Reflections on self-regulating and evolving systems | Dynamic Regulation of Biological Coherence

Core Concept

Biological systems do not preserve coherence through fixed setpoints or simple error correction.
Instead, regulation emerges through a recursively adaptive, phasic cycle, guided by an alignment of internal states with constant external inputs.
Rather than merely reacting to deviations, living systems proactively self-regulate by maintaining internal dynamic homeostasis.
Disruption does not arise from deviation alone, but from a loss of coherence across temporal, spatial, and energetic axes. Biological regulation thus unfolds not as reversion to baseline, but as a nonlinear transition through five interdependent phases.

Phase Loop

This cycle spans five core regulatory phases:

1. Dynamic Predictive Homeostasis: Predictive coherence at minimal energy cost

No fixed “baseline”, only a dynamic coherence sustained by internal alignment. Stability is maintained through phase-aligned processes. Coherence arises from proactive synchronization across subsystems (e.g., cortical-thalamic, glial-neuronal).

  • State: Proactive regulation via minimized prediction error : low-entropy state
  • Mechanisms: Internally aligned rhythms, metabolically efficient coherence
  • Neural correlates: Insula, posterior cingulate, orbitofrontal cortex
  • Behavioral mode: Efficiency, readiness, precision


2. Perturbation & Entropic Disruption: Phase desynchronization across regulatory axes

Minimal Disruptions, internal or external create misalignments. Entropy is introduced not as disorder, but as energetic overload in subsystems. Only when an internal / external cue is larger than the system can handle (“threshold”), does it cause a disruption to the system. Coherence begins to unravel; signal transmission loses precision.

  • Trigger: Unexpected input overwhelms current model or exceeds energetic capacity
  • Entropy rises: Prediction error, neural surprise, or energetic overload
  • Neural correlates: Anterior cingulate, locus coeruleus, salience network
  • Function: Salient detection of mismatch across spatiotemporal axes
  • ↔ Inference pauses; signal amplification initiates

3. Reactive Containment

When disruption exceeds the predictive capacity of the current model, the system enters a containment mode. This phase spans two distinct, nested response strategies, both aimed at temporary stabilization — not yet at structural adaptation.

3A: Low-Cost Reflexive Action

  • Action type: Fast, low-level, autonomic, or habitual
  • Goal: Stabilize the system with minimal energetic expenditure
  • Neural systems:
    • Basal ganglia (habitual loops)
    • Amygdala (emotional tagging)
    • Brainstem and cerebellum (motor reflexes)
    • Vagal tone via parasympathetic pathways
  • Examples:
    • Gaze redirection
    • Reaching for warmth
    • Attentional reallocation
  • Features:
    • Low ATP cost
    • High prior confidence
    • Rapid, non-conscious execution
    • Entropy is managed efficiently without high systemic cost

3B: High-Cost Defensive ResponseEnergetically intensive systemic measures to suppress entropy under high uncertainty

  • Action type: Reflexive but metabolically expensive buffering
  • Goal: Prevent system-wide disintegration and buy time for adaptation
  • Systems involved:
    • Mitochondrial upregulation (ATP for stress response)
    • Hypothalamus → HPA axis (cortisol, adrenaline)
    • Immune-glial interface (inflammation, cytokine signaling)
    • Periaqueductal gray (pain and defensive reactions)
  • Examples:
    • Inflammatory cascade
    • Reactive oxygen species buffering
    • Fever, sickness behavior
  • Features:
    • High ATP cost
    • Precision collapse → broad activation
    • Temporarily suppresses entropy but risks maladaptation
    • Often asynchronous across subsystems → loss of global coherence


4. Adaptive Inference & Structural Reorganization: System-wide reorganization and encoding

If survival is maintained, repatterning begins: Synaptic pruning, receptor scaling, epigenetic modification, glial phenotype shifts. This is not a return to homeostasis but rather a restructuring. Maladaptation may result if realignment fails (e.g., persistent glial priming, maladaptive plasticity).

  • Action type: Deliberative, flexible, conscious model updating
  • Neural Network: DLPFC, mPFC, hippocampus, DMN
  • Mechanisms: Synaptic pruning, gene transcription, mitochondrial biogenesis
  • Behavior: Learning, memory consolidation, reframing, policy restructuring
  • ↔ Inference resumes; free energy minimized through reconfiguration


5. Refined Homeostasis: Stabilization into a new phase-aligned system

The system reaches a new state of coherence, distinct in structure, energetics, and readiness. This state encodes memory of prior disruption, allowing more efficient phase transitions in the future or entrenching dysfunction, depending on trajectory.

  • New attractor: Not a return, but a restructured coherence from prior disruption
  • Features: More efficient predictions, reduced energy cost, broader resilience
  • Possible outcomes:
    • Adaptive memory encoding
    • Maladaptation (e.g., chronic inflammation, PTSD)

Dimensions

Regulation unfolds across three interdependent dimensions:


1. Temporal coordination:

  • Synchronizes ultrafast (ionic) events with slow (genetic or structural) responses
  • Enables regulatory memory, rhythm alignment, phase-dependent readiness.

2. Spatial organization governs the accessibility and influence of components

  • Determines where signals can reach and how they exert influence (e.g., membrane-localized vs endocrine-wide)
  • Defines topology of responsiveness (compartmentalization, gradient signaling, etc.)

3. Energetic demand reflects metabolic cost and activation thresholds

  • Reflects the metabolic cost of regulatory action
  • Determines whether a signal or reorganization is viable, filtering responses through ATP availability, redox state, and activation thresholds.

Systems

Each phase selectively activates distinct regulatory systems, differentiated by their mode of chemical interaction and range of influence. All are rooted in molecular chemistry; what distinguishes them is how, where, and when these molecular signals operate to shape biological coherence:

  • Electrical: Fast, short-range coordination and immediate signal propagation
  • Chemical: Signal amplification and diffusion across local and systemic scales
  • Systemic: Integrative regulation via hormonal, vascular, and inter-organ pathways
  • Genetic: Long-term encoding of regulatory patterns through transcriptional and epigenetic changes

Signals become effective not by static wiring, but through synchronized readiness. They act only when temporally aligned, spatially reachable, and energetically viable.

Adaptability & Progression

Adaptability in biological systems is not a matter of simply raising thresholds or dulling sensitivity. Rather it is about a multidimensional expression of coordinated plasticity. A resilient system can tolerate greater misalignment before tipping into disruption (threshold), but more importantly, it can realign internal rhythms quickly (flexibility), sustaining coherence without excessive energetic cost. It can oscillate across a wider dynamic range (amplitude) without fragmenting and return to coherence efficiently (reversibility), even if that means establishing a new functional baseline rather than reverting to a prior state. Critically, it also encodes the memory of prior disruptions (learning), reducing the energetic and informational cost of future adaptations. Adaptability, then, is not about rigidity or resistance but more so about the system’s ability to explore wider territories and reorganize into stable coherence. In micro-adaptation, this means oscillating more flexibly within an existing basin of attraction; in macro-adaptation, it means redefining the basin itself, emerging into a new regulatory architecture. Thus, adaptability is the dynamic capacity to sustain coherence through movement, and to craft new forms of coherence when old rhythms no longer serve.

References

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

Parr, T., Da Costa, L., & Friston, K. J. (2020). Markov blankets, information geometry and stochastic thermodynamics.
Philosophical Transactions of the Royal Society A, 378(2164).
https://doi.org/10.1098/rsta.2019.0152

Scroll to Top