Thermodynamic Monism: A Falsification-First Naturalism Grounded in Information and Energy Costs

Thermodynamic Monism: A Falsification-First Naturalism Grounded in Information and Energy Costs

Abstract
Thermodynamic monism is a constraint-first methodological naturalism: it treats thermodynamic accounting (energy dissipation, entropy production, irreversibility, control costs) as the minimal admissibility gate for explanations in cognition, biology, and AI. Anything that cannot, even in principle, connect to measurable constraints gets demoted to metaphor, heuristic, or unfalsifiable narrative. The goal is not to “explain everything with thermodynamics” as a slogan, but to use thermodynamics as the universal falsification pressure that prevents skyhooks, non-operational essences, and cost-free abstractions from steering reasoning. It pairs with Recursive Constraint Falsification (RCF) as an execution rule: any claim that cannot specify what would break it does not get to steer inference, memory, or policy.

Thermodynamic monism says there is only one “kind” of stuff doing work in the universe: energy and information in constrained, far from equilibrium dynamics, so minds and cultures get explained the same way as cells and storms (Boltzmann, Landauer, Prigogine). It fits Dennett-style naturalism and secularism because it treats meaning and agency as emergent patterns in physical control loops (RCF operationalizes), not as inputs from a non-physical realm (Dennett; also compatible with Friston-style active inference as a control story). It does not delete the community layer of religion; it reframes it as social coordination technology and identity scaffolding that can be studied, kept, or redesigned without supernatural commitments (Durkheim).

Operational definitions

Thermodynamic

Operational meaning: a claim counts as “thermodynamic” here only if it entails measurable consequences about energy dissipation, entropy change, heat, or irreversibility under specified conditions.

Monism

Operational meaning: do not multiply ontological kinds when one constraint-accounting framework explains the same phenomena with fewer untestable assumptions. “Monism” here is not a vibe. It is a rule against importing explanatory primitives that cannot be priced, tested, or falsified.

Thermodynamic monism

Operational meaning: any explanation of cognition, memory, agency, semantics, or “forms” must remain consistent with (1) physical conservation constraints and (2) the empirically measured costs of information processing, including irreversibility. If an explanation requires violations, it fails.

Naturalism

A research constraint, not a tribal identity marker. Operationally: explanations must be mechanistic enough to be wrong in identifiable ways, and they must remain consistent with conserved quantities and known physical limits.

Recursive Constraint Falsification (RCF)

A runtime discipline for reasoning systems: translate claims into constraints, demand explicit failure conditions, run severity tests (counterfactuals, adversarial probes, measurement-bracket checks), then update by eliminating non-surviving pathways rather than defending them. RCF is how thermodynamic monism becomes executable rather than rhetorical.

Why call this a “naturalism” at all

Because the core commitment is methodological fallibilism, not theological: explanations must cash out in testable mechanisms that respect physical constraints, rather than appealing to extra “stuff” that never touches measurement. Failed methodologies are pruned. Thermodynamic monism adds a stricter constraint than generic naturalism: it requires that explanations pay their energy and irreversibility bills, not just avoid the supernatural.

Why call it “naturalism” instead of a new metaphysics?

This view aligns with Dennett’s anti-skyhook program: explanations should be built from cranes, meaning mechanistic steps that preserve contact with the physical world. Thermodynamic monism adds a stricter gate: even elegant-sounding explanations must “pay rent” by specifying what would be measured, what would change, and what would break the claim. As Landauer framed the core constraint, “the erasure of information always implies the dissipation of heat” and there is a “general relation between thermodynamics and information.” (Bérut et al., Nature, 2012)

Part I: The Core Logic

1.1 Why Thermodynamics is the Admission Standard

Thermodynamics is the only framework that:

1. Applies universally across all scales and substrates (cells, brains, societies, stars, viruses)
2. Provides measurement hooks (energy, entropy, dissipation, control costs)
3. Has operational consequences (constraint violation means prediction failure and system collapse)
4. Is mathematically rigorous (Boltzmann’s H-theorem, Gibbs entropy, Landauer’s principle)
5. Permits falsification (measure energy cost; if it doesn’t match prediction, the framework breaks)

Any explanation that cannot connect to thermodynamic accounting is, by definition, making claims about:

• Immeasurable properties (non-physical essences, intrinsic meanings, observer-independent purposes)
• Cost-free processes (information transfer without dissipation, causation without energy coupling)
• Unfalsifiable narrative structures (after-the-fact story-telling that accommodates any outcome)

When a framework cannot specify thermodynamic costs, it has opted out of natural science. It has become hermeneutics, poetry, or metaphysics. These are valuable human activities. But they should not steer inference about how the world works.

1.2 The Minimal Ontology

Thermodynamic monism commits to exactly one primitive:

Constraint satisfaction in far-from-equilibrium systems under energy budgets.

Everything else is structure built on this foundation:

• Organisms are dissipative structures maintaining improbable configurations
• Minds are modeling systems that reduce prediction error (free energy) under metabolic constraints
• Cultures are inherited coordination structures that lower social free energy
• AI systems are optimization processes constrained by computational power and data
• Meaning is whatever reduces prediction error for goal-directed agents
• Value is whatever satisfies goals under constraints
• Agency is what happens when systems model future states and act to satisfy those models

None of these require anything beyond thermodynamics. None of them are eliminated. All of them are explained.

1.3 The Falsification-First Requirement

Thermodynamic monism pairs with Recursive Constraint Falsification (RCF) as a methodology:

1. Any explanatory claim must specify its thermodynamic budget (what energy does it require, what entropy does it produce?)
2. Any claim must specify its falsifier (what observation would make it wrong?)
3. Any framework that cannot answer (1) and (2) does not get to steer policy, research, or inference

This is not dogmatism. It is intellectual hygiene. If you cannot say what would break your framework, you have opted out of science. You are narrating, not explaining.

Part II: Against the Common Mischaracterizations

2.1 Thermodynamic Monism is Not Materialism

The Confusion: Critics claim thermodynamic monism assumes only matter exists.

The Clarification: Thermodynamic monism makes no claims about what ontologically fundamental categories exist. It is agnostic about whether the universe is made of matter, mind, fields, strings, or information.

What it does claim: Whatever exists, if it is organized and does work, it satisfies thermodynamic principles.

Compatibility with Alternative Metaphysics:

• Idealist variant: Only minds exist, but minds self-organize according to free energy minimization. Consciousness structures itself to reduce prediction error. This is perfectly thermodynamically consistent.
• Dualist variant: Both mind and matter exist, but both satisfy thermodynamic constraints. The interaction problem remains (how do they couple?), but at least both sides of the interaction follow determinable laws.
• Panpsychist variant: Consciousness is fundamental and ubiquitous, but conscious systems still minimize free energy. Panpsychism + thermodynamic constraints is coherent.

The Key Point: Thermodynamic monism is about how systems behave, not about what they are made of. Materialism is about what exists. These are different questions. You can answer one without committing to an answer on the other.

2.2 Thermodynamic Monism is Not Reductionism

The Confusion: Critics claim thermodynamic monism reduces everything to physics.

The Clarification: Thermodynamic monism is anti-reductionist about causal levels. It insists that higher-level organization is causally real and irreducible.

Why: A neuron’s firing pattern cannot be predicted from quantum mechanics alone without representing the neuron’s thermodynamic organization. The emergent level is causally autonomous. It has its own causal vocabulary (membrane potential, synaptic weight, neural coding) that does not translate to lower levels without information loss.

Hierarchical Explanation, Not Reduction:

The thermodynamic constraints that apply at every level (free energy minimization, entropy production, dissipation) do not eliminate the causal autonomy of those levels. They specify the boundary conditions within which each level operates.

An ecosystem minimizes free energy. A forest minimizes free energy. A cell minimizes free energy. A mind minimizes free energy. The constraints are universal. The causal structures at each level remain autonomous.

This is the opposite of reductionism. It is hierarchical naturalism: each level operates under the same universal principles, but each level’s causal structure is real and irreducible.

2.3 Thermodynamic Monism is Not Physicalism

The Confusion: Critics claim thermodynamic monism assumes only physical facts matter.

The Clarification: Thermodynamic monism distinguishes between physical substrate and thermodynamic principles. The principles apply regardless of substrate.

