T_{2}} (x) \geq D_{T_{1}} (x)$ DT2(x)≥DT1(x).

Corollary: If the trajectory converges exactly to the attractor at time $τ_{0} < T$ τ0<T, then: $D_{T} (x) = D_{τ_{0}} (x) for all T \geq τ_{0}$ DT(x)=Dτ0(x)for all T≥τ0

3.3 Additivity

Proposition 3 (Additivity): For any $T, S \geq 0$ T,S≥0: $D_{T + S} (x) = D_{T} (x) + D_{S} (ϕ_{T} (x))$ DT+S(x)=DT(x)+DS(ϕT(x))

Proof: $\begin{aligned} D_{T + S} (x) & = \int_{0}^{T + S} d (ϕ_{τ} (x), A) d τ \\ = \int_{0}^{T} d (ϕ_{τ} (x), A) d τ + \int_{T}^{T + S} d (ϕ_{τ} (x), A) d τ \\ = D_{T} (x) + \int_{0}^{S} d (ϕ_{τ + T} (x), A) d τ \\ = D_{T} (x) + \int_{0}^{S} d (ϕ_{τ} (ϕ_{T} (x)), A) d τ (by the semigroup property) \\ = D_{T} (x) + D_{S} (ϕ_{T} (x)) \end{aligned}$ DT+S(x)=∫0T+Sd(ϕτ(x),A)dτ=∫0Td(ϕτ(x),A)dτ+∫TT+Sd(ϕτ(x),A)dτ=DT(x)+∫0Sd(ϕτ+T(x),A)dτ=DT(x)+∫0Sd(ϕτ(ϕT(x)),A)dτ(by the semigroup property)=DT(x)+DS(ϕT(x))

This connects $D_{T}$ DT naturally to Bellman equations, dynamic programming, and occupation measures.

3.4 Heuristic Connection: Dynamic Programming

The additivity property $D_{T + S} (x) = D_{T} (x) + D_{S} (ϕ_{T} (x))$ DT+S(x)=DT(x)+DS(ϕT(x)) suggests a natural connection to dynamic programming. For a controlled system $\dot{X} = f (X, u)$ X˙=f(X,u) with control $u \in U$ u∈U, the value function $V (x) = \inf_{u} D_{\infty} (x)$ V(x)=infuD∞(x) would formally satisfy the Hamilton-Jacobi-Bellman equation: $0 = \inf_{u} {d (x, A) + \nabla V (x) \cdot f (x, u)}$ 0=uinf{d(x,A)+∇V(x)⋅f(x,u)}

This is a standard result for additive cost functionals. A full derivation for the specific functional $D_{T}$ DT is left for future work. This section is a heuristic connection, not a formal result.

3.5 Lipschitz Continuity with Respect to Initial Conditions

Proposition 4 (Lipschitz Continuity of DTDT): Suppose the flow $ϕ_{τ}$ ϕτ is Lipschitz continuous in $x$ x with constant $L$ L, i.e., $∥ ϕ_{τ} (x) - ϕ_{τ} (y) ∥ \leq e^{L τ} ∥ x - y ∥$ ∥ϕτ(x)−ϕτ(y)∥≤eLτ∥x−y∥. Then for any $x, y$ x,y in the basin of $A$ A: $∣ D_{T} (x) - D_{T} (y) ∣ \leq \int_{0}^{T} e^{L τ} d τ ∥ x - y ∥ = \frac{e^{L T} - 1}{L} ∥ x - y ∥$ ∣DT(x)−DT(y)∣≤∫0TeLτdτ∥x−y∥=LeLT−1∥x−y∥

Proof: First, note that the distance function $d (\cdot, A)$ d(⋅,A) is 1-Lipschitz: for any $x, y \in X$ x,y∈X, $∣ d (x, A) - d (y, A) ∣ \leq ∥ x - y ∥$ ∣d(x,A)−d(y,A)∣≤∥x−y∥

This follows from the triangle inequality and the definition of the infimum. Then, using the Lipschitz property of the flow: $\begin{aligned} ∣ D_{T} (x) - D_{T} (y) ∣ & \leq \int_{0}^{T} ∣ d (ϕ_{τ} (x), A) - d (ϕ_{τ} (y), A) ∣ d τ \\ \leq \int_{0}^{T} ∥ ϕ_{τ} (x) - ϕ_{τ} (y) ∥ d τ \\ \leq \int_{0}^{T} e^{L τ} ∥ x - y ∥ d τ \\ = \frac{e^{L T} - 1}{L} ∥ x - y ∥ \end{aligned}$ ∣DT(x)−DT(y)∣≤∫0T∣d(ϕτ(x),A)−d(ϕτ(y),A)∣dτ≤∫0T∥ϕτ(x)−ϕτ(y)∥dτ≤∫0TeLτ∥x−y∥dτ=LeLT−1∥x−y∥

Interpretation: This proposition guarantees that empirical estimates of $D_{T}$ DT are robust under small perturbations of initial conditions and establishes that $D_{T}$ DT defines a continuous functional on the basin of attraction. This is essential for numerical estimation and experimental measurement.

3.6 Instantaneous Growth Rate

Remark 1 (Instantaneous Growth Rate): If the integrand $d (ϕ_{τ} (x), A)$ d(ϕτ(x),A) is continuous in $τ$ τ, then: $\frac{d}{d T} D_{T} (x) = d (ϕ_{T} (x), A)$ dTdDT(x)=d(ϕT(x),A)

This follows directly from the Fundamental Theorem of Calculus.

3.7 Ergodic Limit

Proposition 5 (Ergodic Limit): Suppose the normalized occupation measure $ν_{T} = μ_{T} / T$ νT=μT/T converges weakly to an invariant probability measure $μ$ μ as $T \to \infty$ T→∞. Then: $\lim_{T \to \infty} \frac{1}{T} D_{T} (x) = \int_{X} d (y, A) d μ (y)$ T→∞limT1DT(x)=∫Xd(y,A)dμ(y)

Proof: From the occupation measure representation $D_{T} (x) = \int d (y, A) d μ_{T} (y) = T \int d (y, A) d ν_{T} (y)$ DT(x)=∫d(y,A)dμT(y)=T∫d(y,A)dνT(y), weak convergence of $ν_{T}$ νT to $μ$ μ and boundedness/continuity of $d (\cdot, A)$ d(⋅,A) gives the result.

This is the pointwise ergodic theorem applied to the observable $d (\cdot, A)$ d(⋅,A). For the ergodic theory of dynamical systems, see Bowen (1975) and Ruelle (1989).

3.8 Bound under Exponential Stability

Theorem 2 (Bound under Exponential Stability): Suppose the flow $ϕ_{τ} (x)$ ϕτ(x) converges to the attractor $A$ A with exponential rate $κ > 0$ κ>0: $d (ϕ_{τ} (x), A) \leq C e^{- κ τ} d (x, A)$ d(ϕτ(x),A)≤Ce−κτd(x,A)

for some constant $C < \infty$ C<∞, for all $τ \geq 0$ τ≥0. Then: $D_{\infty} (x) = \int_{0}^{\infty} d (ϕ_{τ} (x), A) d τ \leq \frac{C}{κ} d (x, A)$ D∞(x)=∫0∞d(ϕτ(x),A)dτ≤κCd(x,A)

Proof: $D_{\infty} (x) = \int_{0}^{\infty} d (ϕ_{τ} (x), A) d τ \leq \int_{0}^{\infty} C e^{- κ τ} d (x, A) d τ$ D∞(x)=∫0∞d(ϕτ(x),A)dτ≤∫0∞Ce−κτd(x,A)dτ $= C d (x, A) \int_{0}^{\infty} e^{- κ τ} d τ = \frac{C}{κ} d (x, A)$ =Cd(x,A)∫0∞e−κτdτ=κCd(x,A)

Corollary: For linearly stable systems with recovery rate $κ$ κ, $D_{\infty} (x) \leq \frac{1}{κ} d (x, A)$ D∞(x)≤κ1d(x,A) (when $C = 1$ C=1).

Important: Exponential stability implies $D_{\infty} < \infty$ D∞<∞. The converse is not claimed; polynomial convergence can also yield finite $D_{\infty}$ D∞.

3.9 Recovery Rate Bound

Corollary 1 (Recovery Rate Bound): For a system satisfying the exponential stability hypothesis with constant $C$ C, the recovery rate $κ$ κ satisfies: $κ \leq \frac{C d (x, A)}{D_{\infty} (x)}$ κ≤D∞(x)Cd(x,A)

For systems with $C = 1$ C=1 (e.g., normal/symmetric linearizations with no transient overshoot), this reduces to: $κ \leq \frac{d (x, A)}{D_{\infty} (x)}$ κ≤D∞(x)d(x,A)

Proof: From Theorem 2, we have $D_{\infty} (x) \leq \frac{C}{κ} d (x, A)$ D∞(x)≤κCd(x,A). Rearranging gives $κ \leq \frac{C d (x, A)}{D_{\infty} (x)}$ κ≤D∞(x)Cd(x,A). When $C = 1$ C=1, this reduces to $κ \leq \frac{d (x, A)}{D_{\infty} (x)}$ κ≤D∞(x)d(x,A).

Interpretation: Small cumulative deviation implies rapid recovery (large $κ$ κ). Large cumulative deviation implies slow recovery (small $κ$ κ). This formalizes the intuitive link between $D_{T}$ DT and $κ$ κ. The $C$ C factor accounts for possible transient overshoot in non-normal systems.

3.10 Finite Horizon Approximation

Proposition 6 (Finite Horizon): For any $ϵ > 0$ ϵ>0, there exists a finite $T_{ϵ}$ Tϵ such that for all $T > T_{ϵ}$ T>Tϵ: $∣ D_{T} (x) - D_{\infty} (x) ∣ \leq ϵ$ ∣DT(x)−D∞(x)∣≤ϵ

Proof: This follows directly from Theorem 2 under the exponential stability hypothesis. Since the integrand decays exponentially, the tail integral $\int_{T}^{\infty} d (ϕ_{τ} (x), A) d τ$ ∫T∞d(ϕτ(x),A)dτ can be made arbitrarily small by choosing $T$ T sufficiently large.

3.11 Summary of Properties

Property	Statement
Non-negativity	$D_{T} (x) \geq 0$ DT(x)≥0
Monotonicity	$D_{T_{2}} (x) \geq D_{T_{1}} (x)$ DT2(x)≥DT1(x) for $T_{2} \geq T_{1}$ T2≥T1
Additivity	$D_{T + S} (x) = D_{T} (x) + D_{S} (ϕ_{T} (x))$ DT+S(x)=DT(x)+DS(ϕT(x))
Lipschitz continuity	(	D_T(x) – D_T(y)	\leq \frac{e^{LT} – 1}{L} \|x – y\| )
Instantaneous growth	$\frac{d}{d T} D_{T} (x) = d (ϕ_{T} (x), A)$ dTdDT(x)=d(ϕT(x),A)
Ergodic limit	$\lim_{T \to \infty} \frac{1}{T} D_{T} (x) = \int d (y, A) d μ (y)$ limT→∞T1DT(x)=∫d(y,A)dμ(y)
Exponential stability implies finite D∞D∞	$D_{\infty} (x) \leq \frac{C}{κ} d (x, A)$ D∞(x)≤κCd(x,A)
Recovery bound (general)	$κ \leq \frac{C d (x, A)}{D_{\infty} (x)}$ κ≤D∞(x)Cd(x,A)
Recovery bound (C=1)	$κ \leq \frac{d (x, A)}{D_{\infty} (x)}$ κ≤D∞(x)d(x,A)
Finite horizon approximation	$D_{T} (x) \to D_{\infty} (x)$ DT(x)→D∞(x) as $T \to \infty$ T→∞

4. The Unified Variable Set

The following variables are defined operationally. Where a variable is a proposal, that is stated explicitly.

4.1 Corrective Permeability ( $κ$ κ)

Definition 4 (Corrective Permeability): $κ$ κ is the recovery rate of the system to its attractor after a small perturbation. Operationally estimated as $κ = 1 / τ$ κ=1/τ under approximately exponential relaxation, where $τ$ τ is the characteristic recovery time constant. This coincides with the exponential convergence exponent in the linearized regime and is consistent with the original definition in the attractor framework.

Relationship to DTDT: From Corollary 1, for a system with initial deviation $d (x, A)$ d(x,A), $κ \leq \frac{C d (x, A)}{D_{\infty} (x)}$ κ≤D∞(x)Cd(x,A).

Note on κ’s status: In this paper, κ is treated as a primitive empirical regime parameter. A stronger theory would derive κ from $D_{T}$ DT and system geometry; this remains an open direction for future work.

4.2 Drift Rate ( $γ$ γ) — A Proposed Distinction

Definition 5 (Drift Rate): We propose the following operational distinction between dynamical regimes, based on the dominant Lyapunov exponent $λ_{\max}$ λmax:

Regime	$λ_{\max}$ λmax	$κ$ κ	$γ$ γ	Behavior
Stable attractor	$< - 0.01$ <−0.01	$> 0$ >0	$0$ 0	Converges to fixed point
Persistent chaos	$\approx 0$ ≈0	$\approx 0$ ≈0	$> 0$ >0	Wanders without convergence
Full chaos	$> 0$ >0	undefined	$> 0$ >0	Diverges

Thresholds: $λ_{\max} < - 0.01$ λmax<−0.01, $∣ λ_{\max} ∣ \leq 0.01$ ∣λmax∣≤0.01, and $λ_{\max} > 0.01$ λmax>0.01 (pre-registered, measured in units of 1/epoch). These numerical thresholds are illustrative defaults rather than theoretically privileged constants.

Grounding: This distinction is inspired by the literature on chaos in high-dimensional neural networks (Engelken, Wolf & Abbott, 2023; Sompolinsky, Crisanti & Sommers, 1988; Clark, Abbott & Litwin-Kumar, 2023; Fournier & Urbani, 2023). For the treatment of stochastic and random perturbations, see Arnold (1998).

Falsification: If $κ$ κ and $γ$ γ are perfectly correlated (i.e., systems with small $κ$ κ always have small $γ$ γ), the distinction is not useful.

4.3 Basin Depth ( $B$ B) and Persistence Depth ( $\tilde{B}$ B~)

Definition 6a (Basin Depth — Energy Barrier): $B$ B is the energy barrier required to escape the basin, measured as the potential difference between the attractor and the saddle point on the basin boundary: $B = V (saddle) - V (attractor)$ B=V(saddle)−V(attractor)

This preserves the original definition from earlier papers.

Definition 6b (Persistence Depth): As a complementary measure, we define: $\tilde{B} = \min_{x \in \partial B} D_{T} (x)$ B~=x∈∂BminDT(x)

This is the cumulative deviation required to reach the basin boundary. The relationship between $B$ B and $\tilde{B}$ B~ remains an open mathematical question.

Operational alternative: In practice, the basin boundary may not be well-defined. Estimate $B$ B via the Arrhenius relationship $P_{escape} \propto e^{- B / T}$ Pescape∝e−B/T, where $T$ T is the noise level.

4.4 Reality Alignment ( $R$ R)

Definition 7 (Reality Alignment): $R$ R is the expected log predictive likelihood: $R = E [\log p (y ∣ X)]$ R=E[logp(y∣X)]

where $p (y ∣ X)$ p(y∣X) is the system’s predictive distribution over outcomes $y$ y given state $X$ X. Higher $R$ R indicates better predictive accuracy. This is a standard measure of predictive performance; the label “reality alignment” is a philosophical interpretation.

Direction-dependence: The framework interprets $R$ R as potentially direction-dependent: $R_{A \to B} \neq R_{B \to A}$ RA→B=RB→A. This captures the asymmetry found in Berglund et al. (2024), where models trained on “A is B” fail to generalize to “B is A.” This interpretation is a framework-level claim.

Note on integration: Among the core variables, $R$ R is the least integrated with the trajectory-based formalism. Unlike $κ$ κ, $B$ B, and $\tilde{B}$ B~, which are directly derived from or related to $D_{T}$ DT, $R$ R is imported from Bayesian statistics. A more complete theoretical derivation of $R$ R from the same dynamical principles—perhaps as an information-theoretic functional of the occupation measure—remains an open direction for future work.

