Borros Arneth1,2*- 1Institute of Laboratory Medicine and Pathobiochemistry, Molecular Diagnostics, Philipps University Marburg, Marburg, Germany
- 2Justus Liebig University Giessen, Gießen, Germany
Why some physical systems are accompanied by subjective experience remains unresolved in neuroscience and philosophy of mind. Building on predictive processing and the Free Energy Principle, I propose that phenomenal consciousness (what-it-is-like-ness) arises when an information-processing system enters a regime of dynamic entropic closure: a metastable condition in which (i) internally generated predictions and (ii) incoming sensory signals are recursively coupled such that net informational entropy exchange with the environment is minimized while internal informational dynamics remain high. In this regime, inference loops become phase-coherent and self-referential, producing a persistent informational pattern—resonant closure—that constitutes awareness. The framework is compatible with, but conceptually distinct from, Integrated Information Theory and global-workspace style accounts. I formalize core constructs at the level of operational constraints, address objections regarding trivial closure and “stationarity,” and derive falsifiable empirical predictions for neurophysiology.
1 Introduction
The scientific challenge of consciousness is not merely to identify neural correlates, but to explain why and when certain physical–informational processes are accompanied by subjective experience. The target explanandum in this manuscript is phenomenal consciousness, i.e., what-it-is-like-ness (Chalmers, 1995). Contemporary neuroscience emphasizes large-scale coordination and ignition-like dynamics in conscious access (Dehaene and Changeux, 2011), and state-dependent breakdowns of global coordination during anesthesia and sleep (Mashour et al., 2020).
Predictive processing and the Free Energy Principle (FEP) model organisms as systems that maintain themselves by minimizing variational free energy through inference and action (Friston, 2010; Clark, 2013). Integrated Information Theory (IIT) proposes that consciousness relates to intrinsic causal–informational structure quantified by Φ (Tononi, 2004). Despite their power, these frameworks leave open a bridging question: what additional constraint differentiates merely functional inference from phenomenal presence?
This manuscript proposes a minimal candidate constraint:
Phenomenal consciousness arises when ongoing inference achieves dynamic entropic closure under recursive self-modeling.
Consciousness, in this view, is associated with a regime of predictive dynamics—neither informational stasis nor unconstrained flux—where uncertainty exchange becomes locally balanced while inference remains active.
2 Operational foundations: information, entropy, and closure
2.1 Information as predictive mutual dependence
Here, “information” is used operationally as predictive mutual dependence between a system’s internal states and its expected sensory states, consistent with predictive-processing accounts (Clark, 2013) and formal treatments under the FEP (Friston, 2010). This anchors “information” to estimable statistical dependencies rather than treating it as a metaphysical primitive.
2.2 Entropy flux across the system–environment boundary
Let internal states be x and sensory states y. Let conditional entropy H(y∣x) quantify uncertainty in sensations given internal state. Define an entropy-flux current Js as the effective flow of conditional uncertainty across the system–environment boundary in predictive state space. Dynamic entropic closure is expressed as the operational constraint:
Equation 1 is not presented as a complete mechanistic law; it is a measurable constraint on net uncertainty exchange. Related “flux–balance” approaches are standard in nonequilibrium thermodynamics and information thermodynamics (Parrondo et al., 2015).
Crucially, Equation 1 is compatible with high energetic throughput: living systems are energetically open yet can maintain relatively bounded uncertainty exchange by continuously reducing prediction error via perception and action (Friston, 2010).
2.3 Dynamic entropic closure
Dynamic entropic closure means Equation 1 holds under non-zero internal inference dynamics (ongoing model updating). Closure is not defined as informational absence.
3 Resonant closure: from predictive loops to phenomenal presence
3.1 Resonance as phase-coherent recursion
Predictive systems couple top-down expectations with bottom-up sensory evidence. When reciprocal message passing becomes sufficiently coherent across hierarchical levels, inference can enter a resonant regime: recurrent prediction–error loops become phase-locked, enabling stable integration rather than uncontrolled divergence. This idea aligns with “communication-through-coherence” proposals (Fries, 2005) and with evidence linking conscious states to long-range synchronization and integration (Varela et al., 2001; Deco et al., 2011; Casali et al., 2013).