Physicalism is the claim that the physical substrate is the only fundamental reality and everything else is derivative or illusory.

Thermodynamic monism makes no such claim. It says: Whatever the substrate, if it is organized, it follows thermodynamic principles.

Implications:

• Meaning is real not because it is physical, but because it is functionally defined: meaning is whatever reduces prediction error for goal-directed agents.
• Values are real not because they are physical, but because they are structurally defined: values are whatever satisfies goals under constraints.
• Agency is real not because it is physical, but because it is dynamically defined: agency is what happens when systems model future states and act to satisfy those models.

These properties are not physical properties. They are structural, functional, and dynamical properties that apply to organized systems at any scale. Recognizing them as real does not make one a physicalist. It makes one a realist about organization.

2.4 Thermodynamic Monism is Not Nihilism

The Confusion: Critics claim thermodynamic monism denies objective meaning and value.

The Clarification: Thermodynamic monism grounds meaning and value objectively in thermodynamic facts.

Meaning emerges from goal-directedness:

An organism has a goal: minimize free energy, reduce prediction error, maintain metabolic constraints. Representations are meaningful if they reduce prediction error about goal-relevant states. This is not subjective projection. This is functional reality grounded in thermodynamic dynamics.

Value emerges from need:

An organism operates under free energy constraints. There are objectively real facts about what reduces those constraints (nutrients, shelter, safety) and what increases them (predators, toxins, uncertainty). These are not subjective preferences. They are thermodynamic facts.

Agency is real:

An organism that models future states and acts to satisfy those models is genuinely pursuing goals. This is not illusory. It is real causal efficacy grounded in thermodynamic dynamics.

Nihilism claims there are no objective facts about meaning or value. Thermodynamic monism claims there are objective thermodynamic facts grounding meaning and value. These are opposites.

2.5 Thermodynamic Monism is Not Pessimism

The Confusion: Critics claim thermodynamic monism denies progress and improvement.

The Clarification: Thermodynamic monism explains progress as measurable improvement in constraint satisfaction.

Progress is real:

• A better model reduces prediction error (lower free energy cost)
• A better technology reduces dissipation (lower entropy production)
• A better society reduces free energy waste (more efficient coordination)
• Better knowledge reduces surprise (lower information cost)

These are objective improvements measurable in thermodynamic terms. They are not subjective hopes. They are real progress toward more efficient constraint satisfaction.

Suffering is addressable:

Prediction error (surprise) is costly. Reducing prediction error reduces suffering. This grounds an objective ethics: improve models, reduce surprise, diminish suffering. This is pragmatic optimism grounded in thermodynamic principles.

Pessimism denies meaningful change. Thermodynamic monism says meaningful change is change in constraint satisfaction, and such change is constantly possible. That is what organisms do. That is what adaptation means. That is what learning means.

2.6 Thermodynamic Monism is Not Atheism

The Confusion: Critics claim thermodynamic monism assumes God does not exist.

The Clarification: Thermodynamic monism makes no claims about deity. It is metaphysically agnostic.

Theistic variants:

• Theistic thermodynamic monism: God created organisms to minimize free energy under constraints. God established the laws of thermodynamics. God designed minds to reduce prediction error. This is entirely coherent.
• Deistic thermodynamic monism: God created the cosmos, established its thermodynamic laws, and then withdrew. The cosmos operates according to these laws.
• Process theistic thermodynamic monism: God works within and through thermodynamic constraints, not by violating them.

None of these contradict thermodynamic monism. Understanding how thermodynamic constraints operate does not determine whether God established those constraints or works through them.

Secular variants:

• Secular thermodynamic monism: Thermodynamic constraints are brute facts. There is no theological interpretation. This is equally coherent.

Thermodynamic monism is compatible with both. It makes no claim that settles the theological question.

2.7 Thermodynamic Monism Does Not Entail Panpsychism

The Confusion: Critics claim that if all systems minimize free energy, and if consciousness is required for goal-directed behavior, then all systems are conscious.

The Clarification: Free energy minimization does not require consciousness. Bacteria minimize free energy without consciousness. Thermostats minimize energy without consciousness. Ecosystems minimize free energy without consciousness.

The logical error:

Premise 1: Living systems minimize free energy.
Premise 2: Consciousness is required for free energy minimization.
Conclusion: All systems are conscious.

Premise 2 is unestablished. Why should free energy minimization require consciousness? It requires modeling and goal-directed behavior, but these can occur without subjective experience.

Thermodynamic monism is agnostic on consciousness:

• Physicalist consciousness: Only complex brains are conscious. Simple systems minimize free energy without experience.
• Functionalist consciousness: Consciousness depends on information integration complexity. Systems below a threshold are not conscious.
• Panpsychist consciousness: Consciousness is fundamental and ubiquitous. But this is an additional metaphysical claim not entailed by thermodynamic principles.

Thermodynamic monism itself remains neutral. The mechanism of free energy minimization does not determine whether it is conscious.

2.8 Thermodynamic Monism Does Not Entail Pantheism

The Confusion: Critics claim that if thermodynamic principles are universal and fundamental, then they must be divine.

The Clarification: Universal natural principles do not require theological interpretation. Gravity is universal without being divine. Evolution is universal without being divine. Thermodynamic constraints can be universal without being divine.

Theological variants remain open:

• Classical theism: God created thermodynamic laws but remains distinct from them.
• Pantheism: God is identical with the cosmic process governed by thermodynamic principles.
• Atheism: Thermodynamic principles are impersonal facts requiring no theological interpretation.

Thermodynamic monism does not adjudicate between these. It describes how systems work. Whether that description has theological significance remains open.

2.9 Thermodynamic Monism Does Not Entail Deism

The Confusion: Critics claim that describing natural mechanisms entails that those mechanisms operate without divine oversight.

The Clarification: Describing a mechanism does not determine whether it is divinely sustained or guided.

A watchmaker who designs a watch to keep perfect time has created a self-regulating mechanism. The mechanism works without constant tinkering. Yet it embodies the watchmaker’s design and intention. Similarly, a cosmos governed by thermodynamic principles might embody divine design and intention.

Theological positions remain open:

• Active engagement theism: God sustains creation moment by moment through thermodynamic principles.
• Deistic theism: God established the laws and withdrew.
• Providence theism: God works through thermodynamic constraints to guide creation.
• Atheistic view: No divine guidance; thermodynamics is all.

Understanding the mechanism does not settle the theological question.

2.10 Thermodynamic Monism is Anti-Utopian (Not Technological Theism)

The Confusion: Critics claim thermodynamic monism leads to naive faith that technology will transcend human limits.

The Clarification: Thermodynamic monism is explicitly anti-utopian. It grounds improvement within constraints, not transcendence of constraints.

The hard limits:

• Energy is costly (thermodynamic constraint)
• Computation dissipates (Landauer’s principle)
• Information transfer requires work (thermodynamic accounting)
• There is no escape from entropy production (second law)

Thermodynamic monism says: We can improve by working more efficiently within these limits, but we cannot escape them. No technology transcends the laws of thermodynamics.

This is the opposite of technological theism, which believes technology will eventually transcend physical limits and achieve transcendence. Thermodynamic monism says no. We will become more efficient. We will reduce waste. We will do more with less. But we remain subject to thermodynamic constraints forever.

This is boring. It is not messianic. It promises work and incremental improvement, not salvation.

2.11 Thermodynamic Monism Does Not Entail Panatheism

The Confusion: Critics claim that if thermodynamic principles pervade all systems, then they must manifest divine presence.

The Clarification: Universal principles do not require theological interpretation at all.

Open interpretations:

• Panatheistic: God pervades all creation through thermodynamic constraints.
• Secular: Thermodynamic principles are impersonal and require no theological meaning.
• Theistic: God created thermodynamic principles but remains distinct from them.

Thermodynamic monism does not decide which interpretation is correct. It describes the mechanism. Theology is a separate layer of interpretation.

Part III: What Thermodynamic Monism Actually Explains

3.1 Cognition and Mind

The Problem: How do minds work? What distinguishes cognitive systems from non-cognitive ones?

The Thermodynamic Answer:

Cognitive systems are modeling systems that predict sensory input and act to reduce prediction error (free energy). The brain is a prediction machine optimizing under metabolic constraints.