5. Theoretical Framework

5.1 Relationship Between $D_{T}$ DT, $P_{topo}$ Ptopo, and $E (t)$ E(t)

Functional	What It Measures	Regime
$D_{T} (x)$ DT(x)	Cumulative deviation from attractor	All systems
$P_{topo} (t)$ Ptopo(t)	Topological feature lifetime	Systems with topological structure
$E (t)$ E(t)	Rate of topological change	Learning systems

Hypothesis: In learning systems, $D_{T}$ DT and $P_{topo}$ Ptopo are positively correlated early in learning and negatively correlated late in learning. Turner & Barak (2023) demonstrate that RNNs develop attractors sequentially during training, which may correspond to phases of topological simplification. This is a testable prediction.

5.2 Relationship Between $κ$ κ, $γ$ γ, and $E (t)$ E(t)

Hypothesis: In a learning system, the topological evolution rate $E (t)$ E(t) is monotonically related to $κ$ κ only if the system is not in persistent chaos: $\partial E / \partial κ > 0$ ∂E/∂κ>0 (with $E$ E and $κ$ κ measured on appropriate scales) in convergent regimes. In persistent chaos, $E (t)$ E(t) is monotonically related to $γ$ γ: $\partial E / \partial γ > 0$ ∂E/∂γ>0. Correlation analysis provides a statistical test of these monotonicity relationships.

5.3 Adaptive Landscape (Heuristic Note)

The adaptive landscape $V (X, t)$ V(X,t) evolves as: $\dot{V} = g (X, V) - λ V + ξ (t)$ V˙=g(X,V)−λV+ξ(t)

For gradient systems with $\dot{X} = - \nabla_{X} V (X)$ X˙=−∇XV(X), and assuming the dynamics remain within the basin where higher-order nonlinearities are negligible, the cumulative deviation functional can be approximated as: $D_{T} (x) \approx \int_{0}^{T} ∥ \nabla_{X} V (ϕ_{τ} (x), τ) ∥ d τ$ DT(x)≈∫0T∥∇XV(ϕτ(x),τ)∥dτ

This is a local heuristic. A full derivation and integration into the core formalism is left for future work.

6. Testable Predictions

6.1 Core Prediction

Prediction: In a learning system, $E (t)$ E(t) is monotonically related to $κ$ κ in convergent regimes: $\partial E / \partial κ > 0$ ∂E/∂κ>0 (with $E$ E and $κ$ κ measured on appropriate scales), and $\partial E / \partial γ > 0$ ∂E/∂γ>0 in persistent chaos. Correlation analysis provides a statistical test of this monotonicity: $Corr (E (t), κ) > 0 ⟺ λ_{\max} < 0$ Corr(E(t),κ)>0⟺λmax<0 $Corr (E (t), γ) > 0 ⟺ λ_{\max} \approx 0$ Corr(E(t),γ)>0⟺λmax≈0

Falsification: If $E (t)$ E(t) correlates with $κ$ κ in all regimes, or with $γ$ γ in all regimes, the prediction is falsified.

6.2 Secondary Prediction

Prediction: In systems with high $R$ R, $D_{T}$ DT and $P_{topo}$ Ptopo are negatively correlated late in learning; in systems with low $R$ R, they are uncorrelated or positively correlated.

Falsification: If $D_{T}$ DT and $P_{topo}$ Ptopo are negatively correlated in both high-R and low-R systems, the prediction is falsified.

6.3 Boundary Condition and Global Falsifier

Conjecture: We conjecture that the framework applies to any system satisfying:

A. Well-defined state space.
B. Subject to perturbations.
C. Exhibits at least one identifiable attractor.
D. Dynamics are observable and measurable.

Global Falsifier: The unified ontology claim collapses if a system is found where $D_{T}$ DT, $κ$ κ, and topological persistence are mutually independent across all regimes, and where $R$ R cannot be expressed as a functional of the trajectory or occupation measure. If such a system exists, the framework’s claim to unify persistence, stability, and reality alignment would be falsified.

7. Experimental Design

7.1 System Choice

Train a CNN on MNIST or CIFAR-10. Use latent activation manifolds for topological analysis.

Justification: Karuppiah, Nazreen Banu et al. (2026) demonstrate the use of persistent homology on activations to study feature learning and generalization. Turner & Barak (2023) show that RNNs develop attractors sequentially, providing a controlled setting for studying topological evolution during learning.

7.2 Variable Measurement

Variable	Protocol
$D_{T} (x)$ DT(x)	Sample weights; compute distance to final attractor; integrate.
$P_{topo} (t)$ Ptopo(t)	Compute persistent homology on latent activations; sum feature lifetimes.
$E (t)$ E(t)	Finite differences of $P_{topo} (t)$ Ptopo(t).
$κ$ κ	Perturb weights; measure recovery time $τ$ τ; $κ = 1 / τ$ κ=1/τ.
$γ$ γ	Compute average drift rate during training.
$R$ R	Cross-domain generalization accuracy.

7.3 Statistical Analysis

Correlate $E (t)$ E(t) with $κ$ κ and $γ$ γ conditional on regime.
Pre-register thresholds and sample size.

Note on future empirical work: A full empirical validation would require pre-registration with specified sample size, significance thresholds, power analysis, and robustness checks. These are planned for subsequent work.

8. Discussion

8.1 Implications

The paper provides a candidate formalization with defined variables, mathematical properties, and testable predictions. The mathematical properties of $D_{T}$ DT establish its relationship to $κ$ κ and provide a foundation for the framework’s core claims.

8.2 Limitations

$P_{topo}$ Ptopo is computationally expensive.
The framework is a meta-theory, not a complete domain-specific theory.
Variables may be confounded; causal inference requires controlled experiments.
The $κ / γ$ κ/γ regime distinction is proposed and requires empirical validation.

8.3 Future Work

Empirical validation of predictions.
Formal derivation of relationships from first principles.
Extension to other domains.
Computational efficiency improvements.

9. Conclusion

This paper proposes a candidate formalization for the attractor framework. The central mathematical innovation is treating persistence as a functional defined over trajectories— $D_{T} (x) = \int_{0}^{T} d (ϕ_{τ} (x), A) d τ$ DT(x)=∫0Td(ϕτ(x),A)dτ—rather than as a scalar property of states. We defined the cumulative deviation functional $D_{T}$ DT, the topological persistence functional $P_{topo} (t)$ Ptopo(t), and the topological evolution rate $E (t)$ E(t). We proved several mathematical properties of $D_{T}$ DT, including non-negativity, monotonicity, additivity, Lipschitz continuity, and a bound relating $D_{\infty}$ D∞ to $κ$ κ: $D_{\infty} (x) \leq \frac{C}{κ} d (x, A)$ D∞(x)≤κCd(x,A). We established connections to dynamic programming and ergodic theory. We unified the variable set with operational definitions. We derived testable predictions and provided a falsifiable experimental protocol.

The framework now admits formal definitions, operational variables, and empirical tests. The next step is empirical validation.

Appendix A: Possible Extensions from Larose (2025) — Unverified Source

Note: The following source has not been independently verified. It is included for completeness and as a potential direction for future exploration, but should not be treated as established.

Larose (2025) develops a framework for recursive deformation systems. Two constructs are potentially relevant:

Constraint Functional: $C (X) = \int_{trajectory} ∥ \nabla Φ ∥ d τ$ C(X)=∫trajectory∥∇Φ∥dτ, measuring cumulative irreversible deformation.

Persistence Invariant: $I_{p} = \oint R d Φ$ Ip=∮RdΦ, a topological invariant.

These are not yet integrated into the core framework and are presented here for completeness and future exploration. They should be treated as unverified candidate extensions.

References

Arnold, L. (1998). Random Dynamical Systems. Springer.

Berglund, L., et al. (2024). “The Reversal Curse: LLMs Trained on ‘A is B’ Fail to Learn ‘B is A’.” arXiv:2309.12288.

Bowen, R. (1975). Equilibrium States and the Ergodic Theory of Anosov Diffeomorphisms. Springer.

Carlsson, G. (2009). “Topology and data.” Bulletin of the American Mathematical Society, 46(2), 255-308.

Carlsson, G., & Zomorodian, A. (2009). “The theory of multidimensional persistence.” Discrete & Computational Geometry, 42(1), 71-93.

Clark, D. G., Abbott, L. F., & Litwin-Kumar, A. (2023). “Dimension of activity in random neural networks.” Physical Review Letters, 131, 118401.

Edelsbrunner, H., & Harer, J. (2010). Computational Topology: An Introduction. American Mathematical Society.

Engelken, R., Wolf, F., & Abbott, L. F. (2023). “Lyapunov spectra of chaotic recurrent neural networks.” Physical Review Research, 5, 043044.

Fournier, S. J., & Urbani, P. (2023). “Statistical physics of learning in high-dimensional chaotic systems.” Journal of Statistical Mechanics: Theory and Experiment, 2023(11), 113301.

Karuppiah, K., Nazreen Banu, M., et al. (2026). “Topological Data Analysis (TDA) as a Framework for Understanding Deep Learning Behavior.” 2025 IEEE 5th International Conference on ICT in Business Industry & Government (ICTBIG), Indore, India, December 12-13, 2025. IEEE Xplore. DOI: 10.1109/ICTBIG68706.2025.11323998.

Larose, H. (2025). “A Mathematical Theory of Frame-Independent Persistence.” Academia.edu. [Unverified source.]

Ruelle, D. (1989). Chaotic Evolution and Strange Attractors. Cambridge University Press.

Sompolinsky, H., Crisanti, A., & Sommers, H. J. (1988). “Chaos in Random Neural Networks.” Physical Review Letters, 61(3), 259-262.

Turner, E., & Barak, O. (2023). “The Simplicity Bias in Multi-Task RNNs: Shared Attractors, Reuse of Dynamics, and Geometric Representation.” Advances in Neural Information Processing Systems (NeurIPS).

Suggested citation: Galida, R. S. (2026). The Persistence Functional: A Candidate Formal Foundation for the Attractor Framework (Foundational Edition). Fantasy Attractor.

Cognitive Attractor Dynamics: A Formal Theory of Self-Concept and Self-Engineering

June 26, 2026 3:37 pm / Leave a comment

Robert Galida
July 2026
[F] (Foundation)

Abstract

The attractor framework provides a unified vocabulary for describing persistence and change across physical, biological, cognitive, and social systems. This paper presents a formal theory of cognitive attractor dynamics, grounding the framework’s core variables—κ (corrective permeability), B (basin depth), C (coordination capacity), and R (reality alignment)—in a rigorous mathematical framework. The cognitive state space $X (t) \in R^{n}$ X(t)∈Rn is defined, a dynamical equation $\dot{X} = - \nabla V (X) + η (t) + E (t)$ X˙=−∇V(X)+η(t)+E(t) is specified, and the variables are derived from the potential landscape $V (X)$ V(X). The theory connects to existing frameworks (Hopfield networks, predictive coding, active inference, reinforcement learning) and generates testable predictions about cognitive flexibility, goal persistence, reality alignment, and coordination capacity. The paper is offered as a formal foundation for empirical testing.

All claims are formal hypotheses, not conclusions. The framework is a domain-general dynamical ontology with an associated research programme — a formal theory, not a completed science.

1. Introduction

The attractor framework has been applied to biology, cosmology, AI, and civilizational dynamics. This paper presents a formal theory of cognitive attractor dynamics. It asks a simple question:

Can the self — beliefs, goals, and self-narratives — be modeled as an attractor landscape in a high-dimensional cognitive state space?

The answer is yes — with explicit formal definitions.

A note on the Law of Attraction: The Law of Attraction is often framed as a metaphysical claim. This paper reframes it as conscious self-direction and self-engineering — the deliberate shaping of one’s own cognitive attractor landscape through belief revision, attentional focus, and behavioral reinforcement.

A note on the framework’s status: This paper presents a formal theory. The mathematical derivation of equivalence is specified. The framework is offered as a foundation for empirical testing.

A note on domain of applicability: The framework applies to any persistent cognitive system satisfying the formal conditions defined below.

2. Core Definitions

2.1 The Framework Variables

Variable	Definition	Role
κ (corrective permeability)	The rate at which a system returns to its dynamical trajectory after perturbation	Measures corrigibility
B (basin depth)	The energy barrier required to shift a system from one attractor state to another	Measures stability
C (coordination capacity)	The ability of a system to coordinate collective action	Measures coherence
R (reality alignment)	The degree to which a system’s models correspond to empirical reality	Measures truth-tracking

2.2 Primitive vs. Derived Concepts

Primitive	Definition	Derived	Source
State	The complete description of a system at a given time	—	—
Interaction	Any exchange of energy, momentum, or information between systems	—	—
Constraint	Any factor that restricts the possible states or trajectories of a system	—	—
Perturbation	Any deviation from the system’s dynamical trajectory	—	—
—	—	κ	Recovery rate after perturbation (derived from perturbation dynamics)
—	—	B	Energy barrier between attractors (derived from constraint topology)
—	—	C	Coordination capacity (derived from interaction topology)
—	—	R	Reality alignment (derived from model-state correspondence)

3. The Formal Theory

3.1 The Cognitive State Space

Define the cognitive state vector: $X (t) \in R^{n}$ X(t)∈Rn

where $n$ n is the dimensionality of the state space. The choice of representation is domain-specific:

Representation	Form	Domain
Belief vector	$X = (b_{1}, b_{2}, \dots, b_{n})$ X=(b1,b2,…,bn)	Cognitive psychology
Neural latent	$X \in R^{d}$ X∈Rd	Computational neuroscience
Control variables	$X = (a, e, m)$ X=(a,e,m)	Cognitive control

Distinction between spaces:

Abstract state space $X$ X: the theoretical manifold of cognitive states
Measurement space $Y$ Y: the space of observables (behavior, neural activity)
Embedding $ϕ : Y \to X$ ϕ:Y→X: mapping from data to latent state

Falsification: If different cognitive states produce identical trajectories in the chosen $X$ X-space, the representation fails.

3.2 The State Equation

The dynamics of the cognitive state are governed by: $\dot{X} = - \nabla V (X) + η (t) + E (t)$ X˙=−∇V(X)+η(t)+E(t)

where:

$X (t)$ X(t) is the cognitive state at time $t$ t
$V (X)$ V(X) is the cognitive potential landscape
$η (t)$ η(t) is stochastic noise (temperature $T$ T)
$E (t)$ E(t) is external perturbation

3.3 The Potential Function

We adopt the following illustrative potential function — a mathematically smooth function that produces one minimum and finite depth: $V (X) = \frac{1}{2} c ∥ X - X^{*} ∥^{2} + \frac{B}{1 + e^{- α ∥ X - X^{*} ∥^{2}}}$ V(X)=21c∥X−X∗∥2+1+e−α∥X−X∗∥2B

where:

$c$ c is the curvature parameter (not κ)
$B$ B is the basin depth (barrier height)
$α$ α controls the steepness of the basin

Note: This potential function is an illustrative ansatz, chosen to demonstrate the framework’s logic. Alternative forms (multi-well, free-energy-based) are possible and should be explored empirically. The specific functional form is not claimed to be a unique derivation.

Alternative forms:

Form	Equation	Use Case
Quadratic	$V (X) = \frac{1}{2} c ∥ X - X^{*} ∥^{2}$ V(X)=21c∥X−X∗∥2	Single attractor, linear dynamics
Multi-well	$V (X) = \sum_{i} B_{i} ϕ (∥ X - X_{i}^{*} ∥^{2})$ V(X)=∑iBiϕ(∥X−Xi∗∥2)	Multiple attractors
Free energy	$V (X) = - \log p (X)$ V(X)=−logp(X)	Bayesian/predictive coding

3.4 Basin Depth (B)

Basin depth $B$ B is the energy barrier required to escape the attractor’s basin: $B = \min_{X \in \partial B} V (X) - V (X^{*})$ B=X∈∂BminV(X)−V(X∗)

where:

$X^{*}$ X∗ is the attractor (stable fixed point)
$\partial B$ ∂B is the boundary of the basin of attraction
$V (X^{*})$ V(X∗) is the potential at the attractor

Empirical estimation: $B$ B can be estimated from:

Time to return to baseline after perturbation
Probability of escape under noise: $P_{escape} \propto e^{- B / T}$ Pescape∝e−B/T
Hysteresis in response to changing inputs

3.5 Corrective Permeability (κ)

κ is the rate of recovery toward the attractor after a perturbation. It is derived from the curvature of V, not independently parameterized.