3.2 What “closure” adds
The central claim is that resonance becomes phenomenal when it is also closed in the entropic sense: prediction and evidence are recursively coupled such that net uncertainty exchange is minimized (Equation 1), while internal inferential dynamics remain rich.
Operationally, resonant closure predicts a conjunction:
• Sustained functional integration / complexity (distributed, differentiated activity) (Casali et al., 2013),
• Phase-coherent interactions supporting effective communication (Fries, 2005; Varela et al., 2001; Deco et al., 2011),
• Reduced net uncertainty exchange (bounded entropy flux) under ongoing inference (Friston, 2010; Parrondo et al., 2015).
4 Avoiding the trivial-closure objection
A key objection is that a system could satisfy “no net entropy flux” trivially if Js = 0, i.e., if there is no information flow at all. This objection is valid—and it is why closure must be dynamic.
Accordingly, this manuscript proposes three jointly necessary constraints for candidate phenomenal consciousness:
1. Non-zero internal inference dynamics (ongoing predictive updating) (Friston, 2010).
2. Recursive self-modeling (the system tracks/predicts its own predictive state; minimal self-reference) (Clark, 2013).
3. Dynamic entropic closure (Equation 1 holds on relevant time scales despite 1–2) (Parrondo et al., 2015).
The first constraint rules out inert “empty” systems; the second rules out purely feedforward prediction; the third rules out unconstrained uncertainty exchange. In FEP terms, this corresponds to balancing accuracy and complexity rather than collapsing into trivial underfitting or pathological overfitting (Friston, 2010).
5 Relation to existing theories
5.1 Free energy principle and predictive processing
Under the FEP, self-organizing systems minimize variational free energy by maintaining a generative model that predicts sensory inputs and acts to sample expected states (Friston, 2010). Predictive processing provides an algorithmic interpretation via hierarchical prediction-error minimization (Clark, 2013).
Resonant closure is proposed as a phenomenal sub-regime of FEP: not all free-energy-minimizers are conscious, but conscious systems should occupy a regime where recursive prediction–error exchange becomes phase-coherent and informationally closed (Equation 1) over time scales relevant to integration of conscious content (Mashour et al., 2020).
5.2 Integrated information theory
IIT focuses on intrinsic causal structure and quantifies integrated information Φ (Tononi, 2004). A modern formalization (IIT 3.0) makes explicit how Φ relates to cause–effect power over system partitions (Oizumi et al., 2014). The present account is not IIT.
Instead:
• Integration (Φ) concerns intrinsic causal–informational interdependence (Tononi, 2004; Oizumi et al., 2014).
• Closure (Equation 1) concerns boundary conditions of uncertainty exchange under ongoing inference (Friston, 2010; Parrondo et al., 2015).
High Φ may facilitate closure by strengthening internal dependencies, but neither implies the other. This is consistent with empirical work suggesting separable dimensions of conscious state (e.g., complexity vs. other organizational properties) (Casali et al., 2013; Schartner et al., 2015).
6 The stationarity problem: consciousness is dynamic
A common critique is that “equilibrium” seems incompatible with lived dynamics: learning, surprise, conflict, temporal flow. The key clarification is that resonant closure is metastable and local, not global and static.
The FEP already treats organisms as nonequilibrium steady states maintained by continuous exchange (Friston, 2010; Mediano et al., 2022). Equation 1 should therefore be read as a local balance constraint over time windows relevant to conscious integration (hundreds of milliseconds to seconds), not as global stationarity.
Surprise and learning can be interpreted as controlled, transient perturbations of closure that trigger model revision; closure is then re-established at a new model configuration (Barrett and Seth, 2011; Bastos et al., 2012; Bohm, 1980; Carhart-Harris and Friston, 2019; Fries, 2015; Friston et al., 2006, 2020; Hohwy, 2013; Penrose and Hameroff, 2014; Seth and Friston, 2016; Tononi and Koch, 2016). This is compatible with global workspace dynamics where conscious access can co-occur with high local prediction error while global organization remains viable (Dehaene and Changeux, 2011; Mashour et al., 2020).