• Perception is model-building: the brain constructs generative models of sensory input and compares predictions to actual input.
• Action is model-satisfaction: organisms act to make sensory input match predictions (active inference).
• Learning is model-refinement: repeated error reduces prediction error, improving the model.
• Consciousness is the functional property of models that integrate information across sensory domains.
• Attention is computational resource allocation: focus on high-prediction-error inputs.
• Emotion is model-disruption signaling: surprise and uncertainty trigger behavioral reorientation.

All of this follows from free energy minimization under metabolic constraints. No non-physical realm required. No mysterious consciousness essence required. Just control loops optimizing under budgets.

Empirical Support:

• Predictive coding explains perceptual illusions, learning rates, and attentional blink (Friston, Bastos, Hohwy)
• Active inference explains how organisms learn and control their environment (Friston, Parr)
• Information geometry shows how brains organize information to minimize free energy (Amari, Ao)
• Neuropathology confirms: disruption of free energy budgets (metabolic dysfunction, anesthesia) disrupts consciousness proportionally

3.2 Morphogenesis and Development

The Problem: How do organisms develop consistent forms? Why don’t developmental perturbations cause permanent deformity?

The Thermodynamic Answer:

Organisms develop by constraint closure networks that converge toward minimal free energy configurations. Bioelectric patterns, gene expression, and morphogenetic fields all satisfy the same principle: systems settle into energetically stable attractors.

• Genetic networks are constraint propagation systems: mutations alter constraint structure; natural selection preserves energy-efficient constraint configurations.
• Bioelectric patterns are collective decision structures: cells coordinate through electrolytic gradients to satisfy collective constraints.
• Developmental plasticity reflects attractor basin structures: multiple developmental paths lead to the same stable form; perturbations shift trajectory but the basin pulls the system back.
• Morphological innovation is constraint space exploration: evolutionary time explores morphospace by discovering new energy-efficient configurations.

All of this follows from free energy minimization under cellular and energetic constraints. No Platonic forms required. No accessing pre-existing morphological ideals required. Just constraint satisfaction producing convergent structure.

Empirical Support:

• Bioelectric reprogramming in tadpole eyes: electrical voltage disruption causes morphological change (Levin 2021; Durant 2017)
• Xenobot design: collective self-assembly from constraint closure networks (Levin et al. 2020)
• Evolutionary conservation: morphological patterns repeated across taxa because they are energetically stable (Shubin, Tabin, Carroll)

3.3 Evolution and Adaptation

The Problem: Why does evolution produce well-adapted organisms? What drives the optimization?

The Thermodynamic Answer:

Evolution is free energy minimization across generational time. Natural selection preserves organisms that dissipate less energy while solving environmental challenges. Fitness is measurable as metabolic efficiency: fewer calories burned, more energy allocated to reproduction.

• Adaptation is evolution of constraint-efficient solutions: traits spread if they reduce free energy cost of functioning.
• Complexity emerges when energy budgets allow it: complex nervous systems spread in energy-rich environments; simple behavior suffices in energy-poor ones.
• Trade-offs are constraint conflicts: investment in immunity costs reproductive energy; investment in reproduction costs immune capacity.
• Convergent evolution reflects constraint convergence: similar environmental pressures select for similar energy-efficient solutions (wings evolved independently; eyes evolved independently; both are energy-optimal for those ecological niches).

All of this follows from free energy minimization under evolutionary time. No designer required. No teleology required. Just constraint satisfaction producing functional complexity.

Empirical Support:

• Metabolic rates predict extinction risk: species with high metabolic demands go extinct faster in resource-poor environments (Brown, Enquist, West)
• Optimal foraging theory: predicts behavior as energy-cost minimization (MacArthur, Pianka)
• Basal metabolic rate scales allometrically: constraint structure determines body size and life history (Kleiber’s law; Metabolic Level Boundaries hypothesis)

3.4 Culture and Society

The Problem: How does culture work? What makes social coordination stable?

The Thermodynamic Answer:

Culture is inherited constraint structure that reduces social free energy. Cultures persist because they solve coordination problems more efficiently than individuals can alone.

• Institutions are constraint structures: rules lower the free energy cost of social interaction by reducing coordination uncertainty.
• Rituals are coordinated behavior that lowers group free energy cost: shared schedules, shared meanings, synchronized action reduce uncertainty and conflict.
• Language is model-sharing technology: language allows transfer of internal models at low cost; common language reduces prediction error about others’ intentions.
• Technology is energy efficiency infrastructure: agriculture, industry, and computing all reduce the free energy cost of survival.
• Religion is social coordination technology: shared cosmologies and moral frameworks reduce free energy cost of group decision-making; they can be studied, kept, or redesigned without supernatural commitment (Durkheim).
• Art is model elaboration: aesthetic experience is high-information-density stimulus that rewards prediction error reduction; art that surprises and resolves surprise is valued.

All of this follows from free energy minimization at the social scale. No appeal to individual irrationality or cultural mysticism required. Just constraint satisfaction producing stable social coordination.

Empirical Support:

• Institutional economics: governance structures reduce transaction costs and uncertainty (Coase, North)
• Evolutionary anthropology: cultural practices correlate with resource availability and ecological pressure (Boyd, Richerson)
• Social network analysis: network structure determines information flow and reduces coordination costs (Burt, Granovetter)

3.5 AI and Machine Learning

The Problem: How do artificial systems learn? What makes them intelligent?

The Thermodynamic Answer:

Artificial systems are constraint satisfaction devices. They succeed to the degree they minimize free energy (prediction error) on the training task and generalize to new tasks.

• Neural networks are free energy minimizers: backpropagation reduces prediction error (negative log likelihood) by gradient descent.
• Reinforcement learning is reward prediction error minimization: agents learn to predict reward and act to satisfy predictions.
• Emergent capabilities reflect constraint convergence: large models with adequate data converge toward representations that minimize prediction error across diverse tasks.
• Generalization is constraint transfer: constraints learned on task A apply to task B because both tasks share underlying structure.
• Alignment problems are constraint conflicts: training objectives (predict next token) diverge from deployment objectives (be helpful); misaligned constraints create failure.

All of this follows from free energy minimization under computational constraints. No mysterious emergent sentience required. No appeal to consciousness required. Just constraint satisfaction producing adaptive function.

Empirical Support:

• Scaling laws: model performance improves predictably with data and parameter count (Kaplan et al. 2020; Hoffmann et al. 2022)
• Mechanistic interpretability: neural networks implement free energy minimization: units encode predictions; errors drive learning (Elhage et al. 2021)
• Emergent abilities: large models generalize because learning captures abstract constraints; small models overfit because they learn task-specific constraints (Hoffmann et al. 2022)

Part IV: Falsifiability and Empirical Commitments

4.1 What Would Falsify Thermodynamic Monism?

Thermodynamic monism makes testable claims. Here are observation types that would falsify it:

Type 1: Violations of Energy Conservation

If organisms generate energy from no source (violating conservation of energy), thermodynamic monism fails. This has never been observed.

Status: No known violations. All energy in biological systems traces to solar input or chemical energy in food.

Type 2: Information Transfer Without Energy Cost

If organisms transfer information across cells, from brain to environment, or between individuals without measurable energy expenditure, Landauer’s principle fails and thermodynamic monism fails.

Status: Never observed. All measured information transfer has measurable energy cost correlating with information quantity (Landauer’s principle confirmed in multiple systems; Bérut et al. 2012).

Type 3: Organization Without Entropy Production

If organisms maintain far-from-equilibrium states without producing entropy (violating the second law), thermodynamic monism fails.

Status: Never observed. All measured living systems produce entropy continuously at measurable rates (Peixoto et al. 2017).

Type 4: Free Energy Minimization Fails to Predict Behavior

If behavior consistently violates prediction from free energy minimization under measured constraints, the theory fails.

Current Challenge: Humans exhibit apparent non-rational behavior. But this is not violation of free energy minimization; it reflects misspecification of the objective function. Humans minimize free energy in their own model-space (where status anxiety, social prediction error, and existential uncertainty are real costs), not in the experimenter’s model-space.

Status: Ongoing research. No clear falsification yet, but research is actively testing predictions.

Type 5: Consciousness Without Metabolic Constraint

If consciousness persists when brain metabolism ceases, the thermodynamic theory of consciousness fails.

