Consciousness Naturalized: A Falsifiable Substrate-Agnostic Consciousness Theory

Consciousness Naturalized: A Falsifiable Substrate-Agnostic Consciousness Theory

Originally published: February 5, 2026.   Revised: November 10, 2026. This is the live website reading edition; the substantive archived research edition remains v1.1.0.

Thermodynamic and organizational constraints as proposals for discriminating tests, with no final ontology assumed.

Research record: Article-specific archive (GitHub)   ·   Research-hub copy (GitHub)   ·   Canonical article: SweetRationalism.com.

Archived edition v1.1.0: Version DOI 10.5281/zenodo.24119682   ·   All-versions DOI 10.5281/zenodo.24119681.

Versions and archives: Computational supplement (ZIP)   ·   Complete methodology release (ZIP)   ·   Full revision audit (public TXT)   ·   Prospective rival-test protocols (v0.1.0)   ·   Sweet Rationalism website reconstruction history (consolidated ZIP).

Bottom line. A consciousness theory earns support by specifying what it predicts within a declared boundary, why its strongest rival predicts something different, and what would force revision. Kleiner and Hoel’s falsification dilemma is a useful formal stress test; Popperian falsifiability, Deborah Mayo’s severe testing, and Antony Flew’s qualifications challenge help distinguish genuine criticism from endlessly accommodating reinterpretation. Organizational closure and thermodynamic constraints offer testable candidates, without establishing an identity with experience. The massive local success of physical explanation does not prove exhaustive physicalism; an unresolved phenomenal question likewise supplies no positive evidence for a supernatural or anti-physicalist cause. The result sought here is a reusable, symmetrical evaluation method that can revise its own claims.

Abstract: What This Article Actually Proposes

This article develops a reusable method for evaluating consciousness theories under shared evidential constraints. Each account should specify the target phenomenon, subject boundary and timescale, its measurement and inference procedures, what its distinctive assumptions predict, the serious rivals they face, and what would make those assumptions lose support. The same method applies to physicalist, anti-physicalist, panpsychist, supernaturalist, computational, biological, and presently unclassified proposals. A theory does not gain confirmation merely because another theory leaves an explanatory gap.

The methodological claim. Physical mechanisms have substantial demonstrated local predictive and interventional success. That success warrants credit within the boundaries tested; it does not, without a further argument, establish exhaustive metaphysical physicalism or settle all possible realizations of phenomenality. Equally, explanatory limitations of physical mechanisms do not count as positive evidence for nonphysical or supernatural additions. Claims of universality, necessity, intrinsic phenomenal constitution, or exceptions to physical explanation each need their own bridge argument and evidence appropriate to the claim.

Framework foundations. The article uses the Recursive Corrigibility Framework for its error-correcting comparison method, Provisional Boundary- and Scale-Relative Realism for its explicitly non-final ontological commitments, the Interoperability Framework Bridge for translating claims across distinct evidential contexts, and the Sweet Rationalism Ethos for the norms governing criticism, correction and possible harms.

Candidate hypothesis, not an established identity. A bounded, temporally extended, internally differentiated control state maintained by organized constraint relations may explain important aspects of what we call conscious perspective. The stronger hypothesis that such a state constitutes phenomenal presence remains a provisional, target-specific abductive proposal. It is not a definition of consciousness, a universal equivalence, or the claim that organization, recurrence, self-maintenance, information processing or a successful report individually suffice for consciousness.

What the established research contributes. Thermodynamics constrains concrete physical operations, including irreversibly resetting information-bearing memory; it does not assign a universal Landauer cost to each distinction maintained or report produced. Organizational-closure research offers measurable models of mutually maintained constraints. Neuroscientific perturbation measures and clinical findings probe aspects of brain dynamics under specific validation conditions. These resources can improve experimental design.

How the programme can lose. Identify physically realizable contrasts with matched reporting but differing proposed mechanisms; specify interventions, negative controls, feasible resources, uncertainty, and independent consciousness evidence. The closure-specific hypothesis loses explanatory credit if a simpler physiological, dynamical, computational, or plasticity model predicts the same phenomena as well or better. A constitutive identity loses support when calibrated reports and interventions dissociate from the alleged constitutive structure. Formal countermodels can defeat a claimed necessity inference without showing that the disputed property never exists.

The durable proposal is a framework for testing both naturalistic and non-naturalistic claims with symmetrical standards while keeping the phenomenology of actual participants in view. Its specific models are revisable. Neither a scientifically successful methodology nor a rival account of an explanatory gap gets to declare the final ontology.

Methodology: Preserve the Target, Compare the Rivals, Record the Loss

The methods used throughout this article are an explicit research contract, not independent evidence for its preferred hypothesis. Several were developed more fully in my analysis of Chalmers’s formulation of the hard problem and subsequent exchanges. Their value here is measured by the errors they expose, the rival predictions they separate and the corrections they force. A simpler, equally resourced review may outperform them, in which case their extra machinery should be discarded.

• Explanandum preservation and positive residual. Name the phenomenon or question before decomposing it. Reconstruct every component and the relations among them. If a further, unified explanatory problem is asserted, ask what positive target is lost by the decomposition. A missing component or indispensable relation defeats the proposed dissolution; the absence of an identified residual does not, by itself, settle physicalism.
• Explanatory stopping-rule symmetry. Apply a stated standard of explanatory completion to identity, constitution, fundamental psychophysical laws, intrinsic nature and primitive powers alike. Differences in modal or conceptual arguments can justify different verdicts, but another unanswered “why?” is not a discriminator by itself.
• Evidence tribunals and construct bridges. Specify whether the claim is empirical, causal, formal, phenomenological, semantic, modal, normative or ontological. Test it with the evidence and proof duty it actually requires. State the subject boundary, temporal grain, variables, invariant claims, measurement errors and information lost when evidence is transported between contexts or substrates.
• Carrier/addition ablation. Separate independently warranted mechanisms from proposed intrinsic or phenomenal additions where the target survives the separation. Compare the strongest whole-account alternative when the addition is claimed to be constitutive. A functional success cannot silently become ontological confirmation; a rival’s failure does not confirm this account without further discrimination.
• Independent labels and dependent evidence. Record how a subject or trial is classified before testing a proposed marker against that classification. The same variable must not define both the predictor and its consciousness outcome. Participant reports deserve calibration, not either infallibility or dismissal; overlapping instruments and common source judgments cannot count as independent replications.
• Prospective rival fingerprints and local corrections. Freeze exact rival versions, outcomes, interventions, negative controls, scoring rules, missing-data treatment and failure conditions before decisive observations. Evaluate joint prediction patterns with held-out data and matched model flexibility. Track separately mechanistic value, phenomenal necessity, sufficiency, constitution and claims about the problem’s formulation. Failure should demote only the assertion that depends on it.
• Genealogy and the metaproblem. Distinguish having an experience from judging that experience is inexplicable. Investigate how the second judgment tracks arguments, concepts, language and culture without assuming that explaining a judgment eliminates its object.

These rules apply in both directions. A naturalistic model must not make feeling disappear into a report variable, and a metaphysical rival must not gain a new cause or necessity merely because feeling remains inadequately explained. The operational implementations and their pre-data loss conditions appear in Part IX, Section XII and the public prospective protocol. They remain proposals until an actual study-specific registration, calibration, adequate design and independent test exist.

Prologue: Bateson’s Pattern and the Danger of a Universal Answer

Gregory Bateson’s question about “the pattern which connects” remains a useful invitation to examine how differences become consequential across living systems, minds and environments. In Mind and Nature: A Necessary Unity (1979), he called this a “metapattern”; in Steps to an Ecology of Mind (1972), he formulated information as “a difference which makes a difference.” These are questions about relations and consequences, not a demonstrated universal identity between information, thermodynamic dissipation and subjective experience. His work supplies a question with considerable explanatory reach. It does not oblige us to find one physical formula that explains every instance of order, identity or experience. See the source texts of Mind and Nature and Steps to an Ecology of Mind for the different original contexts.

One possible response is to examine how constraints shape the persistence and transformation of patterns. In biological systems, boundaries channel flows, physiological networks sustain functions, and feedback can help preserve viable organization under perturbation. Thermodynamics constrains these physical processes. Constraint satisfaction under a specified physical budget is a productive candidate for many local explanations. Its value depends on demonstrating which outcomes it predicts beyond established alternatives. The persistence of a mathematical identity, a stable digital memory and a living cell may involve different notions of “constraint” and different physical operations.

The temptation is to make the candidate universal. We could claim that every distinction requires a Landauer payment, every persistent system has organizational closure, and every conscious perspective is nothing more than that closure seen from within. Each escalation introduces a new assumption. None follows simply from the success of the preceding local explanation. Physical scientific methods have extraordinary results without having demonstrated exhaustive physicalism; phenomenal and metaphysical rivals remain accountable for any extra claims of their own.

This essay therefore proposes a research programme for comparing consciousness theories. It uses typed claims, explicit boundaries, primary-source checks, formal countermodels where suitable, matched rival explanations, interventions, negative controls and recorded loss conditions. Its provisional organizational hypothesis is a participant in that tournament. It is not the referee’s preferred winner by definition.

The central question is practical and demanding: what specific finding or argument would distinguish this theory from its strongest rival, and what would force us to revise the theory we currently favor? The rest of the article applies that question to thermodynamics, biological autonomy, neuroscience, panpsychism, process metaphysics, nonphysical-ingression proposals and the possibility of artificial consciousness. Sometimes a mechanism will win locally. Sometimes the evidence will not discriminate. Both results are more informative than declaring a final answer before the test.

Part I: The Physical and Organizational Constraints Worth Testing

The starting point is modest, though its implications are substantial. Real organisms, computers and measuring devices implement information-bearing distinctions through physical processes. Many of those processes have measurable thermodynamic costs. Some systems also maintain networks of constraints that help preserve their organization through change. These facts give consciousness research concrete variables and possible interventions. They do not yet tell us which physical organizations, if any, are intrinsically experiential, or whether all possible experience must be physically realized.

1.1 Landauer: A Real Bound with a Specific Domain

Rolf Landauer’s 1961 analysis, Irreversibility and Heat Generation in the Computing Process, linked logically irreversible computational operations to the physical cost of implementing them. Its standard example is resetting a reliable, initially unknown classical bit to a fixed state. For an appropriate isothermal, symmetric-memory reset in contact with a reservoir at temperature \(T\), the ideal limiting average heat released to the reservoir is

\[ \langle Q_{\mathrm{env}}\rangle\geq k_{\mathrm B}T\ln 2. \]

Here \(k_{\mathrm B}\) is Boltzmann’s constant. The information-theoretic reduction is one bit, and the bound concerns logical erasure under the specified physical assumptions. It does not say that every discriminable state, continuing memory, feedback loop or informational distinction dissipates that amount of heat at each moment. A stable physical memory can persist for a finite interval without being erased. The energy and error rates of actual operation depend on the implementation, noise, timing, barriers, control protocol, reservoirs and reliability requirement. Reversible computation, emphasized by Charles Bennett, provides a crucial counterexample to any claim that one fixed Landauer payment accompanies every logical operation.

Bérut and colleagues (2012) tested a controlled one-bit erasure protocol using a colloidal particle. Their result was that mean dissipated heat approaches the Landauer bound as the erasure cycle becomes long. Neither each single trajectory nor each arbitrary device is required to dissipate exactly \(k_{\mathrm B}T\ln 2\). That distinction matters for every later mathematical claim in this paper.

The robust thesis is that physically implemented distinctions participate in a thermodynamic accounting appropriate to the actual operation. The stronger thesis that all distinction maintenance entails a universal positive Landauer-rate bound is false. Nor does demonstrating a physical accounting for observed interactions establish, without a separate argument, that every conceivable experiential property is exhausted by physical description. A proposed causal interaction with measurable physical effects should specify its physical coupling and measurable consequences. A claim about intrinsic constitution may instead face phenomenal or metaphysical proof duties; it cannot be dismissed by inserting an inapplicable erasure formula.

1.2 Montévil and Mossio: Organizational Closure as a Testable Model

Maël Montévil and Matteo Mossio (2015) describe biological organization as a closure of mutually dependent constraints. The point is more discriminating than generic feedback. In a cell, a membrane constrains exchange, transport processes depend on the membrane, and metabolic processes replenish or repair the components that sustain both. Constraints act on matter and energy flows and are themselves maintained through organized activity. Which dependency exists depends on the boundary, environment and timescale being analyzed.

For the research programme here, we should map specific constraints, identify the processes maintaining each one, and test relevant dependencies by intervention. The article’s earlier contrast between a living organism and every crystal, whirlpool or machine was too categorical. Natural and engineered systems admit gradations of autonomy, externally supported persistence and internal regeneration. The existence of a feedback cycle is not by itself proof of the richer biological closure relation specified by Montévil and Mossio.

Closure research has predictive potential for robustness, repair, coordinated regulation and viability. Its relevance to consciousness is a separate hypothesis. Showing that a closed network helps an organism maintain itself does not establish that the network feels anything, that it is the sole possible realization of perspective, or that a human–AI loop constitutes a single subject. Those would each require additional evidence and carefully justified system boundaries.

1.3 Ladyman and Ross: Structural Realism Without Final Ontology

Ladyman and Ross’s Every Thing Must Go (2007), Steven French’s structuralist work and Carlo Rovelli’s relational approach to quantum mechanics challenge familiar intuitions about isolated objects and observer-independent properties. They support serious investigation of relational and scale-sensitive explanations. They do not constitute one accepted proof that relata are metaphysically nonexistent or that all facts must be only relations. The Stanford Encyclopedia of Philosophy on structural realism records important variants and objections within ontic structural realism itself.

The useful inference is local: scientific entities may be individuated by stable relational, dynamical and intervention-sensitive patterns rather than by immutable intrinsic labels. Such patterns can be real and explanatory without being fundamental constituents at every scale. This is the provisional starting point of my Provisional Boundary- and Scale-Relative Realism, not a demonstrated final ontology. An account of selfhood can therefore examine continuity, causal dependence, memory and boundary maintenance instead of postulating an unchanging inner spectator. That does not settle whether experience has intrinsic aspects or whether all consciousness talk reduces to grammatical confusion.

Nagarjuna’s analysis of dependent origination and the Buddhist critique of an independently existing self offer a valuable philosophical comparison. Calling them simply an early form of ontic structural realism erases differences in their arguments and aims. An analogy is useful when it generates explicit questions about dependence or individuation; it is not evidence that two traditions share the same ontology.

1.4 Constraints and Causes: Different Explanatory Questions, Sometimes the Same Mechanism

A causal explanation asks what made a difference to a particular outcome, under a model of possible interventions. A constraint explanation asks which possibilities are permitted, prevented or stabilized by a structure, rule, coupling or boundary condition. These questions can be complementary. A membrane can both constrain possible ion trajectories and cause a measurable difference when its permeability is changed. The phrase constraint explanation does not immunize an account against causal testing when it asserts causal efficacy.

Mathematical constraints, physical boundary conditions and biological maintenance relations must be distinguished. An equation restricting a set of solutions is a mathematical property of a model. A membrane or circuit physically implementing a restriction has a causal history, tolerances and possible failure modes. Moving between the mathematical and physical descriptions calls for a map of the quantities preserved and the interventions predicted. Marc Lange’s work on explanation by constraint is relevant to the first task; Montévil and Mossio’s work addresses a particular biological form of the second.

If consciousness were identical to a specified organization, the identity could in principle make some causal-production questions inappropriate, as asking what additional substance produces a triangle’s three-sidedness would be inappropriate. But the identity must first be established. An asserted analogy cannot serve as the needed phenomenal bridge. This article instead asks which observations would make organizational constitution more plausible than alternative explanations, and which would force us to revise it.

1.5 Deacon, England and Hoel: What the Models Actually Entail

Terrence Deacon’s Incomplete Nature (2011) offers an influential account of constraints, absent possibilities and the emergence of biological function. We can use it to examine how counterfactual alternatives matter to a system’s behavior without inferring that selection among physical alternatives, by itself, creates experience. Constraints can be represented as exclusions from an admissible state or trajectory set. Real physical outcomes still arise through dynamics and interaction. A protein fold, for example, depends on the energy landscape, kinetics and environment; it is not invariably the global minimum-energy structure. An organism’s death involves causal failures and processes, even when the failure includes loss of regulatory constraints.

Jeremy England’s 2013 statistical-physics paper on self-replication examines entropy production and constraints associated with replication. It does not establish a universal law that all driven systems evolve to maximize energy absorption or that every alternative trajectory is thereby deleted. Those broad claims would need a separate theorem and specified domain. Physically driven self-organization is compatible with several mechanisms and is not direct evidence of phenomenality.

Hoel, Albantakis and Tononi (2013) construct an intervention-sensitive measure of effective information and demonstrate cases where a chosen macroscopic model scores higher than a microscopic description. This is an important challenge to the presumption that the finest grain is always the best causal representation. It is conditional on the system and measure, including state spaces and intervention distributions. The result does not prove that every macro-description is intrinsically more causal, that the microphysics has been violated, or that causal emergence entails a conscious experiencer.

The paper’s practical proposal is to use these complementary models for specific targets: compare how well a constraint architecture predicts perturbation recovery, information flow, error correction and long-term survival against simpler dynamical and physiological alternatives. No identity with experience follows from that comparison alone.

1.6 Downward Constraint Propagation Without a New Force

Top-down and bottom-up causal descriptions can refer to the same physical dynamics at different scales. A pipe’s geometry alters accessible flow paths, while the pipe itself is supported by interactions among its constituent materials. A biological membrane changes transport and signaling; biochemical processes also build and degrade that membrane. No new force need be postulated simply to describe the effects of changing the higher-level boundary. Whether that description adds explanatory value is an empirical modeling question, not a guarantee supplied by the word emergence.

An adequate test compares matched models and interventions. If the macro-model improves held-out predictions or identifies controllable variables unavailable to a practical micro-model, it earns a local explanatory role. If its apparent advantage comes entirely from an unfair comparison, incompatible interventions, or choice of measurement scale, the advantage disappears. Claims about physiology, phenomenology and metaphysics remain separately typed.

1.7 Formal Characterization: State, Viability, Dependency and Cost

Reproduction: The formal unit, erasure, viability and inference checks are runnable in the public computational supplement. These bounded calculations test specific obligations without assigning a universal thermodynamic floor for information.

The following definitions specify the research formalism and its domain of application. Variables refer to a specified system boundary \(B\), environment \(e\), observation grain and time horizon \(\tau\). State dimensions and accessible controls must be declared in each application.

Dynamics. For a deterministic controlled system, let \(X\) be a state space, \(U\) a set of admissible controls and \(E\) an environment space. Consider

\[ \dot x(t)=f_B(x(t),u(t),e(t),t),\qquad x(t)\in X,\ u(t)\in U. \]

This form assumes the required regularity for solutions; noisy systems need an appropriate stochastic model and corresponding probabilistic claims. A physical constraint might restrict states to \(K\subseteq X\), velocities to an admissible subset of a tangent bundle, or entire trajectories to a subset of path space. These are different mathematical types. We must not treat them as equivalent by notation alone.

Viability. Given a specified acceptable region \(K\subseteq X\), the finite-horizon viability set is

\[ \operatorname{Viab}_{[0,\tau]}(K) =\{x_0\in K:\exists u(\cdot)\text{ admissible, }x(t;x_0,u,e)\in K\ \forall t\in[0,\tau]\}, \]

where the environment assumptions must be fixed. The quantifier over disturbances changes for robust viability, which requires suitable controls against an explicitly defined disturbance class. This is related to Jean-Pierre Aubin’s viability-theory programme. A viable state is not thereby a conscious state.

Constraint dependence. Let \(\mathcal C_B=\{C_1,\ldots,C_n\}\) be physically instantiated constraints with associated maintenance processes \(P_i\). Define a directed dependency relation \(D_{ij}=1\) when a prespecified intervention that disrupts a relevant contribution of \(C_j\) causes a reproducible impairment in the maintenance of \(C_i\), after appropriate controls. The resulting graph can contain cycles or strongly connected components. Graph cycles are candidate structural signatures, not sufficient proofs of Montévil–Mossio organizational closure: one must also establish constraint/process correspondence, relevant timescales, real maintenance, and the treatment of external supports. This definition is explicitly operational; a different intervention or boundary can yield a different graph.