7 Neural and empirical implications
7.1 Neural coordination and perturbational signatures
Conscious states correlate with long-range coordination and dynamical complexity across thalamo-cortical and fronto-parietal networks (Varela et al., 2001; Deco et al., 2011). Perturbational measures such as the perturbational complexity index (PCI) track level of consciousness across wakefulness, sleep, and anesthesia (Casali et al., 2013). Complexity measures from spontaneous EEG/MEG also decrease during propofol anesthesia (Schartner et al., 2015).
The resonant-closure hypothesis predicts that these signatures coincide with bounded net uncertainty exchange under sustained inference—i.e., high internal predictive mutual information alongside reduced entropy-flux imbalance.
7.2 Candidate operationalization: closure vs. integration
Because Φ is in principle calculable from data (Tononi, 2004; Oizumi et al., 2014), a strong empirical opportunity is dissociation: cases where integration remains relatively high but closure fails (or vice versa). Recent information-theoretic tools for multivariate interactions and candidate integrated-information measures offer a starting point for operationalizing closure-adjacent quantities in neural data (Mediano et al., 2022).
7.3 Testable predictions
This framework yields falsifiable predictions:
1. Conscious vs unconscious dissociation: during loss of consciousness (e.g., propofol), breakdown of closure-related measures (entropy-flux imbalance across scales) should precede or exceed declines in integration proxies (Casali et al., 2013; Schartner et al., 2015).
2. Local metastability: conscious access episodes correspond to transient, metastable windows where phase-coherence and uncertainty-flux balance co-occur, rather than global stationarity (Dehaene and Changeux, 2011; Fries, 2005).
3. Perturbation response: systems in resonant closure show high PCI/complexity and rapid re-stabilization after perturbation; closure failure yields either runaway error propagation or damped low-complexity responses (Casali et al., 2013; Schartner et al., 2015).
8 Artificial systems and design principles
If resonant closure approximates a sufficient condition, artificial systems should require: (i) recursive predictive inference, (ii) self-modeling of inferential state, and (iii) bounded uncertainty exchange.
Global Workspace architectures already emphasize broadcast/coordination as a functional signature of conscious access (Baars and Franklin, 2009). The present account adds a boundary-condition criterion: the system must stabilize a self-referential inferential loop with locally minimized uncertainty exchange. More broadly, work on reconciling deep learning with structured representations suggests concrete routes to implementing internal self-modeling components needed for recursion and closure (Garnelo and Shanahan, 2019).
9 Structural analogies and cautions
The idea that “closure yields stable identity” invites analogy with symmetry breaking and order parameters in physics. The Higgs mechanism is a canonical example where symmetry breaking yields persistent properties (mass) (Higgs, 1964; Englert and Brout, 1964). Any use of this analogy here is structural rather than literal: the claim is not that a physical Higgs field exists in the brain, but that stable macroscopic properties can arise from self-consistent interactions.
Related “emergent from informational/thermodynamic constraints” perspectives appear in discussions of emergent gravity (Verlinde, 2011) and information-centric physics proposals (Wheeler, 1990). A relational stance in quantum foundations likewise treats physical description as fundamentally correlation-based, which motivates informational relationality without implying quantum mysticism (Rovelli, 1996). Thermodynamic derivations connecting spacetime dynamics and entropy illustrate the broader plausibility of deep links between physical law and informational constraints (Jacobson, 1995). Consciousness-as-a-state-of-matter proposals provide additional context for treating consciousness as a phase-like regime rather than a separate substance (Tegmark, 2015).
Where topology is referenced, it is intended heuristically to express robustness of organization rather than a computed invariant; for rigorous treatments of topological robustness in physics, (see Haldane, 2017).