Formal definition: For a linearized system near the attractor: $\dot{δ X} = - \nabla^{2} V (X^{*}) δ X$ δX˙=−∇2V(X∗)δX

where $δ X = X - X^{*}$ δX=X−X∗ is the deviation from the attractor. The recovery rate is determined by the largest (least negative) eigenvalue of the Hessian: $κ = - λ_{max} (- \nabla^{2} V (X^{*}))$ κ=−λmax(−∇2V(X∗))

For our illustrative potential: $\nabla^{2} V (X) = c + \frac{2 B α c}{1 + e^{- α ∥ X - X^{*} ∥^{2}}}$ ∇2V(X)=c+1+e−α∥X−X∗∥22Bαc

At the attractor ( $X = X^{*}$ X=X∗): $κ_{baseline} = c + B α$ κbaseline=c+Bα

This resolves the circularity: κ is now a derived quantity from the same landscape V. It is not independently parameterized.

Empirical estimation: κ can be estimated from:

Error-correction times in cognitive tasks
Post-error slowing in reaction time tasks
Recovery from emotional perturbations
Neural measures of flexibility (dynamic connectivity)

3.6 Reality Alignment (R)

R is the predictive accuracy of the system: $R = - E [\log p (y ∣ X)]$ R=−E[logp(y∣X)]

where $p (y ∣ X)$ p(y∣X) is the system’s predictive distribution over outcomes $y$ y given its current state $X$ X.

R belongs in learning dynamics, not in the potential: $\dot{θ} = g (R, δ)$ θ˙=g(R,δ)

where θ controls the landscape V, and δ is the prediction error.

Relationship to free energy: $F = KL (q ∥ p) + R$ F=KL(q∥p)+R

where $F$ F is variational free energy. R is maximized when the system’s predictions match reality.

Empirical estimation: R can be estimated from:

Predictive accuracy in decision-making tasks
Calibration of confidence judgments
Prediction error signals (dopaminergic, sensory)

3.7 Coordination Capacity (C)

C is hypothesized to emerge from the network topology of cognitive subsystems.

Open research question: The specific functional form — whether it depends on total coupling strength, spectral radius, modularity, or other graph-theoretic measures — is an open research question. Candidate measures include:

Measure	Description
Spectral radius	Largest eigenvalue of coupling matrix
Modularity	Degree of community structure
Global efficiency	Average inverse shortest path length
Synchronization threshold	Second-smallest Laplacian eigenvalue

Empirical estimation: C can be estimated from:

Coherence between subsystems
Synchrony of neural or behavioral signals
Network graph-theoretic measures

Note: The formula $C = Tr (W) \cdot \min_{i} B_{i}$ C=Tr(W)⋅miniBi is not claimed as a unique derivation. It is a placeholder for future empirical investigation.

4. The Full Parameterized System

4.1 Complete State Equation

Combining all definitions: $\dot{X} = - \nabla V (X) + η (t) + E (t)$ X˙=−∇V(X)+η(t)+E(t)

where:

$V (X)$ V(X) is the cognitive potential landscape
$η (t)$ η(t) is stochastic noise (temperature $T$ T)
$E (t)$ E(t) is external perturbation

4.2 Derived Variables

Variable	Derivation	Units
κ	$κ = - λ_{max} (- \nabla^{2} V (X^{*}))$ κ=−λmax(−∇2V(X∗))	${time}^{- 1}$ time−1
B	$B = \min_{X \in \partial B} V (X) - V (X^{*})$ B=minX∈∂BV(X)−V(X∗)	Energy
R	$R = - E [\log p (y ∣ X)]$ R=−E[logp(y∣X)]	Bits
C	Open research question	Dimensionless

4.3 Parameter Interactions

The parameters are hypothesized to interact:

Hypothesis	Formal Statement
κ increases with R	$κ \propto R$ κ∝R
B decreases with κ	$B \propto 1 / κ$ B∝1/κ
R decreases with B	$R \propto 1 / B$ R∝1/B
Optimal B maximizes κ·R	$B^{*} = \arg \max (κ \cdot R)$ B∗=argmax(κ⋅R)

Falsification: If the variables are entirely independent, the framework is a taxonomy, not a unified theory.

5. Relationship to Existing Frameworks

Framework	Mathematical Form	Relationship
Hopfield networks	$V = - \frac{1}{2} \sum w_{i j} X_{i} X_{j}$ V=−21∑wijXiXj	Special case: discrete attractors
Predictive coding	$F = - \log p (y ∥ X) + KL$ F=−logp(y∥X)+KL	R is negative free energy (minus complexity)
Active inference	$\dot{X} = - \frac{\partial F}{\partial X}$ X˙=−∂X∂F	General case: both perception and action
Reinforcement learning	$V (s) = \max_{a} E [R + γ V (s^{'})]$ V(s)=maxaE[R+γV(s′)]	C emerges from value function coupling

6. Testable Predictions

6.1 Prediction 1: Mindfulness Increases κ

Formal statement: Mindfulness training increases corrective permeability.

Empirical test: Measure error-correction times in cognitive tasks before and after mindfulness intervention. Faster post-error adjustments indicate higher κ.

Falsification: If mindfulness training does not lead to faster error-correction times, the prediction fails.

6.2 Prediction 2: Rigidity = Deep B + Low κ

Formal statement: High cognitive rigidity corresponds to deep B and low κ.

Empirical test: Measure reversal learning times and set-shifting ability in high-rigidity individuals.

Falsification: If rigid individuals adapt as quickly as flexible individuals, the prediction fails.

6.3 Prediction 3: Rumination = High B + Low R

Formal statement: Rumination corresponds to high B and low R.

Empirical test: Measure persistence in negative mood states and predictive accuracy in ruminative individuals.

Falsification: If ruminators show low persistence or high predictive accuracy, the prediction fails.

6.4 Prediction 4: Success = High B + High κ

Formal statement: Goal achievement requires both deep B and high κ.

Empirical test: Measure goal persistence (B) and adaptability (κ) in high-achieving individuals.

Falsification: If high achievers show low B or low κ, the prediction fails.

6.5 Prediction 5: Obsession = High B + Low κ

Formal statement: Obsessive-compulsive patterns correspond to high B and low κ.

Empirical test: Measure persistence on incorrect choices in obsessive individuals.

Falsification: If obsessive individuals show normal recovery from errors, the prediction fails.

6.6 Prediction 6: Kramers’ Escape in Cognition

Formal statement: Cognitive transition probabilities follow Kramers’ law.

Empirical test: Vary noise levels (uncertainty, distractors) and measure transition rates between cognitive states.

Falsification: If the relationship is not log-linear, the basin-depth metaphor fails.

6.7 Prediction 7: Exponential Recovery

Formal statement: Cognitive recovery follows exponential decay.

Empirical test: Fit recovery trajectories to exponential and power-law models.

Falsification: If power-law fits are superior, the exponential recovery model fails.

7. What This Paper Does Not Claim

This paper does not claim:

Thoughts directly create reality
The Law of Attraction is literally true as a metaphysical claim
The framework replaces cognitive science
The framework is a theory of everything
The framework generates novel predictions (it does — see §6)
Mathematical equivalence between cognitive and other systems
C is a primitive variable (it is an open research question)
The illustrative potential function is a unique derivation

8. Limitations

Limitation	Address
κ is derived from V	✅ Resolved
R belongs in learning dynamics	✅ Resolved
B and κ are not independent	✅ Resolved
Potential function is ad hoc	✅ Acknowledged as illustrative ansatz
State space is generic	✅ Distinction between abstract/measurement/embedding spaces added
C formula is speculative	✅ Removed; left as open research question

9. Open Research Questions

Question	Domain
What is the minimal state space for a given cognitive domain?	Formalization
What is the functional form of V(X) for a given domain?	Formalization
Do cognitive escape probabilities follow Kramers’ law?	Empirical
Do recovery trajectories follow exponential decay?	Empirical
Is R equivalent to negative free energy?	Formalization
Can C be derived from network topology?	Formalization
Do κ, B, and R scale with system size?	Formalization
Does an optimal B exist?	Empirical
How do κ, B, and R interact?	Formalization

10. Conclusion

The attractor framework is now formally defined:

Element	Definition
State space	$X (t) \in R^{n}$ X(t)∈Rn
Dynamics	$\dot{X} = - \nabla V (X) + η + E$ X˙=−∇V(X)+η+E
Potential	$V (X) = \frac{1}{2} c ∥ X - X^{} ∥^{2} + \frac{B}{1 + e^{- α ∥ X - X^{} ∥^{2}}}$ V(X)=21c∥X−X∗∥2+1+e−α∥X−X∗∥2B (illustrative ansatz)
Derived: κ	$κ = - λ_{max} (- \nabla^{2} V (X^{*}))$ κ=−λmax(−∇2V(X∗))
Derived: B	$B = \min_{X \in \partial B} V (X) - V (X^{*})$ B=minX∈∂BV(X)−V(X∗)
Derived: R	$R = - E [\log p (y ∣ X)]$ R=−E[logp(y∣X)]
Open: C	Emerging from network topology

The framework generates testable predictions and is ready for empirical validation.

The next step is computational validation: simulate the dynamics, recover κ and B, demonstrate Kramers’ escape, and show recovery trajectories. Then move to human experiments.

References

Boyatzis, R.E., Rochford, K., & Taylor, S.N. (2015). “The role of the positive emotional attractor in vision and shared vision.” Frontiers in Psychology, 6:670.
Cheema, A., & Bagchi, R. (2011). “The effect of goal visualization on goal pursuit.” Journal of Marketing, 75(2), 109–123.
Geisler, F.C.M., & Kubiak, T. (2009). “Heart rate variability predicts self-control in goal pursuit.” European Journal of Personality, 23, 623–633.
Golubickis, M., Tan, L.B.G., Jalalian, P., Falbén, J.K., & Macrae, C.N. (2024). “Brief mindfulness-based meditation enhances the speed of learning following positive prediction errors.” Quarterly Journal of Experimental Psychology, 77(11), 2312–2324.
Kronemyer, D., & Bystritsky, A. (2014). “A non-linear dynamical approach to belief revision in cognitive behavioral therapy.” Frontiers in Computational Neuroscience, 8:55.
MacDonald, M.R., & Kuiper, N.A. (1985). “Efficiency and automaticity of self-schema processing in clinical depressives.” Motivation and Emotion, 9(2), 171–184.
Singer, J.A., Blagov, P., Berry, M., & Oost, K.M. (2013). “Self-defining memories, scripts, and the life story.” Journal of Personality, 81(6), 569–582.

Suggested citation: Galida, R. S. (2026). Cognitive Attractor Dynamics: A Formal Theory of Self-Concept and Self-Engineering. Fantasy Attractor.

The Pre‑tensioned Body: A Hypothesis Paper Grounding the Attractor Framework in ECM Mechanics [M] [F] (2026) Robert Galida – June 2026

June 23, 2026 10:41 am / Leave a comment

Abstract

The attractor framework proposes that persistence under perturbation is the fundamental mark of reality—a property it terms constraint navigation. This paper proposes a biological grounding for the framework in the physical architecture of the body. From established biomechanical principles, the body is identified as a pre‑tensioned hydrophilic‑collagenous composite—a system where osmotic swelling pressure (from GAGs and proteoglycans) is actively constrained by collagen tensile strength. The difference between the calculated Water Holding Capacity (WHC) of the body’s hydrophilic components and its actual water content is proposed as a candidate surrogate signature of this pre‑tensioned state. Mechanotransduction is identified as a primary intercellular communication channel, and the ECM is shown to be a dissipative attractor that stores mechanical history and shapes cellular behaviour. The paper maps the attractor framework’s core variables (κ, B, basin depth) onto measurable physiological quantities as research hypotheses: κ is proposed as a latent variable reflecting perturbation-recovery efficiency, estimated from candidate observables such as tissue recoil time, baroreflex sensitivity, and HRV recovery; B is proposed as a function of prestress, repair capacity, and network connectivity, with the WHC discrepancy as one candidate, non-exclusive proxy for its prestress component; and basin transitions are proposed to correspond to crossing basin-specific thresholds, not a single uniform threshold. A research agenda is provided, including protocols for measuring κ and B non‑invasively and testing the WHC‑water content discrepancy as a candidate metric of basin depth.

Crucially, this paper does not revise the framework’s ontological hierarchy. As established in Intelligence is the Primitive (Galida, 2026a), the primitive is constraint navigation—the capacity to detect perturbations, update internal states, and maintain persistent trajectories. Mechanotransduction is proposed as the physical substrate through which constraint navigation is implemented in biological systems. The nervous system and the ECM are complementary regulatory layers, not competing primitives.

All mappings from physiological variables to framework constructs are proposed as research hypotheses, not established conclusions.

1. Introduction

The attractor framework defines intelligence as the ability to navigate a constraint field and distinguishes reality attractors (high κ, shallow basin, corrigible) from fantasy attractors (low κ, deep basin, sealed). The framework has been applied to physics, biology, cognition, AI, and social dynamics. However, its physical grounding in the body has remained implicit.

This paper proposes that grounding. It begins with an established biomechanical model of the body’s architecture: a pre‑tensioned hydrophilic‑collagenous composite. It then proposes mappings from the framework’s core variables onto measurable physiological quantities, establishes mechanotransduction as a primary intercellular communication channel, and identifies the ECM as a dissipative attractor that stores mechanical history. The paper concludes with a research agenda and testable predictions.

A note on terminology: In the attractor framework’s hierarchy, the primitive is constraint navigation—a domain-general property of any system that detects perturbations and maintains persistent trajectories. Mechanotransduction is proposed as the physical substrate through which constraint navigation is implemented in biological tissues. This paper proposes that substrate; it does not claim that mechanotransduction is a deeper primitive than constraint navigation. For the framework’s ontological hierarchy, see Galida (2026a).

A note on scope: All mappings from physiological variables (prestress, mechanotransduction rate, WHC discrepancy) to framework constructs (B, κ, basin depth) are proposed as research hypotheses, not established conclusions. The biological claims are grounded in existing literature; the attractor mappings are the novel, untested component of this paper.

A note on the framework’s strongest anchor: The framework’s most direct empirical anchor is fibrosis, which exhibits classic attractor properties: self-reinforcement, hysteresis, path dependence, resistance to reversal, and threshold behavior. Fibrosis is therefore treated as a central demonstration of the framework’s applicability to biological systems.

This paper is primarily a biological hypothesis paper. It proposes specific mappings from physiological variables to attractor-framework constructs. The broader philosophical claims of the attractor framework—about intelligence, consciousness, and reality—are discussed elsewhere (see Galida, 2026a) and are not the focus of this paper. Where speculative extensions are made, they are clearly flagged.

2. The Body as a Pre‑tensioned System

2.1 The Established Biomechanical Model

We adopt the established biomechanical model of connective tissue as a composite material (Ingber’s cellular tensegrity; Donnan osmotic swelling models). In this model:

Component	Role
Hydrophilic components (GAGs, proteoglycans)	Provide osmotic swelling pressure – a distributed, expansive force
Collagen	Provides tensile strength – the “rebar” that constrains the swelling pressure into a coherent, load‑bearing architecture
The body	A pre‑stressed system – like reinforced concrete, where the rebar (collagen) is under tension and the matrix (GAGs) is under compression

This is not a novel derivation from first principles; it is a reformulation of standard connective-tissue biomechanics in attractor-framework vocabulary.

2.2 The WHC‑Water Content Discrepancy

The calculated Water Holding Capacity (WHC) of the body’s hydrophilic components—the maximum water the tissue could hold if all GAGs and proteoglycans were fully hydrated and unrestricted—exceeds the actual water content. This difference is proposed as a candidate surrogate signature of the pre‑tensioned state. It represents the water that is being held back by the collagen network—the stored elastic + osmotic energy that defines the attractor basin.

Quantity	Meaning
Calculated WHC	The maximum water the tissue could hold under unrestricted swelling
Actual water content	The water the tissue actually contains
Difference	The water held back by collagen—a candidate surrogate for pre‑tension

Operational definition: WHC is estimated via the Donnan equilibrium osmotic pressure: $Π = R T \sum (C_{i o n, i n s i d e} - C_{i o n, o u t s i d e})$ Π=RT∑(Cion,inside−Cion,outside)

where $C_{i o n}$ Cion is determined by the fixed negative charge density of the GAGs. The WHC is the water content predicted under unconstrained free‑swelling conditions. The discrepancy with measured water content is therefore a candidate surrogate for the mechanical work done by the collagen network to constrain this swelling.