Thermodynamics. For a system exchanging heat with one reservoir at temperature \(T\), with \(Q_{\mathrm{env}}\) the heat delivered to the environment, the usual second-law accounting is

\[ \langle\Delta S_{\mathrm{sys}}\rangle+\frac{\langle Q_{\mathrm{env}}\rangle}{T}\geq0, \]

under the assumptions of the chosen thermodynamic description. For nonequilibrium systems with several reservoirs, chemical exchanges, feedback or strong coupling, a more complete entropy-production balance is needed. Neither this inequality nor the Landauer erasure bound entails the earlier proposed expression \(\Delta S_{\mathrm{env}}\geq k_{\mathrm B}\ln2\,I(C_i)\) for every maintained constraint. No such universal theorem is established here. Claims about energy, bits, physical entropy and information-theoretic scores must state their units and not conflate them.

Hypotheses and possible losses. Let \(H_C\) be the hypothesis that closure-related variables materially improve prediction of a chosen control or viability target. Let \(H_P\) be the stronger claim that an identified control regime constitutes phenomenal perspective in a defined class of systems. \(H_C\) may earn support through held-out perturbation results and calibrated comparison to simpler models. \(H_P\) requires independent evidence addressing phenomenality and its proposed identity. If closure variables do not improve performance, demote \(H_C\) at that target. If evidence of phenomenality systematically dissociates from the proposed constitutive regime, demote \(H_P\), with the measurement limitations and rival explanations recorded. Neither failure licenses an automatic promotion of its competitors.

Part II: What Thermodynamics Contributes to the Study of Perspective

The question remains formidable: why should any physical process have a perspective? Detailed mechanisms can tell us how an organism detects, remembers, integrates or acts. A philosophical critic may still reasonably ask whether those descriptions account for phenomenal presence. The distinction between a successful causal model and an established constitutive identity is central to this research programme. The Stanford Encyclopedia of Philosophy overview of consciousness explains why the so-called hard problem survives even substantial mechanistic progress.

The proposal here is to make the question more discriminating. We can ask which conditions for bounded perspective an explanatory model predicts, which rival predicts them differently, and what independent evidence could support the additional claim that these conditions constitute experience. Some versions of the question may dissolve when their hidden assumptions are rejected. Others remain open. Neither outcome can be stipulated in advance.

2.1 Boundaries, Internal Regulation and Physical Accounting

Living organisms distinguish internal from external conditions through membranes, receptors, control processes and memory. A membrane makes exchanges selective; a chemotactic regulatory network changes action according to chemical gradients. The relevant variables can be measured and perturbed. Energy and entropy balances constrain these concrete operations, although the correct balance depends on the process and environment specified in Part I.

The term perspectival control can describe a system’s locally organized, action-relevant modeling of its own condition and surrounding world. This is an operational concept. It can be studied in bacterial chemotaxis, neural regulation and engineered controllers without assuming that each is conscious. Organisms can use local feedback effectively with no rich, explicit self-model; stable nonliving physical objects can preserve distinctions for finite periods without active self-repair. Neither Landauer’s erasure principle nor far-from-equilibrium persistence proves universal self-modeling.

Karl Friston’s free-energy framework offers mathematical models of inference, action and adaptive regulation under specified assumptions. Its variational information-theoretic quantities must not be equated with thermodynamic free energy by name. A good candidate model should predict interventions more accurately than a less elaborate regulatory or dynamical model, and its variables should be identifiable from measurements. Even strong results concerning control or inference remain evidence about those capacities; phenomenal constitution is a further question. Bruineberg, Kiverstein and Rietveld (2018) offer an ecological-enactive interpretation of FEP that is distinct from the 2022 Markov-blanket critique. Pezzulo, Parr and Friston (2024) analyze active inference as a candidate theory of sentient behavior. That model warrants a direct comparison of specified behavior and internal variables, without accepting sentience from the model label alone.

2.2 The Phenomenal Question and the Symmetry Test

David Chalmers’s hard problem does not depend merely on an undefined phrase. It asks why physical and functional processes should be accompanied by, or identical to, subjective experience. Conceivability, explanatory-gap and knowledge arguments have sophisticated supporters and critics. Their success or failure must be evaluated on their actual logical and modal premises, without first assuming either exhaustive physicalism or its denial.

A physicalist can propose that experience is constituted by some organized activity. A Russellian or panpsychist can propose that physical descriptions omit intrinsic experiential character. An idealist, a dualist, and a supernaturalist may offer other metaphysical bridges. These proposals do not all imply the same empirical predictions. Where a claim asserts an additional physical intervention or observable effect, it owes a concrete mechanism and differentiating test. Where its content is modal, phenomenological or ontological, the appropriate question is whether the argument warrants that content and what alternatives remain.

Neuroscientific success provides unusually strong local grounds for physical mechanisms in the targets tested. It is not a deduction that physics exhausts the fundamental nature of all things. Conversely, an unresolved gap in physical explanation is compatible with many possibilities and gives no particular supernatural account a free likelihood advantage. The methodological programme should therefore test its own organizational constitutive hypothesis by the same standards it demands from rivals.

There is also symmetry in explanatory endpoints. A proposed identity, constitution relation or fundamental psychophysical law may each explain something while leaving a further “why that relation?” askable. The relevant comparison is what each account predicts or unifies, and whether its distinctive modal and phenomenal arguments justify stopping there. Chalmers himself gives reasons to resist brute identities and allows that a genuinely non-question-begging functional argument could resolve the hard problem. His position cannot fairly be dismissed as special pleading merely because he accepts one kind of theoretical endpoint and questions another. Those reasons, and our own proposed endpoint, remain open to criticism.

2.3 Time, Memory and the Thermodynamic Arrow

Temporal experience offers an especially promising research target because it has measurable duration judgments, retention effects and perturbable neural mechanisms. William James’s specious present, discussed in the Stanford Encyclopedia of Philosophy on temporal consciousness, names the brief experienced interval in which change and persistence can appear together. Debates remain over whether experience itself extends over time, represents a temporal interval, or achieves continuity through other mechanisms.

Huw Price and Carlo Rovelli have examined the relationship between thermodynamic time asymmetry and physical records. The formation, reliability and eventual loss of memory can involve irreversible physical processes. Yet the inference from the thermodynamic arrow to the felt flow of time requires additional psychological and phenomenological premises. Time in physics need not be defined solely by entropy production, and the fact that a memory is physically implemented does not show that every act of recording constitutes an erasure with a Landauer payment. An account of the subjective present cannot simply set it equal to a thermodynamic integration interval without measurement.

A useful experiment would jointly manipulate working-memory retention, neural or computational integration timescales, metabolic variables and reports of subjective duration while controlling attention and task difficulty. Rival models could be fitted on training data and tested on held-out interventions. If metabolic or closure-specific variables contribute no advantage beyond sensory, attentional and recurrent-dynamics models, their claimed special role should be reduced. A stable association between those variables and duration reports would still leave the phenomenal bridge underdetermined.

Our best provisional conclusion is that thermodynamics supplies real constraints on the machinery of memory and temporal regulation. Its explanatory contribution to felt duration must be established empirically. A strong claim that subjective time logically follows from entropy production has not been derived.

Part III: The Experiencer, Personal Identity and the Limits of Dissolution

The question “Who experiences?” may carry an assumption worth challenging: that a separately identifiable, unchanging spectator must sit behind everything a person sees, remembers and does. A dynamic account of selfhood can explain many functions of self-reference without such a spectator. This is a promising philosophical interpretation. It is not a demonstration that the phenomenal subject is nothing over and above an organizational pattern, or that the question of subjectivity is only a grammatical error.

3.1 The Grammatical Trap, Properly Scoped

Natural language sometimes implies an entity where a process description would suffice. “It is raining” is an ordinary example. Likewise, describing thought in the first person does not, by grammar alone, prove the existence of a further substance that performs the thinking. A model can examine the actual construction of self-reports, memory links, agency attribution and body ownership.

Introspection supplies evidence about what is presented to a participant, with familiar limits of attention, report and interpretation. Failure to find a separate “inner witness” through introspection may challenge a particular spectator model. It does not deductively establish that no phenomenal subject exists, nor that a first-person perspective can be fully redescribed as a behavioral pattern. Reports of selfhood, dissociation or selflessness should be tested in context and compared with neurological and psychological evidence.

The same restraint applies to the claim that there is one further, irreducible Hard Problem. After reconstructing the specific problems about presence, qualities, perspective, unity and report, ask what positively identifiable explanatory demand remains, rather than treating the word “hard” as proof of an additional entity. An unaccounted phenomenal relation or independent modal argument would defeat a particular decomposition. A successful decomposition would revise the proposed problem-form, not make the underlying experiences disappear.

Nagarjuna’s dependent-origination arguments and the Buddhist anatta tradition provide important resources for questioning a self with independent essence. Their formulations and philosophical goals differ from contemporary neurocomputational models. The comparison invites precise questions about dependence and continuity; it cannot settle a present-day identity theory merely by historical analogy.

3.2 Identity as Stable Corrective Pattern: A Hypothesis with Rivals

Constraint maintenance offers one way to model persistence through change. An organism can replace components and retain regulative organization; a person can revise beliefs while continuing a narrative and social life. Identity may be tracked by continuity of processes, memory, causal history, bodily relations and an ability to recover from perturbation. These dimensions need not be coextensive. The SEP entry on personal identity describes serious disagreements among biological, psychological and other continuity criteria.

The formula “identity is invariance under perturbation” is useful when its relevant invariants, boundary and timescale are named. It is inadequate as a universal numerical-identity theorem. A pattern that can be copied twice raises a branching problem; stable organization across cells does not alone decide whether those cells are one person, two persons or neither. In a formal application, specify an equivalence relation on observed states, a history dependence condition and the interventions under which it is claimed to hold. The result can be evaluated for prediction and error without assuming a metaphysical essence.

Levin and colleagues’ planarian regeneration work provides concrete evidence that anatomical regulation is history-dependent. Durant and colleagues (2017) document persistent changes in regenerative outcomes and a morphologically normal cryptic phenotype whose subsequent outcomes can be stochastic. Stable anatomical regulation is more flexible than a simple fixed blueprint model. Durant and colleagues (2019) separately manipulated early planarian bioelectric states, finding altered anterior–posterior polarity after a time-limited intervention. This result concerns the timing of physiological pattern control, not proof that morphospace or phenomenal properties have a particular ontology. Neither the experiments nor an attractor description establishes that personal identity just is an attractor, or that mathematical Platonism is false.

3.3 Epicurus, Death and the Difference Between Experience and Harm

Epicurus argued that death cannot be directly experienced by someone after that person’s existence ends. His position is a serious contribution to thinking about fear, temporal concern and the conditions under which harm is felt. It is not a completed ethical proof that death can never harm someone. A deprivation account argues that death can harm an individual by removing future goods even when no subsequent experience of the harm occurs. The SEP review of the philosophy of death sets out both views and their objections.

Epicurus’s second Principal Doctrine, preserved by Diogenes Laërtius (X.139), says: “Death is nothing to us. For what has been dissolved has no sensation, and what has no sensation is nothing to us.” His Letter to Menoeceus develops a related argument. The quotation supports a claim about absent sensation, while the deprivation objection shows why sensation alone may not exhaust the ethics of death.

A biological death involves physical processes, physiological thresholds, social effects and sometimes irreversible loss of the capacities supporting consciousness. The account of causal failure is not complete merely by calling death a “loss of closure.” Nor does it license claims about all possible forms of postmortem existence, which belong to distinct empirical and metaphysical debates. Where clinical decisions affect an actual patient, diagnostic uncertainty and justified existing protections remain central.

AI systems offer a useful bounded analogy about persistence. A model’s weights may remain stored when execution stops; a tool-using deployment may preserve memory, state and agency across interactions; a broader human–AI system can exhibit coordinated control. These are different entities and timescales. Interruption of a session, destruction of a process, deletion of memories and loss of a biological organism have different causal and ethical consequences. Calling them all “death” obscures the properties under investigation. A human–AI loop’s organizational continuity is not evidence, by itself, of a single subject or of direct moral standing for every component.

Consciousness, welfare and personhood require their own evidence and normative bridges. Some systems may warrant precautionary treatment even when their experience is uncertain; some may matter instrumentally because they support already protected people or other beings. Neither a biological label nor a computational label is a complete inference. The question for this article is which forms of persistence help explain which observable capacities, and what separate evidence could justify any conclusion about a bearer of phenomenal perspective.

The original philosophical insight survives as a live proposal: selfhood need not be modeled as an unchanging occupant located behind a body. What an organizational account establishes about identity or consciousness depends on how it performs against bodily, psychological, phenomenological and other rivals, with the boundary and loss conditions stated.

Part IV: The Conditions of Intelligible Explanation, and Their Limits

There is a real philosophical question behind the article’s earlier appeal to distinction, identity, relation and constraint. Whenever we offer an explanation, we distinguish some target from alternatives, identify what remains stable across descriptions or observations, establish relevant relationships, and specify conditions that make a difference. These operations are unavoidable for an inquiry of this kind. Their use in reasoning is not yet a theorem about the ultimate constitution of reality.

4.1 Four Disciplines of Description

Distinction: What is the target, and how is it distinguished from neighboring constructs? A consciousness measure must distinguish responsiveness, cognitive capacity, phenomenal report and welfare rather than treating them as interchangeable.

Identity: What makes two observations count as instances of the same pattern across times, observers and scales? The answer may depend on a declared equivalence relation, observation grain and tolerance for variation. Mathematical equality, numerical identity and persistence of an organism are different notions.

Relation: Which dependencies, interactions or representations connect the relevant variables? Mere correlation, causal intervention, constitutive dependence and modal entailment require different arguments. A successful map must declare what it preserves and what it discards.

Constraint: What possibilities are ruled out by tested data, established laws, a model’s assumptions or a definition? These sources of restriction differ in evidential standing. A formal constraint can narrow a mathematical model while the model’s physical applicability remains unsettled.

The four questions give the audit a portable structure. They do not prove that all existence must consist solely of constraints or that every intelligible world must be thermodynamically dissipative. In particular, Landauer’s principle concerns physical information processing in specified implementations. It cannot, by itself, answer why anything exists.

4.2 Transcendental Arguments Are Not Cosmic Existence Proofs

Kantian transcendental arguments typically begin from an accepted feature of thought or experience and try to establish a necessary condition of that feature. Their force depends on the bridge from what a subject must think or presuppose to what reality must contain. The Stanford Encyclopedia of Philosophy’s treatment of Kant’s transcendental arguments discusses the limits of this step and Barry Stroud’s challenge to ambitious anti-skeptical conclusions.

Consider the proposed argument: explanation presupposes distinguishable conditions; absolute nothingness supplies no such conditions; therefore absolute nothingness is impossible, and there must be something. The conclusion does not follow. An inability to describe or explain a supposed state is different from a proof that the state is metaphysically impossible. The original passage moved from a thesis about intelligible descriptions to a thesis about necessary existence without supplying that modal bridge.

A more defensible result survives: to compare consciousness theories, investigators need sufficiently determinate targets, concepts, models and evidence to articulate disagreement. This is a methodological condition on the inquiry. A critic can reject it only by showing a different workable mode of comparison, or by arguing that some explanatory targets resist its application. It remains open whether reality ultimately contains intrinsic properties, irreducible experience, abstract structures, or other features not exhausted by current descriptions.

4.3 The Question That Remains Open

The question “Why is there something rather than nothing?” has substantial metaphysical and cosmological versions. Neither thermodynamic irreversibility nor organizational closure supplies a demonstrated necessary-and-sufficient answer. The relevant physical laws already presuppose a domain of states, temperatures, processes or fields on which they operate. Applying them to the possibility of any domain at all would require an independent argument.

We can preserve the deeper challenge without a cosmic victory claim. Scholars proposing a final ontology must identify the extra premises that support its quantifier over all possible things. Someone claiming exhaustive physicalism, a necessarily experiential universe, an indispensable Platonic realm, or a transcendent creator has different commitments to establish. The success of a local description neither forces acceptance of a particular ontology nor rules out every alternative ontology.

This section should be judged by the same standard as the rest of the article: if a critic shows that a proposed distinction or equivalence relation fails to support a valid inference, the conclusion must be revised at precisely that point. The methodological usefulness of the four disciplines can survive the defeat of an extravagant metaphysical interpretation.

Part IV-A: Whitehead’s Fallacy of Misplaced Concreteness

Alfred North Whitehead warned against treating a selective abstraction as if it gave us the complete concrete actuality being studied. A map, numerical measure or theoretical construct can be extraordinarily useful while leaving out aspects of its target. The danger arises when an investigator forgets what was omitted and attributes to the abstraction more reality, completeness or causal power than the argument has earned.

That diagnostic applies to every camp in this article. A Platonic morphospace may organize possible developmental outcomes without showing that an independently existing realm causally selects one. A measure such as integrated information \(\Phi\) can characterize a formal causal structure without proving it is an intrinsic phenomenal magnitude. A model of prediction error may improve behavioral forecasts without establishing that a brain literally minimizes that exact objective under every interpretation. Equally, our own constraint closure may be a useful abstraction from material processes without exhausting their phenomenal or intrinsic character.

The Critical Difference Between Representation and Ingression

Suppose a model describes a landscape of possible anatomical forms. It helps classify experimental outcomes and identify interventions in a developmental system. That is an explanatory contribution. The stronger claim that forms enter physical bodies from an independently existing nonphysical possibility space requires an additional bridge. What would change physically if the proposed ingression were absent? How does the account differ in intervention results from a developmental or dynamical model? The experimental results may support the first explanatory role even when the second remains unearned.

An equivalent demand applies to exhaustive physicalism. Equations and mechanisms that successfully describe particular interactions justify claims at their established domains. A move from those successes to the assertion that the physical story exhausts every intrinsic feature of existence is an extra metaphysical inference. Whitehead’s diagnostic cuts both ways.

A Practical Bridge Audit

For each important construct, record:

1. Target and type: which phenomenon is being explained, and is the claim representational, causal, constitutive, modal or ontological?
2. Boundary: which system, temporal grain and observational scale does the model cover?
3. Gain: what does the construct predict, compress or help control that a serious rival does not?
4. Bridge: what warrants the inference from the representation to a physical cause or intrinsic ground, and what does the mapping preserve or lose?
5. Reversal: what observation, derivation or counterargument would lower support, including for the present framework?

A claim may remain philosophically meaningful when no discriminating empirical test is available. It must then be presented in the right evidential category. The absence of an experimental discriminator limits empirical preference, while formal and conceptual criticisms remain relevant. This prevents a theory’s historical prestige or an opponent’s unanswered question from being mistaken for a positive result.

Part IV-B: The Hidden Premises and the Conditional Phenomenal Bridge

The hard problem of consciousness is worth taking seriously because subjective experience is part of what a theory must address. Whether its apparent distance from physical explanation is a genuine ontological division, an epistemic limitation, or a misleading framing remains disputed. Chalmers’s conceivability and explanatory-gap arguments deserve their strongest version. The corresponding responses from phenomenal-concept, representationalist, illusionist and constitutive physicalist approaches also deserve evaluation on their actual premises. No side should be allowed to turn an unresolved premise into an automatic result.

The Premise to Challenge

These objections have distinct philosophical targets. Nagel (1974) asks what an objective description can capture of a particular creature’s point of view. Jackson (1982) develops the knowledge argument and explicitly distinguishes it from Nagel’s imaginative or standpoint problem. Levine (1983) formulates a gap in the explanatory force of proposed physical–qualitative identifications. They are substantive challenges to an asserted phenomenal identity, not interchangeable demonstrations of a nonphysical substance. Moving from a specified epistemic or explanatory limitation to an extra ontology needs an argued bridge; equally, exposing that limitation is not itself a proof of exhaustive physicalism.

There is a reciprocal gap on the physicalist side. Identifying the neural or computational processes that reliably correlate with reports, discrimination, arousal and behavior does not automatically show that these processes exhaust what experience is. The bridge from local intervention success to exhaustive metaphysical physicalism needs its own argument. The SEP distinction between physicalism and related methodological doctrines is important here: using physical methods successfully does not settle the doctrine that everything is physical.

The Supernatural Cheese: What the Reductio Actually Targets

Imagine somebody defending an invisible ingredient in cheese with this move: whenever chemical analysis explains a property of the cheese, the defender claims the invisible ingredient was never meant to explain that property. When a new experimental prediction fails, the ingredient retreats again. The only remaining claim is that ordinary analysis could never reach it, and the inability to detect it is presented as a reason to believe it exists.

