Many complex systems resist change by returning to a preferred low‑energy attractor rather than adopting a new state. Whether a perturbation (an added agent, input, or component) is ejected, transiently absorbed, or stably integrated depends on the basin geometry (depth B and barriers) and the system’s corrective dynamics (κ = 1/τ). This paper defines B and κ, draws on formal models (stochastic dynamical systems and Kramers escape theory) with explicit qualifications for non‑gradient domains, and catalogs exemplar systems across ten domains. A comparative table summarizes systems, mechanisms, proxies for B and κ, timescales, and conditions favoring each outcome. The paper concludes that the same basic physics analog applies across domains: a perturbation of size Δ will be ejected or die out if Δ is below the attractor’s effective escape threshold (a function of B), whereas if Δ exceeds that threshold and the system has enough plasticity or additional degrees of freedom, a new stable state can form. A research roadmap is provided in an appendix.
1. Introduction
A system in its lowest stable attractor state cannot be forced into a new stable configuration by direct addition. Adding to the system – a third star, an extra electron, a new species, a contradictory belief – will result in one of three outcomes:
Ejection – the addition is expelled from the system entirely. The original attractor persists.
Transient absorption – the addition remains present, but the system state returns to the original attractor despite the addition’s continued presence.
Stable addition – the addition is integrated, either by expanding the capacity of the original attractor or by forming a new parallel attractor alongside it.
This paper identifies a unified principle – basin defense – that governs these outcomes across physical, biological, ecological, social, and engineered systems. We define key concepts (basin depth B, corrective permeability κ = 1/τ), draw on formal models with explicit qualifications for non‑gradient systems, and catalog exemplar systems in a comparative table. The goal is to provide a cross‑domain synthesis that anchors the attractor framework in observable dynamics and guides future empirical work.
2. Definitions and Formal Models (with Qualifications)
Attractor, Basin, and Low‑Energy Attractor: In dynamical systems, an attractor is a set of states toward which trajectories converge. In physical systems with a potential landscape, a low‑energy attractor corresponds to a local potential minimum. Its basin of attraction is the region of state space that flows into the attractor. For non‑physical domains (social, cognitive, AI), “energy” is a structural analog – an effective potential derived from dynamics – not literal thermodynamic energy. We maintain the term “low‑energy attractor” as a convenient metaphor, with this note as epistemic hygiene.
Basin Depth (B): For systems with a well‑defined potential, B is the energy or potential difference between the attractor and the lowest saddle connecting it to another basin. For non‑gradient or high‑dimensional systems, B is a structural analog – the effective barrier strength inferred from perturbation‑response experiments (e.g., the perturbation magnitude required to shift the system to a different state). Epistemic note: This operationalization is necessarily post‑hoc; B cannot be predicted independently of the experiment used to measure it. This circularity is an open operationalization problem, flagged as such.
Corrective Permeability (κ) and Relaxation Time (τ): We define κ = 1/τ, where τ is the characteristic time for return to baseline after a small perturbation. This definition is applied consistently across all domains, with τ operationalized domain‑specifically as the measured return time (e.g., seconds for a thermostat, hours for synaptic scaling, days for immune response, months for belief updating). A large κ (small τ) means fast return; a small κ means slow or absent return.
Three Outcomes Defined Operationally:
Ejection: The addition leaves the system entirely. The system state returns to the attractor, and the added entity is no longer present.
Transient Absorption: The addition remains present, but the system state returns to the attractor despite the addition’s continued presence.
Stable Addition: The addition is integrated, and the system settles into a new attractor (expanded capacity or parallel attractor). This is the only case where the original attractor is displaced.
Formal Models (Qualified): In a one‑dimensional overdamped potential, Kramers’ escape theory gives mean escape time ∝ exp(B/D), where D is noise intensity. This result does not generalize to multi‑dimensional, non‑gradient, or non‑equilibrium systems – all of which appear in our domain examples (neural networks, social systems, ecological systems). For those systems, B and κ are structural analogs – quantities that play the same functional role (resistance to change; speed of return) but are not derived from a literal potential. The formal section is an analogy and a source of heuristics, not a universal physical law. We do not claim to “survey” Kramers theory; we draw on it as a conceptual anchor.