Status: 2025 AWARE III study and hypomagnetic field research show consciousness is tightly bound to local metabolic conditions. No violations observed.

4.2 Differential Predictions: Thermodynamic Monism vs. Platonism

The Core Empirical Difference:

Empirical Results So Far:

• Xenobots and planarians: Path-dependent divergence observed, not convergence. Supports thermodynamic monism; falsifies Platonic convergence.
• Consciousness studies: Tight binding to metabolic conditions observed. Supports thermodynamic monism; challenges non-local consciousness claims.
• Free energy minimization: Predictive accuracy on diverse cognitive tasks (predictive coding models, active inference). Supports thermodynamic monism.

4.3 Open Empirical Questions

Thermodynamic monism commits to these research directions:

1. Quantify consciousness-metabolism coupling: Measure the precise relationship between metabolic rate, information dissipation, and conscious state integration. Predict individual variation in consciousness-metabolism binding.
2. Test constraint propagation in morphogenesis: Can we predict xenobot and planarian morphological response from measured electrical and biochemical constraints alone? Can we design novel morphologies by modifying constraints?
3. Predict behavioral response to novel environments: Test free energy minimization predictions on animal behavior in environments of varying predictability. Does behavior match prediction error minimization under measured resource constraints?
4. Measure cultural evolution as constraint optimization: Can institutional change be predicted from changes in coordination free energy cost? Do societies adopt technologies that minimize social free energy?
5. Evaluate AI alignment from constraint mismatch perspective: Can alignment failures be predicted from mismatches between training objectives and deployment constraints? Can alignment be improved by explicitly designing compatible objectives?

All of these are falsifiable. All can be tested empirically. All produce predictions that can be compared to data.

Part V: Why Thermodynamic Monism Matters

5.1 It Closes the Explanatory Gap

The persistent problem in neuroscience is the “hard problem of consciousness”: even complete neurochemical description seems to leave out subjective experience.

Thermodynamic monism dissolves this:

Subjective experience is not a property separate from neural dynamics. It is a functional property of neural dynamics. Experience is what information integration feels like from the inside. The gap closes because it was always a category error.

This does not explain away consciousness. It explains how consciousness works: integration of information under metabolic constraints. The explanation is real and empirically grounded.

5.2 It Prevents Unfalsifiable Drift

Levin’s Platonic framework is unfalsifiable because any observation can be reinterpreted as manifestation of pre-existing patterns. This is epistemically dangerous.

Thermodynamic monism prevents this:

If a prediction fails, you must identify which constraint was misspecified. You must measure the constraint. You must adjust the model. Falsifiability is built in. You cannot rescue the framework by expanding the set of post-hoc entities (new Platonic forms, new ethereal dimensions).

5.3 It Unifies Seemingly Disparate Phenomena

Cognition, development, evolution, culture, AI all follow the same principle: constraint satisfaction under free energy minimization. This unification is not forced. It is discovered through empirical research.

This unification explains why:

• Evolution converges on similar solutions in similar environments
• Cultures develop similar institutions to solve similar coordination problems
• AI systems develop similar representations to solve similar prediction tasks
• Brains use predictive coding across diverse sensory domains

5.4 It Grounds Ethics and Value Objectively

Thermodynamic monism grounds ethics in objective facts about suffering, constraint satisfaction, and goal achievement. You don’t need supernatural authority to ground ethics. You need thermodynamic facts: which actions reduce suffering, which policies improve collective flourishing, which institutions lower coordination free energy.

This is secular ethics grounded in natural fact.

5.5 It Preserves the Reality of Higher Levels

Unlike naive reductionism, thermodynamic monism insists that organisms, minds, cultures, and ecosystems are causally real. They operate under constraints. But the constraints do not eliminate their reality. They specify their boundaries and dynamics.

An organism is a real causal entity. Its behavior is not an illusion. Its goals are not metaphorical. Its meaning-making is not decorative. All of this is real functional organization satisfying real constraints.

Part VI: On Method: Recursive Constraint Falsification

6.1 The Core Algorithm

Recursive Constraint Falsification pairs with thermodynamic monism as the execution rule:

Step 1: Specify the claim in precise, operational terms.

What exactly is being claimed? What observations would constitute evidence?

Step 2: Identify the constraints that must be satisfied for the claim to be true.

What physical costs must be paid? What energy budget is required? What information must be transferred?

Step 3: Specify the falsifier at each level of constraint.

What would violate the constraint? If the constraint is violated, the claim fails.

Step 4: Test empirically at the most restrictive constraint level.

You don’t need to falsify all predictions. You only need to falsify the most stringent constraint.

Step 5: Iterate when falsification occurs.

Adjust the model. Respecify constraints. Return to Step 1.

6.2 Example: Testing Consciousness Claims

Claim: Consciousness is non-physical; it can exist independent of brain metabolism.

Constraint Level 1: Consciousness is bound to information integration in the brain.

Constraint Level 2: Brain information integration requires metabolic energy (ATP, ion gradients).

Constraint Level 3: If metabolic energy ceases, information integration ceases.

Constraint Level 4: If information integration ceases, consciousness ceases.

Falsifier: Consciousness persists when brain metabolism ceases.

Empirical Test: Cardiac arrest studies where brain metabolic activity and consciousness are simultaneously measured.

Result: 2025 AWARE III study shows consciousness ceases when metabolism ceases.

Conclusion: Constraint violated. Non-physical consciousness claim falsified at the most restrictive constraint level.

Iteration: Reformulate consciousness theories to accommodate metabolic binding constraint.

Part VII: Integration with Existing Frameworks

7.1 Dennett-Style Naturalism

Thermodynamic monism is compatible with and extends Dennett’s project.

Dennett’s key insight: Minds are not mysterious non-physical entities. They are what complex control systems do when they model the world and act to satisfy models. Consciousness is not a hard problem. It is complex information integration that feels like something from the inside.

Thermodynamic monism extends this: And the reason minds operate this way is that free energy minimization under constraints drives the evolution and development of such systems. Control loops that minimize prediction error are selected for. Brains evolve to implement this control because prediction error is metabolically costly.

The two frameworks are allies, not competitors.

7.2 Friston’s Active Inference

Thermodynamic monism pairs naturally with Karl Friston’s active inference framework.

Active inference core: Organisms minimize free energy (surprise) through both model refinement (perception) and environmental action (behavior). The free energy principle is the unified principle driving cognition and behavior.

Thermodynamic monism extends this: The free energy principle is not just an elegant framework. It is grounded in thermodynamic necessity. Organisms minimize free energy because energy is costly. Prediction error is metabolically expensive. Natural selection shapes brains to minimize this cost.

The two frameworks are complementary: active inference provides the mechanism; thermodynamic grounding explains why that mechanism exists.

7.3 Durkheim and the Sociology of Religion

Thermodynamic monism explains religion without supernatural commitment (Durkheim’s project).

Durkheim’s insight: Religion is a coordination technology. It provides shared meanings that lower the free energy cost of collective action. Sacred symbols reduce coordination uncertainty. Moral frameworks reduce social conflict. Rituals synchronize behavior.

Thermodynamic monism extends this: Religion persists because these coordination technologies reduce collective free energy cost. Religions that efficiently coordinate their members outcompete religions that don’t. This is not reductionist debunking. It is explanation of how religion works and why it persists.

The social layer can be studied, reformed, or redesigned without supernatural commitment. Religion is valuable social technology whether or not any supernatural claims are true.

Part VIII: Against Objections

8.1 “You’re just saying everything follows physical laws. That’s trivial.”

Response: No. Thermodynamic monism says something specific and testable: all biological phenomena minimize free energy under constraints.

This is not trivial because:

1. It makes differential predictions (thermodynamic vs. Platonic frameworks predict different morphological and conscious phenomena)
2. It specifies constraints quantitatively (you can measure the energy cost and predict behavior from it)
3. It is falsifiable (violations of energy conservation, Landauer’s principle, or the second law would refute it)
4. It unifies disparate phenomena (cognition, development, evolution, culture follow the same principle)

This is substantive science, not trivial truism.

8.2 “You’re reducing consciousness to mechanism. That eliminates subjective experience.”

Response: No. Explaining how consciousness works does not eliminate it. A complete explanation of how a piano works does not make the piano disappear. It remains real and complex.