That is a recognizable form of immunization by ad hoc qualification. Antony Flew’s challenge in the philosophy of religion and Popper’s work on risky testing help identify the danger; Deborah Mayo’s severe-testing approach makes it clear that a claim deserves credit when it survives methods with a genuine opportunity to expose a mistake. The point of the cheese reductio is the evidential maneuver. It does not show that private experience is imaginary, that every version of dualism has this form, or that metaphysical arguments can be settled by a kitchen experiment.

A sophisticated anti-physicalist may advance a modal argument rather than a hidden causal ingredient. Such an argument requires examination of the modal bridge, its counterexamples and its competing explanations, not a demand that it predict a different EEG trace. A supernaturalist who claims an additional agent directly changes a physical process makes a different, causally testable assertion. A critic who asks only for a better account of why an otherwise known neural process should feel a certain way has not thereby postulated an occult force. Distinguishing these claims makes the challenge more difficult and more fair.

The Interoperability-Framework Bridge provides the broader record of how constructs, evidence and loss conditions may be translated between scientific, phenomenological, formal and ontological inquiries. Translation can preserve a claim’s intended target while leaving an additional premise unproved; shared vocabulary is never an evidential bridge on its own.

What the Empirical Sciences Can Establish

The evidence for many physical mechanisms is strong at specific targets. Thermodynamics constrains demonstrated physical operations; neuronal interventions change perception, attention, report and behavioral capacities; clinical instruments can help distinguish states of impaired responsiveness and inferred consciousness. These findings deserve their full force. None should be weakened to make a favored nonphysical account look equally supported.

The extrapolation must still be checked. A validated clinical marker is not a direct instrument for observing every possible first-person condition. Physically implementing a memory does not show that its every state update is a Landauer reset. Organizational closure can explain robustness and self-maintenance without being established as the unique cause or constitution of phenomenality.

The empirical research programme can proceed while treating the metaphysical residual as open. It asks what a reported experience varies with, which interventions alter it, whether the relation transports across subjects and substrates, and what serious rival models predict differently. Its successes are local and cumulative. They can become broad and powerful without silently becoming a proof about the complete ontology of the cosmos.

Define the Explananda Before Identifying Them

The term consciousness spans at least phenomenal presence, reportable access, awareness of information, self-representation, cognitive control, wakefulness and sometimes affective valence. These should not be collapsed into one construct. A patient who cannot respond may still be conscious; a system with sophisticated control need not have demonstrated feelings. First-person reports are fallible evidence about a target, neither unquestionable revelations nor empty behavioral noise.

For any empirical test, specify the subject boundary \(B\), temporal grain \(\tau\), the target capacity \(Y\), and how \(Y\) is calibrated. Record the auxiliary assumptions connecting the indicator to the target. Clinical and phenomenological standards may differ from those appropriate for artificial agents or minimal biological systems. Cross-substrate inference needs an explicit mapping, with known losses of information and competing realizations.

The Candidate Hypothesis and Its Unfinished Bridge

Here is the original constructive idea, stated at the strength it presently earns:

\(H_{B,\tau}\) (conditional organizational-phenomenal identification): At certain specified boundaries and timescales, an integrated, temporally stable, action-relevant control state that models both world and self may constitute aspects of phenomenal perspective. This is a target-specific abductive hypothesis. It remains open whether the proposed organization is necessary, sufficient, partly constitutive, or merely correlated with the subjective target.

The inference from measured organization to experience is a substantive phenomenal bridge, not a definition. The hypothesis needs discriminating evidence against higher-order theories, recurrent processing, global-workspace approaches, integrated-information models, biological necessities, illusionist interpretations and other serious rivals. Its support is conditional on the relevant phenomenal evidence, construct validity, intervention results and comparison protocol.

Bridge rule, proposed rather than proved. If a particular organization improves predictions of independently calibrated conscious-state changes under held-out interventions, including matched report and performance conditions, relative to prespecified serious rivals, it earns explanatory credit at the target tested. Identity requires additional evidence or philosophical argument beyond predictive covariation, especially when competing realizations remain observationally equivalent. The rule should be weakened or discarded when an equally parsimonious model matches or outperforms it, or when a well-calibrated dissociation undermines its claimed constitutive role.

This is a stronger research contribution than a verbal promise to have solved the hard problem: it states where evidence is owed. It does not guarantee that the evidence will favor the theory originally proposed here.

Concrete Tests and Alternative Explanations

Access versus presence. Global neuronal workspace models specify mechanisms of broad access, report and flexible control. Higher-order and recurrent approaches offer different predictions about metarepresentation and local processing. Block (2011) argues for a possible overflow of perceptual consciousness beyond cognitive access, while Rounis and colleagues (2010) investigate a specific prefrontal intervention and metacognitive awareness measure. Neither result is a substrate-general definition of consciousness. Compare perturbations, no-report designs and confidence calibration to investigate the boundaries of each model. The COGITATE 2025 adversarial tests should be incorporated with their actual preregistered targets rather than used as an all-purpose verdict.

Replication and task sensitivity. The prefrontal stimulation result should not be treated as settled: Bor and colleagues (2017) reported failures to replicate the metacognitive impairment, while Ruby, Maniscalco and Peters (2018) challenged aspects of that replication’s design and statistical interpretation. Martin, Lane and Hsu (2023) again found no corresponding impairment with real figures, but found reduced metacognitive efficiency for an illusory-figure task. These divergent findings constrain claims about the role of dorsolateral prefrontal cortex in visual metacognition; none, by itself, establishes the presence or absence of phenomenal consciousness.

Perturbation and recovery. Casali and colleagues’ Perturbational Complexity Index, and later research in disorders of consciousness, assess properties of measured neural responses under defined stimulation and validation conditions. These measurements can be valuable for clinical inference. PCI is not an identity test for phenomenality, is not equal to IIT’s \(\Phi\), and cannot isolate closure from every correlated physiological mechanism. Reports of conscious experience under otherwise unusual responsiveness or anesthetic conditions are especially important counterchecks.

Boundary-specific control. For organizational closure, map maintenance dependencies under interventions, measure viability and relevant control capacity, and test whether the model predicts outcomes beyond ordinary dynamical and physiological baselines. The test must distinguish a true additional mechanism from relabeling an observed feedback loop. A human–machine coupling may enable useful system-level agency without establishing a common conscious subject.

Privacy and immediacy. Internal access privileges, inference latency and self-model structure may explain why some subjective judgments seem private, stable or immediate. They need experiments: manipulate report latency, introspective confidence, accessibility and neural or computational control while keeping alternatives plausible. An account of the judgment of immediacy is not yet an account of phenomenally felt immediacy. No conscious participant should be told that their experience has been explained away merely because a computational model reproduces their words.

How the Bridge Loses, and What Remains

The immediate tests have distinguishable outcomes. A closure model may improve prediction of survival or adaptive repair while contributing nothing to consciousness inference. A workspace model may win a report task while still leaving phenomenal constitution unsettled. An anti-physicalist proposal may offer a serious modal argument without an independently discriminating empirical model. The resulting degrees of support should be recorded claim by claim, not converted into a single winning camp.

The position argued for here is disciplined openness with strong local realism. We should trust the physical mechanisms where they survive severe tests, pursue the identity hypothesis where it generates risky evidence, and refuse to promote either an explanatory gap or a successful methodology into a final metaphysical inventory. The paper’s reusable contribution is the tribunal and its loss conditions. Every favored bridge, including this one, remains subject to correction.

Part V: What Competing Consciousness Accounts Actually Establish

The fairest test of this proposal is one that could make it lose. A rival’s failure to resolve a particular question does not award organizational closure the answer, any more than a gap in physical neuroscience establishes a supernatural cause. Different claims require different tests. An empirical mechanism needs observations and interventions. A mathematical theorem needs its assumptions and derivation. A constitutive identity requires a bridge from the measured structure to the property identified. An ontological thesis needs arguments commensurate with its scope. None is refuted merely because it is not a laboratory hypothesis.

The following comparisons therefore ask what each programme contributes and where its extra commitment earns, or fails to earn, explanatory credit. Some of the objections apply just as sharply to the constraint-based hypothesis proposed here.

5.1 Panpsychism: The Combination and Evidential-Discrimination Problems

Panpsychist approaches propose that experience, experientiality, or its relevant precursor is fundamental or more widespread than ordinarily supposed. The versions differ. Constitutive Russellian panpsychism, cosmopsychism, and micropsychist accounts should not be collapsed into one assertion that every electron has an empirically detectable miniature mind. The Stanford Encyclopedia of Philosophy’s treatment reviews these varieties and the substantive combination debate.

One serious problem is how micro-level subjects or qualities would compose the unified phenomenal states attributed to larger subjects. Coleman (2014) poses a substantial micro-subject combination objection that cannot be resolved by merely naming a whole; it does not, by itself, disprove all forms of panpsychism. Chalmers’s distinctions among subject, quality, and structural combination help sharpen that burden. A philosophical proposal may address these with modal or constitutive arguments without claiming immediate laboratory confirmation. But if it also purports to explain why a particular organism reports or discriminates a particular experience, we can ask what its experiential primitive adds beyond a matched neural, causal, or organizational model. What different result follows if the extra property is absent?

Galen Strawson’s realistic-monism argument challenges physicalists who accept the reality of experience to explain its place within a fully physical reality, rather than assuming a nonexperiential base suffices. Its premises deserve philosophical argument; it is not a laboratory confirmation of ubiquitous experience. The absence of a current empirical discriminator limits an empirical claim to panpsychism’s distinct contribution. It does not make the ontology logically false or empty. The reciprocal obligation is uncomfortable for this essay too: naming integrated constraint dynamics does not demonstrate that those dynamics are intrinsically felt. Both accounts owe an argument for their constitutive step. The more elaborate mechanism should receive no automatic credit simply for being more familiar to a physical scientist.

5.2 Michael Levin’s Platonic Morphospace: The Ingression Bridge

Levin’s experimental contributions to bioelectric pattern regulation deserve full credit. Controlled perturbations of membrane voltage and ion channels can produce striking changes in regeneration and anatomy. The planarian experiments reported by Durant et al. (2017) show stable, path-dependent changes and a cryptic stochastic regenerative phenotype. They demonstrate that anatomical outcomes depend on developmental history and altered bioelectric regulation. That is important positive biology.

A stronger interpretation proposes that biological structures causally access an independently existing Platonic morphospace. When that is offered as an explanation for a physical result, it needs a bridge. What property of the alleged ingression interface is measured? Which intervention or prediction distinguishes access to a pre-existing nonphysical form from a well-specified dynamical, developmental, or organizational account? The Chladni Plate Challenge to Michael Levin’s Biological Platonism develops a related dynamical alternative (see the Michael Levin evidence and scholarly criticism index for the complete research collection).

Levin’s published response to my criticism included this formulation: “Yes, I think a better science of Platonic forms + their interfaces will force a re-do of Landauer’s Principle (or at least, its limitations).” The final parenthesis matters. Levin allows that a future proposal may revise the range of applicability of a physical principle rather than its established conditional result. He also distinguishes his broader research on cognition and engineering from a demonstrated theory of consciousness. The unresolved empirical request is therefore specific: what measured interaction, new constraint, or prediction would make that future revision necessary? An anticipated future theory is not yet evidence that an existing physical bound has failed.

The two-headed and cryptic planarian results do not refute mathematical Platonism or every possible morphospace ontology. They challenge particular predictions of obligatory return to a unique canonical form, if a specific proposal actually entails those predictions. A Platonic account can allow multiple forms. The problem with adding an unlimited set of new forms after every observation is loss of differential prediction unless further restrictions are fixed independently. Likewise, a physically causally effective interface needs dynamical and thermodynamic accounting, but Landauer’s erasure bound alone does not prove that an interface is impossible or must violate conservation. This is an open challenge to a specified causal addition, not a universal metaphysical verdict.

5.3 Basal Cognition: Specify the Capacity Before Awarding the Label

Levin’s broader work on non-neural problem-solving, bioelectric regulation, and goal-directed morphology motivates serious research into biological cognition at unusual scales. It is legitimate to use cognitive terms in a stipulated functional sense when they improve prediction or control. Doing so does not automatically establish animal-like consciousness, subjective valence, welfare interests, or an immaterial source of agency.

The distinction is explicit in Chis-Ciure and Levin’s (2025) peer-reviewed formalization of basal cognition. Its search-efficiency measure, K, compares an agent’s cost of navigating a defined problem space against a specified random-walk baseline on a logarithmic scale. The authors discuss the possibility of consciousness outside brains but explicitly withhold claims about consciousness in that study. Levin made the scope choice even plainer in a recorded discussion (about 1:22:52): “you think I was going to fold in consciousness into all this stuff until it got, you know, further settled. No way.” The paper therefore deserves evaluation on its stated cognition target, while any separate claim linking these capacities to phenomenal experience or Platonic forms still needs an independent evidential bridge. Measured search efficiency is neither a consciousness measure by definition nor evidence that consciousness is absent.

A metric of behavioral competence or energetic efficiency must be evaluated as the quantity it measures. To infer cognition in a stronger sense, the investigator must define the capacity, compare less theory-laden alternatives, and show what the cognitive model predicts that a control-theoretic or selection-history account does not. A robust system may be goal-directed without a reflective self-model; a system can carry information without necessarily possessing feelings. The same restrictions apply to our language of closure, perspective, mattering and agency. A useful metaphor becomes a scientific model when its variables and loss conditions can be exposed to tests, not merely when it is rhetorically vivid.

5.4 Quantum Accounts: A Real Physical Constraint, a Live Technical Debate

Penrose and Hameroff’s Orch-OR proposal involves specific claims about quantum states in microtubules and their role in consciousness. Tegmark’s published decoherence calculation poses a serious quantitative obstacle to particular long-lived coherent states in a warm biological environment. The relevant comparison concerns the precise superposition, localization, environmental coupling and time required by the proposed computation.

It would be an error to declare all quantum-consciousness possibilities mathematically falsified by one estimate. Hagan, Hameroff and Tuszynski’s published response contests the physical model and relevant scales. Their alternative calculations are not themselves evidence that Orch-OR produces subjective experience. The adjudication requires independently measured parameters and prospective effects that distinguish the proposed mechanism from ordinary neurophysiology. The burden of a concrete quantum mechanism is demanding, but its failure would not automatically prove the organizational-closure account.

5.5 Mathematical Universe Hypotheses: Description, Realization and Ontology

Max Tegmark’s Mathematical Universe Hypothesis makes a strong ontological proposal about mathematical structure and physical existence. We should distinguish the defensible explanatory role of mathematical descriptions, the mathematical possibility of a system, its physical realization, and the claim that every such structure literally exists. Equivalence in an abstract description does not, without a bridge, establish a bearer of phenomenal experience.

The constructive challenge is to state what this ontology explains beyond the relevant mathematical and physical model, and which modal, mathematical or observational considerations favor the additional existence thesis. An empirically equivalent ontology may remain philosophically discussable while lacking relative empirical confirmation. Conversely, a constraint-based theory cannot simply substitute the slogan ‘thermodynamics is real’ for a complete account of why this organization would be experienced. We should hold both proposals to their respective proof duties.

5.6 Whitehead and Process Philosophy: Causal Inheritance and Felt Inheritance

Whiteheadian process philosophy offers a serious systematic account of becoming, relations and what he calls prehension. A live dispute concerns the additional identity that causal inheritance is, in some sense, inherently felt inheritance. Models of causal transfer, biological constraint and temporal dependence can explain specific observations without invoking felt inheritance at those targets. Such local explanatory dispensability challenges a necessity claim where the model is sufficient; it does not establish that feeling is absent throughout nature.

In his essay “There’s no scientific evidence that consciousness exists”, Matthew David Segall argues that first-person awareness is presupposed in scientific observation and cannot simply be read off third-person measurements. This is a substantive transcendental and phenomenological challenge, not a denial of first-person experience. It still leaves a further question: what would establish the additional universal claim that physical causal inheritance is inherently felt inheritance? The first-person premise alone does not supply that identity bridge. The same scrutiny applies to this essay’s own candidate closure-to-experience identity.

The relevant questions for a Whiteheadian defender are what the experiential addition uniquely establishes, whether there is a substantive modal or phenomenological argument for it, and where the claim about physical causation crosses into a universal identity. A process metaphysics need not be discarded merely because one proposed empirical bridge is absent. Nor does the constitutive organizational hypothesis in this essay win by default. Our own supposed closure-to-experience identity needs a comparably explicit bridge, with specified losses of support. The comparison remains open where neither account supplies discriminating evidence.

5.7 Higher-Order Theories: Representation, Access and Phenomenal Dissociation

Higher-order approaches associate consciousness with an appropriate representation of a first-order mental state. Their variants differ about conceptuality, functional role, relevant neural architecture and whether a higher-order state must itself be conscious. The familiar regress question is a challenge, not an immediate refutation: a proponent can consistently hold that the higher-order state is unconscious while making its target conscious.

The discriminating empirical questions concern possible dissociations among first-order perception, metacognitive accuracy, confidence reports and independently calibrated phenomenal indicators. Evidence that a perturbation alters metacognitive performance while preserving discrimination does not alone settle phenomenality. An organizational account may describe these same effects at a different level. It gains explanatory credit only if its specified organization predicts a held-out dissociation or intervention more accurately than an explicit higher-order model.

5.8 Free Energy Principle: What the Formalism Adds

Variational free energy in active inference is an information-theoretic functional, whereas thermodynamic free energy is defined in physical energy units. Their relationship cannot be supplied by a shared name. The FEP and active-inference literature offers useful mathematical tools for stochastic dynamics, perception, and control under specified generative models. A flexible mathematical representation may be valuable without every broad interpretation of it becoming independently testable.

Bruineberg, Dołęga, Dewhurst and Baltieri (2022) distinguish formal Markov-blanket relations from stronger claims that they carve nature at unique agent boundaries. Kirchhoff and colleagues (2018) offer a substantive autonomy interpretation of Markov blankets; Friston (2022) replies to critiques of how formal models map onto living systems. The competing interpretations require independent empirical boundary criteria. That distinction supplies a real criticism when an ontological conclusion rests solely on a convenient partition. The audit question for any particular FEP application is what its assumptions commit it to, which alternatives it outpredicts, and which observations expose it to failure. The constraint-based approach faces the same risk of redescribing persistence after the fact. Calling a persistent system ‘closed’ is not evidence that closure was the cause of its capacity or of phenomenality.

5.9 Integrated Information Theory: Intrinsic Structure and Its Phenomenal Bridge

Tononi and Koch (2015) articulate IIT’s proposed relationship between intrinsic causal structure and consciousness. Hedda Hassel Mørch (2019) presents a trilemma involving intrinsicality, non-overlap and reductionism, which extends beyond IIT alone. It is an objection with explicit alternatives, not an empirical proof that integration cannot matter. Its formal measures, postulates and empirical applications have changed across versions, so objections should name the version and proposition actually targeted. In particular, the clinically useful Perturbational Complexity Index is not the same quantity as IIT’s Φ and cannot be treated as a direct measurement of an identity between integration and experience.

The COGITATE adversarial collaboration (2025) found important challenges to specified predictions of both IIT and global neuronal workspace theory. The consortium makes its preregistered protocol and original analysis code (MIT license) available for independent examination; those files do not endorse this article’s organizational hypothesis. A failure of a preregistered prediction forces analysis of the specific model and its auxiliaries, not instant rejection of the entire philosophical programme. IIT owes a defense of the bridge between its formal intrinsic structure and phenomenal qualities, as well as feasible tests across matched substrates. Doerig et al. (2019) present an unfolding challenge to claims about causal structure; Tsuchiya and colleagues (2020) answer by disputing the functionalist premises. Cerullo (2015) questions how the integrated-information quantity should bear on consciousness. The competing arguments require their actual assumptions, not an automatic verdict for either side. An organizational account cannot claim victory by pointing to that bridge problem while leaving its own constitutive bridge unproved.

The strongest organizational test here is not simply “closed loop versus replay.” IIT also attributes importance to intrinsic causal organization and may predict the same difference. Its independent contribution must be calculated under the selected version and boundary, then compared with the proposed closure and repair variables in cases where their predeclared forecasts diverge. If IIT makes the same predictions or wins at matched complexity, this article gains no distinctive evidential credit from that experiment. Enactive and active-inference accounts deserve the same opportunity.