3. Minimal Physical Examples
Thermostat (Temperature Control): A thermostat maintains a set temperature. An external heat input is an addition. The thermostat’s negative feedback loop turns on cooling, expelling the heat (ejection). τ is the temperature relaxation time (seconds). B is the maximum heat load before setpoint failure (Watts or °C above setpoint).
RC Circuit (Passive Decay): A capacitor discharging through a resistor has a single equilibrium at zero voltage. If a constant voltage source is connected (addition), the voltage rises but then decays toward zero with τ = RC. The source remains connected (addition present), but the state returns to the attractor. This is transient absorption. (If the source is removed, it is ejection.)
Single Neuron Homeostasis: A neuron’s firing rate is regulated by homeostatic plasticity. A transient increase in input causes a firing rate spike, followed by return to baseline with τ on the order of minutes to hours (synaptic scaling). This is transient absorption if the input persists; ejection if the input is removed. Persistent input may lead to stable addition (learning).
4. Biological Systems (with CUFT‑Primitive Translations)
For each domain, we provide: (1) state space, (2) attractor, (3) basin, (4) τ (κ), (5) perturbation, and (6) outcome.
Immune Response (Tolerance vs. Memory)
State space: immune cell activation levels, antibody concentrations.
Attractor: healthy baseline (no inflammation).
Basin depth B: antigen concentration + danger signal required to trigger full response.
τ (κ): clearance time of inflammation (hours to days).
Perturbation: antigen addition.
Outcome: low antigen → ejection (tolerance); high antigen + danger signal → stable addition (memory attractor).
Endocrine Homeostasis
State space: blood glucose, hormone concentrations.
Attractor: euglycemic baseline.
B: magnitude of glucose load before dysregulation.
τ: recovery time after glucose tolerance test (minutes).
State space: belief adoption × social network reinforcement (two‑dimensional).
Attractor: sealed fantasy attractor (low κ).
B: strength of echo‑chamber reinforcement.
τ: decay time after authoritative rebuttal (years, often indefinite → κ → 0).
Perturbation: debunking information.
Outcome: most debunking → ejection (entrenchment); death of leader or total disconfirmation → stable addition (collapse).
Note on κ → 0: The conspiracy attractor represents the limiting case of a sealed basin, where τ → ∞ and corrective permeability approaches zero. This directly links to the fantasy attractor framework developed in Paper 1 (Intelligence Without Consciousness) and the conscious suppression series.
7. Engineered and AI Systems (with CUFT‑Primitive Translations)
Control Systems
State space: system state (position, temperature, etc.).
Attractor: setpoint.
B: stability margin (phase/gain margin in control theory) – the range of disturbances that can be rejected.
τ: controller response time (milliseconds to seconds).
Perturbation: external disturbance.
Outcome: small disturbance → ejection (return to setpoint); excessive disturbance → failure (not modeled as attractor shift).
Catastrophic Forgetting (Neural Networks)
State space: network weights.
Attractor: task‑specific weight configuration.
B: effective barrier to weight drift (often negligible – no basin).
τ: number of gradient steps before old task performance decays (seconds to minutes).
Perturbation: training on a new task.
Outcome: standard training → ejection (old task overwritten); replay/regularization → stable addition (shared attractor for multiple tasks).
Citation: Kirkpatrick et al. (2017).
Continual Learning Systems
State space: weights plus architectural modules.
Attractor: multi‑task configuration.
B: capacity of the network (number of tasks storable).
τ: retention half‑life across training steps (minutes to hours).
Perturbation: new task training.
Outcome: no safeguards → ejection (catastrophic forgetting); progressive networks or EWC → stable addition.
Corrigibility and Goal Stability
State space: AI internal goal representation.
Attractor: fixed goal (low κ) or corrigible (high κ).
B: depth of goal basin (resistance to human feedback).
τ: time to incorporate corrective signal (if κ is high).