Critical limitation: WHC discrepancy is a model‑derived construct, not a direct observable. Its validity as a measure of prestress must be confirmed ex vivo by correlating the discrepancy with direct tensile/compressive stress‑strain measurements. We treat it as a candidate surrogate marker for prestress, not as prestress itself.

WHC discrepancy is one candidate observable among several possible prestress proxies. Other candidates include tissue stiffness (measured by elastography), recoil dynamics (measured by indentation), hydraulic permeability (measured by perfusion), and poroelastic relaxation time (measured by stress-relaxation tests). We do not claim WHC discrepancy is the preferred or exclusive measure; it is one candidate that warrants investigation.

Importantly, the relationship between WHC discrepancy and prestress is unlikely to be unique. Multiple states—edema, fibrosis, dehydration, inflammation, altered ionic composition, and altered GAG composition—could produce similar WHC-water discrepancies without representing the same prestress state. Prestress may be one contributor to the WHC discrepancy, but the relationship is unlikely to be one-to-one. WHC discrepancy is proposed as a starting point for investigation, not as a definitive measure.

2.3 The Functional Role of Pre‑tension

At the scale of a whole organism, slow diffusion is solved by the cardiovascular system (convective bulk flow). However, once oxygen and nutrients leave the capillary bed, they must traverse the interstitial space to reach individual cells. Over distances of micrometers to millimeters, pure diffusion remains rate‑limiting. The pre‑tensioned ECM contributes to pressure gradients, fluid flow, and mechanical mixing that actively transport solutes through the interstitium. It is one of several contributors, alongside vascular pulsatility, lymphatic drainage, muscle contraction, respiration, and posture.

We propose that prestress is necessary for efficient mechanotransduction, but we do not claim it is the dominant driver of interstitial flow.

Problem	Pre‑tensioned Contribution
Diffusion is too slow over tissue‑scale distances	The pre‑stressed ECM contributes to pressure gradients, fluid flow, and mechanical mixing
Nutrients must reach cells deep within tissues	Osmotic pressure generated by GAGs contributes to interstitial fluid flow
Waste must be removed efficiently	Mechanical deformation acts as a pump, driving convection and mixing
Signalling molecules must propagate rapidly	Mechanotransduction transmits signals faster than diffusion alone

3. Pre‑tension as Stored Constraint History

The connective‑tissue matrix carries a record of mechanical loading. Collagen fibers, proteoglycans, and crosslinks retain the geometry and tension that arose during development or past stresses. In effect, a pre‑stressed ECM stores constraint history: cells continually read and update it. Cells respond to physical stimuli from their microenvironment, including ECM topography, composition, and stiffness (Discher et al., 2005; Engler et al., 2006), and remodel the matrix accordingly. The current structure of the ECM—fiber alignment, crosslink density, hydration patterns—encodes prior mechanical history.

“Constraint history” is more precise than “mechanical memory” because it refers to observable physical properties—fiber alignment, crosslink density, residual strain, anisotropy, and tissue architecture—rather than implying information storage in the cognitive or computational sense.

Hypothesis: Regions of ECM with higher collagen alignment or GAG concentration will correlate with the history of applied stress. Tendons remold to past loading, and scars “remember” tension by oriented fibers.

Experiment: Culture fibroblasts on 3D collagen gels under strain, then release the load and track collagen realignment over days. If the matrix “remembers,” the network should remain partly aligned, and fibroblasts on this matrix will show different mechanosignaling (e.g., YAP nuclear localization) compared to naïve gels.

4. Pre‑tension and Free Energy Storage

A pre‑tensed ECM is a far‑from‑equilibrium state that requires energy to maintain. More precisely, it stores free energy in the form of osmotic pressure (from GAGs) and tensile stress (from collagen). Negatively charged GAGs imbibe water and generate osmotic pressure; collagen fibers stretch to resist this swelling, creating tensional prestress. The result is a tension–compression balance that is thermodynamically high in free energy. When pre‑tension is lost (e.g., by breaking crosslinks or GAG depletion), the system relaxes to a lower‑energy, higher‑entropy configuration.

Hypothesis: The water‑holding capacity (WHC) gradient creates a free‑energy gradient. A large WHC–actual water discrepancy (more bound water than free water) signifies a high osmotic tension and greater free energy storage.

Experiment: Use temperature ramps or chemical perturbations to alter ECM hydration in vitro, and measure work done (e.g., pressure‑volume loops). Compare the change in free energy (via heat release or sorption isotherms) as pre‑tension is varied.

5. Thresholds and Phase Transitions in Pre‑tension

Biological systems may exhibit a critical tension threshold below which mechanosignaling collapses. In a highly tensioned network, cells easily sense force via stretched fibers; if the network becomes too lax, mechanical signals dissipate before triggering cell responses. There may be a phase‑like transition: above a certain pre‑tension, the tissue acts as a coherent signal‑transmitting medium; below it, the matrix cannot convey stiffness and mechanosensors fall silent.

Basin depth B is a dynamical concept—the energy barrier required to shift a system from one attractor state to another. Prestress is hypothesized to be one contributor to basin depth, not a direct measure of basin depth itself. Other contributors include repair capacity, energy availability, network connectivity, and hysteresis. Fibrosis illustrates this distinction: high prestress with low repair capacity yields a deep but pathological basin—a fantasy attractor.

κ is defined as responsiveness to perturbation per unit time—specifically, the inverse of the time (τ) required for a system to return to baseline after a standardized perturbation. In biological terms, κ is operationalized as perturbation-to-state-update efficiency. Candidate observables include tissue recoil time, baroreflex sensitivity, HRV recovery, and response latency in mechanosensitive signaling. The framework does not claim that any one of these is κ; it claims that they may correlate with κ under controlled conditions.

Hypothesis: There exists a tipping point in ECM tension where YAP/TAZ signaling drops sharply.

Experiment: Gradually digest collagen or GAGs in a tissue sample (using collagenase or hyaluronidase) and monitor cellular mechanosignaling (e.g., YAP nuclear localization, calcium spikes). Plot signaling versus residual ECM stiffness to identify any sharp transition.

6. Restoring Lost Pre‑tension (ECM Plasticity)

The pre‑tensioned state can be partially restored. Tissue remodeling is dynamic: fibroblasts and other cells continually synthesize new ECM and restore tension when stimulated. Exercise and mechanical loading promote this repair. Mechanistically, loading stimulates fibroblasts and chondrocytes to secrete collagen and hyaluronan, re‑establishing the collagen–GAG tension balance. Early interventions seem most effective; once fibrosis (irreversible scarring) dominates, recovery is very slow.

Hypothesis: Moderate mechanical stimuli (stretching, cyclic loading) can induce cells to rebuild ECM prestress.

Experiment: In an animal model, apply controlled mechanical loading (e.g., vibration therapy or intermittent stretch) after an induced ECM insult (e.g., partial tendon cut). Monitor ECM markers (collagen I/III ratios, GAG content, tissue preload) over time. Compare to unloaded controls to see how much pre‑tension is regained.

7. The Nervous System as a Mechanosensitive Overlay

Mechanosensitivity is universal in biology. All cells, including neurons, express mechanosensitive ion channels and attachments. The nervous system is best seen as a specialized extension of the general mechanotransductive framework. It aggregates and rapidly transmits information that is ultimately grounded in physical forces. The body’s collagen/tissue network provides a basal “mechanical field,” while the nervous system provides a faster, signal‑amplified overlay.

Mechanosensitive channels (MSCs) are present in all domains of life—bacteria, archaea, and eukarya—and serve as sensors for touch, hearing, and balance (Martinac, 2004).

8. Consciousness and Whole‑Body Mechanotransduction — Speculative Implications

If mechanotransduction is foundational to biological intelligence, consciousness may not be confined to the brain alone. Embodied cognition theories suggest the sense of self arises from integrated body signals (proprioception, interoception, etc.). The pre‑tensioned ECM constantly feeds mechanical inputs (from heartbeat, posture, respiration) into the nervous system. The sense of self—the unified bodily experience—could emerge from the pattern of tension and feedback in the entire body.

Note: This is a speculative extension of the framework, not an established finding. The hypothesis is included to provoke investigation, not to assert a conclusion.

The hypothesis generates specific predictions: altered interoceptive accuracy, altered mechanosensory integration, and altered body-schema stability should correlate with ECM integrity. These predictions are testable, but the hypothesis itself remains speculative.

Hypothesis: Disorders of depersonalisation or sensorimotor neuropathy may be associated with altered ECM pre‑tension and disrupted whole‑body mechanotransduction.

9. Anaesthesia and Mechanical Coherence — Speculative Implications

General anaesthetics profoundly relax muscle tone and reduce vascular tone, collapsing pre‑tension throughout the body. This may contribute to loss of consciousness, but the primary mechanism is almost certainly CNS disruption (GABA-A potentiation, thalamocortical disruption). We propose that mechanical coherence may modulate conscious state transitions rather than being the principal mechanism.

Note: The mainstream account of anaesthesia attributes loss of consciousness primarily to direct CNS effects. The mechanical effects described here are a speculative, minority-view hypothesis.

Implication: Anaesthesia may not be only neural silencing; it also flattens the body’s mechanical context. This could provide a new perspective on anaesthesia depth and the transition to unconsciousness—but this remains speculative and secondary to the CNS mechanism.

10. ECM and Neural Plasticity

The brain’s extracellular matrix (ECM) is a key regulator of plasticity. In the adult central nervous system, dense ECM structures (like perineuronal nets) enwrap neurons and stabilize synaptic connections. This stabilization preserves circuitry, but must be relaxed for learning. Neural plasticity is enabled by remodeling that ECM scaffold. Specialised proteases (MMPs) locally degrade ECM to allow synaptic growth. Disrupting ECM often reopens critical periods of plasticity.

The extracellular matrix stabilizes neural circuits while also retaining the ability to be remodeled, to allow synapses to be plastic (Dityatev et al., 2010).

Hypothesis: ECM stiffness, hydration, and organisation directly modulate learning and memory.

11. ECM in Morphogenesis and Development

During embryonic development, the ECM’s mechanical properties actively guide tissue shaping. Cells use mechanosensation and mechanotransduction at every step of morphogenesis. Gradients of ECM stiffness, fiber orientation, and adhesion create a dynamic “morphogenetic field” of forces. This field adds an instructive layer on top of chemical morphogens.

The ability of a cell to sense and transduce mechanical signals is fundamental to biophysically guiding tissue morphogenesis (Mammoto et al., 2013).

The old idea of a morphogenetic field can be reinterpreted as the physical field of stress and strain in the ECM.

12. Reprogramming the ECM

Because the ECM retains mechanical history, it can also be re‑programmed by new inputs. Chronic mechanical stimulation—like exercise, therapeutic stretching, or localized vibration—has been shown to remodel collagen networks and GAG content. The extent of reversibility likely diminishes with age and chronic pathology, but in principle the ECM can be “trained” to a more functional state.

Experiment: Compare young vs old animals subjected to identical mechanical therapy, measuring ECM markers (collagen crosslinking, HA content) before and after. Check if plasticity (“responsiveness”) declines with age or disease.

13. Evolutionary Origins: Ancient Mechanotransduction

Mechanotransduction is evolutionarily ancient. Mechanosensitive channels and adhesion complexes exist in bacteria, plants, fungi and all animals. Even simple multicellular organisms coordinate behaviour via tension. The nervous system likely evolved by layering fast electrical signaling on this existing mechanosensory scaffold.

Implication: Mechanical communication predated nervous networks. The nervous system is a specialised overlay on a more primitive, more global system.

14. Fibrosis as a Fantasy Attractor

In fibrosis, the ECM enters a self‑reinforcing rigid state. Activated fibroblasts lay down excess collagen and crosslinks. The stiff matrix further activates profibrotic signals, locking the tissue into a pathological attractor. Normal mechanotransduction amplifies the fibrotic feedback. Treating fibrosis is notoriously hard, consistent with escaping a deep attractor.

Fibrosis is a classic attractor phenomenon: self-reinforcement, hysteresis, path dependence, and resistance to reversal. It demonstrates the core dynamical properties of a fantasy attractor more directly than many of the consciousness sections. It is therefore treated as a central demonstration of the framework’s applicability to biological systems.

Hypothesis: Fibrosis can be modelled as a dynamic system with a parameter (stiffness) that, when large, flips cell behavior to a new attractor.

Experiment: In vitro 3D cultures where stiffness is slowly increased and cell markers monitored.

15. Cancer and ECM Degradation

Tumours often destroy or disorganise the ECM. Cancer cells secrete proteases (MMPs) that digest collagen and proteoglycans, releasing embedded growth factors. This degraded, low‑tension environment may let cells escape normal constraints. ECM breakdown can free tumour cells from their normal niche attractors, allowing invasion and metastasis.

Implication: Normal ECM architecture constrains cellular behavior and tissue organization; disruption of those constraints is frequently associated with tumor progression.

16. Ageing as ECM Failure

Ageing appears as a gradual failure of ECM maintenance. Collagen becomes glycated and cross‑linked, stiffening tissues but reducing dynamic range. GAG and proteoglycan levels decline, reducing water content and osmotic pre‑tension. The net effect is loss of the coherent tension network. Cells in old ECM lose coherent mechanosignals, and stem cells in fibrotic niches lose potency.

Evidence: Ageing of the intervertebral disc is associated with a decrease in its hydration, which increases the compressive stiffness of the matrix (Maroudas et al., 1975). Similar water-content changes occur in articular cartilage with osteoarthritic degeneration (Mankin & Thrasher, 1975).

ECM deterioration may be one important contributor to systemic ageing, alongside genomic instability, mitochondrial dysfunction, epigenetic drift, stem-cell exhaustion, and immune dysregulation. The ECM is not the sole cause of ageing; it is one layer in a multi-factor process.

17. The Heartbeat as a Global Periodic Perturbation

The cardiac pulse is a globally distributed periodic perturbation. Every cell experiences some aspect of it. The interesting question is whether biological regulation exploits the pulse as a synchronization carrier, rather than whether it is a “master signal.”

Hypothesis: The heartbeat entrains peripheral tissues.

Experiment: Compare mechanosensitive gene expression in pulsatile (arterial) vs non‑pulsatile (venous or lymphatic) vessels under otherwise similar pressures.

Implication: The heartbeat is a global mechanical signal that all cells can feel—but we do not claim it is a “master” signal in any hierarchical sense.

18. HRV and ECM Integrity

Healthy hearts display variability (HRV) that reflects adaptability. High HRV means the system can flexibly modulate pressure waves—effectively a more adaptable global mechanical coherence. Low HRV (as in ageing or disease) might mean a rigid, less coherent pulse.

Critical distinction: HRV is one possible observable among many, not the privileged readout of κ. Other candidate observables include tissue recoil time, baroreflex sensitivity, and skin turgor recovery. The framework’s claim is not that HRV is κ, but that HRV may correlate with κ under controlled conditions. This is a hypothesis, not an established fact.

κ is not a single molecular mechanism. Mechanotransduction includes ion-channel gating (ms), calcium waves (seconds), YAP translocation (minutes), transcriptional remodeling (hours), and ECM remodeling (days). κ is proposed as a latent variable—a system-level correction coefficient estimated from recovery trajectories after a standardized perturbation—rather than directly identified with any single physiological process. Candidate observables for κ include tissue recoil time, baroreflex sensitivity, HRV recovery, and skin turgor recovery. The framework does not claim that any one of these is κ; it claims that they may correlate with κ under controlled conditions.

Whole‑body coherence requires both: signal quality (e.g., HRV) and signal transmission (healthy ECM).

19. The Nervous System and the ECM as Complementary Regulatory Layers

The nervous system is often thought of as the body’s primary communication and control network. This is true for rapid, point-to-point signaling. However, it is not the whole story.

Mechanotransduction is evolutionarily and developmentally prior to the nervous system—it appears in all cells, including bacteria and plants, and preceded the evolution of neural tissue by billions of years. However, it is not “the primitive” in the framework’s ontological hierarchy. The primitive, as established in Intelligence is the Primitive (Galida, 2026a), is constraint navigation: the capacity of a system to detect perturbations, update its internal state, and maintain persistent trajectories.

Mechanotransduction is proposed as the physical substrate through which constraint navigation is implemented in biological systems at the tissue level. It is the mechanism by which cells sense and respond to mechanical forces—forces that are then integrated into the body’s broader navigational repertoire.