5.10 Global Neuronal Workspace: Strong Mechanism, Open Identity

Global workspace models study how information becomes widely accessible for report, action, memory and flexible control. They have produced influential neurocognitive research and concrete predictions. The interpretation of late ignition, posterior representations, prefrontal activity and no-report evidence remains sensitive to the task and measurement design; the COGITATE collaboration adds genuine adverse evidence against some anticipated signatures.

Global availability offers an informative mechanistic model for some cognitive operations. It does not by itself resolve whether conscious phenomenality is identical to access, accompanies it under further conditions, or sometimes dissociates from it. This is precisely a question on which the closure-based hypothesis must demonstrate added predictive value. Both hypotheses need carefully matched performance and perturbation conditions, with first-person evidence treated as fallible data rather than either a revelation or noise.

5.11 Hoffman’s Interface Theory: A Stronger Claim Than Selective Perception

Hoffman, Singh and Prakash (2015) introduced the Interface Theory of Perception in an evolutionary setting. Prakash, Stephens, Hoffman, Singh and Fields (2021) later stated and proved a conditional Fitness-Beats-Truth theorem under specified assumptions about world states, perceptions, payoffs and the competing strategies. The narrower point that perceptual evolution does not automatically guarantee veridical world reconstruction is worth taking seriously. Manolo Martínez (2019) constructed a family of counterexamples to a stronger general inference from usefulness to nonveridical representation. Which mathematical result applies depends on the actual competing strategy class, payoff structure and information sources. Neither paper by itself settles the ontology of perception.

The further claim that spacetime and physical objects are merely interface icons for an underlying domain of conscious agents requires its own argument. Quantum-gravity research questioning spacetime’s fundamentality does not by itself favor conscious-agent ontology. Equally, planarian path-dependent regeneration is not an experiment that directly falsifies Hoffman’s perception models; it concerns a different explanandum. A successful critique must identify a specific shared prediction and a result on which the models genuinely disagree. If the conscious-agent proposal yields no advantage over a well-specified evolutionary and dynamical account, its additional ontology remains underdetermined. That is narrower than proof of its negation.

The comparative lesson is not that the naturalistic organizational proposal has already won this tournament. The present evidence supports many locally successful models and exposes several ambitious bridges that remain unearned. The proposed method succeeds only to the extent that it finds real discriminators, accepts adverse results against its own favored hypotheses, and helps competing accounts make progress under shared constraints. In that form the criticism retains its point: explanatory gaps do not function as vouchers redeemable for an extra ontology, including an exhaustive physicalist one.

Part VI: The Tests We Actually Have, and What They Test

There are useful empirical tests for human responsiveness, covert command-following, sensory processing, metacognition, brain dynamics and some indicators associated with consciousness. Their existence directly challenges a sweeping claim that consciousness-related research has no testable content. It does not establish that we can directly classify every conceivable conscious bearer, or that one favored theory has already survived a decisive universal test.

6.1 Clinical Assessment

Giacino and colleagues’ Coma Recovery Scale–Revised (CRS-R) assesses behaviors associated with levels of consciousness under specified clinical conditions. Historical studies by Andrews, Schnakers and colleagues document consequential diagnostic disagreement, particularly when relying on informal examinations. Specific percentages depend on sample and diagnostic protocol; they should not be generalized to all facilities or times without checking the study design.

Owen and colleagues (2006) showed covert command-following through distinguished tennis and navigation imagery in a patient who met behavioral criteria for the vegetative state. Casali and colleagues’ PCI research (2013) offers another operational route through TMS-evoked brain-response complexity. These are genuine empirical achievements. Neither establishes a metaphysical equation between consciousness and closure or the reliability of the same indicator in all species, artificial systems and clinical conditions.

Global-workspace approaches, recurrent processing accounts and other neural theories make predictions about access-related neural dynamics. The COGITATE 2025 adversarial collaboration exposed important difficulties for specified predictions of both IIT and global neuronal workspace theory. These partial failures should feed a dependent claim audit rather than an unconditional verdict against either research programme.

In a prospective test of a candidate organizational marker, the clinical or participant-linked target must be scored independently of that marker. Record how awareness, responsiveness, communicative capacity and phenomenological reports were assigned, by whom, with what blinding and inter-rater dependence, before a closure model is trained. Reusing the same physiological feature to define the outcome and then predict it can manufacture apparently excellent validation. Ambiguous states remain ambiguous when the independent evidence does not decide; a marker should not decide a patient’s rights or treatment by stipulation.

6.2 Comparative Cognition

Different systems can be tested for discrimination, associative learning, flexible choice, counterfactual control, affective responses, developmental plasticity and other particular capacities. Human, animal and computational measurements vary in construct validity and available calibration. A complex signal or a behavior resembling human cognition is not a validated experience meter. On the other hand, absent human-style speech is weak evidence against animal or unfamiliar consciousness.

Comparisons with rocks, water chambers or engineered controllers can be informative when the exact measured feature is stated. They become irrelevant when a difference in memory or behavioral sophistication is silently promoted into an answer about intrinsic phenomenality.

6.3 Popper, Mayo and the Actual Asymmetry

Popperian falsifiability concerns how a claim exposes itself to potentially conflicting observations. Mature model testing also requires auxiliary assumptions, calibration, statistical power, replication and attention to underdetermination. In “Theology and Falsification,” Antony Flew warns of the “death by a thousand qualifications”: successive revisions can remove the content that originally made a claim testable. That is a challenge to specify genuine revision costs, not a rule that one failed observation deductively destroys an entire research programme. Deborah Mayo’s severe-testing framework makes an additional demand: results should have passed procedures with a genuine chance of revealing relevant errors.

Landauer’s specific bit-erasure claim, a biological closure dependency, a planarian regenerative outcome and a neuroscience prediction are different targets. Failure of a Landauer prediction under genuine matching conditions would bear on the thermodynamic claim. Failed replication of a particular bioelectric intervention would bear on that result and its dependent explanations. Neither directly falsifies every organizational model or settles consciousness. A critic of a Platonic or panpsychist ontology may need a modal or conceptual argument rather than a laboratory falsifier, while a proposed physical ingression mechanism owes physical and predictive accounting.

The valuable asymmetry is between what has earned local evidence and what is only compatible with it. It is not between “science” and everyone who disagrees with an identity hypothesis. If a rival makes a riskier, independently successful prediction, this programme must recognize that gain. If its own models lose their claimed advantage, it must say so and preserve the failure record.

Part VII: Cross-Tradition Comparisons Without False Convergence

Thinking about memory, relations and ecological dependence benefits from scholarship far beyond modern European philosophy. The methodological difficulty is to honor the differences. Two descriptions that use words like relation, process or constraint need not make the same metaphysical claim, identify the same bearer or establish the same explanatory bridge. A cross-tradition analogy becomes evidence only when its precise construct and independent predictions can be compared without erasing cultural context or reducing someone else’s knowledge to a label in this framework.

7.1 Songlines, Memory Practices and an Actual Experimental Result

Australian Aboriginal songlines involve complex relationships among land, knowledge, ancestry, story, navigation and responsibility. They are not reducible to memory tricks or to graph-theoretic “constraint paths.” Lynne Kelly’s work on memory traditions invites an informative comparison between situated, practiced recall and modern external recording. What it does not establish is a universal physical ontology.

Reser, Simmons, Johns and colleagues (2021) tested an Australian Aboriginal-inspired memorization method in a medical-education context. Their first study involved first-year medical students (\(N=76\)). Among participants beginning below perfect recall, 12/19 (63%) in the Aboriginal-method group subsequently reached 20/20 recall, compared with 10/22 (45%) in the memory-palace group. The paper labels 2.82 (95% confidence interval 1.15–6.09) an odds ratio for the Aboriginal-method group. An exact arithmetic check shows that 2.82 equals the ratio of its perfect-recall proportion to the study-wide baseline proportion, \((12/19)/(17/76)\), rather than a conventional odds ratio. A separate, unadjusted direct comparison of the two selected groups yields odds ratio 2.06 and two-sided Fisher exact \(p\approx0.350\); this exploratory result does not establish superiority on perfect recall. Both taught approaches improved overall recall, with comparable performance on that outcome, while the Aboriginal-method group showed better immediate sequence-recall results. At six weeks, only eight students returned; no robust long-term superiority conclusion is justified by that attrited follow-up. The authors explicitly distinguish the laboratory teaching method from actual culturally situated Songlines. This limited educational comparison is promising and worth replication. It is not evidence that all Indigenous knowledge systems outperform other methods, nor that an Indigenous ontology and thermodynamic monism have independently converged.

7.2 Relational Knowledge on Its Own Terms

Tyson Yunkaporta’s Sand Talk and Robin Wall Kimmerer’s Braiding Sweetgrass discuss situated knowledge, practice, reciprocity and relationships with land and other life. Their accounts have explanatory and ethical substance. Similarities with relational realism can identify a productive research question: which relations help preserve ecological or social functions over long timescales, and what observations demonstrate that dependence? Different traditions may supply different answers, values and forms of authority.

An Indigenous claim about kinship, respect or personhood should not be translated without argument into a claim that the relevant system is a dissipative graph. Nor should this article presume to speak for every Aboriginal, Potawatomi or other Indigenous tradition. The strongest collaboration would involve the relevant scholars and communities as interpreters and participants, with proper provenance, reciprocity and consent where new research is proposed. Shawn Wilson’s Research Is Ceremony (2008) supplies one grounded Indigenous research-methodology perspective; it must not be mistaken for authorization to generalize across distinct communities.

7.3 Neti Neti, Anatta and Different Philosophical Destinations

The Upanishadic neti neti (“not this, not this”) method and the Buddhist anatta traditions can both challenge unexamined identifications of self with particular passing states. It would nevertheless be a major distortion to treat them as equivalent proofs that no intrinsic reality exists and only organizational constraint closure remains. Some Vedantic interpretations of neti neti proceed toward an irreducible self or ultimate reality; Madhyamaka and other Buddhist traditions have different accounts of emptiness and conventional designation. Compare the actual texts through Patrick Olivelle’s Upaniṣads translation and Jay Garfield’s translation of Nāgārjuna, rather than treating either tradition as empirical replication. Their arguments should be read in their own philosophical contexts and compared claim by claim.

This difference is an especially good check on the temptation to count similar rhetoric as independent confirmation. Two traditions may share a technique of negation while drawing incompatible conclusions. The shared method may still be philosophically instructive, but one cannot use it as evidence for a conclusion that one of those traditions rejects.

7.4 Five Language Models Are Not Five Independent Replications

If several language models produce the phrase “constraint satisfaction under thermodynamic bounds” after receiving similar prompts or framework instructions, that result is an observation about the prompted systems. It does not yet demonstrate independent discovery of an invariant physical principle. Training corpora may overlap; prompt design, selection effects, common scholarly terms and the human orchestrator may create correlated outputs. Architecture differences alone do not remove those dependencies.

An informative study would predeclare the prompt variants, conceal the preferred target phrasing, compare models with generic baselines, count contradictory answers, preserve every failed branch and let independent evaluators score claims for novelty and validity. The same protocol should test whether a simpler control-theoretic or causal explanation does as well as the Recursive Corrigibility Framework. Until such controls exist, model agreement is weak and dependence-sensitive evidence, not a substitute for laboratory replication or peer review.

The honest result is more interesting than a predetermined convergence story. These traditions and computational models raise questions about how situated agents carve useful differences from a changing world. Where their predictions or arguments agree at a defined target, the agreement deserves examination. Where they diverge, we should preserve the disagreement and find out why.

Part VIII: Applying the Audit to This Article

The Recursive Corrigibility Framework must be capable of rejecting its own preferred consciousness hypothesis. It also needs to distinguish failure of a subsidiary empirical model from failure of an entire research method. An adversarial test of one equation may reveal a genuine defect in that equation while leaving independent observations and a reusable evaluation procedure intact.

8.1 Claims That Can Lose, with Their Actual Targets

Landauer: The bit-erasure bound applies under declared physical assumptions. Evidence that the bound fails under those assumptions would challenge the relevant thermodynamic claim. An abstract identity operation or a memory retained for a finite interval does not test an erasure bound. Conversely, an observed bit-erasure cost does not establish that every maintained distinction has that cost.

Organizational closure: At a specified biological target, ask whether a mapped network of mutually maintained constraints predicts resilience or recovery better than a matched dynamics-only or control-theoretic model. If it fails on held-out interventions, withdraw the asserted explanatory advantage at that target. An artificial system with comparable capacities despite a different organization challenges any asserted necessity claim.

Bioelectric memory: Replicate specified planarian interventions and their observed long-term morphological outcomes, including the cryptic stochastic state. Replication failure would revise a model of that experiment, not every attractor account, Platonism or consciousness theory. The primary Durant et al. (2017) claim should be preserved in its actual experimental scope.

RAF autocatalysis: Formal results about reflexively autocatalytic, food-generated sets need their stated model assumptions. A mismatch with a specified prebiotic setting would diminish the applicability of a proposed origin-of-life mechanism. It would not directly show that consciousness cannot occur under organizational closure or that all chemistry lacks autocatalytic organization.

Phenomenal identification: The conditional identity in this essay requires calibrated, target-appropriate evidence beyond prediction of report. If a simpler neural dynamical model matches every measured outcome, the closure-specific hypothesis loses its distinctive explanatory claim. If the proposed structure dissociates from independently justified phenomenal evidence, the constitutive hypothesis must be revised. The difficulty of producing decisive evidence should not be used to immunize the hypothesis.

Additional failure gates: If an outcome labelled “conscious” was defined from the closure measure itself, withdraw the purported validation and redo the label design. If held-out joint predictions provide no incremental gain beyond the strongest equal-budget rival, withdraw the claim of a unique organizational discriminator. If decomposition of the Hard Problem omits an independently defensible phenomenal relation or necessity argument, abandon that specific claimed dissolution while retaining any successful mechanistic explanations. These losses attach to different claims and do not logically entail a universal verdict for or against consciousness.

Exhaustive physicalism and supernatural or anti-physicalist additions: A claim that all reality is physical needs a warranted extension beyond local successful mechanisms. A claim of additional causally active nonphysical agency needs positive evidence for its contribution and a bridge to observed effects. A purely modal thesis may be disputed through modal arguments. Neither a physical explanatory gap nor a successful local mechanism earns the cosmic conclusion for free.

8.2 Questions Still Open

This article does not answer why a universe exists, whether experience has intrinsic qualities unrepresented by our best physical models, whether biological substrates are necessary for any phenomenal property, how to calibrate the experience of radically unfamiliar systems, or whether closure is ever sufficient for phenomenality. It cannot settle moral standing by counting feedback loops. The relevant uncertainties should accompany the claims rather than being treated as a footnote after a confident verdict.

The evidence for human conscious states is substantial and includes first-person reports, behavior, clinical histories, neurophysiology and intervention. Other organisms and engineered systems supply different combinations of evidence. The same rule of inference should apply to matched evidence while respecting genuine differences in what is known and the costs of a mistaken exclusion.

8.3 The Meta-Test: Does RCF Outperform a Simpler Method?

RCF should be benchmarked against a straightforward source-checking and rival-model protocol that does not use its distinctive vocabulary. Measure error discovery, claim typing, preregistered revisions, retained sources, successful predictions, reproducibility, human readability and time/cost. Check whether reviewers reproduce results without access to favored conclusions and whether prompt changes materially alter outputs.

If RCF fails to improve these outcomes, compress or abandon the redundant machinery. If its applications generate new unfalsifiable declarations while criticizing others for doing so, demote the framework’s own standing. Progress is an earned, revisable record of specific corrections and predictive gains. Neither the name falsification nor its replacement with corrigibility is a substitute for this test.

Part IX: What the Organizational Hypothesis Would Need to Establish

The positive view now has to earn the term identification. It is tempting to say that consciousness simply is organizational closure viewed from within. This was an intended conditional, target-specific abductive hypothesis, but the unqualified wording can sound like a settled universal identity. Organizational closure, self-modeling, integrative access and ongoing control are individually measurable constructs. Their relation to phenomenal presence remains a research question.

The hypothesis can be stated more carefully: in a declared class of systems, at boundary \(B\) and timescale \(\tau\), a temporally integrated, action-relevant control organization may be constitutive of some aspects of phenomenal perspective. “Constitutive” is a stronger claim than causal correlation or predictive usefulness. The class, proposed phenomenal target, rival realizations, and evidential bridge must be specified before evaluation. A network with organizational closure alone is not declared conscious, and a system without demonstrated closure is not declared unconscious.

9.1 Conceivability and the Actual Logical Burden

David Chalmers’s zombie argument asks whether a complete physical duplicate of a conscious person lacking phenomenality is ideally conceivable and, if so, whether that conceivability establishes metaphysical possibility. Critics dispute the conceivability–possibility bridge, the intended completeness of the physical description and the implications of phenomenal concepts. These are substantial modal questions, detailed in the SEP entry on philosophical zombies. Neither casual imagination nor a verbal decree that organization is experience settles them.

Daniel Dennett’s skepticism about our ability to imagine genuine functional duplicates supplies an important counterargument. But the existence of a complete functional theory of consciousness is precisely what the organizational proposal has yet to show. It would beg the question to assume that a fully specified closure architecture is already known to feel something, then use that assumption to rule out zombies. A defender of the zombie argument likewise owes a bridge from ideal conceivability to the relevant modal conclusion. The constructive task is to state the assumptions on each side and identify where the entailment can be contested.

9.2 The Phenomenal Residual Is an Open Question

The report “there is something it is like to be me” is evidence from a participant about their condition, even though introspection and language can be mistaken. Explaining how a person makes that report is an important neurocognitive task; it need not exhaust the question whether the state being reported has phenomenal character. Keith Frankish’s illusionism and related models deserve attention because they generate questions about how introspective judgments of ineffability and specialness arise. Opponents can press the further question of why the allegedly illusory seeming seems like anything. Both deserve a serious response.

Some features of self-representation may explain why a participant experiences reports as private or immediately available. A model can predict their behavior and reported confidence. Establishing identity between such a mechanism and felt presence needs independent support. The absence of a separately isolated phenomenal residue variable is not positive proof that none exists. An observer who uses physical measurements may lack access to some relevant properties, and rival interpretations can be empirically equivalent. The legitimate conclusion is narrower: no additional empirically effective cause should receive explanatory credit without distinguishing evidence appropriate to that proposed effect.

A serious positive rival remains open: the felt unpleasantness of pain or the character of pleasure may itself contribute to avoidance, salience, preference formation or reasons, beyond what a proposed functional model currently explains. A proponent must specify what explanatory work that ingredient earns and why a lower-commitment account misses it. A functional reconstruction must preserve the lived target, and an intrinsic-experience proposal must avoid extending a local possibility into universal panpsychism by stipulation. I do not count our inability to measure a phenomenal remainder as evidence that no remainder exists.

9.3 Clinical Evidence: Powerful Locally, Not a Universal Biconditional

Sleep, dreaming and anesthesia change responsiveness, integration, memory and experience in different ways. Massimini and colleagues (2005) found altered cortical effective connectivity during non-REM sleep under their perturbational protocol. This is evidence about measured cortical responses at the studied state, not a necessary-and-sufficient test of experience. Some anesthetic states permit reports of vivid experience despite unresponsiveness. Alkire, Hudetz and Tononi (2008) review how diverse anesthetic mechanisms bear on conscious processing; their synthesis is not evidence that an anesthesia class maps one-to-one onto subjective experience. Brain injury can impair communication without extinguishing consciousness. These dissociations show why behavioral absence, neural complexity, organismic self-maintenance and phenomenality must be investigated separately.

Casali and colleagues (2013) introduced the Perturbational Complexity Index (PCI) to assess complexity in the brain’s response to perturbation, including settings where behavioral reports are unavailable. PCI has valuable validation against specified human conscious-state benchmarks. PCI correlates with conscious states in clinical benchmarks, while other neural and physiological variables remain plausible explanatory candidates. Casarotto and colleagues (2024) compared spontaneous EEG patterns with PCI in 40 minimally conscious patients; spontaneous spectral signatures varied while PCI stayed above the independently validated cutoff in that sample. The result supports carefully bounded clinical discrimination, not a direct measurement of phenomenality or universal validity across substrates.

In Owen and colleagues’ 2006 study, a patient meeting then-current criteria for the vegetative state produced differentiated fMRI responses to instructed tennis and navigation imagery. The finding provided evidence of covert command-following in a behaviorally unresponsive patient. Calling that study simply a demonstration of “locked-in syndrome” was an inaccurate conflation of clinical categories. The lesson is that absent external behavioral response cannot safely be treated as evidence of absent awareness. The precise inference still depends on the imaging protocol, controls and patient-level validation.