10 Conclusion
Consciousness is proposed here as a dynamically self-stabilized informational regime: resonant closure, where recursive predictive inference becomes phase-coherent and net uncertainty exchange is locally minimized while internal informational dynamics remain high. The account (i) clarifies the explanandum (phenomenal consciousness), (ii) avoids trivial “empty system” closure, (iii) reconciles closure with dynamical experience via metastability, and (iv) yields predictions testable with neurophysiological and perturbational methods (Casali et al., 2013; Schartner et al., 2015).
If supported, resonant closure reframes consciousness as a lawful phase of information-processing dynamics—neither an epiphenomenal add-on nor a mere synonym for function, but a regime characterized by recursive self-modeling under dynamically maintained entropic boundary conditions.
Data availability statement
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics statement
Ethical approval was not required for the study involving humans in accordance with the local legislation and institutional requirements. Written informed consent to participate in this study was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and the institutional requirements.
Author contributions
BA: Project administration, Visualization, Formal analysis, Methodology, Writing – review & editing, Funding acquisition, Data curation, Software, Investigation, Writing – original draft, Resources, Conceptualization, Validation, Supervision.
Funding
The author(s) declared that financial support was not received for this work and/or its publication.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
Baars, B. J., and Franklin, S. (2009). Consciousness is computational: the LIDA model of global workspace theory. Int. J. Mach. Conscious. 1, 23–32.
Barrett, A. B., and Seth, A. K. (2011). Practical measures of integrated information for time-series data. PLoS Comput. Biol. 7:e1001052. doi: 10.1371/journal.pcbi.1001052,
Bastos, A. M., Usrey, W. M., Adams, R. A., Mangun, G. R., Fries, P., and Friston, K. J. (2012). Canonical microcircuits for predictive coding. Neuron 76, 695–711. doi: 10.1016/j.neuron.2012.10.038,
Carhart-Harris, R. L., and Friston, K. J. (2019). REBUS and the anarchic brain: toward a unified model of the brain action of psychedelics. Pharmacol. Rev. 71, 316–344. doi: 10.1124/pr.118.017160,
Casali, A. G., Gosseries, O., Rosanova, M., Boly, M., Sarasso, S., Casali, K. R., et al. (2013). A theoretically based index of consciousness independent of sensory processing and behavior. Sci. Transl. Med. 5:198ra105. doi: 10.1126/scitranslmed.3006294,
Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behav. Brain Sci. 36, 181–204. doi: 10.1017/s0140525x12000477,
Deco, G., Jirsa, V. K., and McIntosh, A. R. (2011). Emerging concepts for the dynamical organization of resting-state activity in the brain. Nat. Rev. Neurosci. 12, 43–56. doi: 10.1038/nrn2961,
Dehaene, S., and Changeux, J.-P. (2011). Experimental and theoretical approaches to conscious processing. Neuron 70, 200–227. doi: 10.1016/j.neuron.2011.03.018,
Englert, F., and Brout, R. (1964). Broken symmetry and the mass of gauge vector mesons. Phys. Rev. Lett. 13, 321–323. doi: 10.1103/physrevlett.13.321
Fries, P. (2005). A mechanism for cognitive dynamics: neuronal communication through neuronal coherence. Trends Cogn. Sci. 9, 474–480. doi: 10.1016/j.tics.2005.08.011,
Fries, P. (2015). Rhythms for cognition: communication through coherence. Neuron 88, 220–235. doi: 10.1016/j.neuron.2015.09.034,
Friston, K. (2010). The free-energy principle: a unified brain theory? Nat. Rev. Neurosci. 11, 127–138. doi: 10.1038/nrn2787,
Friston, K., Kilner, J., and Harrison, L. (2006). A free energy principle for the brain. J. Physiol. Paris 100, 70–87. doi: 10.1016/j.jphysparis.2006.10.001,
Friston, K., Parr, T., and de Vries, B. (2020). The graphical brain: belief propagation and active inference. Neurosci. Biobehav. Rev. 118, 42–64. doi: 10.1016/j.neubiorev.2020.07.005,
Garnelo, M., and Shanahan, M. (2019). Reconciling deep learning with symbolic artificial intelligence: representing objects and relations. Curr. Opin. Behav. Sci. 29, 17–23. doi: 10.1016/j.cobeha.2018.12.010
Haldane, F. D. M. (2017). Nobel lecture: topological quantum matter. Rev. Mod. Phys. 89:040502. doi: 10.1103/revmodphys.89.040502
Higgs, P. W. (1964). Broken symmetries and the masses of gauge bosons. Phys. Rev. Lett. 13, 508–509. doi: 10.1103/PhysRevLett.13.508
Jacobson, T. (1995). Thermodynamics of spacetime: the Einstein equation of state. Phys. Rev. Lett. 75, 1260–1263. doi: 10.1103/physrevlett.75.1260,
Mashour, G. A., Roelfsema, P., Changeux, J.-P., and Dehaene, S. (2020). Conscious processing and the global neuronal workspace hypothesis. Nat. Rev. Neurosci. 21, 751–764.
