Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

The KB-Flux Formalism: A Computational Epistemology of Quantum Mechanics and Active Inference

Tim Siu Ming Shum

Published Mar 20, 2026• 0 citations
Publisher

Abstract

The measurement problem in quantum mechanics resists resolution within traditional substance ontologies. This paper proposes the Kantian-Bayesian Flux (KB-Flux) formalism, an operationalist framework that reinterprets physical laws not as descriptions of an observer-independent reality, but as necessary data-compression and belief-updating protocols. By mapping the quantum wave function () to the Bayesian belief state () of an observer operating under the Free Energy Principle, we structurally align quantum superposition with latent epistemic uncertainty and measurement collapse with discrete Bayesian updates. This framework formally navigates quantum no-go theorems, synthesizes Relational Quantum Mechanics (RQM) with Active Inference, delineates its departures from Quantum Bayesianism (QBism), and yields heuristic predictions for the architectural evolution of Artificial General Intelligence and neurophysics. For a broader, accessible exploration of this framework and its implications for the nature of reality and mind, follow the Substack publication: The Shum Briefing.

DOI
10.5281/zenodo.19126312

Related papers

The Kantian-Bayesian Interpretation of Quantum Mechanics: A Cognitive Dissolution of the Measurement Problem

Shum, Tim Siu Ming

[Author’s Note (April 2026)]Supersession Notice: This 2025 paper explores the early KB-Flux framework. It has been structurally overhauled and formally superseded by “The Light & Flux (L&F) Framework (2026)”. While the 2025 version relies on cognitive metaphors—treating the wave function as a “generative model” —the 2026 L&F framework provides a rigorous mathematical proof deriving complex Hilbert space from finite-inference constraints. It replaces “phenomenal grammar” with the Simplex Projection (𐍀), grounded in metabolic and thermodynamic mandates. Researchers seeking the formal Non-Derivability Principle and a stratified three-tier ontology should refer to the 2026 version. Original AbstractReaders are advised to treat the present paper strictly as a historical record of the framework’s initial development. For the current, accurate formulation, please refer to the new publication.The measurement problem—why quantum superpositions yield definite outcomes upon observation—has persisted as the central interpretative challenge of quantum mechanics for nearly a century. This paper proposes that the problem dissolves when viewed through the joint lens of Kantian epistemology and predictive processing (Bayesian brain) theory. Drawing on Kant’s distinction between phenomena (experience) and noumena (reality-in-itself), we argue that the quantum wave function (Ψ) does not represent a physical field in the noumenal world, but rather the generative model of a prediction-generating cognitive agent. In this framework, "superposition" represents a state of uninitialized probabilistic expectation (Latent Space), while "collapse" is reinterpreted as a Bayesian belief update—the discrete transition from a probability distribution to a definite value, necessitated by the information-processing constraints of the observer. By treating the observer not as a passive witness but as an active inference engine, this framework: Naturalizes the collapse postulate: Measurement is the update of an internal model, not a physical discontinuity in nature. Resolves nested-observer paradoxes: Wigner's Friend is resolved by treating reality as a distributed system with eventual consistency, where distinct agents hold distinct local cache states. Grounds the Born Rule: We argue that the squared-amplitude probability rule ($P=|\Psi|^2$) serves as a transcendental constraint (akin to Gleason’s Theorem) required for the internal consistency of any probabilistic agent. This interpretation suggests that the "weirdness" of quantum mechanics reflects the necessary architectural constraints of any finite mind attempting to model an inaccessible reality, effectively treating physics as the "User Interface" of reality.

•Dec 12, 2025•0 citations

Physics as the User Interface of Consciousness: A Kantian-Bayesian Synthesis of Zen Impermanence and Quantum Mechanics

Tim Siu Ming Shum

[Author’s Note]This paper is retained as a historical record. The informal ‘Kantian-Bayesian’ framework presented here has been entirely superseded by the Light & Flux (L&F) framework, which corrects the temporal substrate circularity, introduces a three-tier formal ontology, and provides a rigorous deductive derivation of complex Hilbert space from inferential constraints. See [DOI of L&F V5]. For a broader, accessible exploration of this framework and its implications for the nature of reality and mind, follow the Substack publication: Mind & Machine. Original AbstractModern interpretations of Quantum Mechanics, particularly Quantum Bayesianism (QBism) and the Free Energy Principle, increasingly suggest that physical reality is not an observer-independent ontology but a generative model constructed to minimize predictive error. This paper proposes a formal synthesis between this "Physics-as-Interface" paradigm and the phenomenology of Eastern Zen philosophy. We argue that the Zen concept of "Subtle Impermanence" (Anicca) is mathematically equivalent to a Latent Markovian Substrate, while the "illusion of constancy" maps rigorously to the Belief State (q(x)) maintained by the observer to facilitate inference under partial observability. By framing the Wave Function not as a physical object but as a computational strategy to mitigate Variational Free Energy, we demonstrate that the laws of physics are essentially stability protocols encoded by consciousness. This synthesis moves beyond metaphorical comparisons, offering a rigorous, information-theoretic grounding for the convergence of Eastern phenomenology and Western computational mechanics. Key Theoretical Contributions: The Markovian Noumena (Hardware): Reinterpreting Kant’s Noumena as a Latent Markov Process (x_t)—a generative substrate of discrete, conditionally independent moments that lacks intrinsic object permanence. The Non-Markovian Interface (Software): Defining conscious perception as a Belief State ($q(x)$) that acts as an "Active Buffer." This interface imposes a non-Markovian constraint (memory/history) onto the latent data to construct a navigable user interface. Wave Function as Belief State: Aligning with QBism, we formally equate the quantum wave function (Ψ) with the observer's internal belief state, where "collapse" is redefined as a standard Bayesian belief update minimizing prediction error. Free Energy & Complexity: Utilizing Friston’s Free Energy Principle, we argue that "Causality," "Time," and "Objects" are optimal compression algorithms derived to minimize the Complexity Cost (KL Divergence) of the internal model. The Thermodynamics of Time: Proposing that the "Arrow of Time" is an artifact of Inference Asymmetry, where predicting a divergent future requires higher computational entropy than retrodicting a convergent past.