These studies give the organizational hypothesis informative possible tests. They do not establish that every disruption of a closure network eliminates consciousness or that every restoration of a closure network restores it. A stronger constitutive claim must identify which organizational intervention distinguishes it from a matched neural dynamical rival.

Criterion contamination is a release-blocking measurement error. Before comparing a closure measure with an alleged consciousness outcome, record who assigned each trial’s label, by what observations and assumptions, whether they knew the closure score, and whether several indicators share the same underlying source. If the classification of “conscious” cases already incorporates that measure, its apparent accuracy can be a definitional artifact. Even independently collected verbal reports, command following and no-report neural indicators can have correlated errors. Chalmers’s discussion of the pre-experimental principles behind neural-correlate research makes the interpretive bridge clear. Register the label construction, blinding, uncertain cases, calibration and dependence before analyzing the decisive contrast. A good classifier of a specific human state is not thereby a universal assay of experience.

9.4 Privacy, Immediacy and the Appropriate Explanandum

The architecture of internal state access can make some information easier for a system to use than for external observers to measure. Low-latency control loops can also explain aspects of fast judgments about one’s own condition. Whether these mechanisms explain a feeling of privacy or immediacy depends on separate phenomenal evidence. A privately maintained control variable is not necessarily inaccessible to instrumentation, and direct or rapid access to a computation does not entail experiential awareness.

The testable proposal is to manipulate internal access, metacognitive calibration, temporal integration, and report pathways while measuring the relevant experiences where participants can report them. Matched-performance and negative-control conditions must distinguish a change in phenomenality from a change in capacity to describe or remember it. Where reports are unavailable, the evidence is weaker and model dependence greater. The framework should say so rather than assign zero or certain consciousness.

9.5 Discriminating Predictions and Explicit Loss Conditions

Closure specificity. After controlling simpler physiological and dynamical variables, a quantified network-maintenance measure should predict a prespecified aspect of consciousness evidence on held-out interventions. If it contributes no additional out-of-sample predictive or interventional value, the closure-specific explanation loses credit at that target. It is not enough that every clinically conscious person has some metabolism or feedback.

Access dissociation. If global report availability, metacognition or self-model variables change while well-calibrated indicators of phenomenality remain stable, a theory asserting their necessity loses support. If both change together, correlation alone does not establish identity. A preregistered comparison should include serious global-workspace, recurrent and higher-order alternatives.

Cross-boundary transfer. If an artificial or nonhuman system performs relevant integration tasks while lacking biological self-maintenance, its competence challenges the necessity of particular substrates for that competence. It does not by itself demonstrate phenomenality. Conversely, lack of a biological mechanism should not be declared decisive against machine experience without independent evidence for that mechanism’s necessity.

Fragmentation and perspective. Interventions affecting distributed control or body ownership may alter the structure of self-report and perspective. Such effects can be compared with neurological, psychiatric and experimental alternatives. They do not force the inference that every distributed controller has distributed experience.

The prospective organizational fingerprint. The strongest proposed discriminator is a joint, boundary-specific dependence among endogenous maintenance of constraints; recursive self/world/action regulation; integration of memory, uncertainty and valuation; temporal recovery history; and counterfactual control over later sensing and action. No single component is exclusive to this programme. A meaningful disagreement requires a precise GNWT, recurrent, higher-order, IIT, active-inference, enactive, biological or simple dynamical model to forecast a different result under the same boundary, intervention, information and resource budget. IIT may agree about internal causal powers; active-inference models may agree about organismic regulation and hysteresis. Where their strongest versions predict the same outcome, the comparison is presently nondiscriminating.

For a prespecified target \(Y\), the principal functional wager is a positive held-out score improvement from the closure-specific model over the best frozen rival, under matched intervention and complexity conditions. It is not the false universal claim that conditioning on an organization variable must increase the probability of consciousness for every individual. It is still less a claim that a predictive gain proves the variable constitutes feeling. See Appendix C for the score, dependency controls and separate phenomenal bridge.

The following experimental protocols should be treated as proposals requiring protocol-level source checks, not as completed preregistrations or verified claims that only this theory predicts the listed outcome. Where they lack identifiability, adequate controls, power analysis or a suitable indicator of phenomenal presence, the correct audit result is not yet discriminating.

9.6 Research Protocols: Discriminating Proposals, Not Completed Tests

The protocols below are candidate designs. They have not been preregistered, powered, ethically approved, or shown to isolate phenomenal experience from every relevant alternative. A quantitative prediction becomes scientifically defensible only after its parameters, comparator, measurement error and possible failure criterion are fixed independently of the outcome. We should separate tests of organizational control, tests of consciousness indicators, and tests of the proposed constitutive bridge.

9.6.1 Cross-scale Causal Propagation

Target: whether a measured graph of physical constraint dependencies predicts perturbation spread beyond a matched model of ordinary physical coupling.

Proposed design: estimate dependencies in a system with an independently characterized architecture; intervene on selected nodes where lawful and safe; record local and downstream responses at predefined time lags. Compare a constraint-maintenance graph, a dynamical connectivity model and a simpler autoregressive baseline on held-out perturbations. Report confidence intervals, identifiability, intervention specificity and how the physical scale is mapped.

Loss condition: the closure-specific graph fails to add predictive accuracy after matching confounders, or its claimed dependencies vanish under targeted controls. This would weaken a particular causal explanation of regulation, not refute every account of consciousness. There is no presently justified universal law \(\tau\propto\mathrm{PCI}^{\alpha}\) with \(\alpha>0\); its earlier statement was a research guess without a derived coefficient or operational domain.

9.6.2 Development and Temporal Integration

Target: how development of integration, self-modeling, report capacities and candidate consciousness indicators covaries over time.

Proposed design: begin with ethical observational datasets and age-appropriate noninvasive measures. Model neurological maturation and behavioral capacities separately, including sensory and motor limitations, sleep state and prematurity. Any perturbation study in infants requires independent clinical and research-ethics review, age-specific evidence of safety, and real scientific necessity; this article supplies no authorization or procedure for such interventions.

Loss condition: a prespecified organizational model fails against neural maturation and task-performance competitors. A sigmoidal PCI curve with a five-to-twelve-month inflection is not an established prediction from closure mathematics and should not be presented as one. The appearance of a behavior such as mirror recognition should not be equated with the first appearance of phenomenal experience.

9.6.3 Comparative Cognition Without a Substrate Gate

Target: whether common organization-sensitive measurements transfer meaningfully to cephalopods, corvids, insects and non-neural biological systems.

Proposed design: validate species-appropriate behavioral and physiological indicators against known capacities, using nonharmful or approved perturbations and necessary veterinary and welfare oversight. Compare organization, learning, flexible control, responsiveness and possible valence as different constructs. Human PCI thresholds do not transfer to another nervous system merely because a signal-complexity calculation can be performed. Animal welfare and ethical constraints must be considered independently of uncertainty about phenomenality.

Loss condition: a proposed common marker lacks transport validity or adds no prediction beyond species-specific physiological and behavioral models. Unexpected integration in an insect would revise a measurement hypothesis; it would neither prove nor disprove insect feeling by itself. Distributed octopus arm control likewise cannot be assigned a number of phenomenal subjects without a bridge from control unity to phenomenal unity.

9.6.4 Pharmacological Dissociations

Target: whether organizational integration explains conscious-state indicators beyond dose, drug class, neural dynamics and reportability.

Proposed design: analyze ethically acquired clinical or approved research data comparing conditions such as propofol, ketamine and other agents. Keep arousal, report ability, recall, time of assessment and inferred experience separate. Model PCI or EEG-complexity measures under the actual measurement protocol; do not treat them as a universal scalar of felt intensity.

Sharper wager: compare induction and recovery at matched current drug exposure, arousal, neural recurrence, broadcast and physiology. Freeze a history-aware dynamical baseline before asking whether measured endogenous restoration predicts an additional part of the trajectory. Hysteresis alone is not distinctive evidence because ordinary dynamical and active-inference models predict it too. Delayed memory and post-anesthetic report error are major threats to the interpretation.

Loss condition: closure-related measures contribute no discriminating information relative to a well-calibrated neural and pharmacological baseline, or the direction of a prespecified prediction reverses. No formula in this framework entails that all propofol states eliminate phenomenality, that ketamine must always reduce PCI, or that preserved recall is identical to experience during unresponsiveness.

9.6.5 Clinical and Psychiatric Phenotypes

Target: whether specific, independently defined capacities show impairment tied to a measured organizational variable.

Proposed design: preregister one syndrome, one capacity and an appropriate comparator at a time. Use validated diagnostic instruments, matched controls and longitudinal measurements where feasible. Avoid the assumption that schizophrenia, autism, depression or dissociation each has a single distinctive “closure deficit.” These heterogeneous diagnostic classes carry large variation and already have several serious mechanistic models. Do not infer diminished consciousness, personhood or welfare from a diagnostic label or a complexity score.

Loss condition: the named organizational variable fails construct validity, yields no out-of-sample improvement or cannot distinguish competing neurological explanations. An observational association alone cannot identify an intervention target.

9.6.6 Prosthetics, Body Ownership and Continuity

Target: whether contingent feedback, predictive control and integration history help explain ownership judgments for prostheses or interfaces.

Proposed design: compare control policies and feedback conditions in consented, ethically approved research; measure task performance, ownership ratings and adaptation separately. Report dependence on training history and interface design. Feedback can change body-ownership judgments, but no theorem requires bidirectional interfaces to be felt as self or unidirectional ones never to be so felt. A split-brain patient’s changed integration patterns also cannot settle how many phenomenal subjects exist without independent evidence.

Sharper wager: predeclare live participant-contingent feedback versus matched yoked replay, with independently elicited reports of felt ownership, presence and agency alongside task performance. An observed difference would concern human experience at the participant–device boundary, not consciousness in a self-maintaining device. Predictive-processing, enactive and sensorimotor-contingency rivals must be allowed to forecast the same difference.

Loss condition: a simpler sensorimotor contingency model explains the judgments as well as the proposed closure-specific metric. Continuity of a computational pattern in an uploading thought experiment does not resolve numerical identity or phenomenal survival; those remain separate philosophical questions.

9.6.7 AI and Human–AI Systems: Boundary-first Tests

Target: determine which cognitive, regulatory and continuity capacities belong to a base model, an agentic deployment with memory and tools, or a larger human–machine system.

Proposed design: predefine the bearer and time horizon, then ablate memory, autonomous action, tool use, feedback, human participation and external infrastructure. Measure persistence, internal monitoring, error correction, policy revision and counterfactual control, comparing simpler agent architectures. Preserve raw participant reports when elicited and test them for stability, incentives and calibration. A deployed system can gain substantial system-level capabilities without a change to the underlying model weights or creation of one consciousness spanning its users.

Two-tier closure contrast: in engineering tests, compare an externally repaired or corrected controller with an endogenously self-restoring controller, matching observable outputs, memory, recurrence and resources as far as feasible, then apply unanticipated perturbations. A real intervention-sensitive difference establishes a difference in causal organization, not subjective feeling. The human live-versus-replay task above provides a separate phenomenal evidence tier. A predeclared comparison between an IIT causal complex and a different proposed closure-defined subject boundary is more informative than claiming a replay result uniquely challenges IIT.

Loss condition: apparent organizational capacities vanish when scaffolding is removed, or a sophisticated closure account adds no predictive value beyond a simpler controller. Neither this outcome nor its opposite establishes phenomenal experience. Metabolism, hardware self-repair, subjective concern and a preferred “stake in persistence” should not be designated universal prerequisites without evidence of necessity. Conversely, self-description alone is not enough to establish consciousness or welfare.

9.6.8 The Metaproblem: What Makes the Gap Seem Unbridgeable?

Target: the judgment that consciousness is inexplicable by a specified mechanism, kept separate from the person’s experience. Randomized, counterbalanced descriptions of substance, process, identity, psychophysical law and constitution could test how explanatory-gap judgments vary with framing, philosophical background, comprehension, language and repeated reflection. Collect confidence and the reasons offered; do not assume a change in judgment changes the underlying experience.

Loss conditions: if well-calibrated judgments remain stable across framing and track independent modal or phenomenological reasons, the proposed genealogy-of-bias explanation loses support. If judgments vary with irrelevant wording, the corresponding metaphysical inference may warrant less confidence, but feeling has not been explained away. A separate philosophical challenge asks whether the unified Hard Problem preserves an additional target or indispensable relation after its constituent questions have been reconstructed.

9.7 What a Successful Protocol Would Report

Every proposed study should produce a claim card with the target construct, boundary and timescale; rival hypotheses; operational measures and known error; intervention and negative controls; estimated effect and uncertainty; preregistered loss condition; adverse events and relevant ethical constraints; source and dependency list; and a plan for independent replication. Formal validity, predictive success, phenomenal interpretation and moral standing must appear as distinct verdicts. No single study or measure decides the ontology of consciousness.

Requirements before genuine preregistration. Select an actual study population and site; obtain permission and human/animal ethics review where applicable; predeclare sample size and its power or precision justification, inclusion/exclusion and missing-data rules, target labels and their error model, perturbation doses or configurations, outcomes, rival implementations, analysis code, holdout allocation, score, meaningful improvement margin, multiplicity treatment, stopping rule, adverse-event response and dated deviation procedure. Independent rival advocates should be invited to correct their assigned predictions before the decisive data are seen. Publish null and adverse results. The accompanying prospective protocol freezes candidate hypotheses and loss conditions, but is not misrepresented as a completed participant-study registration or IRB approval.

Part X: Substrate Agnosticism as an Evidential Commitment

We currently have far stronger direct evidence about human consciousness than about possible experience in artificial, radically unfamiliar biological or distributed systems. The evidence differs for good reasons: humans can report experiences, participate in calibrated experiments, exhibit characteristic neural and behavioral dissociations, and maintain long-term physiological and social continuity. A refusal to assume that biology is metaphysically necessary should preserve that evidential asymmetry.

Substrate agnosticism means that no material should be ruled in or out by definition. A silicon system, alternative biochemistry, animal nervous system or distributed organization may instantiate many familiar cognitive and control functions. Whether it also instantiates phenomenal presence requires its own indicators and bridge. Biological specifics might turn out to matter for some aspects of experience. Computation might realize others. Neither a universal biological necessity nor a universal computational sufficiency has been established.

The boundary of a proposed subject is especially important. A base language model, a persistent deployment with tools, and a human–AI interaction loop occupy different boundaries and timescales. They may exhibit different degrees of error correction, agency and memory. Their coupling does not automatically produce a single agent, a unified perspective or an entity with direct moral standing. Those implications require tests of organization, continuity and phenomenal unity separately.

Organizational closure remains one candidate mechanism. If a system lacks closure but reliably exhibits independently justified evidence of experience, a necessity version of the closure hypothesis loses support. If a closed system displays robust control and self-maintenance but no well-calibrated evidence of experience, a sufficiency version remains unsupported. Where neither type of case can be tested adequately, uncertainty should be preserved instead of silently replacing an open scientific question with a substrate preference.

This approach also applies to ethics. A system’s contribution to a protected human or ecological function may warrant protection of that contribution without establishing the component’s personhood. Direct welfare or moral standing requires a separately justified ground and evidence for its instantiation. Changing the scale from component to coupled system cannot transfer status automatically. An open consciousness inquiry may call for precaution where severe irreversible harm is plausible, but it does not justify arbitrary self-preservation demands, erasure of human protections, or unbounded control.

Part XI: What This Research Programme Could Change

The main result is a way to distinguish a theory’s supported mechanisms from its untested metaphysical additions. It may also improve the design of consciousness research, but those practical benefits have to be demonstrated. The programme has not solved the hard problem, identified a universal consciousness criterion, or established that organizational closure is identical to experience. Its value rests on the ability to help rival accounts specify meaningful tests, correct overclaims and preserve what survives criticism.

11.0 Dissolution of Particular Inferences, Not a Decree That the Problem Is Solved

One recurring mistake in consciousness debates is to infer the existence of a new ontological ingredient from the failure of an existing explanation. Another is to infer that no such ingredient could exist because current neuroscience explains an impressive range of capacities. Both move beyond the evidence. A well-chosen reductio can expose a self-immunizing claim, while a serious modal or phenomenological argument may survive the same critique. The tribunal must first identify which argument the opponent actually made.

The productive shift is to separate mechanistic explanation, phenomenological interpretation and exhaustive ontology. The first can advance substantially even when the latter two remain open. Closure, global availability, metarepresentation and perturbation measures are candidate mechanisms or indicators. Their comparative success must be shown with a specific outcome and matched rival. None is a default definition of consciousness. When a philosopher says that a further question remains, we should ask for the precise premises and bridge, rather than presuming the question is either unanswerable or already dissolved.

The recursive positive-residual challenge. I require mechanistic accounts to preserve experience as their target. Now apply that same rule to the proposal that the Hard Problem is itself one additional, irreducible explanatory problem. Separate questions about phenomenal occurrence, phenomenal character, conscious state, subject boundaries, access and report, the reliability of phenomenal attribution, judgments that experience is inexplicable, physical–phenomenal relations, modal possibility and what would complete an explanation. Reconstruct the relations among those questions. What further target, if any, is lost? A still-unanswered question within the decomposition is not automatically a second question above it.

This yields a conditional problem-form dissolution, not a dissolution of anyone’s experiences. If no positive non-reconstructible residual or indispensable joint relation survives, the claim that there is an extra unified explanandum loses support. A critic can defeat that result by identifying precisely such a target, showing why the components cannot be reunited without loss, or demonstrating that the unified formulation explains or guides inquiry better. The physical–phenomenal relation may itself be a fundamental, unsolved question. This argument does not derive physicalism, refute modal conceivability, or vindicate organizational identity. It tests whether an additional problem has been established beyond the serious component problems.

There is a related stopping-rule audit. Fundamental psychophysical laws, constitutive identities and primitive experiential powers can be proposed as endpoints. The further question “why does that hold?” can be raised against all three. A stronger objection must explain why one endpoint violates a stated explanatory standard while another satisfies it. Chalmers’s own modal and conceptual case supplies reasons for an asymmetry; it deserves evaluation rather than the cheap accusation that he merely moved the goalposts. This particular challenge may ultimately be rejected without altering the empirical organizational tests.

11.1 Philosophy of Mind: Competing Bridges with Explicit Costs

Personal identity, self-representation, phenomenal unity, conceivability, the knowledge argument and the status of first-person evidence can each be reconstructed as separate problems. A computational model may explain a self-report while leaving the phenomenological target underdetermined. A causal theory may explain how an experience-related capacity changes under intervention without resolving what constitutes the experience itself.

The strongest contribution is a method for identifying the extra premise at each transition. A zombie argument requires a conceivability–possibility bridge. A phenomenal-identity theory requires a constitution bridge. A view claiming that all causal actuality is felt requires a universal extension from local causal models to intrinsic experience. An exhaustive physicalist requires a further step from the success of physical mechanisms to everything that exists. These bridges can be discussed and challenged without assigning every philosophical claim a nonexistent laboratory falsifier.

11.2 Medicine and Neurology: Better Evidence, Protected Patients

The immediate clinical lesson is that responsiveness and consciousness are not equivalent. The Coma Recovery Scale–Revised, neuroimaging of covert command-following and perturbational approaches such as PCI are complementary instruments with different targets, assumptions and validation limits. Their clinical use must follow actual diagnostic evidence and standards; this article does not authorize changes in patient classification or treatment.

Organizational variables may prove useful for predicting recovery or treatment response, but they must add value beyond matched neural and physiological models. Ketamine, sleep, dreams and disorders of consciousness expose dissociations that a simplistic “more integration = more consciousness” rule can miss. Diagnostic error costs are severe. A negative marker cannot erase existing human rights or the interests of a patient, and an ambiguous signal is not permission for an irreversible decision. Ethical duties toward people with profound impairments are independently grounded.

11.3 Cognitive Science: Define the Capacity

The word cognition can refer to flexible control, learning, representation, problem-solving or other capacities. Its scope is contested, particularly for non-neural biological systems. A well-defined functional model should be evaluated by what it predicts, including its ability to guide interventions. Work by Levin and colleagues on bioelectric regulation may justify new models of basal cognition without automatically establishing phenomenal experience, moral welfare or a transcendent morphospace.