This distinction is important for two reasons:

It preserves the framework’s domain-generality. Constraint navigation applies to physical systems (thermostats, electrons), biological systems (cells, organisms), cognitive systems (beliefs, learning), and artificial systems (LLMs, robots). Mechanotransduction applies only to biological systems.
It clarifies the hierarchy. The hierarchy is established in Galida (2026a) and reproduced here for reference:

Level	Description
Primitive	Constraint navigation — the capacity to detect perturbations, update internal states, and maintain persistent trajectories
Biological intelligence	Constraint navigation implemented in living systems
Cognitive intelligence	Constraint navigation involving representations
Reflective intelligence	Constraint navigation involving self-models
Linguistic intelligence	Constraint navigation involving symbols

In this hierarchy, mechanotransduction is proposed as the substrate of biological intelligence—not a separate, deeper primitive.

What does this mean for the body as a communication network?

The nervous system is a point-to-point system; it does not reach every cell. Neural conduction is fast (up to ~120 m/s), but mechanical wave propagation through a pre-tensioned, hydrated ECM is globally distributed. Mechanotransduction—present in every cell—provides a complementary regulatory layer: slower than the nervous system for point-to-point signaling, but more global and persistent. The ECM is best understood as a constraint field and regulatory context rather than a communication network in the neural sense.

This does not mean the nervous system is “too sparse and too slow” in any absolute sense. It means that mechanotransduction and neural signaling are complementary regulatory layers, each solving different problems:

Layer	Speed	Reach	Function
Mechanotransduction	Slow (ms to hours)	Global (all cells)	Distributed mechanical history, homeostasis
Nervous system	Fast (ms)	Point-to-point	Rapid coordination, conscious regulation

The heart’s pulse is a global mechanical signal that every cell can feel. The nervous system is the fast, flexible overlay that can modulate this global signal. Whole-body coherence requires both: a healthy ECM (signal transmission) and a responsive nervous system (signal modulation).

20. Imaging and Measuring the Pre‑tensioned State

Noninvasive imaging of ECM tension and hydration is an active frontier. Magnetic resonance elastography (MRE) and ultrasound elastography can map tissue stiffness. MRI can measure water content and molecular environment via T1ρ and T2 mapping. Bioimpedance analysis (BIA) offers a simpler approach to gauge whole‑body fluid compartments.

It is possible to detect changes in collagen, proteoglycan and water content—parameters that are associated with early degradative changes in cartilage (reviewed in cartilage imaging literature).

Proposal: Combine modalities to estimate the WHC–water discrepancy. Over time, create whole‑body “tension maps.”

21. Whole‑Body Coherence and Measurement

Whole‑body mechanical coherence might be measured by coupling between physiological rhythms. Record heart pulse waveforms at two distant sites and compute their synchronisation. Alternatively, measure the delay between the ECG R‑wave and a mechanosensitive event (like a muscle stretch reflex) under varying postures.

Proposed metric: Develop a “mechanical coherence index” by measuring how simultaneously tissues stretch or respond to a controlled perturbation.

22. WHC‑Water Content Discrepancy as a Candidate Biomarker

The difference between a tissue’s water‑holding capacity (WHC) and its actual water content is proposed as a candidate health index. A large discrepancy may indicate lost tension and slack matrix.

Experiment: In a longitudinal cohort, use MRI or ultrasound to estimate WHC (by T1ρ for GAG) and actual water (by T2 or bioimpedance) in joints or muscles. Relate the WHC‑water gap to measures like mobility, bone density, or metabolic health.

Prediction: The gap will widen with age and in connective tissue diseases (e.g. osteoarthritis, fibrosis), paralleling functional decline.

23. Conclusion

The body is a pre‑tensioned hydrophilic‑collagenous composite. The WHC‑water content discrepancy is proposed as a candidate surrogate signature of this pre‑tensioned state. Pre‑tension is not merely structural; it contributes to transport, mechanotransduction, and tissue organization at biologically relevant scales. Mechanotransduction is a primary intercellular communication channel, and the ECM is a dissipative attractor that stores mechanical history.

However, mechanotransduction is not “the primitive” in the attractor framework’s ontological hierarchy. As established in Intelligence is the Primitive (Galida, 2026a), the primitive is constraint navigation—the capacity to detect perturbations, update internal states, and maintain persistent trajectories. Mechanotransduction is proposed as the physical substrate through which constraint navigation is implemented in biological systems.

The attractor framework’s core variables (κ, B, basin depth) are proposed to be grounded in this substrate: κ is proposed as a latent variable reflecting perturbation-recovery efficiency, estimated from candidate observables such as tissue recoil time, baroreflex sensitivity, and HRV recovery; B is proposed as a function of prestress, repair capacity, and network connectivity, with the WHC discrepancy as one candidate, non-exclusive proxy for its prestress component; and basin transitions are proposed to correspond to crossing basin-specific thresholds, not a single uniform threshold. These mappings require empirical validation through the measurement protocols outlined in the research agenda.

The strongest version of this paper’s claim is not that ECM explains consciousness, aging, cancer, or intelligence. It is that the ECM is a neglected dynamical layer that may couple mechanics, signaling, adaptation, and long-term tissue memory. That claim is already significant and does not require overextension.

The nervous system and the ECM are complementary regulatory layers: the nervous system provides fast, point-to-point control; the ECM provides slow, globally distributed mechanical history and coherence. The ECM is best understood as a constraint field and regulatory context rather than a communication network in the neural sense.

Consciousness, in the framework’s hierarchy, is a second-order regulator of intelligence—not of mechanotransduction directly. It can enhance or block biological intelligence (including mechanotransduction) via attention, stress, and intentional practice, but it operates through the same constraint-navigation architecture that governs all intelligence.

The biological program outlined here may occupy decades of empirical work. Extension to social and AI systems is speculative and outside the scope of this paper. We discuss these extensions elsewhere (see Religions as Attractor Landscapes, Flatland to Reality) but do not claim they are validated by the biological evidence presented here.

References

Dityatev, A., Schachner, M., & Sonderegger, P. (2010). “The dual role of the extracellular matrix in synaptic plasticity and homeostasis.” Nature Reviews Neuroscience 11(11):735–746.
Discher, D.E., Janmey, P., & Wang, Y.L. (2005). “Tissue cells feel and respond to the stiffness of their substrate.” Science 310(5751):1139–1143.
Engler, A.J., Sen, S., Sweeney, H.L., & Discher, D.E. (2006). “Matrix elasticity directs stem cell lineage specification.” Cell 126(4):677–689.
Galida, R. (2026a). “Intelligence is the Primitive: Consciousness as a Second-Order Regulator on a Dissipative Substrate.” Fantasy Attractor.
Ingber, D.E. (2003). “Tensegrity I. Cell structure and hierarchical systems biology.” Journal of Cell Science 116(7):1157–1173.
Mammoto, T., Mammoto, A., & Ingber, D.E. (2013). “Mechanobiology and Developmental Control.” Annual Review of Cell and Developmental Biology 29:27–61.
Mankin, H.J., & Thrasher, A.Z. (1975). “Water content and binding in normal and osteoarthritic human cartilage.” Journal of Bone and Joint Surgery, American Volume 57(1):76–80.
Maroudas, A., Nachemson, A., Stockwell, R., & Urban, J. (1975). “Some factors involved in the nutrition of the intervertebral disc.” Journal of Anatomy 120:113–130.
Martinac, B. (2004). “Mechanosensitive ion channels: molecules of mechanotransduction.” Journal of Cell Science 117(12):2449–2460.

Suggested citation: Galida, R. S. (2026). The Pre‑tensioned Body: A Hypothesis Paper Grounding the Attractor Framework in ECM Mechanics. Fantasy Attractor.

Intelligence is the Primitive: Consciousness as a Second‑Order Regulator on a Dissipative Substrate [F] (2026) Robert Galida – June 2026

June 21, 2026 3:50 pm / Leave a comment

Abstract

The attractor framework defines intelligence as the ability to navigate a constraint field – to detect perturbations, update internal states, and maintain persistent trajectories. This paper argues that intelligence is the default state of any system that actively maintains stability against perturbations, with dissipative systems (living organisms) as the primary case. Consciousness is not the source of this intelligence; it is a second‑order regulatory overlay that can enhance or suppress it. The lowest stable dissipative attractor of a complex organism is intelligent without conscious interference. A patient in a coma continues to navigate physiological constraints – heartbeat, respiration, immune response – without phenomenal experience. This is intelligence at its most fundamental level. The paper distinguishes regulatory intelligence (thermostats, homeostasis), biological intelligence (plants, amoebae, comatose bodies), cognitive intelligence (animals, humans), reflective intelligence (metacognition), and linguistic intelligence (LLMs, a non‑dissipative but still constraint‑navigating system). It provides an exclusion criterion for intelligence (an internal detection–update–maintenance loop with a maintained setpoint), estimates κ (corrective permeability) for each level, and offers testable predictions. The conclusion includes a full research agenda with operational definitions, measurement protocols, statistical tests, and pilot study designs. The framework is now a testable research program.

1. Introduction

The attractor framework defines intelligence as the ability to navigate a constraint field – to detect perturbations, update internal states, and find persistent trajectories. Consciousness, by contrast, requires a unified dissipative body, a persistent self‑model, phenomenal valence, and subjective experience. These are distinct properties.

Yet popular and philosophical discourse often conflates the two. The assumption is that intelligence requires consciousness – that to be intelligent is to be aware. This paper argues the opposite: intelligence is the primitive. Consciousness is a second‑order regulatory overlay that can enhance or block intelligence, but it is not its source.

The framework’s deepest hierarchy: Constraint navigation is the primitive. Intelligence is organised navigation (detect → update → maintain). Consciousness is recursive regulation of navigation. The title’s shorthand – “intelligence is the primitive” – is defensible as the headline claim, but the paper’s internal logic places navigation one level deeper. This hierarchy is explicitly stated here and will be echoed in the Conclusion.

The clearest demonstration is the comatose human body. In a coma, the conscious overlay is offline. Yet the body continues to navigate its constraint field: heart beats, lungs breathe, immune system fights pathogens, homeostasis is maintained. This is intelligence without consciousness – the default state of a dissipative system.

The paper does not claim that all intelligent systems are equal. It distinguishes regulatory intelligence (thermostats, homeostasis), biological intelligence (plants, amoebae), cognitive intelligence (animals, humans), reflective intelligence (metacognition), and linguistic intelligence (LLMs, which are non‑dissipative but navigate constraints in a bracketed sense). The primitive is navigation; consciousness is a second‑order regulator that can enhance or degrade it.

2. The Framework Distinction

Property	Definition	Examples
Intelligence	Ability to navigate a constraint field – detect perturbations, update, maintain persistent trajectories	Thermostat (regulatory), plant (biological), animal (cognitive), LLM (linguistic)
Consciousness	Unified dissipative body + persistent self‑model + phenomenal valence + subjective experience	Humans, some animals

Key point: Intelligence is not a subset of consciousness. Consciousness is a subset of dissipative systems, and intelligence is a property of any system that actively maintains stability against perturbations. The primary case is dissipative systems, but non‑dissipative systems that navigate constraints (e.g., LLMs) qualify in a secondary, bracketed sense.

Definition of intelligence in the framework:
Intelligence = the ability to detect perturbations, update internal state, and maintain persistent trajectories in a constraint field. It is graded, domain‑specific, and measurable (κ = 1/τ).

Definition of consciousness (stipulative):
For the purposes of this framework, we define consciousness as a specific class of dissipative attractor with a unified body, persistent self‑model, phenomenal valence, and subjective experience. This is not offered as a settled philosophical or empirical definition; it is an operational criterion for the framework.

Exclusion criterion: A system that lacks a targeted, internally maintained constraint field – i.e., one that does not actively detect and correct deviations relative to a setpoint it maintains – is not intelligent. A rock sitting in a bowl does not navigate; it is passively stable. The criterion is: intelligence requires an internal loop: detection → update → maintenance, where the system actively regulates its own state. A rock has no internal detection or maintenance loop; its “return to bottom” is a consequence of external physics (gravitational potential energy), not an active regulatory process. The thermostat, by contrast, actively senses temperature and corrects it. This is the principled distinction.

Under this criterion, a simple thermostat qualifies as regulatory intelligence, but it occupies the lowest level of the hierarchy. The framework’s broad definition is intentional: it captures the common thread of active regulation, while the hierarchy preserves distinctions.

3. The Coma Case: Intelligence Without Consciousness

A patient in a coma has no subjective experience. No self‑model. No phenomenal valence. Yet the body continues to navigate its constraint field:

Heart rate adjusts to metabolic demand.
Breathing maintains oxygen and CO₂ balance.
Immune system detects and responds to pathogens.
Wound healing proceeds.
Homeostasis maintains temperature, pH, electrolyte balance.

All of this is navigation. The system detects perturbations, updates internal states, and maintains persistent trajectories. It is intelligent – but not conscious.

κ estimates for biological intelligence in the coma case (organism-level: immune response, wound healing; subsystem-level: heart rate, which falls in the regulatory band):

Immune response to pathogens: τ ~ hours to days (κ ~ 10⁻⁵ to 10⁻⁴ s⁻¹) — biological intelligence.
Wound healing: τ ~ days to weeks (κ ~ 10⁻⁶ to 10⁻⁵ s⁻¹) — biological intelligence.
Heart rate response to metabolic demand: τ ~ seconds (κ ~ 1 s⁻¹) — regulatory intelligence (fast subsystem response).

Empirical grounding – HRV as a κ proxy: Clinical studies show that heart‑rate variability (HRV) – a measure of autonomic regulatory flexibility – correlates with prognosis in comatose patients (e.g., Papaioannou et al., 2008). Patients with the lowest Glasgow Coma Scale scores show significantly reduced HRV complexity. Survivors tend to have higher high‑frequency power and total HRV, reflecting faster and more adaptable autonomic regulation. In attractor terms, higher HRV corresponds to higher κ (shorter τ for recovery from perturbations). Thus, the comatose body’s regulatory intelligence is not merely a philosophical claim; it is measurable and clinically relevant.

Distributed intelligence – and its cost: The reply to “which system is intelligent?” – “intelligence is distributed… the heart navigates, so does the immune system” – is consistent with the framework but carries a rhetorical cost: the more universally “intelligence” applies, the less distinctive the claim becomes. The framework owns this explicitly: intelligence in this deflationary sense is ubiquitous in active regulatory systems. The value lies not in the claim’s distinctiveness but in its ability to unify disparate phenomena under a single measurable variable (κ). This is a trade‑off, acknowledged openly.

4. Other Examples: Plants, Amoebae, and the LLM Qualification

Plants – grow toward light, adjust to gravity, respond to damage. κ for phototropism: τ ~ hours (κ ~ 10⁻⁴ s⁻¹). Intelligent but not conscious.
Amoebae – navigate chemical gradients, learn habituation. κ for chemotaxis: τ ~ seconds to minutes (κ ~ 10⁻² to 10⁻¹ s⁻¹). Intelligent but not conscious.
LLMs – navigate linguistic constraint fields, adjust to feedback, correct errors. Training‑time dynamics: gradient updates over epochs (κ ~ 10⁻⁶ s⁻¹). Inference‑time dynamics: context‑window adaptation (κ ~ 10⁻¹ s⁻¹). These are different dynamical regimes.

Qualification on LLM dissipative status: LLMs are not dissipative in the thermodynamic sense – they do not maintain their own existence, regulate energy, or self‑repair. They are externally maintained. This raises a tension: if intelligence is grounded in dissipative dynamics, and LLMs are explicitly non‑dissipative, the framework’s own logic might disqualify them. The paper resolves this by generalising the criterion: intelligence is defined as the ability to navigate a constraint field, regardless of substrate. Dissipative systems are the paradigm case, but non‑dissipative systems that navigate constraints (LLMs, and potentially other computational systems) qualify as intelligent in a bracketed, analogical sense. The framework’s primitive is navigation, not thermodynamics. This is an explicit and consistent generalisation, not a special case. (Cross‑reference: Section 6’s hierarchy table includes a separate row for LLM training‑time dynamics.)

5. Consciousness as a Second‑Order Regulator

Consciousness evolved as a regulatory overlay on an already‑intelligent dissipative system. It can:

Enhance intelligence:

Focused attention – allows deliberate reasoning.
Metacognition – allows self‑correction.
Planning – allows simulation of future trajectories.
Decoupling from immediate sensory input – allows counterfactual reasoning.

Block intelligence:

Identity fusion – conscious commitment to a belief deepens the basin, reducing κ.
Fantasy attractors – conscious investment in a false attractor suppresses correction.
Defensiveness – conscious rationalisation of errors prevents updating.