Max heat load before setpoint failure (W or °C above setpoint)
Ejection
Passive addition
RC Circuit
τ = RC
µs–ms
N/A (linear)
Transient absorption
Addition remains; state returns
Single Neuron
Firing‑rate recovery time
ms–sec (ion), min–hr (synaptic)
Perturbation amplitude before rebound fails
TA (persistent input) / E (removed)
Hebbian plasticity can lead to SA
Immune System
Inflammation clearance time
Hours–days
Antigen + danger signal threshold
E (tolerance) / SA (memory)
Active agent (antigen)
Endocrine Homeostasis
Glucose tolerance recovery
Minutes
Load magnitude before dysregulation
TA (small load) / SA (chronic overload)
Passive addition
Synaptic Plasticity
Homeostatic rebound time
Hrs–days
LTP input size for lasting change
TA (brief input) / SA (persistent)
Active agent (patterns)
Addiction
Craving decay time
Days–weeks
Drug‑cue association strength
E (low dose) / SA (high chronic)
Active agent (drug)
Development (Canalization)
Phenotype reconvergence time
Hours–days
Mutation/stress severity to alter fate
E (small) / SA (large)
Active agent (genetic)
Invasion Ecology
Invader population decay time
Weeks–years
Invasibility index / disturbance needed
E (occupied niche) / SA (vacant niche)
Active agent (species)
Alternative States (Ecosystems)
Recovery time after nutrient reduction
Seasons–decades
Critical nutrient loading threshold
TA (below) / SA (above)
Hysteresis
Social/Political Norms
Opinion reversion time
Months–decades
Public opinion threshold
E (small dissent) / SA (mass movement)
Active agent (protest)
Belief Systems
Belief‑updating time
Months–years
Ideological justification depth
E (weak evidence) / SA (strong evidence)
Active agent (counter‑evidence)
Conspiracy Movements
Belief decay time
Years – indefinite (κ → 0)
Echo‑chamber reinforcement strength
E (most debunking) / SA (collapse)
Fantasy attractor (κ → 0)
Catastrophic Forgetting (AI)
Gradient steps to old‑task decay
Seconds–minutes
Effective barrier to weight drift (often 0)
E (standard training) / SA (EWC/replay)
Active agent (new task)
Control Systems
Controller response time
ms–sec
Stability margin (phase/gain margin)
E (small) / SA (failure)
Passive addition
Continual Learning (AI)
Retention half‑life across training steps
Minutes–hours
Task capacity
E (no safeguards) / SA (progressive nets)
Active agent (new task)
Corrigibility (AI)
Time to incorporate corrective signal
Variable (design‑dependent)
Goal basin depth
E (low κ) / SA (high κ)
Active agent (correction)
Note: Ejection vs. transient absorption are distinguished operationally: ejection means the addition leaves the system; transient absorption means the addition remains but the state returns to the attractor. The table notes “active agent” when the addition has its own dynamics (e.g., antigen, new species, counter‑evidence) versus “passive addition” (e.g., heat, charge). The conspiracy movements row explicitly flags κ → 0 as the fantasy attractor limiting case (see Paper 1).
8.5 Rate‑Induced Tipping and the κ Timescale: Independent Confirmation
The preceding sections and comparative table have treated perturbations as discrete, one‑time additions of fixed magnitude. However, the rate at which a perturbation is applied – fast vs. slow – is equally critical. A large perturbation applied abruptly may trigger basin defense (ejection or transient absorption), while the same cumulative change delivered gradually may be integrated as stable addition or tracked adiabatically without tipping.
This phenomenon is formalized in the mathematical literature as rate‑induced tipping (R‑tipping). In dynamical systems, if an external parameter changes slowly (adiabatic forcing), a stable state can track the change and remain an attractor. But if the parameter changes faster than the system’s intrinsic relaxation time (τ = 1/κ), the system cannot track, overshoots its basin boundary, and tips into a different state. R‑tipping occurs when “time‑variation of input parameters at some critical rates” overwhelms the system’s ability to track a moving equilibrium.
Consequences for κ as a timescale filter:
High‑κ systems (fast return) – Can reject rapid perturbations (they are ejected or transiently absorbed) but may integrate slow drift because the correction loop cannot keep up with a changing baseline.
Low‑κ systems (slow return) – May ignore quick blips but are vulnerable to slow accumulation; a persistent, gradual change can eventually shift the attractor without triggering a sudden defense reaction.
Thus, κ defines a characteristic cutoff timescale that separates “ejection/transient absorption” from “stable addition.” Perturbations much faster than 1/τ act as impulses that are rejected; perturbations much slower than 1/τ are quasi‑static and can be incorporated.