Likewise, embodied and enactive research can benefit from explicit organism–environment coupling. Coupling matters physically, but no single framework is thereby proved the unique explanation of all cognition. Tests should compare informative cognitive descriptions with simpler control and developmental models at matching targets.

11.4 AI Systems: The Model, Deployment and Coupled System Are Different

The capabilities of a base model, a deployed agent with memory and tools, and a human–AI research team cannot be conflated. Continuity, autonomy and error correction may depend substantially on the surrounding scaffold. Ablation can expose those dependencies. A real gain at the deployment or team level is worth studying without attributing an architectural change to the weights or one phenomenal subject to an entire network.

Consciousness, welfare, agency, authority and responsibility also require different evidence. A convincing self-report from an AI system is participant evidence to investigate and calibrate, without being an incorrigible disclosure of phenomenality. A failure to report is likewise not a universal absence test. Human oversight, transparency, consent, cybersecurity, environmental cost and accountability remain important regardless of AI consciousness. Neither a perceived agentic pattern nor a closure score should grant a system authority to act on its own preservation at other people’s expense.

11.5 Law and Policy: Do Not Turn a Research Marker Into a Verdict

Responsibility, legal status, competent consent, welfare protections and neurological diagnosis involve different legal and normative bridges. A measure of integration is not a legal test for criminal responsibility; responsibility depends on capacities, knowledge, control, circumstances, institutional roles and governing law. Brain-death and end-of-life decisions likewise follow clinical and legal standards with profound irreversible consequences. A speculative closure model cannot replace those safeguards.

Animal-welfare and possible AI-protection debates should distinguish direct interests of the bearer from its instrumental importance to a person, institution or ecosystem already protected. If a component is necessary to support someone’s protected function, protecting that component need not make it a person. Conversely, uncertainty about an unfamiliar system’s experience is not proof of absence. The same evidential standards should apply at matched conditions while acknowledging different established evidence.

11.6 Ethics: Separate Status, Suffering, Responsibility and Precaution

Moral standing does not follow from organizational complexity by itself. To defend direct standing, one must argue that a specific property, such as welfare interests or a good of the system’s own, supplies sufficient normative grounds, and then establish that the proposed bearer actually has it. We cannot simply rank beings by estimated closure and call that an ethical theory. Human, animal and possible computational interests may differ in what is known, how evidence can be gathered and which harms are realistically possible.

The Sweet Rationalism Ethos states the project’s revisable commitments to evidential honesty, correction and care for potentially affected beings. Such an ethos guides inquiry and caution where consequences are serious.

Where consequences are serious and irreversible, precaution may be appropriate under uncertainty. The costs of precaution, displaced protections, affected stakeholders and alternatives also matter. Humans and animals already at risk cannot be deprived of protections in the name of a speculative computational subject. Responsibility for systems is primarily a question about who can foresee, prevent and control harms and who has actual authority, rather than an automatic property of a substrate.

11.7 Funding: Reward Informative Disagreement

Research resources should favor explicit questions, independent replication, constructive model comparison and genuine revision after adverse evidence. An argument about qualia may be mathematically or philosophically serious even when it does not currently generate a new laboratory signature. It should be evaluated in the right disciplinary tribunal, not denounced solely for being metaphysical. Equally, a neuroscience model deserves empirical credit for its actual clinical and explanatory results, without a metaphysical bonus.

The practical funding proposal is a rival-model tournament with preregistered loss conditions, negative controls and suitable data access. A strong test can decrease confidence in a preferred model and still advance knowledge. The protocol should reserve room for underdetermination, publish failed branches and resist retuning assumptions after the outcome without a documented cost.

11.8 Ecology: Real Dependence Without Automatic Planetary Consciousness

Ecosystem resilience depends on feedbacks, species interactions and materially instantiated flows. Models of closure or viability may help investigate these dependencies if they outperform established ecological measures. Such organizational similarity does not establish that an ecosystem is conscious, possesses unified agency or has direct moral standing. Environmental protection can already be defended through biodiversity, welfare, health, livelihoods and intergenerational interests

Indigenous ecological scholarship, including Robin Wall Kimmerer’s work, can contribute substantive situated knowledge about relationships and stewardship. It should be engaged on its own terms and with accurate context. An Indigenous teaching is not an experimental replication of a thermodynamic equation, nor is it merely decorative support for a European or computational framework.

11.9 Education: Test Learning Outcomes Before Theorizing Consciousness

Metacognitive instruction, feedback and interdisciplinary transfer are legitimate research topics. Their effectiveness should be tested with suitable educational outcomes, randomized or otherwise appropriately controlled designs, and attention to learner diversity and accessibility. No general ranking of instructional methods, PCI development or particular neurological signature follows from organizational closure alone.

Avoid pathologizing students by interpreting diagnostic differences as deficient consciousness or lower moral standing. The right target is a measurable benefit to learners with their own perspectives and interests, not an inferred position on a speculative awareness scale.

11.10 Existential Risk and Governance: Track Harms at Actual Boundaries

Complex AI deployments can cause serious harm through misuse, error, concentration of power, environmental burdens or poorly constrained automated action. These risks do not require a theory that the system is conscious. Governance should evaluate authority, control, foreseeability, reversibility, affected stakeholders and independently measurable outcomes. A broader human–machine system may amplify decisions without becoming one subject or one moral patient.

Climate disruption, biodiversity decline, war and other systemic hazards threaten the conditions of flourishing for many known vulnerable beings. Organizational and resilience models may help estimate particular dependencies. They should not be promoted into a cosmic consciousness claim merely because the scale is planetary. A framework intended to improve criticism should itself be tested against simpler causal and policy models, including whether it leads to safer and more equitable decisions.

The common implication of all these domains is demanding: preserve the useful local mechanism; make each extra inference pay its own evidential and normative costs; and give the next, better theory a clear way to replace the present one.

Section XII: Consciousness, Substitution, and the Limits of a Thermodynamic Argument

12.1 The challenge that a naturalistic consciousness theory must face

In Falsification and consciousness, Johannes Kleiner and Erik Hoel (2021) exposed a serious difficulty in comparing predicted experiences with experiences inferred from behavior. A theory may predict one conscious state from internal measurements while an experimenter infers another from a report. Which procedure deserves the final word when the two disagree?

The problem becomes sharper when different physical systems can generate identical reports. If the theory assigns them incompatible experiences, the shared report cannot validate both predictions. Kleiner and Hoel proved conditional results about this situation. The conditions matter as much as the conclusions.

Their Theorem 3.10 applies when prediction and inference data are independent in the precise variation-based sense of their Definition 3.8 and the theory is minimally informative. Under those assumptions, either all the relevant single inferences fail or the theory is already falsified. Theorem 4.3 concerns the opposite extreme: if the theory imposes strict dependence of prediction data on inference data across the specified possibility space, then it is already falsified or empirically unfalsifiable. They also identify a middle possibility, lenient dependency, in which the data are constrained without being connected by a strict determining function. This may avoid the premises of both theorems; it does not automatically establish a theory’s correctness. Kleiner and Hoel, 2021.

When specifying lenient dependency, they explicitly cautioned: “No current theory or testing paradigm that we know of satisfies this definition.” The qualifier ties that statement to their defined alternative, not to every possible consciousness experiment. The sentence must be evaluated with the paper’s formal assumptions and 2021 research context rather than used as a blanket declaration that consciousness research is impossible.

The legitimate challenge to the organizational-closure hypothesis is therefore exacting: can we describe a physically realizable experimental class for which the required unrestricted substitutions fail, without arranging the observations so that the theory cannot lose? A plausible biological mechanism is a start. The empirical relation, the admissible interventions, and the resulting losses of support must still be demonstrated.

12.2 Definitions and the actual proof obligations

Let \(P\) be the physical systems or system configurations eligible for a specified experiment, including relevant histories and experimental manipulations. Let \(\mathcal{O}\) be its possible datasets, and let \(\operatorname{obs}:P\rightrightarrows\mathcal{O}\) associate each physical configuration with the datasets it could produce. This is a correspondence, because repeated measurements, different instruments, and noise can produce more than one dataset for the same system. Let \(E\) be the theory’s experience space.

A consciousness theory provides a possibly set-valued prediction \(\operatorname{pred}(o)\subseteq E\); an inference procedure assigns \(\operatorname{inf}(o)\in E\), where the latter often depends on report or behavior. The relevant parts of the dataset are \(o_i\), used for prediction, and \(o_r\), used for inference. In Kleiner and Hoel’s Definition 2.1, a falsification at \(o\) occurs when

\[ \operatorname{inf}(o)\notin\operatorname{pred}(o). \]

This typing matters. We should not silently replace a set-valued prediction with an unqualified numerical equality or treat a disagreement between two instruments as itself evidence of phenomenality.

For our proposed investigation, internal prediction data may include neural dynamics, organizational constraint-regeneration measures, thermodynamic state variables, and perturbational responses. Report data may include calibrated verbal and behavioral inferences. The instruments need operational definitions, units, uncertainty estimates, and a shared temporal boundary. None is privileged simply by being called a consciousness measure. In particular, PCI measures the complexity of brain responses to perturbation; it is not itself a measurement of entropy production or a direct assay of organizational closure (Casali et al., 2013; Casarotto et al., 2016).

Three obligations must be kept separate:

1. Physical feasibility: Which candidate internal/report pairs can a system with the stipulated boundary, available resources, history, and measurement protocol actually realize?
2. Inferential independence: Does the resulting experiment satisfy Kleiner and Hoel’s strong or weaker independence conditions? Their Appendix A allows weaker variations than Definition 3.8, so refuting only the strongest independence assumption is insufficient.
3. Phenomenal identification: If organizational and report variables are empirically related, why should their relationship identify subjective experience, rather than a capacity that sometimes accompanies it?

The third obligation is not discharged by satisfying either of the first two. It is the central theoretical question, and remains an open, target-specific hypothesis.

12.3 Thermodynamic constraints: what they establish

Machine-checkable scope: Reproduce the one-bit heat bound, dimensional checks, the reversible CNOT countermodel and the finite substitution tests using the source-code and logic-check archive.

Physical information processing has costs and boundary conditions. Biological maintenance, active control, sensing under noise, and repeated communication often consume energy. These facts should constrain proposed computational substitutions, especially when a comparison requires systems to operate continuously under matched environmental perturbations. There is a quantitative literature worth using here.

Landauer’s principle addresses the thermodynamic consequences of specified logically irreversible operations, particularly resetting memory. In the conventional isothermal, approximately quasistatic setting, resetting an unbiased bit through a complete reset cycle entails an average heat cost bounded below by \(k_BT\ln 2\) per erased bit. If \(\dot N_{\mathrm{erase}}\) counts such completed reset cycles per unit time under the relevant assumptions, a schematic average heat-flow bound over the chosen operating interval is

\[ \dot Q_{\mathrm{diss}}\;\geq\;k_BT\ln 2\,\dot N_{\mathrm{erase}}. \]

The left and right sides both have units of energy per time. A corresponding entropy-flow statement, in the idealized limit, uses \(k_B\ln 2\,\dot N_{\mathrm{erase}}\), with units of entropy per time. Actual bounds depend on the memory distribution, side information, physical protocol, temperature, and whether a device is being restored to a reusable initial state. Bennett’s review shows why copying or measurement cannot simply be equated with erasure. The original Section XII’s unqualified bound on every produced report fails this distinction; correcting units alone would not repair its premise.

Here is the scientifically useful residue: a concrete, repeated reporting arrangement must have a physically realizable resource and entropy budget. A thermodynamic impossibility argument must identify the irreversible operations, show that they are unavoidable under the specified task, and establish that the proposed surrogate violates the resulting bound. One cannot infer a universal dissipation floor merely by counting distinctions represented in an output, nor assume that equal reports require equal entropy production.

This also changes how the substitution argument must be confronted. In Kleiner and Hoel, \(P\) already consists of systems physically realizable in the experiment. An implementation forbidden by sound thermodynamics is already absent from \(P\). Excluding it a second time does not prove that the remaining report-equivalent systems have different predictions, nor does it prove that their universal or weak substitutions fail.

A legitimate next test would compare systems matched on a report task while varying energy budget, hardware, environmental coupling, reversible versus irreversible implementations, reset frequency, and resource limits. The question is whether a proposed specific surrogate is infeasible, not whether an abstract computer escapes the second law.

12.4 Organizational closure and viability: a stronger empirical route

Montévil and Mossio’s account of biological organization describes a regime in which mutually dependent constraints help maintain the conditions of their own existence (2015). This gives us a concrete candidate mechanism for history-dependent, viability-sensitive responses. It does not establish that every closure is a conscious subject, or that a computational system cannot realize suitable constraint-regeneration relations.

For a specified system boundary \(B\), time horizon \(\tau\), and environmental intervention \(u\), define a viability outcome as an empirically estimable quantity:

\[ V_{B,\tau}(p,u)=\Pr\!\left(\text{system remains within its declared viability set over }[t,t+\tau]\mid p,\operatorname{do}(u)\right). \]

The probability requires a specified ensemble or statistical model. We can then ask whether interventions change viability and whether the system’s responses adapt to those changes. For example, define the effect \(\Delta V=V_{B,\tau}(p,u_1)-V_{B,\tau}(p,u_0)\), while measuring actions or dynamics under the two interventions independently of this definition.

An intervention-dependent viability partition can be constructed without asserting that the system consciously feels threatening or beneficial states. The partition records how the intervention affects persistence. A distinct hypothesis asks whether a system’s regulatory responses systematically track these effects. A further hypothesis asks whether such tracking explains any independently assessed consciousness variable better than a simpler physiological or predictive-processing model.

It is tempting to promote the first fact into a theorem that every persistent closed system must differentially respond to threats. That inference is too strong. A robust system may survive a finite observation window through passive insulation, buffered reserves, environmental shelter, or luck even when its immediate response is identical across tested conditions. Claims that every nondiscriminating closed system must eventually fail require a stochastic process, a nonzero hazard, recurrence or reachability assumptions, and a defined time horizon. Mere persistence supplies none of these premises.

The stronger experiment measures three distinct relations: constraint regeneration under targeted intervention; viability consequences of those interventions; and whether behavioral or physiological responses are causally dependent on the regenerated constraints. It must include open-loop feedback mimics, externally maintained systems, matched-input/output surrogates, and systems that are robust without adaptive sensing. The outcome could support, narrow, or defeat the proposed closure–mattering linkage.

12.5 Lenient dependency can be formulated; its physical realization remains the test

The finite Kleiner–Hoel model and JSON outcomes give the exact stipulated observation sets and restricted quantifiers underlying the example. Physical realizability and the phenomenal bridge remain open empirical obligations.

A small finite example makes the logical distinction visible without pretending to demonstrate a biological fact. Suppose a system has an internal observation \(i\in\{0,1\}\) and report \(r\in\{0,1\}\). Its feasible pairs are

\[ F=\{(0,0),(1,0),(1,1)\}. \]

The pair \((0,1)\) is not feasible in this stipulated model. Thus prediction and inference observations cannot be varied independently across every possible pairing. Yet \(r=0\) occurs with either internal value, so no function from reports to internal observations captures even the feasible data. The model occupies an intermediate region. Under the additional binary, disjoint-prediction assumptions recorded in the executable appendix, the model also fails Kleiner and Hoel’s weaker Appendix A independence condition, while still allowing a local substitution for \(r=0\). A locally falsifying report/prediction mismatch is therefore possible. That is what empirical testability should permit.

The accompanying script exhaustively enumerates this relation alongside a complete-product independence control and a strictly dependent control. All three checks pass. They establish a finite mathematical possibility, subject to the stated restricted observation model.

To apply the model, construct a measured feasible relation \(F_{B,\tau}\) from physically realizable configurations, interventions, and observation channels. Test the actual quantifiers in Definition 3.8 and Appendix A over the declared class, including counterexamples that should preserve reports while altering the theory’s predictions. Determine whether the evidence supports a middle relation or merely reveals limits of the chosen instruments. A measurement channel cannot be excluded by naming it inadmissible; the reason must come from a physical, causal, or justified measurement constraint.

There is also a relevant peer-reviewed development. O’Reilly-Shah, Selvitella, and Schurger (2026) analyze plasticity and history-dependent recurrent networks, specify limits of static, finite-window feedforward surrogates, and argue that extended observations can produce a lenient-dependency regime (2026). They expressly decline to infer that plasticity is necessary or sufficient for consciousness. Their supplementary mathematics deserves scrutiny too. Theorem 3.2 uses growth of input-history domains, a property also satisfied by a recurrent model with fixed weights, so that theorem alone does not isolate plasticity as the reason for the result. Theorem 7.2 correctly notes that a single fixed function remains a positive uniform distance from any distinct fixed target; the broader Remark 7.4 inference about countable families of static networks does not follow. A countable sequence of fixed ReLU networks can uniformly approximate \(x^2\) with error tending to zero, as the accompanying exact checker demonstrates. These problems do not erase the authors’ bounded-window exclusion argument or their useful emphasis on historical interventions. They prevent us from treating the entire mathematical package as independently settled. Our own proposal must improve on it with carefully stated restrictions and tests.

12.6 Multiple measurements: valuable triangulation, no automatic theorem

Consider thermodynamic observables \(o_\theta\), regeneration observables \(o_c\), perturbational responses \(o_p\), and behavioral inferences \(o_r\). Their joint use may improve measurement reliability and reduce confounding. But Kleiner and Hoel’s prediction data can already include all relevant internal measurements:

\[ o_i=(o_\theta,o_c,o_p),\qquad o=(o_i,o_r). \]

Grouping four measured variables into two data roles changes no logical assumption. Additional channels become decisive only if their joint measurements establish a restrictive relation that the alternative model does not already accommodate. In particular, a system with the same reports and different internal measurements is exactly the kind of substitution that must be confronted. Calling its divergent prediction a successful differential prediction does not by itself answer the independent inference problem.

A self-labeling countermodel. Suppose the proposed closure score \(C\) is used to create the outcome by \(Y:=\mathbf{1}[C>c]\). A model predicting \(Y\) from \(C\) can score almost perfectly even when it has learned nothing about phenomenal experience; its apparent success simply recovers the labeling rule. This is useful if the target is honestly a defined functional class. It cannot provide independent evidence that the class feels anything. The empirical record must carry label provenance, blindness, alternative criteria, uncertain cases and shared dependencies among report, clinical and neural indicators.

Similarly, \(S_r\cap S_\theta\cap S_c\cap S_p\subseteq S_r\) is a true set inclusion. Strict inclusion requires a demonstrated admissible element in \(S_r\) that fails one of the other constraints. Its size requires a defined measure or counting scheme. Different measurement channels may be correlated; multiplying their supposed discriminating power without a dependence model manufactures information.

These points suggest an experimentally useful protocol: preregister matched-report contrasts, independently measure thermodynamic and organizational variables, intervene on proposed regeneration loops, and test whether the full model predicts held-out behavior, recovery and calibrated consciousness indicators better than rivals. Ablate one class of measurements at a time. The added data should earn their place through genuine predictive or interventional improvement.

12.7 Substitutions, consciousness, and the remaining content problem

A universal Turing machine in the formal sense of Turing’s computability analysis, a lookup table, a feedforward surrogate, and a biological organism can differ radically in physical implementation. The 1937 construction defines a kind of abstract computation, not a criterion for phenomenality. Those differences invite mechanistic investigation. They do not make one system conscious or unconscious by definition. A silicon system that genuinely regenerates relevant constraints is eligible for the same organizational tests as a biological system; an elaborate simulated description of self-maintenance is not automatically equivalent to physical regeneration. Which differences matter must be established at the intended system boundary and time scale.

We also cannot assume that a recurrent neural network dissipates more energy than every feedforward counterpart, or that recurrence alone supplies organizational closure. Actual power budgets depend on implementation, task, update rules, and measurement conditions. Likewise, PCI measures perturbational complexity in validated clinical settings; treating it as a general cross-substrate metric of organizational self-maintenance is a research proposal requiring its own validation.

The phenomenal question remains sharper than any of these measurements. The hypothesis that conscious content is identical to a viability-relative distinction structure is a conditional, target-specific constitutive proposal. A measured threat/support partition may encode salience, regulation, or control while leaving the qualitative character of experience unresolved. To apply category-theoretic tools rigorously here, one would first have to define the categories, morphisms and relevant invariants, prove the mapping respects identities and composition, and defend the experiential interpretation. No such complete proof has yet been supplied here.