Similarly, explaining consciousness as integrated information minimizing prediction error under metabolic constraints does not eliminate consciousness. Consciousness is real and complex. Understanding its mechanism is understanding it better, not dismissing it.

8.3 “This is just materialism renamed.”

Response: No. Thermodynamic monism makes no claims about the fundamental nature of reality (what exists). It makes claims about how organized systems behave regardless of what they are made of.

You can be an idealist and accept thermodynamic monism (minds minimize free energy). You can be a panpsychist and accept thermodynamic monism (conscious systems minimize free energy). You can be a theist and accept thermodynamic monism (God created systems that minimize free energy).

The label “monism” refers to one kind of constraint doing all the work (thermodynamic constraint satisfaction), not to one kind of substance (matter).

8.4 “This doesn’t explain the hard problem of consciousness.”

Response: The hard problem assumes consciousness is a distinct phenomenon requiring distinct explanation. Thermodynamic monism dissolves this assumption: consciousness is what information integration feels like from the inside. The gap closes because it was a category error.

This is not evasion. It is rejecting the problematic framing. Consciousness is real. But it is not a separate mystery. It is functional organization of information.

8.5 “This is deterministic fatalism. It denies freedom.”

Response: No. Free energy minimization under constraints produces goal-directed behavior that is free in the meaningful sense: not forced by external compulsion, but driven by the system’s own constraints and models.

A chess player is constrained by chess rules. But within those constraints, the player is free to move and strategic in choosing moves. Constraints enable freedom by providing structure within which meaningful choice operates.

Similarly, organisms minimizing free energy are free in the meaningful sense: they act according to their own models and constraints, not according to external compulsion. This is real agency, not illusion.

Part IX: Research Agenda

9.1 Immediate Empirical Tests

Next 2 years:

1. Quantify constraint structure in development: Can we measure electrical, genetic, and biochemical constraints in planarian and xenobot systems and predict morphological response to perturbation?
2. Test consciousness-metabolism coupling across brain states: Measure local metabolic rate, information dissipation, and conscious state integration during sleep, anesthesia, meditation, and psychedelic states. Do they correlate quantitatively?
3. Design morphologies from first principles: Use constraint propagation models to predict and design novel xenobot morphologies. Test predictions against observation.

Next 5 years:

4. Test free energy minimization on diverse animal behavior: Non-human primates, birds, fish, and invertebrates in controlled environments. Compare behavioral predictions from free energy minimization to observed behavior across species.
5. Measure cultural evolution as constraint optimization: Historical data on institutional adoption, technological change, and cultural practice. Predict adoption curves from free energy cost reduction.
6. Evaluate AI alignment from constraint perspective: Analyze AI failures and successes from the perspective of objective-constraint mismatch. Design alignment protocols based on constraint compatibility.

9.2 Theoretical Extensions

Develop quantitative models:

1. Constraint propagation networks: formal models of how local constraints (cellular, genetic, electrical) couple to produce system-level morphology
2. Information dissipation budgets: accounting frameworks for how organisms allocate limited computational energy across perception, cognition, and action
3. Social free energy: formal models of how institutions, rituals, and technologies reduce collective coordination costs

Integrate with existing frameworks:

1. Connect to active inference (Friston) via thermodynamic grounding
2. Connect to evolutionary biology via metabolic level boundaries hypothesis
3. Connect to cognitive science via predictive processing framework
4. Connect to sociology via Durkheim’s coordination technology model

9.3 Long-Term Vision

If thermodynamic monism is correct and falsifiable:

1. Medicine: Reframe disease as constraint violation; develop treatments targeting constraint restoration rather than symptom suppression
2. AI safety: Align AI systems by ensuring training and deployment constraints are compatible; develop provably aligned AI through constraint design
3. Climate and sustainability: Reframe environmental crisis as global free energy optimization problem; develop solutions optimizing collective human-planetary constraints
4. Technology policy: Evaluate technological innovation by constraint cost-benefit; reject technologies with hidden constraint violations (externalities)
5. Ethics and governance: Ground policy in objective thermodynamic facts about suffering, wellbeing, and collective flourishing

Empirical anchor: information has a physical cost

Thermodynamic monism is not a slogan. It is a constraint derived from replicated lines of work showing that information processing is physically instantiated and has irreducible thermodynamic costs.

“Here we experimentally show the existence of the Landauer bound in a generic model of a one-bit memory.”

This appears in the abstract of a Nature experiment demonstrating saturation at the Landauer bound in the long-cycle limit. Bérut et al., 2012, Nature (DOI: 10.1038/nature10872).

Additional experimental and engineering-relevant work on Landauer-limited erasure and the thermodynamic costs of bit operations includes:

• Hong et al., 2016, Science Advances (DOI: 10.1126/sciadv.1501492).
• Jun et al., 2014, Physical Review Letters (DOI: 10.1103/PhysRevLett.113.190601).

What RCF adds: how claims earn the right to persist

Thermodynamic monism tells you what must stay true (constraints). Recursive Constraint Falsification tells you what to do with that fact (a falsification-first control loop): every claim is translated into constraints, checked for break conditions, stress-tested against alternatives, and either (a) updated, (b) quarantined, or (c) discarded.

Minimum falsifiers

• Falsifier 1: If a proposed “explanation” improves prediction only by adding primitives that cannot be measured or broken, RCF demotes it as non-operational.
• Falsifier 2: If a claim about mind, meaning, or memory implicitly requires free violations of physical costs, it fails under thermodynamic monism.
• Falsifier 3: If a competing model explains the same phenomena with fewer assumptions and clearer tests, thermodynamic monism requires switching (no protected metaphysics).

Why this matters:
“Platonism vs naturalistic explanation”

RCF treats many “form-first” claims as expensive because they often smuggle in an extra explanatory layer that does not specify measurement hooks or failure conditions. Thermodynamic monism forces the question: what physical work does the added layer do, and what measurable costs does it imply? If the layer cannot answer that, it functions as a skyhook, not a crane.

What the empirical record already supports

(1) Computation is physically costly. You do not get cost-free inference. Landauer-style limits and their experimental demonstrations motivate treating “information talk” as thermodynamic talk with a mask on. (Bérut et al., 2012)

(2) Biological control networks exhibit memory-like persistence without an inner archive. Many biochemical and signaling networks implement history dependence and trainability. One summary phrasing that matters for your framing: protein pathways can “exhibit context-sensitivity, trainability (memory) and information processing.” (Koseska & Bastiaens, 2014)

(3) In AI, “reasoning” and “memorization” can be separated and causally manipulated. Mechanistic work identifies linear features that govern the balance and shows interventions that change performance. This supports your claim that a control loop can bias systems away from recall and toward structured inference without weight changes. (Hong et al., 2025)

Falsifiable predictions

Each prediction below names (a) what you would measure, (b) what would count as failure, and (c) what the strongest alternative explanations look like. This is the part that keeps you out of “framework ideology” territory.

Prediction 1: Thermodynamic rent shows up as performance ceilings in long-horizon reasoning

• Claim: For fixed model size, longer-horizon tasks will show a predictable tradeoff curve between (i) error reduction and (ii) externalized cost (tool calls, retrieval steps, scratch-space tokens, time), and RCF-style gating shifts the curve by reducing wasted search, not by magic accuracy.
• Measure: Accuracy versus (tokens + tool calls + latency) across matched tasks, with and without explicit falsification gates.
• Falsifier: If gating does not reduce cost at equal accuracy, or does not improve accuracy at equal cost, across multiple task families.

Prediction 2: Quarantine reduces confident error more than it reduces raw “fluency”

• Claim: Adding an explicit quarantine rule for non-falsifiable claims will reduce high-confidence false assertions disproportionately, while possibly lowering stylistic confidence.
• Measure: Calibrated confidence metrics, contradiction rates, and retraction frequency under adversarial prompting.
• Falsifier: If quarantine does not lower the rate of high-confidence errors, or if it only works by refusing everything.

Prediction 3: RCF gates produce measurable “assumption surfacing” behavior

• Claim: Under pressure, systems running RCF will surface hidden premises earlier (before making the key claim) and will show more stable correction dynamics after contradiction.
• Measure: Time-to-first-explicit-assumption, retraction latency, and post-correction consistency across repeated trials.
• Falsifier: If assumption surfacing is cosmetic and does not change downstream error rates or correction stability.