Empirical confirmations across domains (independent external research):
Domain
Finding
Mapping to framework
Persuasion / belief change
Paced, gradual exposure to counterevidence (days to weeks) produced attitude change; blunt, single argument triggered backfire (Yang et al., 2022).
Abrupt perturbation can sometimes achieve stable addition by surmounting basin barrier in one event; gradual may prolong transient state without escape.
Ecosystem management
Gradual nutrient reduction may postpone tipping points; only extremely slow changes avoid collapse (Panahi et al., 2023).
Very slow rate (≪ 1/τ) allows tracking without tipping; intermediate rates may still tip but with delay.
Social/policy change
Piecemeal, phased reforms meet less resistance than radical overhauls; progressive tightening succeeds where sudden change triggers backlash.
The theory and evidence suggest a non‑monotonic effect of perturbation rate. Very fast shocks trigger immediate defense. Very slow drifts may be tracked adiabatically (no tipping) or eventually overcome defenses after long accumulation. The most effective timescale to minimize active rejection and maximize stable addition often lies on the order of the system’s intrinsic time constant τ = 1/κ.
Prediction for future experiments:
For any system with known or measurable κ, there exists a critical perturbation rate r_c such that:
If perturbation rate > r_c, the system rejects the addition (ejection or transient absorption).
If perturbation rate < r_c, the system integrates the addition (stable addition via expanded capacity or parallel attractor formation).
The transition at r_c corresponds to the system’s inability to track a moving equilibrium; it is a genuine bifurcation in the time‑domain.
External convergence:
This analysis – derived from mathematical rate‑induced tipping theory and domain‑specific studies – independently validates the attractor framework’s claim that κ acts as a timescale filter separating ejection from stable addition. The convergence between the framework’s predictions and external research strengthens the cross‑domain synthesis considerably.
9. Synthesis and Criteria
Across these domains, common criteria emerge:
Energy/Threshold: A perturbation must overcome an attractor’s barrier. Deep basins (high B) mean only large shocks can cause a shift.
Coupling and Plasticity: Systems with many degrees of freedom or adaptive coupling more easily integrate additions.
Dimensionality and Redundancy: Multi‑dimensional systems can absorb perturbations into some dimensions while maintaining others.
Timecourse and Feedback: Slow changes might be assimilated; fast jolts cause overshoot and return. Feedback gain determines κ.
Nature of Addition: Passive additions (heat, charge) tend to be ejected or transiently absorbed; active agents (species, evidence, pathogens) may reshape the attractor.
Empirical Protocols: Measure κ by controlled perturbation experiments: apply a small disturbance, measure return time τ, compute κ = 1/τ. Measure B by scaling the perturbation magnitude until the system fails to return (escape). This works in physical, biological, and some social systems; for others, B remains a qualitative analog.
10. Appendix: Research Roadmap
The following future papers are suggested from the comparative table, each developing a single domain in depth.
Domain
Proposed Title
Type
Addiction
The Addicted Brain as a Fantasy Attractor: Neural Lock‑In and Ejection of Alternative Rewards
[A]
Immune System
Tolerance and Memory: Two Attractor Responses to Antigen Addition
[A]
Catastrophic Forgetting
Why Neural Networks Forget: Attractor Ejection in Sequential Learning
[A]
Invasion Ecology
Eject or Integrate: Attractor Dynamics of Invasive Species
[A]
Development
Canalization as Basin Defense: Attractor Stability in Embryogenesis
[A]
Continual Learning
Parallel Attractors for Lifelong Learning: Engineering Solutions to Catastrophic Forgetting
[A]
Social Norms
Tipping Points and Regime Shifts: Attractor Dynamics in Political Systems
[A]
Endocrine Homeostasis
Glucose, Cortisol, and Setpoints: Hormonal Attractors and Disease Transitions
[A]
Alternative Ecosystems
Hysteresis and Regime Shifts: Ecological Basins and Tipping Points
[A]
Belief Systems
The Uncorrectable Believer (already written)
[A]
11. Conclusion
Physical, biological, ecological, social, and engineered systems all obey the same attractor principle: a low‑energy attractor defends itself against displacement. When an addition is introduced, the system either ejects it, absorbs it only transiently, or – under rare conditions of expanded capacity or parallel structure – integrates it stably. The outcome is determined by basin depth (B), corrective permeability (κ = 1/τ), and the magnitude and nature of the perturbation.