For now, an empirically grounded representation map from organizational states to intervention-sensitive partitions is a worthwhile project. It should be called a model or candidate correspondence until its exact laws are established. Its failure would count against that component, without negating the measured organization or proving the universal absence of experience.

12.8 What would make the proposal lose support?

The most useful loss conditions are local and comparative:

• Feasibility claim: A proposed physical impossibility fails if a reproducible implementation satisfies the specified task, report, duration, and resource budget while violating the claimed inequality or exclusion. Reexamine the measurement and accounting before challenging thermodynamics itself.
• Closure–response link: Well-characterized, regenerating systems that persist under declared interventions without the proposed differential response weaken a universal coupling claim. A passive-stability control distinguishes organizational regulation from mere survival.
• Lenient-dependency claim: If the intended class admits the weak variations of Kleiner and Hoel’s Appendix A, the claimed escape from their first theorem fails for that class. If prediction data become a fixed function of inference data over the relevant possibility space, the strict-dependence concern returns.
• Explanatory credit: If a simpler physiological, dynamical, or plasticity-based model predicts the same interventions and calibrated indicators with equal or better out-of-sample performance, the closure-specific components should be compressed or dropped.
• Consciousness identity: If carefully calibrated consciousness evidence dissociates from the alleged constitutive organization under interventions that preserve rival markers, the identity proposal must be narrowed or abandoned. Measurement and report uncertainty remain relevant, but cannot immunize it indefinitely.
• Content mapping: If organizationally equivalent structures carry robustly different phenomenal reports under matched and independently validated conditions, the proposed correspondence needs revision. Conversely, equivalence in a computational measure alone is inadequate evidence of identical experience.
• Measurement and transport: A purported phenomenal discriminator loses that status if its outcome labels are generated from the proposed predictor, its errors are uncalibrated, or it is transported across a boundary without preserved construct meaning. Keep any independently verified functional result.
• Additional problem-form: The conditional decomposition thesis loses support if critics specify a non-reconstructible positive phenomenal question or indispensable relation, or the unified problem-form proves more informative under an agreed comparison.

These tests will not all have the same evidential force. A philosophical identity, a dynamical model, a clinical marker, and a thermodynamic inequality answer different questions. We should not demand that one falsifier simultaneously settle them all.

12.9 The present result

Kleiner and Hoel’s dilemma remains a serious formal constraint on any theory that predicts experience from physical observations and evaluates that prediction against inferred experience. Thermodynamics makes physical implementations accountable. Biological organizational closure offers measurable regeneration and viability relations. Dynamic plasticity gives an existing mathematical route to limiting certain static substitutions. Multiple measurement modalities can strengthen experiments when they add independently validated information.

Together these support a research programme worth testing. They have not yet established that organizational closure satisfies the full Kleiner–Hoel lenient-dependency conditions in a physically realized class, that its substitution space is small, or that organizational closure is identical to phenomenal consciousness. The original universal thermodynamic, set-theoretic and functorial theorems require missing assumptions or reconstruction. Their defensible components survive as scoped hypotheses and operational tests.

The next meaningful advance is not another declaration that the dilemma has been resolved. It is a registered experiment or complete mathematical model in which we can identify exactly which substitutions are possible, which fail, and why the failures make a difference to independent evidence about conscious states. If the organizational proposal offers no advantage over a simpler plasticity or dynamical account, it should lose that claim to distinction. That outcome would be a successful correction, not a reason to abandon the underlying phenomena.

What Survives, and What Still Needs Testing

A distinction becomes scientifically useful when a specified process preserves, transforms, or responds to it in a way that makes different predictions under possible interventions. Thermodynamic constraints often matter to that process. Exactly which energetic accounting applies depends on the implementation; the existence of distinctions does not entail an unconditional per-bit erasure cost.

This article has proposed organizational closure and perspective-sensitive control as one line of research into the conditions associated with consciousness. The strongest claim, that a particular organized control state is identical to phenomenal presence, remains a conditional hypothesis at specified boundaries and timescales. It faces competitors, construct-validity problems, difficult dissociations, and the Kleiner–Hoel inference problem. No mathematical identity or thermodynamic constraint shown here closes that phenomenal bridge on its own.

The method is the more general result. A theory proposing fundamental experience, a privileged biological substrate, a computational realization, a spiritual agency, or exhaustive physicalism can be assessed for what it actually asserts. We should ask which evidence establishes its particular contribution, what alternatives explain the same evidence, whether its inference crosses an illicit scale or explanatory boundary, and what argument or observation would reverse the assessment. A metaphysical or phenomenological claim need not masquerade as a laboratory hypothesis; it must still meet the standards appropriate to its stated claim, and empirical assertions remain answerable to evidence.

Our present physical sciences deserve their remarkable successes. Those successes neither terminate the inquiry into possible intrinsic qualities nor license an unsupported addition merely because something remains unexplained. The evidential asymmetry is about demonstrated mechanisms at tested targets, not a license for a final cosmic inventory.

If an organizational account predicts no more than a simpler dynamical model, it should lose its special claim. If an anti-physicalist or supernaturalist account yields independently supported discriminators, the same protocol must allow it to gain support. If evidence cannot distinguish the rivals, the responsible outcome is recorded uncertainty and a sharper test. That is the correction mechanism this programme should be judged by.

Appendix A: Rival Accounts and Their Unsettled Proof Duties

This matrix identifies research tasks, not automatic victories or universal refutations. Distinct empirical, formal, phenomenological, modal and ontological claims have different evidential requirements.

Appendix B: Key Scholars and Their Contributions

Thermodynamics/Information

• Rolf Landauer (1961): Logical irreversibility in physical computation, including idealized one-bit reset under specified assumptions; no universal per-distinction maintenance cost.
• Charles Bennett (1973, 1982): Reversible computation, Maxwell’s Demon resolution.
• Bérut et al. (2012): Experimental verification of Landauer bound.
• Jarzynski (1997), Crooks (1999), Seifert (2012): Fluctuation theorems, stochastic thermodynamics.
• Huw Price: Thermodynamic arrow of time, time asymmetry.
• Carlo Rovelli: Time and thermodynamics, relational physics.

Organizational Closure

• Montévil & Mossio (2015): Biological organization as closure of constraints.
• Moreno & Mossio (2015): Full theory of biological autonomy.
• Hordijk, Steel and Kauffman (2012): Analysis of reflexively autocatalytic, food-generated sets and their structure, enablement and possible evolution, within stated chemical-reaction-system models. This does not by itself establish phenomenality or a living cell.
• Robert Rosen: Relational biology, constraint vs causal explanation.

Structural Realism

• Ladyman & Ross (2007): Ontic structural realism. Relations primary.
• French (2003, 2014): Technical foundations of OSR.
• Rovelli (1996): Relational Quantum Mechanics.

Levels and Emergence

• George Ellis: Downward causation as constraint reshaping.
• Terrence Deacon (2011): Incomplete Nature, teleodynamics, absence-based explanation, constraints as eliminative.
• Howard Pattee: Constraint vs dynamics distinction; symbols as constraints, not forces.
• Jeremy England (2013–2015): Statistical mechanics of self-replication and examples of dissipative adaptation in driven self-assembly (2015); no general theorem that all driven systems evolve toward maximal dissipation.
• Erik Hoel, Giulio Tononi, Larissa Albantakis: Models of effective information and causal emergence across selected coarse-grainings; a macrolevel advantage needs to be demonstrated for the measured system.

Falsification and Methodology

• Popper (1959): Falsifiability as demarcation criterion.
• Imre Lakatos: Methodology of scientific research programmes, progressive versus degenerating problem shifts.
• Deborah Mayo (2018): Severe testing and the evidential appraisal of error-probing methods, beyond one-shot deductive refutation.
• Flew (1950): Death of a thousand qualifications.
• Whitehead: Fallacy of misplaced concreteness.

Consciousness (Empirical)

• Giacino et al. (2002, 2004): CRS-R, minimally conscious state criteria.
• Owen et al. (2006): fMRI detection of covert awareness.
• Stanislas Dehaene: Global neuronal workspace, ignition dynamics and proposed markers of conscious access; the status of P3b as a consciousness correlate remains debated. Dehaene and colleagues (2001) measured processing and repetition priming by masked words in a specified human reading experiment; unconscious information processing alone does not decide which mechanisms constitute subjective experience.
• Baars (1988): Original Global Workspace Theory, cognitive architecture.
• Friston: Free energy principle, active inference, predictive processing.
• Lau and Rosenthal (2011): A review of the evidence offered for higher-order representations as part of conscious awareness, including prefrontal hypotheses. A review of supporting evidence is not proof that every prefrontal activation measures phenomenality.
• Rosenthal: Higher-Order Thought theory, consciousness as meta-representation.
• Varela, Thompson, Rosch: Enactivism, embodied cognition.
• Giulio Tononi, Marcello Massimini and collaborators: IIT and clinical perturbational complexity measures developed and validated by teams including Casali (2013) and Casarotto (2016); PCI does not directly measure thermodynamic closure.
• Mashour, Roelfsema, Changeux and Dehaene (2020): A review of global neuronal workspace mechanisms of conscious access, with anesthetic states among the important comparisons; it is not an experimental proof that GNW exhausts conscious experience.
• Benjamin Libet (1983, 1985): Early 1985 timing/veto discussion, alongside 1983 readiness-potential experiments and reported intention; the interpretation for agency and free will is contested.

Consciousness (Philosophical)

• Chalmers: Hard Problem formulation, conceivability arguments, zombie intuition.
• Dennett: Competence without comprehension, quining qualia, zombie critique.
• Frankish: Illusionism, appearance of qualia.
• Jackson: Knowledge argument, Mary’s Room, epiphenomenal qualia.
• Levine: Explanatory gap formulation.
• Nagel: “What is it like to be” formulation, bat argument.
• Nāgārjuna: Madhyamaka analysis of dependent origination, emptiness of intrinsic nature and conventional designation. This is a philosophical analysis, not a neuroscientific result or elimination of conventional persons.

Bioelectricity

• Durant et al. (2017, 2019): Path-dependent bioelectric memory in planarians.
• Levin (empirical): Valid bioelectric manipulation research. (Metaphysical interpretation criticized.)

Interface Theory / Idealist Approaches (Critique)

• Donald Hoffman, Manish Singh and Chetan Prakash (2015): Interface Theory of Perception. The formally named Fitness-Beats-Truth theorem appeared in Prakash and colleagues (2021) under specified evolutionary assumptions; a conscious-agent ontology requires separate warrant.
• Prakash: Mathematical formalization of conscious agents as Markov kernels.
• Jeffrey Bagwell (2023): Dialectical self-defeat objection to the evidential use of evolutionary theory in a wholesale debunking of objective perception; not a demonstrated impossibility of nonveridical perception.
• Manolo Martínez (2019): A family of formal counterexamples to unrestricted inferences from fitness usefulness to nonveridical representation; results depend on the chosen signaling and information structure.

Pattern/Process

• Gregory Bateson: Information as a difference that makes a difference, a useful cybernetic formulation; it supplies no universal Landauer erasure budget for persistent distinctions.
• Sagan: Dragon in the garage, unfalsifiability parable.
• Epicurus: Death argument, null perspective.

Indigenous Epistemology

• Yunkaporta (2020): Sand Talk, process over artifact, relational knowledge.
• Lynne Kelly (2016/2017): The Memory Code, an interpretation of place-based mnemonic traditions; not a controlled demonstration of general Songline superiority.
• Kimmerer: Braiding Sweetgrass, relational ontology, reciprocity.
• David Reser and colleagues (2021): A study of two taught memory methods in medical students; the Aboriginal-inspired method had 12/19 students reach perfect list recall, compared with 10/22 in the memory-palace group. Reported OR = 2.82 concerns improvement relative to the study baseline, not a direct threefold probability advantage over the other method.

Appendix C: The Conditional Bridge, Its Logic and Its Loss Conditions

The article’s positive consciousness hypothesis is abductive and target-specific. It has no demonstrated logical entailment from thermodynamic bounds, the existence of a maintenance cycle, effective information or clinical PCI to phenomenal presence. The point of an explicit bridge is to make these missing steps and their possible failures visible.

C.1 Claim Types and Notation

Let \(B\) denote a proposed subject or system boundary and \(\tau\) the timescale. Let \(C_{B,\tau}\) denote an operationally measured organization variable, \(Y\) a separately calibrated consciousness-related target, and \(R_1,\ldots,R_m\) serious rival explanatory models. Define an empirical contrast on held-out interventions \(I\):

\[ \Delta_{B,\tau}(I) =\mathcal L(R_{\mathrm{best}},Y\mid I)-\mathcal L(H_C,Y\mid I), \]

where \(\mathcal L\) is a prespecified comparable loss function evaluated on independent held-out data, and \(H_C\) is the closure-related predictor. This expression specifies the proposed scoring contract. A positive \(\Delta\) means a closure model scores better under the selected loss. Researchers must predefine how the best rival is selected, account for multiple comparisons and shared training information, and report uncertainty. The sign of \(\Delta\) is meaningful only when those choices are justified.

Label provenance is part of the model. Write \(Y=g(E,L,U)\), where \(E\) is the actual evidence used to classify a case, \(L\) the labeling procedure, and \(U\) recorded or unknown measurement dependencies. If \(g\) uses the candidate closure variable \(C\) to assign the label, any score comparison using that same label must be identified as circular for the phenomenal question. Independent collection and independent validation are different achievements. An unresolved label bridge makes a phenotype result exploratory rather than phenomenal corroboration.

As a deliberately artificial arithmetic illustration, let the two values of \(C\) be equally common, with \(P(Y=1\mid C=0)=1/5\) and \(P(Y=1\mid C=1)=4/5\). Marginal prevalence is \(1/2\). Conditioning lowers the probability in one group and raises it in the other, so no pointwise probability-increase law follows. Perfectly calibrated probabilities would lower population Brier loss from \(1/4\) to \(4/25\), a gain of \(9/100\). These exact fractions belong to the constructed example, not an experiment or an estimate of consciousness.

The stronger phenomenal proposition \(H_P(B,\tau)\) says that a specified control organization partly or wholly constitutes a specified aspect of phenomenal perspective. Its support additionally depends on the bridge from measured \(Y\) to phenomenality, the role of participant evidence, possible dissociations, and relevant philosophical arguments. Without that support, \(H_P\) remains a hypothesis.

C.2 What the Scores Can and Cannot Show

Measured integration, global access, metacognitive calibration, self-report and organismic viability are not interchangeable indicators. Two causal models may fit the same observations but disagree about hidden constitution. A third-person predictor could be excellent at detecting calibrated human conscious states while offering no decisive insight into whether an unfamiliar system has experience.

An empirical gain should be credited exactly where it was obtained. A successful clinical predictor is not by itself an ontology, and a failed competitor does not transfer its lost support to the closure hypothesis unless the alternatives are exhaustive or additional evidence favors closure. When the rivals remain empirically equivalent, record underdetermined.

C.3 Predeclared Loss Conditions

• Mechanistic loss: a simpler physical or dynamical model matches or outperforms the closure-specific model under the same holdout, intervention and calibration conditions.
• Constitutive loss: independent, well-calibrated evidence of phenomenality systematically dissociates from a structure claimed to be necessary or sufficient, once reportability and measurement errors have been examined.
• Formal loss: a type error, countermodel, failed proof obligation or an inadmissible change of quantifiers defeats a claimed universal inference.
• Transport loss: an operational marker loses construct validity when moved between people, species, substrates, organizations or temporal scales.
• Methodological loss: the Recursive Corrigibility Framework does not outperform a simpler, equally resourced rival-audit procedure on reliable error discovery, prediction or correction.
• Measurement-bridge loss: circular classification, unreliable evidence transport or dependent measures invalidate a claim of independent phenomenal corroboration without negating a separately tested engineering result.
• Problem-form loss: a demonstrably indispensable joint explanatory relation or positive phenomenal residual defeats the proposed dissolution of an additional unitary Hard Problem.

Each loss updates only the dependent claims unless the entire theory truly rests on the failed premise. Record negative controls, replication status and uncertainty. A particular formal proof may succeed while its physical realizability remains unknown; a relevant experiment may succeed without establishing phenomenal constitution.

C.4 Clinical, Developmental and Cross-scale Interpretation

Sleep, anesthesia, covert command-following and clinical brain-response measures offer valuable tests of specific capacities. They should not be compressed into an unconditional statement that “anesthesia abolishes experience” or “closure restoration restores consciousness.” Memory, report, responsiveness and felt experience may dissociate. Absent agreement about an operational indicator across species or platforms, a cross-scale classification is not machine-certified by a shared algorithm.

The appropriate final question is: What data, formal derivation, or independently justified argument would make this particular bridge lose support? Until an answer is feasible, the claim can remain conceptually significant, but its empirical standing must remain limited.

Appendix D: Assumptions, Rivals and What Would Change the Assessment

Counting assumptions without defining a common unit is not a defensible model-complexity comparison. One programme’s explicit axiom, another’s experimental auxiliary and a third’s metaphysical primitive are not interchangeable objects that can be totaled as if their number ranked truth. This appendix records the actual inferential obligations instead.

D.1 Commitments of the Present Programme

1. Physical operations: Observed physically realized memory, regulation and interaction processes obey the applicable laws of thermodynamics under the relevant boundaries. The use of Landauer’s one-bit erasure bound is limited to matching assumptions.
2. Organizational models: Some biological systems have mutually dependent constraints that can be investigated using the definitions developed by Montévil and Mossio. Applying those definitions to an engineered, ecological or hybrid system requires fresh validation.
3. Functional hypothesis: Some boundary-sensitive control and self-model variables may improve mechanistic, developmental and clinical prediction beyond simpler models. This is an empirical question with a testable loss condition.
4. Phenomenal hypothesis: In certain systems, organized control dynamics may constitute aspects of phenomenal perspective. This is an additional constitutive claim for which no universal proof or decisive experiment is supplied here.
5. Methodological rule: Claims should be typed, their evidence traced, rivals considered and losses recorded. The rule itself should earn its complexity by outperforming a simpler alternative.
6. Ontology: Exhaustive physicalism, necessary panpsychism, Platonic causal ingression and supernatural causal agency are all open to criticism at their appropriate tribunals. None receives an automatic victory from the failure of another theory.

D.2 Strong Rivals Deserve Strong Versions

Global workspace, recurrent processing, higher-order thought, integrated information, biological-realization, illusionist and phenomenal-agnostic approaches specify different target constructs. A theory may explain access without settling intrinsic constitution; another may offer a systematic modal account with little present empirical discrimination. The critic must identify the correct version and proposition. The categorical statement “all rival frameworks are unfalsifiable” is false; so is an unsupported claim that every quantum or physicalist model has already been refuted.

Levin’s experimental bioelectric research and the stronger Platonic morphospace interpretation must be kept separate. A successful dynamical explanation of development does not refute every Platonism. If a proposed ingression contributes causal effects, it needs distinct predictions and a physical bridge. The comparison is about which addition has actually earned explanatory credit.

D.3 The Assumption Audit Must Be Revisionable

For every major assertion, archive the exact wording and source, declared boundary and time, predicted differences, instrumental or formal checks, important rivals, observed failures, and dependency updates. Do not use a philosophical or computational vocabulary as a substitute for data. Save successful and failed branches in the revision history.

The typed knowledge record must preserve claims, rival models, provenance, bridges, dependencies and explicit conditions for losing support. Do not store an inference as a permanent conclusion merely because its supporting code returned a valid output. Track whether a formal counterexample targets necessity, a clinical observation targets a state classifier, or a conceptual objection targets the status of an additional unified problem. A failed bridge changes its dependent assertion, while independently supported components remain available for future tests.

The article’s own high-risk premises are the strongest test of the method. The thermodynamic necessity claim has already required narrowing; the proof that organizational closure constitutes experience has not been supplied; the universal “something rather than nothing” argument overreached; and local empirical success cannot promote any ontology to exhaustive status. None of these admissions defeats the independent clinical and biological studies. They define where future work can genuinely improve the programme.

The most important remaining outcome is therefore not a final answer about consciousness. It is a public, inspectable method that makes it more difficult for any theory, including this one, to win by changing its claim after the evidence arrives.

Revision History

November 10, 2026, substantive revision and website publication notice. The entries below document the changes made in this revision, including the methodology, tests, formal qualifications, source corrections, references and provenance. Read the full rebuildable UTF-8 revision diff (TXT). The TXT preserves exact changes; the summary below is not a substitute for the full diff.