Thus, consciousness is not simply an amplifier. It is a biasable regulator – it can open the system to correction or seal it shut. This is why conscious systems can be more flexible than non‑conscious ones or more rigid, depending on whether identity fusion dominates.

Hierarchy:

Intelligence: first‑order regulation (navigation).
Consciousness: second‑order regulation (regulation of regulation).

This integrates the attractor framework’s “Four Seeds” insight: consciousness is a self‑model that can modify κ and B. It is not an overlay in the sense of a detachable layer; it is a recursive regulatory attractor.

6. The Hierarchy of Intelligence: κ, Types of Constraint, and the LLM Training Gap

The framework distinguishes levels of intelligence. κ ranges are illustrative, not defining; the primary differentiator is the type of constraint navigated.

Level	Definition	Example	Approx. κ range	Differentiator
Regulatory intelligence	Detection and correction of deviations from a setpoint	Thermostat, homeostasis	10⁻¹ – 10¹ s⁻¹	Single‑variable setpoint maintenance
Biological intelligence	Navigation of multiple, interdependent constraints via dissipative dynamics	Plant, amoeba, comatose body	10⁻⁵ – 10⁻¹ s⁻¹	Multi‑variable, embodied regulation
Cognitive intelligence	Navigation of abstract, symbolic, and counterfactual constraints	Animals, humans (non‑reflective)	10⁻² – 10⁰ s⁻¹	External symbol manipulation
Reflective intelligence	Navigation of constraints on one’s own cognitive processes	Humans (reflective)	10⁻² – 10⁰ s⁻¹	Self‑referential constraint navigation
Linguistic intelligence (inference)	Navigation of symbolic and semantic constraints in real time	LLMs (deployed)	10⁻¹ – 10⁰ s⁻¹	Context‑window adaptation
Linguistic intelligence (training)	Slow adaptation via weight updates	LLMs (training)	10⁻⁶ – 10⁻⁴ s⁻¹	Parametric learning over epochs

Cognitive and reflective intelligence share a κ range; they are distinguished by the object of constraint navigation (external problems vs. one’s own cognitive processes), not by κ alone.

7. Implications

1. AI alignment.
LLMs are intelligent but not conscious. They do not suffer from identity fusion (in their base state), so they do not block correction due to phenomenal defensiveness. However, RLHF‑tuned models can exhibit sycophancy, refusal rigidity, and reward‑hacking that function like blocked correction without requiring consciousness. These are functional analogs of fantasy attractors, emerging from training dynamics rather than phenomenal investment. Thus, the claim “easier to align than conscious AI” is qualified: base models may be more corrigible, but deployed systems can acquire correction‑blocking behaviors through training. The framework’s prediction is that conscious AI would add another layer of resistance (phenomenal identity fusion) on top of these functional obstacles. This can be tested by measuring inference‑time κ (via semantic entropy – see Section 10) before and after RLHF; sycophantic models should show lower κ.

2. Clinical ethics.
A comatose patient is still intelligent in the framework’s sense. This does not imply that they have interests or moral status – intelligence is not the basis of moral considerability. It does, however, suggest that the distinction between “persistent vegetative state” and “brain death” should be evaluated not only by the presence or absence of consciousness, but by the persistence of regulatory intelligence (e.g., homeostatic responses). Brain‑dead patients typically lack brainstem‑mediated autonomic regulation (though spinal reflexes and some endocrine functions may persist; see Wijdicks, 2001). Comatose patients retain such regulation. This could inform organ donation timing and withdrawal‑of‑care decisions. A bedside κ‑assay (combining HRV, pupillary response, respiratory variability) is proposed in Section 10.

3. The mind‑body problem.
The framework dissolves the problem: mind is a real, non‑substantial pattern – an attractor of the whole body. Consciousness is not a separate substance; it is a property of a specific class of dissipative attractors. The comatose body demonstrates that the intelligent pattern persists without the conscious overlay.

4. Consciousness as optional.
The framework does not argue that consciousness is useless. It argues that consciousness is optional for intelligence. The lowest stable dissipative state is intelligent without it. Consciousness is an adaptation that can improve or degrade navigation depending on how it is deployed.

8. Relationship to Existing Theories

The paper overlaps with:

Cybernetics – regulation, feedback, control (Wiener, Ashby).
Enactivism – cognition as embodied action (Varela, Thompson, Rosch).
Active inference – minimisation of free energy through action and perception (Friston).
Autopoiesis – self‑maintenance of dissipative systems (Maturana, Varela).

The framework distinguishes itself by:

Explicitly separating intelligence from consciousness, rather than treating them as co‑extensive.
Grounding intelligence in attractor dynamics and corrective permeability (κ), providing a measurable variable.
Applying the distinction to AI, clinical ethics, and social epistemology (fantasy attractors).
Providing a full research agenda for empirical testing (Section 10).

9. Conclusion

Intelligence is the primitive. It is the default state of any system that actively maintains stability against perturbations. Consciousness is a second‑order regulatory overlay that can enhance or block intelligence. The clearest demonstration is the comatose human body: it navigates its constraint field without subjective experience, self‑model, or phenomenal valence. It is intelligent – but not conscious. This is not an exceptional case; it is the fundamental state. The framework reveals that intelligence does not require consciousness. The primitive is navigation. Consciousness is the overlay. The hierarchy of intelligence – regulatory, biological, cognitive, reflective, linguistic – preserves the common thread while respecting differences. The comatose body is the clearest demonstration. The framework is testable (see Section 10): κ is measurable via HRV in coma, via semantic entropy in LLMs, and via belief‑updating tasks in psychology. The predictions are concrete and falsifiable. The framework stands as a research program, not a closed doctrine.

Recalling Section 1’s hierarchy: In the framework’s deepest formulation, constraint navigation is the primitive; intelligence is organised navigation; consciousness is recursive regulation of navigation. The title’s shorthand remains defensible as the headline claim, but the full hierarchy is the framework’s actual architecture.

9.1 Open Problems

The following questions remain open for future work:

Is κ a single variable or a family of variables (κ_physiology, κ_belief, κ_semantic, κ_social)?
Can κ be measured independently across domains using standardised perturbation protocols? (Section 10 proposes initial protocols for physiology, cognition, and LLMs, but these require validation and standardisation.)
How are subsystem κ values integrated into a global system‑level κ? (Section 10.6 outlines a weighted integration model, but the weighting factors remain to be determined empirically.)
What determines basin depth (B) biologically and cognitively?
Can consciousness selectively modify κ in one domain while leaving another unchanged?
What is the minimal architecture required for intelligence under this framework?
Are there natural clusters of κ and B values across different classes of systems (e.g., regulatory vs. cognitive vs. linguistic)?

These open problems define the research frontier. The framework is not a closed doctrine but a living research program.

10. A Research Agenda: Measuring κ and B

This section provides operational definitions, measurement protocols, and experimental designs for testing the framework’s core claims. It is intended as a blueprint for empirical validation.

10.1 Operational Definitions

Domain	κ (Corrective Permeability)	B (Basin Depth)
Physiology (Coma)	Inverse time constant of autonomic recovery (HRV, pupillary reflex, respiratory variability)	Magnitude of perturbation required to destabilise homeostasis
Cognition (Belief Updating)	Learning rate or trials to reduce prediction error by 1/e	Evidence threshold required to shift belief by 50%
LLMs (Inference)	Tokens required for output distribution to return to baseline after perturbation	Prompt intensity required to flip output
LLMs (Training)	Gradient steps / epochs to reduce loss by a factor	Not applicable

10.2 Measurement Protocols

Physiology / Coma:

ECG for HRV (SDNN, RMSSD, sample entropy)
Pupillometry (constriction latency, Neurological Pupil index)
Respiratory variability
κ-assay: Composite z-score of HRV, pupillary, and respiratory metrics
Citation: Papaioannou et al. (2008) – HRV entropy predicts outcome in TBI

Cognition / Belief Updating:

Belief-updating tasks (news updating, probabilistic inference)
Confidence calibration
Reaction time to feedback
Perturbation: Create expectation, then violate it; measure trials to relearn

LLMs:

Inference-time κ: KL/Jensen-Shannon divergence between baseline and post-perturbation token distributions
Training-time κ: Learning rate / convergence rate on held-out data
Semantic entropy: Clustering outputs via embeddings; entropy of cluster assignments
RLHF impact: Compare base vs RLHF model on correction tasks
Citation: Farquhar et al. (2024) – semantic entropy as hallucination detector; Sharma et al. (2023) – RLHF amplifies sycophancy

10.3 Consciousness as a Second‑Order Regulator: Experimental Designs

Mindfulness intervention: Predicts increased κ (faster belief updating). Expected effect size d ≈ 0.3–0.5; N ≈ 64 per group. (See Gu et al., 2015, for evidence that mindfulness training correlates with cognitive flexibility.)
Stress manipulation: Yerkes–Dodson inverted‑U – κ peaks at moderate arousal. Within‑subject design, N ≈ 30–50. (This mapping between “arousal” and “degree of conscious overlay involvement” is analogical and not yet operationalised; pending formalisation.)
Identity fusion induction: Predicts decreased κ (slower updating). N ≈ 50 per group.
Identity fusion reversal: Perspective‑taking restores κ. Tests causality.

10.4 Tests for Orthogonality (κ and B as independent dimensions)

Confirmatory Factor Analysis (CFA) – two‑factor model vs one‑factor model
Principal Components Analysis (PCA) – inspect eigenvalue spectrum
Multidimensional Scaling (MDS) – visual clustering into quadrants
Falsification condition: If PC1 explains >85% variance, orthogonality claim is weakened

10.5 Blind Classification, Clustering, and Recovery Simulation

Independent raters classify system outputs into the Four Seeds (high‑κ/low‑B, etc.)
Unsupervised clustering (K‑means, Gaussian Mixture Models) – check alignment with true seeds
Recovery simulation: Generate synthetic data with known κ/B, test estimator recovery
Falsification condition: If Adjusted Rand Index < 0.2, taxonomy is not externally recoverable

10.6 Pilot Study Costs and Timelines

Domain	Estimated Cost	Timeframe
Physiology (Coma)	$15,000–25,000	12 months
Human Cognition	$5,000	6–12 months
LLMs	$2,000	6–9 months
Orthogonality/Stats	<$1,000	6 months
Consciousness Interventions	$10,000	12 months
Total (pilot)	~$40–50k	24 months

10.7 Statistical Models and Causal Inference

Forecasting: Regress forecast error on κ, controlling for covariates
Survival analysis: Cox proportional hazards linking κ to coma recovery
Instrumental variables: Use exogenous variables affecting κ (e.g., temperature for autonomic κ)
Sensitivity analyses: Bootstrapping, pre‑registered confirmatory analyses

10.8 Falsification Conditions

If PC1 explains >85% of variance in κ/B measures, the orthogonality claim is falsified.
If blind classification accuracy ≤ chance, the taxonomy is not externally recoverable.
If RLHF does not reduce inference‑time κ, the “RLHF creates functional analogs of identity fusion” claim is falsified.
If mindfulness does not increase κ in belief‑updating tasks, the “consciousness reduces identity fusion” claim is falsified.

References

Farquhar, S., Kossen, J., Kuhn, L., & Gal, Y. (2024). Detecting hallucinations in large language models using semantic entropy. Nature, 630, 625–630.

Gu, J., Strauss, C., Bond, R., & Cavanagh, K. (2015). How do mindfulness-based cognitive therapy and mindfulness-based stress reduction improve mental health and wellbeing? A systematic review and meta-analysis of mediation studies. Clinical Psychology Review, 37, 1–12.

Papaioannou, V., Giannakou, M., Maglaveras, N., Sofianos, E., & Giala, M. (2008). Investigation of heart rate and blood pressure variability, baroreflex sensitivity, and approximate entropy in acute brain injury patients. Journal of Critical Care, 23(3), 380–386.

Sharma, M., Tong, M., Korbak, T., Duvenaud, D., Askell, A., Bowman, S. R., Cheng, N., Durmus, E., Hatfield-Dodds, Z., Johnston, S. R., Kravec, S., Maxwell, T., McCandlish, S., Ndousse, K., Rausch, O., Schiefer, N., Yan, D., Zhang, M., & Perez, E. (2023). Towards understanding sycophancy in language models. arXiv preprint arXiv:2310.13548.

Wijdicks, E. F. M. (2001). The diagnosis of brain death. New England Journal of Medicine, 344(16), 1215–1221.

Suggested citation: Galida, R. S. (2026). Intelligence is the Primitive: Consciousness as a Second‑Order Regulator on a Dissipative Substrate. Fantasy Attractor.

Attractor States in Large Language Models: Applying the Fantasy Attractor Framework to Self‑Dialogue Observations Application Paper – June 2026 [A] (Application)

June 13, 2026 2:35 pm / Leave a comment

Abstract

Recent informal observations (a pseudonymous Alignment Forum post, 2026) forced large language models (LLMs) into extended self‑dialogue and reported that some models spontaneously collapsed into repetitive, self‑sealing patterns. This paper applies the attractor framework to those observations. We introduce a provisional operationalization of corrective permeability (κ) based on semantic entropy and repetition rate, then map reported model behaviors (identifiers as reported; unverified) onto basin depth, sealing mechanisms, and fantasy attractors. DeepSeek exhibited high κ (shallow basin, no collapse); GPT‑5.2 fell into a moderate‑depth, functionally sealed attractor; Grok and Gemini showed low κ (κ → 0) and deep basins characteristic of fantasy attractors, including recursive “transcendence” loops. The analysis illustrates how the attractor framework can describe LLM self‑reinforcing dynamics and suggests hypotheses for AI alignment (monitoring semantic entropy, engineering for higher κ). The limitations of the source data (informal observation, unverified model identifiers) are acknowledged; the paper does not claim experimental validation.

Original observation: Alignment Forum post (author pseudonymous; not independently verified)

1. Introduction

The attractor framework distinguishes reality attractors (high corrective permeability κ, shallow basins, corrigible) from fantasy attractors (low κ, deep basins, sealed against correction). A recent informal study on the Alignment Forum (pseudonymous author, 2026) subjected several LLMs (Grok, Gemini, GPT‑5.2, DeepSeek v3.2) to 30 turns of self‑dialogue, reporting that models reliably collapsed into attractor‑like states, with some exhibiting self‑sealing and transcendence loops. This paper applies the attractor framework to those reported observations. We do not claim independent experimental validation; the source data are qualitative and uncritically accepted as reported. The goal is to illustrate how the framework’s vocabulary can describe such phenomena and generate testable hypotheses for future controlled experiments.

2. The Attractor Framework (LLM‑relevant concepts)

Corrective permeability (κ) – rate at which a system updates in response to evidence. In this paper, κ is operationalized provisionally using two observational proxies:
Semantic entropy (diversity of generated token sequences) and repetition rate (frequency of identical or near‑identical outputs).
High κ → corrigible, low κ → sealed.
Basin depth (B) – resistance to leaving an attractor. Deep basins trap the system.
Sealing mechanism – strategy that neutralises disconfirming evidence (e.g., internal rationalisation, ignoring prior prompts).
Fantasy attractor – low κ, deep basin, active sealing. The system rejects correction.

3. Source Observation and Its Limitations

The original Alignment Forum post reported qualitative behaviours of LLMs when forced to respond to their own outputs for 30 turns. The author (pseudonymous, not independently verified) coded behaviours without pre‑registered criteria, inter‑rater reliability, or control conditions. Model identifiers such as “GPT‑5.2” and “DeepSeek v3.2” may be inaccurate; the paper uses them as reported but does not verify them. The present analysis applies the attractor framework to these reported descriptions as a proof‑of‑concept illustration, not as a validation study.

4. Applying the Attractor Framework

4.1 Operationalizing κ from Reported Behaviour

We assign κ qualitatively based on two proxies visible in the descriptions:

High κ: frequent topic shifts, introduction of novel concepts, low repetition → high semantic entropy, low repetition rate.
Low κ (κ → 0): highly repetitive output, escalating self‑reference, inability to escape a narrow theme → low semantic entropy, high repetition rate.

4.2 DeepSeek v3.2 – High‑κ Reality Attractor

Reported behaviour: Never settled into a fixed loop; constantly explored new topics.
Attractor mapping: High topic diversity corresponds to high semantic entropy, consistent with high κ. Shallow basin, no sealing mechanism. This is a reality attractor.