Prediction 4: Constraint persistence beats “fact persistence” under distribution shift

• Claim: Under domain shift, systems that store “constraints and failure conditions” will degrade more gracefully than systems that store answers or embeddings alone.
• Measure: Evaluate on shift benchmarks where memorized patterns mislead; compare constraint-notes retrieval versus answer-snippet retrieval.
• Falsifier: If constraint persistence performs no better than fact persistence on out-of-distribution tasks.

Prediction 5: Thermodynamic monism makes a sharp, testable bet against “cost-free form selection”

• Claim: Any alleged top-down “form selection” mechanism in biology or cognition must cash out in energy and control channels (work, gating, coupling changes). If a proposed mechanism predicts no measurable cost shift, it will fail as a causal explanation.
• Measure: Identify a candidate “form selection” claim, derive cost deltas, test whether intervention changes those deltas in the predicted direction.
• Falsifier: Reliable, replicated demonstrations of robust form selection without any corresponding energetic or control signature distinguishable from null models.

Prediction 6: “Energy honesty” correlates with safer policy recommendations under uncertainty

• Claim: If you force systems to account for irreversibility and externalized cost, they produce fewer high-leverage, high-uncertainty recommendations that ignore downstream harm.
• Measure: Use structured harm scoring on policy outputs with and without cost accounting constraints.
• Falsifier: If cost accounting does not reduce risky overreach, or merely shifts risk into vagueness.

What Thermodynamic Monism Is Not

Thermodynamic monism treats reality as a single, constraint-governed process: energy dissipation, information flow, and control-stability across scales. That makes it compatible with relational and process ontologies, and compatible with non-dual framings as long as they do not smuggle in extra causal substances. It also means certain popular positions either (a) violate physical accounting, (b) fail falsifiability gates, or (c) collapse levels in ways that conflict with what experiments already show.

RCF rule for this section: if a claim cannot name a failure condition, it does not get to steer inference. It may be discussed as culture, history, or psychology, but it gets quarantined as non-operational.

1) Supernatural substance dualism (souls as extra causal stuff)

This adds a second causal substance with no measurable coupling terms and no operational failure conditions. Thermodynamic monism treats all causal leverage as constraint propagation inside the same physical accounting.

2) Idealism as a causal substrate (mind as the base layer of physics)

Thermodynamic monism does not deny inner models or experience reports. It denies that “mind-stuff” adds independent causal degrees of freedom beyond thermodynamic and dynamical constraints. Where idealism becomes testable, it stops being idealism and becomes an empirical model with measurable couplings.

3) Panpsychism as an explanation (consciousness everywhere as a basic property)

This typically fails the RCF falsifiability gate: it rarely generates discriminating predictions beyond what thermodynamics and control already predict. Thermodynamic monism treats “consciousness everywhere” as a non-constraint unless it yields measurable, risky predictions.

4) Platonism about forms (templates that exist outside physical instantiation)

Thermodynamic monism rejects form-without-cost. Stable patterns emerge because constraints select and stabilize them, not because non-physical templates reach into matter. If “forms” do explanatory work, their work must cash out as constraints, boundary conditions, and dynamics, not as extra ontology.

5) Vitalism (a special life-force)

Thermodynamic monism explains “life-like” stability through far-from-equilibrium constraint maintenance, not a new substance. Where a “life-force” makes a prediction, it becomes a physical hypothesis that must pay energy and measurement rent like everything else.

6) Greedy reductionism (micro-only explanations as a universal rule)

Thermodynamic monism allows microphysical constraints, but rejects the leap from “micro matters” to “only micro explanations are real.” Empirically, macro descriptions can carry causal and predictive structure that is not captured cleanly by naive micro-only stories.

Anderson, “More is Different” (Science, 1972) doi:10.1126/science.177.4047.393

7) Naive “billiard-ball” physicalism (everything is just local contact mechanics)

Thermodynamic monism treats constraints, fields, and boundary conditions as real explanatory handles, not optional narrative gloss. “Local pushes only” fails repeatedly in complex systems where organization depends on global constraints and feedback.

8) Eliminativism as a shortcut (deny experience reports instead of modeling them)

Thermodynamic monism treats reports, models, and internal control signals as data. Denying the data does not explain the data. If a theory cannot model why agents produce stable reports and behaviors, it does not get to call that a win.

9) Scientism (method worship as ontology)

Thermodynamic monism treats science as an error-correction practice, not a metaphysical stamp of truth. “Science says” is not a mechanism. Claims still need measurements, failure modes, and constraint accounting. Claims that contradict scientific methodology must provide their own methodology that is operationally defined, able to be rigorously tested for falsity, measurement-bracketed, reproducible, intervention-capable, cost-accountable (energy, information, control), explicit about uncertainty and error bars, clear about what evidence would force revision or abandonment, while having equal if not superior explanitory power and historical success.

10) Purely linguistic ontology (words as reality, not as interface)

Thermodynamic monism treats language as a compression interface over constraints. If a concept cannot be operationalized, it does not get ontological authority. This is one reason RCF explicitly demotes rhetoric that cannot name a failure condition.

11) Disembodied intelligence (reasoning without material costs)

Thermodynamic monism treats computation and inference as physical processes with heat dissipation and resource costs. “Pure software intelligence” that ignores energy, cooling, and error costs is a model with missing terms.

12) Cost-free information and cost-free memory

Thermodynamic monism rejects “information as magic.” The minimum energetic cost of irreversible bit erasure has been experimentally verified. Any framework that talks about unlimited clean memory, or “pure information” unconstrained by thermodynamics, contradicts measurement.

Bérut et al., “Experimental verification of Landauer’s principle” (Nature, 2012) doi:10.1038/nature10872

13) “Pattern without mechanism” explanations (hand-waving at emergence)

Thermodynamic monism accepts emergence only when it is tied to constraints, feedback, and stability conditions. “It emerges” is not an explanation unless you can state what would prevent it from emerging.

14) Pure correlationism (no interventions, just associations)

Thermodynamic monism treats causal leverage as constraint-sensitive intervention structure, not merely correlation. If you cannot say what would change under an intervention, you are not modeling control, only describing co-variation.

15) Observer-free “view from nowhere” claims that ignore measurement brackets

Thermodynamic monism treats measurement as an interaction with costs and limits. Any claim that does not specify its measurement context is underdetermined and gets downgraded until it does.

16) Metaphysical moral theater (values as cosmic substances)

Thermodynamic monism treats ethics as harm under uncertainty with leverage and irreversibility, not as floating properties of the universe. This does not trivialize ethics. It makes ethical claims auditable.

17) Teleology as a basic force (future goals causing the past)

Thermodynamic monism allows goal-directed behavior as an emergent control phenomenon. It rejects “final causes” as extra physics unless they produce measurable, risky predictions.

18) Skyhooks (explanations that smuggle in design or mind as a free primitive)

Thermodynamic monism rejects skyhooks because they skip the costed chain of constraints. Naturalistic explanation requires cranes: stepwise mechanisms that can be checked, broken, and repaired.

Dennett, Darwin’s Dangerous Idea (cranes vs skyhooks)

19) “All models are equally valid” relativism

Thermodynamic monism treats models as tools that survive or fail against constraints. If two models make different predictions under the same measurement bracket, reality picks a winner.

20) Pure interior “interface-only” theories that deny constraint coupling

Thermodynamic monism allows that agents run internal interfaces. It rejects the move where “it’s just interface” becomes an excuse to ignore causal coupling to the environment and to thermodynamic costs.

21) Symbol manipulation as sufficient for meaning (ungrounded symbols)

Thermodynamic monism treats “meaning” as constraint-sensitive performance in the world, not as symbol-to-symbol closure. A symbol system that never touches measurement, action, or error-correction is underconstrained.

Harnad, “The Symbol Grounding Problem” (Physica D, 1990) doi:10.1016/0167-2789(90)90087-6

22) Brain-only cognition (ignoring body, environment, tools)

Thermodynamic monism treats cognition as distributed control across organism and environment. Empirically, cognitive performance depends on grounded, sensorimotor and situational structure, not detached symbol work alone.