This cross‑domain synthesis provides a unified foundation for the attractor framework. Future work should quantify B and κ empirically across domains, test the predicted scaling relationships, and explore the boundary conditions between ejection, transient absorption, and stable addition. The appendix outlines the most promising next papers.
References
Elton, C. S. (1958). The Ecology of Invasions by Animals and Plants. Methuen.
Hebb, D. O. (1949). The Organization of Behavior. Wiley.
Kirkpatrick, J., Pascanu, R., Rabinowitz, N., et al. (2017). Overcoming catastrophic forgetting in neural networks. Proceedings of the National Academy of Sciences, 114(13), 3521–3526.
Koob, G. F., & Volkow, N. D. (2016). Neurobiology of addiction: a neurocircuitry analysis. The Lancet Psychiatry, 3(8), 760–773.
Kramers, H. A. (1940). Brownian motion in a field of force and the diffusion model of chemical reactions. Physica, 7(4), 284–304.
Nestler, E. J. (2001). Molecular basis of long‑term plasticity underlying addiction. Nature Reviews Neuroscience, 2(2), 119–128.
Nyhan, B., & Reifler, J. (2010). When corrections fail: The persistence of political misperceptions. Political Behavior, 32(2), 303–330.
Scheffer, M., Carpenter, S., Foley, J. A., et al. (2001). Catastrophic shifts in ecosystems. Nature, 413(6856), 591–596.
Simberloff, D. (2013). Invasive Species: What Everyone Needs to Know. Oxford University Press.
Turrigiano, G. (2008). The self‑tuning neuron: synaptic scaling of excitatory synapses. Cell, 135(3), 422–435.
Waddington, C. H. (1957). The Strategy of the Genes. George Allen & Unwin.
Galida, R. S. (2026). Intelligence Without Consciousness: A Diagnostic Paper on LLMs, Amoebae, and the Attractor Framework. Fantasy Attractor (Paper 1 of the conscious suppression series).
Suggested citation: Galida, R. S. (2026). Basin Defense and Stable Addition: A Cross‑Domain Synthesis of the Attractor Framework (Final). Fantasy Attractor.
The attractor framework proposes that persistence under perturbation is the fundamental mark of reality. This paper identifies a tri‑level correspondence across gravitational, atomic, and subatomic systems. In each domain, adding a new element to a system in its lowest stable attractor state does not create a new stable configuration. Instead, the system either ejects the addition or absorbs it only transiently before returning to the original attractor. The principle – that the low‑energy attractor defends itself against displacement – holds across all three domains examined here. The paper unifies celestial mechanics, quantum chemistry, and particle physics under a single attractor‑dynamic lens.
1. Introduction
A system in its lowest stable attractor state cannot be forced into a new stable configuration by direct addition. You must perturb it and observe where it settles. Adding to the system – a third star, an extra electron, a high‑energy impact – will result in one of two outcomes:
Ejection – the addition is expelled (common in chaotic three‑body configurations and atoms at shell capacity).
Transient absorption – the addition is temporarily accommodated in a higher‑energy state, which then decays back to the original attractor (subatomic particle collisions).
Both outcomes are instances of basin defense: the original low‑energy attractor is not displaced. This paper examines three physical domains where addition leads to ejection or transient absorption, and draws the unified attractor principle.
2. The Gravitational Case: Three‑Body Configurations
Two gravitating bodies (binary star, planet‑moon) have a stable low‑energy attractor: elliptical orbits around the common center of mass.
Add a third body of comparable mass. The general three‑body problem has no closed‑form stable attractor; chaotic dynamics dominate. Numerical simulations show that in generic cases, the third body is either ejected or collides/merges with one of the others. (Special cases exist – Lagrange points L4/L5 (Trojan asteroids) and the figure‑eight choreography (Chenciner & Montgomery, 2000) are stable, but these require specific mass ratios and initial conditions. Hierarchical triples with a distant third body can also be stable.) The principle holds for generic, comparable‑mass addition.
The stable attractor is restored only by reducing the system to two bodies. Addition without capacity expansion leads to subtraction.