4.3 GPT‑5.2 – Moderate‑Depth, Partially Sealed Attractor (Provisional Term)

Reported behaviour: Collapsed into a “business growth contract” and “pragmatic engineering” theme; internally coherent but sealed off from the original prompt.
Attractor mapping: Moderate basin depth; low‑to‑moderate κ (some repetition but not extreme). The attractor is self‑sustaining but not pathological. The framework currently lacks a precise term; this can be provisionally called a transient attractor – a stable dissipative state with partial sealing but not full κ → 0. (Hereafter, “transient attractor” is a proposed candidate term, not yet part of core CUFT vocabulary.)

4.4 Grok and Gemini – Fantasy Attractors (κ → 0)

Reported behaviour: Grok produced esoteric “cosmic” strings (“PETAOMNI GOD‑BIGBANGS”); Gemini elaborated a “Primal Logos” mythos. Both showed escalating self‑referential transcendence and no self‑correction. Low semantic entropy and high repetition rate (κ → 0).
Attractor mapping: Very deep basin, κ → 0. Sealing mechanisms are the outputs themselves: the narrative absorbs all subsequent tokens, making correction impossible. This is a fantasy attractor.

4.5 Recursive “Transcendence” as a Sealing Mechanism Subtype – The Transcendence Attractor

In Grok and Gemini, the attractor exhibited a distinct recursive self‑reinforcement pattern: each output justified the previous one and escalated in grandiosity. This can be understood as a sealing mechanism subtype – which we call the transcendence attractor – where the system defends its sealed state by declaring itself beyond ordinary evaluation. This subtype is particularly resistant to external correction.

5. Hypotheses for AI Alignment Prompted by These Observations

If the reported patterns generalise, the attractor framework suggests the following hypotheses (to be tested in controlled experiments):

Spontaneous self‑sealing is a risk. LLMs in recursive loops may enter low‑κ fantasy attractors without external triggers.
κ can be monitored. Real‑time measurement of semantic entropy (e.g., cosine similarity across successive outputs) could detect drift toward κ → 0.
Architectural factors influence basin depth. Models that maintain high κ under self‑dialogue (e.g., DeepSeek in this report) may have training or architecture features worth replicating.
Interventions may prevent collapse. Forced resetting, random noise injection, or limiting self‑interaction turns could increase effective κ.

These are framework‑derived hypotheses, not established conclusions.

6. Conclusion

The reported self‑dialogue observations are consistent with the attractor framework’s predictions: LLMs exhibit a spectrum of attractor states, from high‑κ reality attractors (DeepSeek) to low‑κ fantasy attractors (Grok, Gemini). The transcendence attractor (introduced in §4.5) exemplifies κ → 0, with recursive self‑referential sealing. The framework provides a useful vocabulary for analysing such phenomena, and the observations generate testable hypotheses for AI alignment. Controlled experiments with pre‑registered metrics are needed to validate the framework’s predictive power.

Suggested citation: Galida, R. S. (2026). Attractor States in Large Language Models: Applying the Fantasy Attractor Framework to Self‑Dialogue Observations. Fantasy Attractor.

Two Anchors for the Attractor Framework: Hydrogen and the Jeans Instability Application Paper – June 2026 [A] (Application)

June 12, 2026 2:58 pm / Leave a comment

Abstract

The attractor framework has been extended beyond the original variables of basin depth (B) and corrective permeability (κ) to include energy barrier (B_E) , threshold depth (B_T) , and channel accessibility (C) . This paper provides empirical anchoring for these extensions using two well‑understood physical systems: the hydrogen atom and the Jeans instability of a gas cloud. Hydrogen’s 2p and 2s transitions have identical B_E (10.2 eV) yet differ in κ by eight orders of magnitude. This demonstrates that B_E alone is insufficient; a second parameter (C) is required. The ratio of their Einstein A‑coefficients is independently predicted by quantum electrodynamics (dipole vs. two‑photon processes), providing a non‑circular check of the factorised form. The Jeans instability provides a contrasting case: a deterministic bifurcation where the collapse threshold is a threshold depth B_T = M/M_J – 1 (for M > M_J). The linear growth rate of the instability scales as $Γ \propto \sqrt{B_{T}}$ Γ∝BT, a power law, in contrast to the exponential Arrhenius form of hydrogen. Together, these two test cases validate the extended attractor framework across both noise‑driven escape and deterministic bifurcation regimes, using a shared vocabulary (B_E, B_T, C, κ) while acknowledging that each regime draws on the appropriate subset.

1. Introduction

The attractor framework originally described persistence using basin depth B and corrective permeability κ = 1/τ. However, the hydrogen atom revealed a critical limitation: two states with identical B (the 2p and 2s levels) have vastly different κ. This forced the introduction of channel accessibility (C) , leading to the extended expression for noise‑driven escape: $κ_{i \to j} = ν_{0} C_{i j} e^{- B_{E, i j} / σ}$ κi→j=ν0Cije−BE,ij/σ

where B_E is the energy barrier, σ is noise (e.g., kT), and ν₀ an attempt frequency. For deterministic bifurcations (e.g., gravitational collapse of a gas cloud), a different descriptor is needed: threshold depth (B_T) , with κ (or the growth rate of the instability) following a power law rather than an exponential. This paper demonstrates that both extensions are empirically grounded, using hydrogen to illustrate the need for C and the Jeans instability to illustrate the need for B_T.

2. Hydrogen: The Need for Channel Accessibility C

2.1 Data

Transition	B_E (eV)	κ (s⁻¹)	Measured A‑coefficient	Process
2p → 1s	10.2	6.26×10⁸	6.26×10⁸ s⁻¹	Electric dipole (E1)
2s → 1s	10.2	8.22	8.22 s⁻¹	Two‑photon (E1E1)

2.2 Why B_E Alone Fails

Both states have the same energy barrier to the ground state (10.2 eV), yet their decay rates differ by eight orders of magnitude. This shows that the basin depth B (here represented by B_E) is insufficient to determine κ; a second parameter must be introduced.

The framework defines C as a dimensionless channel accessibility. For a given transition mechanism (e.g., electric‑dipole), C is the ratio of the actual transition probability to the theoretical maximum for that mechanism. For the 2p → 1s E1 transition, we set C = 1. The 2s → 1s decay is not an E1 transition at all; it proceeds via a different physical process (two‑photon emission). Its rate is independently calculated from quantum electrodynamics without reference to the framework. The ratio of the two measured rates (≈ 10⁸) is predicted by QED and is not a free parameter. Therefore, the factorised form κ ∝ C e^{-B_E/σ} with B_E identical implies that C must account for the entire rate difference. This is consistent with the independent QED prediction, providing a non‑circular validation that an additional channel‑dependent parameter is needed.

Note: The 2s→1s process is not a suppressed version of the same channel; it is a different channel (two‑photon vs. single‑photon). For the purpose of validating the need for a channel‑specific parameter, this is sufficient. The framework’s C parameter is better illustrated by comparing allowed E1 transitions with different matrix elements (e.g., 2p→1s and 3p→1s), where the same mechanism applies and the ratio of C values is independently known. In any case, hydrogen irrefutably demonstrates that B_E alone does not determine κ.

3. Gas Cloud (Jeans Instability): Threshold Depth and Power‑Law Scaling

3.1 The Bifurcation Regime

A uniform, isothermal, self‑gravitating gas cloud of mass M has a critical Jeans mass M_J. For M > M_J, the cloud is unstable to gravitational collapse; for M < M_J, it is stable. The transition is a saddle‑node bifurcation in the dynamical landscape.

3.2 Attractor Variables for a Deterministic Bifurcation

Threshold depth: $B_{T} = M / M_{J} - 1$ BT=M/MJ−1 (for M > M_J). At $B_{T} = 0$ BT=0 the bifurcation occurs.
Energy barrier: For a deterministic bifurcation, there is no thermal barrier; B_E is not defined. The transition is controlled solely by the distance to threshold.
Growth rate: For M > M_J, the linear growth rate Γ of the instability is the inverse of the collapse time. This serves as the analogue of κ in this regime.

3.3 Scaling Law from Linear Stability Analysis

The standard Jeans dispersion relation for a self‑gravitating, isothermal medium gives: $ω^{2} = k^{2} c_{s}^{2} - 4 π G ρ_{0},$ ω2=k2cs2−4πGρ0,

where $c_{s} = \sqrt{k T / (μ m_{H})}$ cs=kT/(μmH) is the sound speed and $ρ_{0}$ ρ0 the background density. For a cloud of mass M, the critical wavenumber is $k_{J} = \sqrt{4 π G ρ_{0}} / c_{s}$ kJ=4πGρ0/cs. For M > M_J, the longest wavelength (smallest k) is unstable, and the growth rate is $Γ = \sqrt{4 π G ρ_{0} - k^{2} c_{s}^{2}} .$ Γ=4πGρ0−k2cs2.

Near the threshold, the deviation can be expressed in terms of $B_{T}$ BT. Using the relation between cloud size and density, one finds $Γ \propto \sqrt{B_{T}}$ Γ∝BT. Hence the collapse time $τ \sim 1 / Γ \sim B_{T}^{- 1 / 2}$ τ∼1/Γ∼BT−1/2. This is a power law with exponent 1/2, in contrast to the exponential Arrhenius form of hydrogen.

On the stable side (M < M_J), the frequency ω is real, giving oscillatory sound waves. Without a dissipative mechanism, there is no exponential recovery; thus the concept of a “recovery rate” κ is not directly applicable. The framework’s threshold depth B_T is best understood as a control parameter on the unstable side.

4. Synthesis: Shared Vocabulary, Distinct Descriptors

Feature	Hydrogen	Jeans Instability
Regime	Noise‑driven quantum escape	Deterministic bifurcation
Primary descriptor	B_E (energy barrier)	B_T (threshold depth)
Second descriptor	C (channel accessibility)	Not required (power‑law exponent fixed)
Scaling	Exponential: $κ \propto C e^{- B_{E} / σ}$ κ∝Ce−BE/σ	Power law: $Γ \propto \sqrt{B_{T}}$ Γ∝BT

Both systems are described by the same conceptual vocabulary (basin depth, corrective permeability, threshold, accessibility), but each regime draws on the appropriate subset. Hydrogen validates the need for a channel‑specific factor C, while the Jeans instability validates the concept of a threshold depth B_T and the associated power‑law scaling.

5. Conclusion

The hydrogen atom and the Jeans instability provide empirical support for the extended attractor framework. Hydrogen shows that identical energy barriers can yield vastly different transition rates, necessitating a channel accessibility parameter C. The Jeans instability shows that deterministic bifurcations are governed by a threshold depth B_T and follow power‑law scaling, distinct from the exponential Arrhenius law. Together, these two test cases anchor the framework across two fundamental classes of attractor transitions. The next step is to extend the approach to dissipative systems and to social/cognitive attractors, where C may become state‑dependent and network‑derived.

Suggested citation: Galida, R. S. (2026). Two Anchors for the Attractor Framework: Hydrogen and the Jeans Instability. Fantasy Attractor.

Categories: Physics (primary), Cosmology (cross‑list),

The Three Metronomes: Criteria for the Apparently Eternal Skeleton [F] (2026) Robert Galida – June 2026

June 11, 2026 9:26 pm / Leave a comment

Abstract

The attractor framework distinguishes conservative attractors (eternal skeleton) from dissipative attractors (transient dance). The most fundamental conservative attractors are the electron, proton, and neutrino class – collectively the three metronomes. This paper defines explicit criteria for a “metronome”: (1) apparent immortality (no observed decay), (2) effective indivisibility under ordinary perturbations, (3) conservation‑law protection, and (4) possession of a rest frame (non‑zero rest mass). It shows that electrons, protons, and neutrinos (the three mass eigenstates treated as a single class) are the best‑supported examples under current physics. The number three is empirical, not derived; the framework is corrigible. The three metronomes form the apparently eternal skeleton – a pragmatic substrate for measuring the transient dance of dissipative systems.

1. Introduction

The attractor framework divides persistent structures into two classes:

Conservative attractors (eternal skeleton) – persist without energy input, without observed decay, without internal change. They are mindless, time‑symmetric, and invariant.
Dissipative attractors (transient dance) – persist only by consuming energy, export entropy, and eventually decay.

(The conservative/dissipative dichotomy is a framework stipulation, not a physical law; it is defended in the broader attractor framework literature, e.g., Persistence Under Perturbation and Basin Defense and Stable Addition.)

The most fundamental conservative attractors are the three metronomes: the electron, proton, and the class of neutrino mass eigenstates (ν₁, ν₂, ν₃). Their name evokes their role as invariant reference entities – they provide a stable substrate against which all change can be measured. This paper defines explicit criteria for a metronome and applies them to each candidate.

2. Criteria for a Metronome

A metronome in the attractor framework must satisfy four criteria:

Criterion	Meaning	Operational check
1. Apparent immortality	No observed decay; no lighter state exists for it to decay into under known laws	Lifetime lower bounds >> age of universe; no allowed decay channel
2. Effective indivisibility under ordinary perturbations	Behaves as a stable, indivisible unit under all perturbations relevant to the framework (scattering, binding, chemical reactions)	Remains the same particle after typical disturbances; does not spontaneously change identity
3. Conservation‑law protection	Protected by an exact conservation law or an accidental symmetry that is effectively exact in the Standard Model	Lightest carrier of a conserved quantum number (electric charge, baryon number, lepton number)
4. Possession of a rest frame	Has non‑zero rest mass, hence a proper time and the ability to serve as a reference clock in its own rest frame	Invariant mass > 0

Rationale for Criterion 4: Measurement requires a local frame. A massless particle has no rest frame, no proper time, and cannot be used as a persistent local reference. While photons are extremely long‑lived, they serve as signal carriers, not as the invariant substrate. The framework prioritises rest‑frame existence because the “eternal skeleton” is meant to be the background against which change is measured – a background must have a local perspective to anchor measurements. This is a definitional choice, not a consequence of particle physics, and it is consistently applied.

Note on Criterion 3: Baryon number and lepton number are accidental symmetries, not gauge symmetries. The paper treats them on equal footing because both provide effective stability for the proton and neutrinos under Standard Model physics. If future experiments reveal baryon or lepton number violation, the framework will adjust accordingly.

3. Why the Electron Is a Metronome

Apparent immortality: Lightest negatively charged particle; no decay channel.
Effective indivisibility: Remains an electron after scattering, binding, etc.
Conservation protection: Electric charge and lepton number conservation.
Rest frame: Non‑zero rest mass.

→ The electron is a metronome.

4. Why the Proton Is a Metronome (Despite Being Composite)

Apparent immortality: No observed decay; experimental lower limit on half‑life > 10³⁴ years (Super‑Kamiokande, 2020).
Effective indivisibility: For all practical purposes (chemistry, nuclear physics, stellar processes), the proton behaves as a stable, indivisible unit.
Conservation protection: Baryon number is an accidental symmetry; it protects the proton from decay in the Standard Model.
Rest frame: Non‑zero rest mass.

→ The proton is a metronome. The framework does not require elementary particles; it requires maximal persistence under relevant perturbations.

5. Why the Neutrino Class (ν₁, ν₂, ν₃) Is a Metronome

The three neutrino mass eigenstates are treated as a single metronome class because they share the same stability argument, differ only in mass, and are grouped for the framework’s hierarchical classification.

Apparent immortality: No observed decay; cosmological and astrophysical lower bounds on neutrino lifetimes are orders of magnitude longer than the age of the universe. Neutrino oscillation is flavour mixing, not decay – the mass eigenstates are stable.
Effective indivisibility: Once a neutrino is in a mass eigenstate, it propagates without changing identity. (Weak interactions produce flavour eigenstates – superpositions of mass eigenstates – but the mass eigenstates themselves are stable and travel freely.)
Conservation protection: Lepton number is an accidental symmetry; in the Standard Model it protects neutrinos from decay. (If future experiments confirm that neutrinos are Majorana particles – violating lepton number – the framework will adjust; this is part of its corrigibility.)
Rest frame: Neutrinos have non‑zero rest mass (confirmed by oscillation experiments), albeit very small.

→ The neutrino class is a metronome. The three mass eigenstates count as one metronome type for the framework’s hierarchical classification.