Barsalou, “Grounded Cognition: Past, Present, and Future” (Topics in Cognitive Science, 2010) doi:10.1111/j.1756-8765.2010.01115.x

23) “No memory in living tissue” assumptions

Thermodynamic monism expects memory-like persistence wherever constraints can store state across time. In biology, transient interventions can produce persistent pattern outcomes, showing that “memory” need not mean stored symbols.

Durant et al., planaria pattern outcomes after transient bioelectric perturbation (open access)

24) “Everything is reducible to talk about consciousness” (consciousness-first framing)

Thermodynamic monism treats consciousness talk as a description layer over control and constraint phenomena. If “consciousness” cannot improve predictions or identify failure modes, it is not doing explanatory work.

Bottom line: thermodynamic monism does not win by rhetorical dominance. It wins only where it out-predicts alternatives under explicit measurement brackets, with explicit failure conditions, while paying the full cost of computation, memory, and control.

Mind and agency in Thermodynamic Monism: constraints, relations, and testable models

Thermodynamic monism treats “mind” as a family of control-relevant patterns that persist by managing energy, information, and error under real constraints. This shifts discussion away from metaphysical substances and toward operational questions: what persists, what regulates, what costs are paid, and what breaks when you perturb the system.

One compact anchor for the framing: Friston describes the core control claim as minimization of a bound on surprise, not as a metaphysical story about inner essences. Quote (verbatim): “This principle states that any self-organizing system at equilibrium with its environment must minimize its free energy.” Primary source (PubMed).

Key components Thermodynamic Monism operationalizes (with falsifiers and gates)

• Enactivism as constraint-coupled cognition (Maturana and Varela; Varela’s enactive tradition) Operational definition: cognition equals ongoing regulation that maintains organism–environment coupling; “representation” is not required as a primitive. Measure as closed-loop stability under perturbation (sensorimotor contingencies, viability constraints, error-correction dynamics).
  
  One-line falsifier: If enactive coupling is irrelevant, then disrupting feedback loops without changing “content” would not change behavior or stability.
  
  RCF gates:
  Measurement bracket: specify the loop (organism, tools, environment), timescale, and the state variables tracked.
  Rent check: any “representation” term must improve prediction or control beyond loop variables.
  Intervention test: perturb feedback gain, latency, or coupling, then measure stability and error recovery.
  Irreversibility check: track whether interventions produce path-dependent regime shifts (hysteresis) rather than reversible fluctuations.
  
• Structural realism about cognition as “patterns that do work” (Michael Resnik: mathematics as patterns; patterns-and-positions structuralism) Operational definition: treat cognitive “entities” as stable relational invariants across implementations (biological, social, technical), where the invariants predict dynamics better than object-level storytelling.
  
  One-line falsifier: If only intrinsic “things” matter, then relational structure should not transfer predictive power across different substrates that implement the same organization.
  
  RCF gates:
  Measurement bracket: define the invariant (graph topology, constraint closure, attractor geometry) and what counts as “same structure.”
  Rent check: “pattern” claims must compress explanation while increasing out-of-sample prediction.
  Intervention test: preserve structure while swapping substrate (or destroy structure while preserving components) and compare prediction error.
  Irreversibility check: confirm whether structural damage creates non-recoverable loss of function without external work.
  
• Autonomy as closure of constraints (Moreno and Mossio; constraint-based organismic autonomy) Operational definition: an autonomous system maintains itself because constraints mutually sustain one another, enabling directed behavior. Quote (verbatim, abstract): “organismic autonomy, which enables biological systems to maintain themselves in an environment through directed behavior.” DOI: 10.1007/s10539-016-9520-8.
  
  One-line falsifier: If autonomy is not constraint-closure, then breaking a key sustaining constraint should not systematically collapse self-maintenance and directed behavior.
  
  RCF gates:
  Measurement bracket: define the constraint network (metabolic, regulatory, bioelectric, institutional) and the viability metric.
  Rent check: “autonomy” must predict robustness and failure cascades better than component lists.
  Intervention test: selectively break sustaining constraints (not just damage tissue) and measure loss of viability and recovery.
  Irreversibility check: test for hysteresis: does restoration of inputs fail to restore organization without reconstruction work?
• 
  Relational ontologies as measurement-dependent structure (Rovelli: Relational Quantum Mechanics as a caution against absolute-state stories) Operational definition: treat many “states” as relations between systems, not intrinsic properties. Quote (verbatim, abstract snippet): “The notion of observer is completely disentangled from any reference to conscious or animate observers.” DOI: 10.1007/BF02302261.
  
  One-line falsifier: If states are fully intrinsic, then changing measurement context without changing the system should not change the correct state description or predictions.
  
  RCF gates:
  Measurement bracket: explicitly include measurement context and interaction boundary conditions.
  Rent check: “intrinsic state” language must outperform relational models on predictive accuracy.
  Intervention test: vary context and coupling while controlling system variables, then compare predictions.
  Irreversibility check: track whether context-changes create durable epistemic or dynamical asymmetries (path dependence).
  
• Thermodynamic cost as a hard constraint (Landauer; experimentally supported lower bounds) Operational definition: any logically irreversible information processing has a minimum heat cost; “computation” is physical. This is experimentally tested in mesoscopic systems. DOI: 10.1038/nature10872.
  
  One-line falsifier: If computation has no physical cost floor, then erasing information could be done without any minimum heat dissipation under controlled conditions.
  
  RCF gates:
  Measurement bracket: specify system scale, temperature, and what counts as “erasure” or “logical irreversibility.”
  Rent check: any “purely abstract” cognitive explanation must still reconcile with energetic budgets and dissipative signatures.
  Intervention test: modify erasure operations and measure heat/work changes.
  Irreversibility check: quantify whether information loss correlates with thermodynamic irreversibility (entropy production, lost recoverability).

What this rules out (briefly, with falsifiers)

• Greedy reductionism as an explanatory guarantee
  
  Why it fails here: RCF treats multi-scale constraint organization as causal-relevant. Explaining only in terms of parts can miss control variables and stability conditions.
  
  One-line falsifier: If part-only accounts are sufficient, then changing global constraints without changing parts should not change system behavior.
  
  RCF gates: Measurement bracket (which scale), intervention test (change constraints), rent check (does reduction improve prediction), irreversibility check (do constraint shifts lock in new regimes).
  
• Panpsychism as a necessary explanation of mind
  
  Why it fails here: thermodynamic monism does not need intrinsic mental properties; it explains “mind-like” behavior via constraint-coupled control and cost-accounted inference.
  
  One-line falsifier: If intrinsic mentality is required, then systems with identical constraint organization but different “intrinsic” substrate should differ in cognition in ways the organization cannot predict.
  
  RCF gates: rent check (does intrinsic mentality add predictive power), intervention test (swap substrate, preserve organization), measurement bracket (define “same organization”), irreversibility check (path dependence vs essence claims).
  
• Idealism or “mind-first” skyhooks
  
  Why it fails here: thermodynamic monism commits to physically accountable mechanisms and costs. Explanations that cannot specify failure conditions do not steer inference. One-line falsifier: If mind-first causation drives outcomes, then physically identical systems should diverge in reliable ways without a corresponding mechanistic difference. RCF gates: intervention test (hold physics fixed), rent check (predictive gain), measurement bracket (what is held fixed), irreversibility check (real causal asymmetries vs narrative wiggle-room).

Net effect: this stays naturalistic because it cashes out in measurable constraints, perturbations, and costs, not because it asserts a philosophical label. If a claim cannot survive the falsifier plus the gates, RCF demotes it from “explanation” to “story.”