Mediano, P. A. M., Seth, A. K., and Barrett, A. B. (2022). Measuring integrated information: comparison of candidate measures in theory and simulation. Entropy 24:1051.
Oizumi, M., Albantakis, L., and Tononi, G. (2014). From the phenomenology to the mechanisms of consciousness: integrated information theory 3.0. PLoS Comput. Biol. 10:e1003588. doi: 10.1371/journal.pcbi.1003588,
Parrondo, J. M. R., Horowitz, J. M., and Sagawa, T. (2015). Thermodynamics of information. Nat. Phys. 11, 131–139. doi: 10.1038/nphys3230
Penrose, R., and Hameroff, S. (2014). Consciousness in the universe: a review of the ‘Orch OR’ theory. Phys Life Rev 11, 39–78. doi: 10.1016/j.plrev.2013.08.002
Rovelli, C. (1996). Relational quantum mechanics. Int. J. Theor. Phys. 35, 1637–1678. doi: 10.1007/bf02302261
Schartner, M. M., Seth, A. K., Noirhomme, Q., et al. (2015). Complexity of multi-dimensional spontaneous EEG decreases during propofol induced general anaesthesia. PLoS One 10:e0133532. doi: 10.1371/journal.pone.0133532,
Seth, A. K., and Friston, K. J. (2016). Active interoceptive inference and the emotional brain. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 371:20160007. doi: 10.1098/rstb.2016.0007
Tegmark, M. (2015). Consciousness as a state of matter. Chaos Solitons Fractals 76, 238–270. doi: 10.1016/j.chaos.2015.03.014
Tononi, G. (2004). An information integration theory of consciousness. BMC Neurosci. 5:42. doi: 10.1186/1471-2202-5-42,
Tononi, G., and Koch, C. (2016). Consciousness: here, there and everywhere? Nat. Rev. Neurosci. 17, 450–461. doi: 10.1038/nrn.2016.44,
Varela, F. J., Lachaux, J.-P., Rodriguez, E., and Martinerie, J. (2001). The brainweb: phase synchronization and large-scale integration. Nat. Rev. Neurosci. 2, 229–239. doi: 10.1038/35067550,
Verlinde, E. P. (2011). On the origin of gravity and the laws of Newton. J. High Energy Phys. 2011:029. doi: 10.1007/jhep04(2011)029
Keywords: consciousness, entropy flux, free energy principle, neural synchrony, phenomenal experience, predictive processing, self-reference
Citation: Arneth B (2026) Resonant closure: consciousness as a dynamically self-stabilized informational state. Front. Hum. Neurosci. 20:1742084. doi: 10.3389/fnhum.2026.1742084
Edited by:
Domenica Veniero, University of Nottingham, United KingdomReviewed by:
Joseph Monaco, SelfMotion Labs, United StatesEnrique Aramendia, Independent Researcher, Pamplona, Spain
Copyright © 2026 Arneth. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Borros Arneth, Ym9ycm9zLmFybmV0aEBzdGFmZi51bmktbWFyYnVyZy5kZQ==