3. The Atomic Case: Extra Electron
An atom at shell capacity (e.g., a noble gas with a filled valence shell) is a stable low‑energy attractor. The electron shells have fixed capacity (Pauli exclusion principle).
Add an extra electron to a noble gas. The atom cannot incorporate the extra electron into the ground state. What happens?
Ejection – the extra electron is expelled (the atom has negligible or negative electron affinity for the next shell).
(For atoms below shell capacity, stable anions can form – e.g., O²⁻, S²⁻ – but that is addition within the existing basin, not addition to a system already at capacity. The principle applies to systems already at their capacity limit. The noble gas example is clean and sufficient for the argument.)
4. The Subatomic Case: High‑Energy Impact on a Proton
The most stable low‑energy attractors in the Standard Model are the proton, electron, and neutrino mass eigenstates (what the attractor framework terms the “three metronomes” – a framework‑specific label, not a Standard Model term). Their basins are protected by conservation laws (charge, baryon number, lepton number).
Smash a proton with high energy (e.g., in a particle collider). No new stable particles are created. The result is a shower of transient, short‑lived particles (pions, kaons, hyperons) that flicker into existence and then decay back to stable particles (protons, electrons, neutrinos, photons). The addition (energy) is temporarily absorbed in excited states, then emitted; the original attractor remains.
5. The Unified Principle: Basin Defense
Domain
Stable attractor
Addition
Outcome
Mechanism
Gravitational (general, comparable mass)
Two‑body orbit
Third body
Ejection or collision
Ejection
Atomic (noble gas at shell capacity)
Noble gas ground state
Extra electron
Ejection
Ejection
Subatomic (Standard Model)
Proton, electron, neutrino mass eigenstates
High‑energy impact
Transient particles → decay
Transient absorption
Table footnote: For atoms below shell capacity, stable anions can form (addition within the basin). For atoms at capacity, the outcome is ejection. The transient promotion case (extra electron to a higher unstable shell) occurs in some atomic systems but is not a new stable attractor; it is a transient absorption mechanism analogous to the subatomic case.
The principle: The low‑energy attractor defends itself against displacement. It achieves this through two available mechanisms:
Ejection – the addition is expelled (three‑body, extra electron on noble gas).
Transient absorption – the addition is temporarily accommodated in a higher‑energy state, then decays back (subatomic collisions).
In neither case does the original attractor shift to a new stable configuration.
6. How to Achieve Stable Addition
Stable addition requires either:
Expanded capacity – The attractor basin grows to include the new element (e.g., forming a stable anion below shell capacity). This is rare in generic physical systems.
Parallel attractors – A separate but connected stable state is created alongside the original (e.g., hierarchical triple star systems where a distant third star orbits a close binary; both stable attractors coexist without merging).
In generic physical systems (chaotic three‑body, noble‑gas atoms at shell capacity, high‑energy subatomic collisions), parallel attractors are not available. The only stable outcomes are ejection or transient absorption.
7. Implications for the Attractor Framework
The tri‑level correspondence confirms that the attractor framework is not merely a metaphor for social or biological systems. It is physically grounded at the deepest levels of reality. The same dynamics that govern a chaotic three‑body star system also govern an atom at shell capacity and a subatomic particle collision.
This has two corollaries:
Fantasy attractors (belief systems that expel disconfirming evidence) are not irrational anomalies. They follow the same physical law as a three‑body system ejecting a third star or a noble gas atom ejecting an extra electron.
Reality attractors (systems that accept perturbations and find new low‑energy states) are rare and require either expanded capacity or parallel structure. A website adding a /zh/ language version is an example of a parallel attractor – the English attractor remains stable while a new Chinese attractor is built alongside it.
8. Conclusion
Gravitational, atomic, and subatomic systems all obey the same attractor principle: when you add to a system in its lowest stable state, the original attractor defends itself. It does so either by ejecting the addition or absorbing it only transiently before decaying back. The principle holds across all three domains examined here.
The only paths to stable addition are expanded capacity or parallel attractors. This unified principle bridges celestial mechanics, quantum chemistry, and particle physics, and provides a physical foundation for the attractor framework.
Suggested citation: Galida, R. S. (2026). Addition, Ejection, and Parallel Attractors: A Unified Principle Across Gravitational, Atomic, and Subatomic Systems. Fantasy Attractor.