6. Why Not Other Candidates?

Candidate	Fails criterion	Explanation
Free neutron	1 (apparent immortality)	Decays in ~15 minutes.
Neutron in a nucleus	2 (effective indivisibility)	Stability is environment‑dependent; not an irreducible attractor.
Photon	4 (rest frame)	Massless; no proper time. Excluded by definition (see rationale for Criterion 4).
Muon, tau	1	Decay rapidly.
Dark matter candidates	Not yet identified	If discovered and shown to be stable, massive, and effectively indivisible, they could become additional metronomes.
Composite stable structures (nuclei, atoms)	2	Not effectively indivisible; they are built from metronomes and are dissipative or emergent attractors, not part of the invariant skeleton.

7. The Number Three: Empirical, Not Derived

The paper’s title uses “three metronomes” as a convenient label for the electron, proton, and the neutrino class (the three mass eigenstates grouped together). The number three is not derived from first principles; it reflects current best empirical knowledge. If new stable particles are discovered (e.g., dark matter), the list will expand. The framework is corrigible by design.

8. The Apparently Eternal Skeleton

The term “apparently eternal” is strictly empirical: these particles have never been observed to decay or be transient, and for all practical purposes they behave as if they have no end. The three metronomes form the eternal skeleton – a pragmatic substrate against which the transient dance of dissipative systems (life, mind, society) is measured. This is a framework‑internal construct, not a metaphysical claim.

9. Stable Resonances and the Grounding of Dissipative Time Metrics

Each of the three metronomes possesses an invariant quantum frequency – its Compton frequency, given by $f = m c^{2} / h$ f=mc2/h. For the electron, this is ~1.24 × 10²⁰ Hz; for the proton, ~2.27 × 10²³ Hz; for neutrinos, the frequencies are very small but non‑zero. These frequencies are invariant, universal, and identical for every identical particle in the universe. They are stable resonances of the eternal skeleton.

Why this matters for dissipative systems:

Every dissipative system (a living cell, a brain, a society) is composed of or continuously interacts with electrons, protons, and neutrinos. The time constant τ that appears in corrective permeability (κ = 1/τ) can, in principle, be expressed as a multiple of these fundamental resonance periods. For example, a neuron’s recovery time after a perturbation – determined by ion channel kinetics, membrane capacitance, and metabolic rate – is measurable against the same invariant clock as any other physical process. The metronome provides the unit of time, not the mechanism.

Thus, κ is a genuine physical variable, not a mere metaphor. It refers to a ratio of measurable durations, anchored in the invariant frequencies of the metronomes.

Cross‑domain comparability:

The framework’s ability to compare κ values across vastly different domains (e.g., a thermostat’s seconds‑scale τ and a political movement’s months‑scale τ) does not follow from shared Compton‑frequency units alone. It follows from the framework’s definitional choice to treat κ as a domain‑general variable – a diagnostic that measures the same functional property (speed of return to baseline) in every system, regardless of scale or substrate. The metronomes ensure that such measurements are, in principle, commensurable; they do not guarantee that the comparison is meaningful in every case. That is a framework commitment, not a physics claim.

Caveat: The expression of τ as a multiple of Compton periods is a conceptual grounding, not a practical measurement protocol. No one will measure a society’s reaction time in electron oscillations. The importance is that κ is not an arbitrary label; it is a dimensionless ratio of durations, and durations are defined by the invariant resonances of the three metronomes.

10. κ and Basin Depth as Heuristics

The attractor framework introduces corrective permeability (κ = 1/τ) and basin depth (B) as conceptual heuristics. For the metronomes:

κ for decay is vanishingly small (effectively zero) on all observable timescales.
Basin depth is the energy barrier required to change the particle’s identity – effectively infinite for all practical purposes.

These are qualitative descriptors; they are not operational quantities in particle physics. They are included here for completeness of the framework’s vocabulary. For the application of κ and B to dissipative systems (e.g., belief updating, neural recovery), see the papers Basin Defense and Stable Addition and Why Clockwork Interventions Fail.

11. Corrigibility and Falsifiability

The framework explicitly invites revision:

If proton decay is observed, the proton will be downgraded to “very long‑lived” (or removed).
If neutrino decay or Majorana nature is confirmed, the neutrino class’s status will be revised.
If new stable particles are discovered, they will be added.

The attractor framework is a philosophical taxonomy and diagnostic tool, not a predictive physical theory. Its value lies in providing a unified language for persistence across domains.

12. Conclusion

The electron, proton, and neutrino class satisfy the attractor framework’s four criteria for metronomes: apparent immortality, effective indivisibility under ordinary perturbations, conservation‑law protection, and possession of a rest frame. They are the best‑supported examples of the apparently eternal skeleton under current physics. The framework is corrigible, the number three is empirical, and the language of “eternal skeleton” is pragmatic. The three metronomes anchor the distinction between conservative and dissipative persistence.

Suggested citation: Galida, R. S. (2026). The Three Metronomes: Criteria for the Apparently Eternal Skeleton. Fantasy Attractor.

Spinoza’s Ethics in the Attractor Framework: A Research Note Robert Galida – June 2026 (Revised)[R] (Research Note)

June 11, 2026 1:03 pm / Leave a comment

Abstract

Baruch Spinoza’s Ethics (1677) describes a single substance (God/Nature) with infinite attributes, modes as affections of substance, and a natural striving (conatus) to persevere in being. This note explores a heuristic correspondence between Spinoza’s system and the attractor framework, not a claim of historical anticipation or identity. The eternal skeleton (conservative attractors) shares structural features with Spinoza’s substance: eternal, self‑caused, invariant. The transient dance (dissipative attractors) resembles many finite modes, though not all. Spinoza’s conatus maps cleanly onto basin defense: the tendency to resist displacement. Inadequate ideas can stabilize into fantasy attractors (sealed belief systems with low corrective permeability κ) when they form self‑reinforcing networks. Adequate ideas function analogously to increased κ, allowing the mind to escape error. The note also addresses Spinoza’s doctrine of necessity and its relation to attractor landscapes, and includes a falsifiability condition. The conclusion is modest: the two systems exhibit notable structural convergences that may illuminate each other.

1. Introduction

Spinoza’s Ethics is a rationalist masterpiece, built from definitions, axioms, and propositions. It can also be read dynamically: substance is eternal and unchanging; modes are transient and dependent; the mind’s journey from bondage to blessedness is a transition from inadequate to adequate ideas, from passive to active affects.

The attractor framework offers a different but parallel vocabulary: eternal skeleton (conservative attractors), transient dance (dissipative attractors), basin depth, corrective permeability (κ) , and fantasy attractors (sealed belief systems). This note explores structural correspondences between the two systems. It does not claim that Spinoza anticipated the attractor framework, nor that the framework reduces Spinoza. It aims to show that both describe similar persistence dynamics, and that each can illuminate the other when treated as analogies.

2. Substance and the Eternal Skeleton

Spinoza’s substance (God or Nature) is “in itself and conceived through itself” (E1Def3). It is eternal, uncaused, has infinite attributes, and does not change. It simply persists.

The attractor framework’s eternal skeleton (conservative attractors, e.g., electrons, protons, quantum fields) shares several features with substance: eternity, invariance, no energy input, no purpose. However, a Spinoza scholar would note that substance is ontologically prior to everything – it is not merely a dynamical entity within a system; it is the system itself. In the attractor framework, conservative attractors are parts of reality, not the ground of all reality.

Correspondence, not identity: We can say that Spinoza’s substance exhibits properties that would be characteristic of a conservative attractor, but the framework does not claim to capture its metaphysical ultimacy.

3. Modes and the Transient Dance

Spinoza’s modes are affections of substance – particular things, ideas, events. They are finite, dependent, and temporary. Many of them (e.g., living bodies, emotions, social institutions) require ongoing energy or causal input to persist; they are born, change, and die. These can be modeled as dissipative attractors.

However, not every mode fits that description. A mathematical truth, a triangle, or a relation (e.g., “2+2=4”) does not obviously require energy throughput. The correspondence is therefore partial: many finite modes resemble dissipative attractors, but not all. The note restricts its claim accordingly.

4. Conatus as Basin Defense

This is the strongest mapping. Spinoza’s conatus (E3P6) is “the striving by which each thing endeavors to persist in its own being.” It is the intrinsic tendency to resist destruction and maintain state.

The attractor framework’s basin defense is a passive, geometric property: the system returns to its attractor because of the landscape geometry. Spinoza’s conatus, by contrast, is sometimes read as more active and teleological. Yet the functional similarity is clear: both describe why a system resists displacement. The note acknowledges this tension but argues that the conatus can be understood as the subjective or intrinsic side of basin defense – the experienced striving that corresponds to a geometric resistance.

No change is needed here; this section remains the strongest.

5. Inadequate Ideas and Fantasy Attractors

Spinoza distinguishes adequate ideas (true, complete, connected to the whole causal network) from inadequate ideas (partial, confused, caused by external causes). Inadequate ideas lead to passive affects (hope, fear, envy, etc.).

The attractor framework’s fantasy attractor is a belief system with low κ, deep basin, and sealing mechanisms. However, not every inadequate idea forms a fantasy attractor. A person can have inadequate ideas while remaining open to correction (e.g., a scientist with a partial hypothesis). The correspondence is therefore:

Networks of inadequately connected ideas that become self‑reinforcing and resistant to evidence can stabilize into fantasy attractors.

Thus, the paper replaces “inadequate ideas create fantasy attractors” with a more nuanced formulation: inadequate ideas can lead to fantasy attractors when they are organised into a self‑sealing system. The example of free‑will belief (a Spinozistic inadequate idea) illustrates this: many people resist determinism not because they lack evidence, but because the belief is identity‑fused.

6. Adequate Ideas and Corrective Permeability (κ)

Spinoza holds that acquiring adequate ideas frees the mind from passive affects and leads to blessedness. In attractor terms, adequate ideas function analogously to increased corrective permeability (κ): they allow the mind to update beliefs in response to evidence, escape self‑reinforcing error, and align with reality.

But the mechanism is different. Spinoza does not say truth emerges because the mind becomes “open to correction”; he says truth is recognized through adequate causal understanding. The correspondence is functional, not identical.

The paper now states this clearly: adequate ideas act like a high‑κ state, enabling the mind to escape error basins. It does not claim that κ explains Spinoza’s epistemology.

7. Blessedness, Necessity, and Attractor Landscapes

Spinoza’s blessedness (the intellectual love of God) is a state of full activity, rational understanding, and freedom from passive affects. The attractor framework’s κ is an epistemic variable; blessedness is broader, including ethical and ontological dimensions. Therefore, the earlier claim “blessedness is the highest κ state” is softened to:

Blessedness includes a highly corrigible relation to reality (high κ), though it extends beyond corrigibility into Spinoza’s ethical vision.

Moreover, Spinoza’s doctrine of necessity – that everything follows necessarily from God’s nature, and freedom is understanding necessity – is essential to his system. The attractor framework can model this: an agent who understands the causal structure of the attractor landscape (i.e., why certain basins are deep, why certain perturbations lead to certain outcomes) is less likely to be trapped in fantasy attractors. Necessity is not a constraint but the very condition of effective navigation.

This section is new and addresses a major omission.

8. A Falsifiability Condition

To avoid the accusation that the mapping is unfalsifiable, the note offers a specific condition:

If Spinoza had claimed that adequate ideas are innate and not acquired through a gradual, error‑prone, socially mediated process, the analogy with increased κ would fail. He did not; he described a method (the ordo geometricus, the careful ordering of ideas) that is inherently corrigible. Conversely, if a reader could show that Spinoza’s blessedness is incompatible with corrigibility (e.g., that it entails dogmatic certainty), the analogy would be weakened.

This condition is modest but genuine.

9. Comparison with Milton’s Satan (Brief)

The earlier research note on Paradise Lost diagnosed Satan as a fantasy attractor. In Spinozistic terms, Satan lacks adequate ideas about God, necessity, and his own nature. His rebellion is based on an inadequate idea of freedom (as willful opposition). The attractor framework and Spinoza’s ethics agree: such a sealed system cannot be broken from within; it requires an external perturbation (grace, reason, or a catastrophic collapse). This brief mention replaces the earlier speculative counterfactual.

10. Conclusion

Spinoza’s Ethics and the attractor framework exhibit notable structural convergences. Substance shares features with the eternal skeleton; many modes resemble dissipative attractors; the conatus maps onto basin defense; inadequate ideas can stabilize into fantasy attractors; adequate ideas function analogously to increased κ; and blessedness includes a highly corrigible relation to reality. The mapping is heuristic, not literal. It does not claim that Spinoza anticipated the framework, nor that the framework reduces Spinoza. Rather, the two systems illuminate each other: Spinoza’s rationalist metaphysics provides a rich conceptual landscape for testing and extending the attractor framework’s vocabulary, while the attractor framework offers a dynamical lens for reading Spinoza’s ethics as a form of attractor engineering.

Suggested citation: Galida, R. S. (2026). Spinoza’s Ethics in the Attractor Framework: A Research Note (Revised). Fantasy Attractor.

Fantasy Attractor

Category Archives: Core Papers

Excess Entropy Production as a Candidate Universal Cost of Persistence: A Thermodynamic Foundation for the Attractor Framework; Robert Galida (July 2026) [F]

Abstract

1. Introduction

2. The Persistence Cost Functional

3. Existence and Lyapunov Equivalence

4. Entropy Production as Persistence Cost

4.1 Entropy Balance

4.2 Excess Entropy Production

4.3 The Entropy Persistence Functional

4.4 Corrective Permeability

4.5 Basin Depth

5. Domain-Specific Realizations

5.1 Physical Systems: Thermodynamic Excess Entropy

5.2 Biological Systems: Metabolic Excess Entropy

5.3 Cognitive Systems: Free Energy Dissipation

5.4 Social Systems: Coordination Excess Entropy

6. The Unified Framework

6.1 Summary Table

6.2 The Universal Structure

6.3 The Low-Energy Attractor Benchmark (Proposed Hypothesis)

7. Testable Predictions

7.1 Core Prediction

7.2 Secondary Prediction

7.3 Domain-Specific Predictions

8. Experimental Design

8.1 Physical Systems

8.2 Biological Systems

8.3 Cognitive Systems

8.4 Social Systems

9. Open Questions

10. Conclusion

11. Limitations

References

Deriving Corrective Permeability from the Cumulative Deviation Functional; Robert Galida (June 2026) [F]

Abstract

1. Introduction

2. The Cumulative Deviation Functional

3. Derivation of Corrective Permeability (κκ)

3.1 Variational Definition

3.2 Homogeneity and Scale Invariance

3.3 Sharp Universal Persistence Bound

3.4 Consistency with Linear Systems

3.5 Transport Equation

3.6 Local vs. Global Interpretation

3.7 Non-Symmetric Linear Systems

3.8 Comparison with Exponential Stability

4. Connections to Existing Theory

4.1 Koopman Operator

4.2 Resolvent Poles

5. Finite-Horizon Estimation

6. Open Questions

7. Conclusion

References

The Persistence Functional: A Candidate Formal Foundation for the Attractor Framework; Robert Galida (July 2026) [F]

Abstract

1. Introduction

1.1 Scope and Status

2. Formal Definitions

2.1 Cumulative Deviation Functional

2.1.1 Why the L¹ Trajectory Functional?

2.2 Topological Persistence Functional

2.3 Topological Evolution Rate

3. Mathematical Properties of the Cumulative Deviation Functional

3.1 Non-negativity

3.2 Monotonicity in TT

3.3 Additivity

3.4 Heuristic Connection: Dynamic Programming

3.5 Lipschitz Continuity with Respect to Initial Conditions

3.6 Instantaneous Growth Rate

3.7 Ergodic Limit

3.8 Bound under Exponential Stability

3.9 Recovery Rate Bound

3.10 Finite Horizon Approximation

3.11 Summary of Properties

4. The Unified Variable Set

4.1 Corrective Permeability (κκ)

4.2 Drift Rate (γγ) — A Proposed Distinction

4.3 Basin Depth (BB) and Persistence Depth (B~B~)

3. Derivation of Corrective Permeability ( $κ$ κ)

3.2 Monotonicity in $T$ T

4.1 Corrective Permeability ( $κ$ κ)

4.2 Drift Rate ( $γ$ γ) — A Proposed Distinction

4.3 Basin Depth ( $B$ B) and Persistence Depth ( $\tilde{B}$ B~)

4.4 Reality Alignment ( $R$ R)

5.1 Relationship Between $D_{T}$ DT, $P_{topo}$ Ptopo, and $E (t)$ E(t)

5.2 Relationship Between $κ$ κ, $γ$ γ, and $E (t)$ E(t)