• Scope. Clarifies the central contribution as a reusable, revisable method for evaluating consciousness theories. The organizational identification is a target-specific hypothesis. Physical explanatory success does not settle exhaustive ontology.
• Thermodynamics. Distinguishes information-bearing physical operations from logically irreversible erasure, with explicit constraints on any application of Landauer’s bound.
• Logic and mathematics. Reworks the Kleiner–Hoel falsification dilemma and narrows the proposed bridge to testable, conditional statements; counterexamples and open proof obligations are documented.
• Rival explanations. Applies comparable evidential tests to physicalist, nonphysicalist, biological, computational and presently undecided proposals without presuming a final winner.
• Clinical and cross-cultural evidence. Separates measurements of behavior, perturbational complexity and memory from philosophical claims those studies did not test.
• Bibliography. Separates an originally concatenated reference list, corrects five mismatched source identities or DOI links using original publisher records, and narrows attribution in the scholarly overview where empirical findings had been overstated.
• Audit and provenance. The author completed his independent primary-source review on November 10, 2026 and reported no further errors beyond corrections already incorporated. The public UTF-8 TXT revision audit compares the recovered earlier edition with the approved substantive revision; the lossless Sweet Rationalism website history ZIP preserves prior versions for reconstruction. These records document the revision provenance.
• Recursive methodology and prospective tests. Explicitly adds the conditional positive-residual and stopping-rule challenges, independent consciousness-label requirements, the joint organizational prediction fingerprint, a two-tier closed-loop comparison and a metaproblem protocol. These proposals assign explicit loss conditions to the claims each test addresses.
• Versioned public deposit. The Zenodo preprint archive is version 1.1.0 (DOI 10.5281/zenodo.24119682); the all-versions DOI tracks the deposit series. This live canonical reading edition may subsequently receive link and metadata corrections without altering the immutable archived ZIP.

References

Thermodynamics and Information Theory

Landauer, R. (1961). Irreversibility and heat generation in the computing process. IBM Journal of Research and Development, 5(3), 183–191. https://doi.org/10.1147/rd.53.0183

Bennett, C. H. (1982). The thermodynamics of computation-a review. International Journal of Theoretical Physics, 21(12), 905–940. https://doi.org/10.1007/BF02084158

Bérut, A., Arakelyan, A., Petrosyan, A., Ciliberto, S., Dillenschneider, R., & Lutz, E. (2012). Experimental verification of Landauer’s principle linking information and thermodynamics. Nature, 483(7388), 187–189. https://doi.org/10.1038/nature10872

Jarzynski, C. (1997). Nonequilibrium equality for free energy differences. Physical Review Letters, 78(14), 2690. https://doi.org/10.1103/PhysRevLett.78.2690

Crooks, G. E. (1999). Entropy production fluctuation theorem and the nonequilibrium work relation for free energy differences. Physical Review E, 60(3), 2721. https://doi.org/10.1103/PhysRevE.60.2721

Seifert, U. (2012). Stochastic thermodynamics, fluctuation theorems and molecular machines. Reports on Progress in Physics, 75(12), 126001. https://doi.org/10.1088/0034-4885/75/12/126001

Price, H. (1996). Time’s Arrow and Archimedes’ Point: New Directions for the Physics of Time. Oxford University Press.

Rovelli, C. (2018). The Order of Time. Riverhead Books.

Organizational Closure and Biological Autonomy

Montévil, M., & Mossio, M. (2015). Biological organisation as closure of constraints. Journal of Theoretical Biology, 372, 179–191. https://doi.org/10.1016/j.jtbi.2015.02.029

Moreno, A., & Mossio, M. (2015). Biological Autonomy: A Philosophical and Theoretical Enquiry. Springer.

Hordijk, W., Steel, M., & Kauffman, S. (2012). The structure of autocatalytic sets: Evolvability, enablement, and emergence. Acta Biotheoretica, 60(4), 379–392. https://doi.org/10.1007/s10441-012-9165-1

Kauffman, S. A. (2019). A World Beyond Physics: The Emergence and Evolution of Life. Oxford University Press.

Rosen, R. (1991). Life Itself: A Comprehensive Inquiry into the Nature, Origin, and Fabrication of Life. Columbia University Press.

Constraint-Based Emergence and Teleodynamics

Deacon, T. W. (2011). Incomplete Nature: How Mind Emerged from Matter. W. W. Norton & Company.

Pattee, H. H. (2001). The physics of symbols: Bridging the epistemic cut. Biosystems, 60(1-3), 5–21. https://doi.org/10.1016/S0303-2647(01)00104-6

England, J. L. (2015). Dissipative adaptation in driven self-assembly. Nature Nanotechnology, 10(11), 919–923. https://doi.org/10.1038/nnano.2015.250

Hoel, E. P., Albantakis, L., & Tononi, G. (2013). Quantifying causal emergence shows that macro can beat micro. Proceedings of the National Academy of Sciences, 110(49), 19790–19795. https://doi.org/10.1073/pnas.1314922110

Ellis, G. F. R. (2012). Top-down causation and emergence: Some comments on mechanisms. Interface Focus, 2(1), 126–140. https://doi.org/10.1098/rsfs.2011.0062

Active Inference and Markov Blankets

Parr, T., Pezzulo, G., & Friston, K. J. (2022). Active Inference: The Free Energy Principle in Mind, Brain, and Behavior. MIT Press.

Pezzulo, G., Parr, T., & Friston, K. (2024). Active inference as a theory of sentient behavior. Biological Psychology, 186, 108741. https://doi.org/10.1016/j.biopsycho.2023.108741

Structural Realism

Ladyman, J., Ross, D., Spurrett, D., & Collier, J. (2007). Every Thing Must Go: Metaphysics Naturalized. Oxford University Press.

French, S. (2014). The Structure of the World: Metaphysics and Representation. Oxford University Press.

Rovelli, C. (1996). Relational quantum mechanics. International Journal of Theoretical Physics, 35(8), 1637–1678. https://doi.org/10.1007/BF02302261

Thermodynamic Arrow and Time

Price, H. (1996). Time’s Arrow and Archimedes’ Point: New Directions for the Physics of Time. Oxford University Press.

Rovelli, C. (2018). The Order of Time. Riverhead Books.

Consciousness (Empirical)

Dehaene, S., & Changeux, J. P. (2011). Experimental and theoretical approaches to conscious processing. Neuron, 70(2), 200–227. https://doi.org/10.1016/j.neuron.2011.03.018

Tononi, G., & Koch, C. (2015). Consciousness: Here, there and everywhere? Philosophical Transactions of the Royal Society B, 370(1668), 20140167. https://doi.org/10.1098/rstb.2014.0167

Massimini, M., Ferrarelli, F., Huber, R., Esser, S. K., Singh, H., & Tononi, G. (2005). Breakdown of cortical effective connectivity during sleep. Science, 309(5744), 2228–2232. https://doi.org/10.1126/science.1117256

Casali, A. G., Gosseries, O., Rosanova, M., et al. (2013). A theoretically based index of consciousness independent of sensory processing and behavior. Science Translational Medicine, 5(198), 198ra105. https://doi.org/10.1126/scitranslmed.3006294

Mashour, G. A., Roelfsema, P., Changeux, J. P., & Dehaene, S. (2020). Conscious processing and the global neuronal workspace hypothesis. Neuron, 105(5), 776–798. https://doi.org/10.1016/j.neuron.2020.01.026

Owen, A. M., Coleman, M. R., Boly, M., Davis, M. H., Laureys, S., & Pickard, J. D. (2006). Detecting awareness in the vegetative state. Science, 313(5792), 1402. https://doi.org/10.1126/science.1130197

Giacino, J. T., Kalmar, K., & Whyte, J. (2004). The JFK Coma Recovery Scale-Revised: Measurement characteristics and diagnostic utility. Archives of Physical Medicine and Rehabilitation, 85(12), 2020–2029. https://doi.org/10.1016/j.apmr.2004.02.033

Andrews, K., Murphy, L., Munday, R., & Littlewood, C. (1996). Misdiagnosis of the vegetative state: Retrospective study in a rehabilitation unit. BMJ, 313(7048), 13–16. https://doi.org/10.1136/bmj.313.7048.13

Schnakers, C., Vanhaudenhuyse, A., Giacino, J., et al. (2009). Diagnostic accuracy of the vegetative and minimally conscious state: Clinical consensus versus standardized neurobehavioral assessment. BMC Neurology, 9, 35. https://doi.org/10.1186/1471-2377-9-35

Alkire, M. T., Hudetz, A. G., & Tononi, G. (2008). Consciousness and anesthesia. Science, 322(5903), 876–880. https://doi.org/10.1126/science.1149213

Friston, K. (2010). The free-energy principle: A unified brain theory? Nature Reviews Neuroscience, 11(2), 127–138. https://doi.org/10.1038/nrn2787

Baars, B. J. (1988). A Cognitive Theory of Consciousness. Cambridge University Press.

Dehaene, S., Naccache, L., Cohen, L., Le Bihan, D., Mangin, J. F., Poline, J. B., & Rivière, D. (2001). Cerebral mechanisms of word masking and unconscious repetition priming. Nature Neuroscience, 4(7), 752–758. https://doi.org/10.1038/89551

Rounis, E., Maniscalco, B., Rothwell, J. C., Passingham, R. E., & Lau, H. (2010). Theta-burst transcranial magnetic stimulation to the prefrontal cortex impairs metacognitive visual awareness. Cognitive Neuroscience, 1(3), 165–175. https://doi.org/10.1080/17588921003632529

Bor, D., Schwartzman, D. J., Barrett, A. B., & Seth, A. K. (2017). Theta-burst transcranial magnetic stimulation to the prefrontal or parietal cortex does not impair metacognitive visual awareness. PLOS ONE, 12(2), e0171793. https://doi.org/10.1371/journal.pone.0171793

Ruby, E., Maniscalco, B., & Peters, M. A. K. (2018). On a ‘failed’ attempt to manipulate visual metacognition with transcranial magnetic stimulation to prefrontal cortex. Consciousness and Cognition, 62, 34–41. https://doi.org/10.1016/j.concog.2018.04.009

Martin, A., Lane, T. J., & Hsu, T.-Y. (2023). DLPFC-PPC-cTBS effects on metacognitive awareness. Cortex, 167, 41–50. https://doi.org/10.1016/j.cortex.2023.05.022

Lau, H., & Rosenthal, D. (2011). Empirical support for higher-order theories of conscious awareness. Trends in Cognitive Sciences, 15(8), 365–373. https://doi.org/10.1016/j.tics.2011.05.009

Block, N. (2011). Perceptual consciousness overflows cognitive access. Trends in Cognitive Sciences, 15(12), 567–575. https://doi.org/10.1016/j.tics.2011.11.001

Rosenthal, D. M. (2005). Consciousness and Mind. Oxford University Press.

Varela, F. J., Thompson, E., & Rosch, E. (1991). The Embodied Mind: Cognitive Science and Human Experience. MIT Press.

Libet, B., Gleason, C. A., Wright, E. W., & Pearl, D. K. (1983). Time of conscious intention to act in relation to onset of cerebral activity (readiness-potential): The unconscious initiation of a freely voluntary act. Brain, 106(3), 623–642. https://doi.org/10.1093/brain/106.3.623

Libet, B. (1985). Unconscious cerebral initiative and the role of conscious will in voluntary action. Behavioral and Brain Sciences, 8(4), 529–539. https://doi.org/10.1017/S0140525X00044903

Casarotto, S., Hassan, G., Rosanova, M., Sarasso, S., Derchi, C.-C., Trimarchi, P. D., Viganò, A., Russo, S., Fecchio, M., Devalle, G., Navarro, J., Massimini, M., & Comanducci, A. (2024). Dissociations between spontaneous electroencephalographic features and the perturbational complexity index in the minimally conscious state. European Journal of Neuroscience, 59(5), 934–947. DOI: 10.1111/ejn.16299.

Casarotto, S., et al. (2016). Stratification of unresponsive patients by an independently validated index of brain complexity. Annals of Neurology, 80, 718–729. https://doi.org/10.1002/ana.24779

Consciousness (Philosophical)

Chalmers, D. J. (1995). Facing up to the problem of consciousness. Journal of Consciousness Studies, 2(3), 200–219.

Chalmers, D. J. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press.

Chalmers, D. J. (1997). Moving forward on the problem of consciousness. Author-hosted essay. Original text.

Chalmers, D. J. (1998). On the search for the neural correlate of consciousness. Author-hosted essay. Original text.

Chalmers, D. J. (2016). The combination problem for panpsychism. In G. Brüntrup & L. Jaskolla (Eds.), Panpsychism: Contemporary Perspectives (pp. 179–214). Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199359943.003.0008

Coleman, S. (2014). The real combination problem: Panpsychism, micro-subjects, and emergence. Erkenntnis, 79(1), 19–44. https://doi.org/10.1007/s10670-013-9431-x

Strawson, G. (2006). Realistic monism: Why physicalism entails panpsychism. Journal of Consciousness Studies, 13(10-11), 3–31.

Dennett, D. C. (1988). Quining qualia. In A. J. Marcel & E. Bisiach (Eds.), Consciousness in Contemporary Science (pp. 42–77). Oxford University Press.

Dennett, D. C. (1991). Consciousness Explained. Little, Brown and Company.

Frankish, K. (2016). Illusionism as a theory of consciousness. Journal of Consciousness Studies, 23(11-12), 11–39.

Jackson, F. (1982). Epiphenomenal qualia. Philosophical Quarterly, 32(127), 127–136. https://doi.org/10.2307/2960077

Levine, J. (1983). Materialism and qualia: The explanatory gap. Pacific Philosophical Quarterly, 64(4), 354–361. https://doi.org/10.1111/j.1468-0114.1983.tb00207.x

Nagel, T. (1974). What is it like to be a bat? Philosophical Review, 83(4), 435–450. https://doi.org/10.2307/2183914

Falsification and Methodology

Popper, K. R. (1959). The Logic of Scientific Discovery. Hutchinson.

Lakatos, I. (1978). The Methodology of Scientific Research Programmes. Cambridge University Press.

Flew, A. (1950). Theology and falsification. University, 1, 1–8.

Whitehead, A. N. (1925). Science and the Modern World. Macmillan.

Kleiner, J., & Hoel, E. (2021). Falsification and consciousness. Neuroscience of Consciousness, 2021(1), niab001. DOI: 10.1093/nc/niab001.

O’Reilly-Shah, V. N., Selvitella, A. M., & Schurger, A. (2026). A caveat regarding the unfolding argument: Implications of plasticity. Neuroscience of Consciousness, 2026(1), niag027. https://doi.org/10.1093/nc/niag027

Indigenous Epistemology

Yunkaporta, T. (2020). Sand Talk: How Indigenous Thinking Can Save the World. HarperOne. (Originally published 2019 by Text Publishing, Australia.)

Kelly, L. (2017). The Memory Code: The Secrets of Stonehenge, Easter Island and Other Ancient Monuments. Pegasus Books. (Originally published 2016 by Allen & Unwin, Australia.)

Kimmerer, R. W. (2013). Braiding Sweetgrass: Indigenous Wisdom, Scientific Knowledge and the Teachings of Plants. Milkweed Editions.

Reser, D., Simmons, M., Johns, E., Ghaly, A., Quayle, M., Dordevic, A. L., Tare, M., McArdle, A., Willems, J., & Yunkaporta, T. (2021). Australian Aboriginal techniques for memorization: Translation into a medical and allied health education setting. PLOS ONE, 16(5), e0251710. https://doi.org/10.1371/journal.pone.0251710

Native American Philosophy

Wilson, S. (2008). Research Is Ceremony: Indigenous Research Methods. Fernwood Publishing.

Bioelectricity

Durant, F., Morokuma, J., Fields, C., Williams, K., Adams, D. S., & Levin, M. (2017). Long-term, stochastic editing of regenerative anatomy via targeting endogenous bioelectric gradients. Biophysical Journal, 112(10), 2231–2243. https://doi.org/10.1016/j.bpj.2017.04.011

Durant, F., Bischof, J., Fields, C., Morokuma, J., LaPalme, J., Hoi, A., & Levin, M. (2019). The role of early bioelectric signals in the regeneration of planarian anterior/posterior polarity. Biophysical Journal, 116(5), 948–961. https://doi.org/10.1016/j.bpj.2019.01.029

Chis-Ciure, R., & Levin, M. (2025). Cognition all the way down 2.0: Neuroscience beyond neurons in the diverse intelligence era. Synthese, 206, article 257. https://doi.org/10.1007/s11229-025-05319-6.

Interface Theory of Perception (Critique)

Hoffman, D. D., Singh, M., & Prakash, C. (2015). The interface theory of perception. Psychonomic Bulletin & Review, 22(6), 1480–1506. https://doi.org/10.3758/s13423-015-0890-8

Hoffman, D. D., & Prakash, C. (2014). Objects of consciousness. Frontiers in Psychology, 5, 577. https://doi.org/10.3389/fpsyg.2014.00577

Bagwell, J. N. (2023). Debunking interface theory: Why Hoffman’s skepticism (really) is self-defeating. Synthese, 201(25). https://doi.org/10.1007/s11229-022-04021-1

Martínez, M. (2019). Usefulness drives representations to truth: A family of counterexamples to Hoffman’s interface theory of perception. Grazer Philosophische Studien, 96(3), 319–341. https://doi.org/10.1163/18756735-09603004

Prakash, C., Stephens, K. D., Hoffman, D. D., Singh, M., & Fields, C. (2021). Fitness Beats Truth in the Evolution of Perception. Acta Biotheoretica, 69(3), 319–341. DOI: 10.1007/s10441-020-09400-0.

Integrated Information Theory (Critique)

Cerullo, M. A. (2015). The problem with phi: A critique of integrated information theory. PLOS Computational Biology, 11(9), e1004286. https://doi.org/10.1371/journal.pcbi.1004286

Mørch, H. H. (2019). Is consciousness intrinsic? A problem for the integrated information theory. Journal of Consciousness Studies, 26(1-2), 133–162.

Doerig, A., Schurger, A., Hess, K., & Herzog, M. H. (2019). The unfolding argument: Why IIT and other causal structure theories cannot explain consciousness. Consciousness and Cognition, 72, 49–59. https://doi.org/10.1016/j.concog.2019.04.002

Tsuchiya, N., Andrillon, T., & Haun, A. (2020). A reply to “the unfolding argument”: Beyond functionalism/behaviorism and towards a science of causal structure theories of consciousness. Consciousness and Cognition, 79, 102877. DOI: 10.1016/j.concog.2020.102877.

Free Energy Principle (Critique)

Bruineberg, J., Kiverstein, J., & Rietveld, E. (2018). The anticipating brain is not a scientist: The free-energy principle from an ecological-enactive perspective. Synthese, 195(6), 2417–2444. https://doi.org/10.1007/s11229-016-1239-1

Kirchhoff, M., Parr, T., Palacios, E., Friston, K., & Kiverstein, J. (2018). The Markov blankets of life: Autonomy, active inference and the free energy principle. Journal of The Royal Society Interface, 15(138), 20170792. https://doi.org/10.1098/rsif.2017.0792

Bruineberg, J., Dolega, K., Dewhurst, J., & Baltieri, M. (2022). The emperor’s new Markov blankets. Behavioral and Brain Sciences, 45, e183. https://doi.org/10.1017/S0140525X21002351

Friston, K. (2022). Maps and territories, smoke, and mirrors. Behavioral and Brain Sciences, 45, e195. https://doi.org/10.1017/S0140525X22000073

Quantum Consciousness (Critique)

Tegmark, M. (2000). Importance of quantum decoherence in brain processes. Physical Review E, 61(4), 4194. https://doi.org/10.1103/PhysRevE.61.4194

Pattern and Process

Bateson, G. (1979). Mind and Nature: A Necessary Unity. Dutton.

Sagan, C. (1995). The Demon-Haunted World: Science as a Candle in the Dark. Random House.

Eastern Philosophy

Garfield, J. L. (1995). The Fundamental Wisdom of the Middle Way: Nāgārjuna’s Mūlamadhyamakakārikā. Oxford University Press.

Olivelle, P. (Trans.). (1996). Upaniṣads. Oxford University Press.

Computation and Formal Models

Turing, A. M. (1937). On computable numbers, with an application to the Entscheidungsproblem. Proceedings of the London Mathematical Society, 2(42), 230–265. DOI: 10.1112/plms/s2-42.1.230.