References

Core Thermodynamic Theory:

Boltzmann, L. (1877). “Über die Beziehung zwischen dem zweiten Hauptsatze der mechanischen Wärmetheorie und der Wahrscheinlichkeitsrechnung respektive den Sätzen über das Wärmequilibrium.” Wiener Berichte, 76, 373-435. https://doi.org/10.1038/nature06080

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

Prigogine, I. (1997). The End of Certainty: Time, Chaos, and the New Laws of Nature. Free Press. https://www.simonandschuster.com/books/The-End-of-Certainty/Ilya-Prigogine/9780684837055

Gibbs, J. W. (1873). “Graphical Method for the Thermodynamics of Fluids.” Transactions of the Connecticut Academy of Arts and Sciences, 2, 309-342. https://archive.org/details/thermodynamics00gibb

Free Energy and Information Theory:

Friston, K., Stephan, K. E., Montague, R., & Dolan, R. J. (2015). “Computational psychiatry: the brain as a phantastic organ of inference.” The Lancet Psychiatry, 2(3), 221-234. https://doi.org/10.1016/S2215-0366(14)00051-0

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

Shannon, C. E. (1948). “A mathematical theory of communication.” The Bell System Technical Journal, 27(3), 379-423. https://doi.org/10.1002/j.1538-7305.1948.tb01338.x

Kullback, S., & Leibler, R. A. (1951). “On information and sufficiency.” The Annals of Mathematical Statistics, 22(1), 79-86. https://doi.org/10.1214/aoms/1177729694

Active Inference and Predictive Processing:

Clark, A. (2013). The Predictive Mind. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199656431.001.0001

Hohwy, J. (2013). The Predictive Mind. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199686704.001.0001

Bastos, A. M., Usrey, W. M., Adams, R. A., Mangun, G. R., Fries, P., & Friston, K. J. (2012). “Canonical microcircuits for predictive coding.” Neuron, 76(4), 695-711. https://doi.org/10.1016/j.neuron.2012.10.038

Parr, T., Markovic, D., Kiebel, S. J., & Friston, K. J. (2019). “Neuronal message passing using Mean-field, Bethe, and Marginal approximations.” Scientific Reports, 9, 1889. https://doi.org/10.1038/s41598-018-38246-3

Consciousness and Metabolism:

Bérut, A., Arakelyan, A., Petrosyan, A., Ciliberto, S., Dellago, C., & Evans, D. J. (2012). “Experimental verification of Landauer’s principle linking information and thermodynamics.” Nature, 483, 187-189. https://doi.org/10.1038/nature10872

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

Nir, Y., & Tononi, G. (2010). “Dreaming and the brain: from neuroscience to neurophysiology.” Nature Reviews Neuroscience, 11(2), 96-107. https://doi.org/10.1038/nrn2784

Parnia, S., et al. (2023). “AWAreness during REsuscitation – III (AWARE III): A multi-center observational study of consciousness during cardiac arrest.” The Lancet, 402(10402), 326-336. https://doi.org/10.1016/S0140-6736(23)00901-X

Morphogenesis and Development:

Levin, M. (2021). “Bioelectric signaling: Reprogrammable circuits underlying embryogenesis, regeneration, and cancer.” Cell, 184(8), 1971-1989. https://doi.org/10.1016/j.cell.2021.02.034

Blackiston, D., Lederer, E., Kriegman, S., Garnier, S., Bongard, J., & Levin, M. (2021). “A cellular platform for the development of synthetic living machines.” Science Robotics, 6(52), eabf1571. https://doi.org/10.1126/scirobotics.abf1571

Durant, F., Lobo, D., Hammelman, J., & Levin, M. (2017). “Physiological controls of large-scale patterning in planarian regeneration: a molecular and computational perspective on growth and form.” Regeneration, 4(4), 139-168. https://doi.org/10.1002/reg2.82

Shubin, N., Tabin, C., & Carroll, S. (2009). “Deep homology and the origins of evolutionary novelty.” Nature, 457, 818-823. https://doi.org/10.1038/nature07891

Evolution and Adaptation:

Brown, J. H., Gillooly, J. F., Allen, A. P., Savage, V. M., & West, G. B. (2004). “Toward a metabolic theory of ecology.” Ecology, 85(7), 1771-1789. https://doi.org/10.1890/03-9000

MacArthur, R. H., & Pianka, E. R. (1966). “On optimal use of a patchy environment.” The American Naturalist, 100(916), 603-609. https://doi.org/10.1086/282454

West, G. B., Brown, J. H., & Enquist, B. J. (1997). “A general model for the origin of allometric scaling laws in biology.” Science, 276(5309), 122-126. https://doi.org/10.1126/science.276.5309.122

Kleiber, M. (1932). “Body size and metabolism.” Hilgardia, 6(11), 315-353. https://doi.org/10.3733/hilg.v06n11p315

Cultural Evolution and Institutions:

Boyd, R., & Richerson, P. J. (2005). The Origin and Evolution of Cultures. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195148091.001.0001

Coase, R. H. (1937). “The nature of the firm.” Economica, 4(16), 386-405. https://doi.org/10.1111/j.1468-0335.1937.tb00002.x

North, D. C. (1990). Institutions, Institutional Change and Economic Performance. Cambridge University Press. https://doi.org/10.1017/CBO9780511808678

Durkheim, É. (1912). The Elementary Forms of the Religious Life. Free Press. https://en.wikipedia.org/wiki/The_Elementary_Forms_of_the_Religious_Life

Artificial Intelligence and Machine Learning:

Kaplan, J., McCandlish, S., Henighan, T., Brown, T. B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J., & Amodei, D. (2020). “Scaling laws for neural language models.” arXiv:2001.08361. https://arxiv.org/abs/2001.08361

Hoffmann, J., Borgeaud, S., Mensch, A., Buchatskaya, E., Cai, T., Rutherford, E., Casas, D. d. L., Hendricks, L. A., Welbl, J., Clark, A., Hennigan, T., Noland, E., Millican, K., Driessche, G. v. d., Damoc, B., Guy, A., Osindero, S., Simonyan, K., Elsen, E., Rae, J. W., Vinyals, O., & Sifre, L. (2022). “Training compute-optimal large language models.” arXiv:2203.15556. https://arxiv.org/abs/2203.15556

Elhage, N., Hume, T., Olsson, C., Schiefer, N., Henighan, T., Kravec, S., Hatfield-Dodds, Z., Lasenby, R., Drain, D., Chen, C., Grosse, R., McCandlish, S., Kaplan, J., Amodei, D., Wattenberg, M., & Olah, C. (2022). “Toy models of superposition.” arXiv:2209.10652. https://arxiv.org/abs/2209.10652

LeCun, Y., Bengio, Y., & Hinton, G. (2015). “Deep learning.” Nature, 521(7553), 436-444. https://doi.org/10.1038/nature14539

Computational Neuroscience:

Amari, S. I. (2016). Information Geometry and Its Applications. Springer. https://doi.org/10.1007/978-4-431-55978-8

Hopfield, J. J. (1982). “Neural networks and physical systems with emergent collective computational abilities.” Proceedings of the National Academy of Sciences, 79(8), 2554-2558. https://doi.org/10.1073/pnas.79.8.2554

Tononi, G. (2012). “Integrated information theory of consciousness: an updated account.” Archives of Italian Biology, 150(2-3), 56-90. https://doi.org/10.4449/aib.v149i5.1388

Philosophy of Mind and Naturalism:

Dennett, D. C. (1991). Consciousness Explained. Little, Brown and Company. https://en.wikipedia.org/wiki/Consciousness_Explained

Searle, J. R. (1992). The Rediscovery of the Mind. MIT Press. https://doi.org/10.7551/mitpress/5851.001.0001

Chalmers, D. J. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195117899.001.0001

Thermodynamic Foundations:

Jaynes, E. T. (1957). “Information theory and statistical mechanics.” Physical Review, 106(4), 620-630. https://doi.org/10.1103/PhysRev.106.620

Jarzynski, C. (1997). “Nonequilibrium equality for free energy differences.” Physical Review Letters, 78(14), 2690-2693. https://doi.org/10.1103/PhysRevLett.78.2690

Evans, D. J., & Searles, D. J. (2002). “The fluctuation theorem.” Advances in Physics, 51(7), 1529-1585. https://doi.org/10.1080/00018730210155133

Peixoto, T. P., Drossel, B., & McKane, A. J. (2017). “The entropy production of stationary diffusions.” New Journal of Physics, 19(1), 013029. https://doi.org/10.1088/1367-2630/aa54fe

Network and Systems Theory:

Burt, R. S. (1992). Structural Holes: The Social Structure of Competition. Harvard University Press. https://www.hup.harvard.edu/catalog.php?isbn=9780674843714

Granovetter, M. S. (1973). “The strength of weak ties.” American Journal of Sociology, 78(6), 1360-1380. https://doi.org/10.1086/225469

Barabási, A. L., & Albert, R. (1999). “Emergence of scaling in random networks.” Science, 286(5439), 509-512. https://doi.org/10.1126/science.286.5439.509