•Dec 27, 2025•0 citations

Quantum Mechanics as a Theorem of Finite Inference: The Light & Flux Framework

Tim Siu Ming Shum

[Author’s Note]This paper introduces the Light & Flux (L&F) framework, which formally supersedes the earlier "KB-Flux" formalism. The previous ontological mappings have been entirely replaced by the rigorous three-tier architecture and the Non-Derivability Principle presented herein. For a broader, accessible exploration of this framework and its implications for the nature of reality and mind, follow the Substack publication: Mind & Machine. AbstractQuantum mechanics is not an empirical physical law; it is the geometric theorem of coherent inference under fundamental uncertainty. The explanatory deadlock of modern quantum foundations stems from a flat ontology that falsely presumes classicality as the default state of reality. This paper introduces the Light & Flux (L&F) framework, a three-tier formal architecture demonstrating that incomputable complex Hilbert space is the universal inferential baseline. The macroscopic classical world, conversely, is a metabolically constrained, lossy compression rendered by finite biological agents. We formalize this irreversible compression via the Simplex Projection and establish the Non-Derivability Principle: demanding a unitary dynamical account for a specific classical measurement outcome is a mathematical category error. Finally, we map the precise deductive gap between pure inferential logic and generalized probabilistic theories (GPTs), confronting the Gödelian limits of our own cognitive codec.

•Apr 7, 2026•0 citations

Ze and Relational QM

Jaba Tkemaladze

The measurement problem in quantum mechanics challenges our understanding of reality, demanding explanations beyond both Copenhagen's "collapse" and the Many-Worlds' ontological multiplicity. This paper introduces and formalizes the Ze framework as a novel synthesis of Relational Quantum Mechanics (RQM) and the Active Inference paradigm from theoretical neuroscience. Ze posits that quantum states are relational, defined by the posterior beliefs of interacting generative models engaged in variational free energy minimization. Within this framework, quantum superposition is formalized as high compatibility (ℐ ≈ 1) between competing models, characterized by low free-energy conflict (ΔF < θ). Conversely, the transition to a localized state—the physical correlate of "collapse"—is reconceived not as a metaphysical event but as an optimization-driven phase transition. This occurs when model conflict exceeds a critical threshold (ΔF > θ), a process objectively driven by interactions like which-path marking. We demonstrate that matter-wave interferometry with complex molecules provides a direct experimental testbed for these principles, where which-path information and quantum erasure actively manipulate ΔF. Extending the isomorphism, we propose that transitions in human cognition—from focused wakefulness to dreaming and psychedelic states—are governed by analogous shifts in the brain's inferential threshold (θ). Thus, Ze offers a unified, testable architecture bridging quantum foundations, statistical physics, and the neuroscience of consciousness.

Longevity Horizon•Jan 15, 2026•7 citations

Reformulation of quantum mechanics and strong complementarity from Bayesian inference requirements

William Heartspring

This paper provides an epistemic reformulation of quantum mechanics (QM) in terms of inference consistency requirements of objective Bayesianism, which include the principle of maximum entropy under physical constraints. Physical laws themselves are understood in terms of inference and physical consistency requirements. Strong complementarity - that different observers may "live" in separate Hilbert spaces - follows as a consequence, which resolves the firewall paradox. Other clues pointing to this reformulation are analyzed. The reformulation, with the addition of novel transition probability arithmetic, resolves the measurement problem completely, thereby eliminating charge of subjectivity of measurements from quantum mechanics. An illusion of collapse comes from Bayesian updates by observer's continuous outcome data. Spacetime is to be understood in epistemic sense, instead of existing independently of an observer, following spirits of black hole complementarity. Dark matter and dark energy pop up directly as entropic tug-of-war in the reformulation.

•Apr 7, 2019•0 citations

A strict epistemic approach to physics

Per Östborn

The general view is that all fundamental physical laws should be formulated within the framework given by quantum mechanics (QM). In a sense, QM therefore has the character of a metaphysical theory. Consequently, if it is possible to derive QM from more basic principles, these principles should be of general, philosophical nature. Here, we derive the formalism of QM from well-motivated epistemic principles. A key assumption is that a physical theory that relies on entities or distinctions that are unknowable in principle gives rise to wrong predictions. First, an epistemic formalism is developed, using concepts like knowledge and potential knowledge, identifying a physical state $S$ with the potential knowledge of the physical world. It is demonstrated that QM emerges from this formalism. However, Hilbert spaces, wave functions and probabilities are defined in certain well-defined observational contexts only. This means that the epistemic formalism is broader than QM. In the fundamental layer of description, the physical state $S$ is a subset of a state space $\mathcal{S}=\{Z\}$, such that $S$ always contains many elements $Z$. These elements correspond to unattainable states of complete knowledge of the world. The evolution of $S$ cannot be determined in terms of the individual evolution of the elements $Z$, unlike the evolution of an ensemble in classical phase space. The evolution of $S$ is described in terms of sequential time $n\in \mathbf{\mathbb{N}}$, which is updated according to $n\rightarrow n+1$ each time an event occurs, each time potential knowledge changes. Sequential time $n$ can be separated from relational time $t$, which describes distances between events in space-time. There is an entire space-time associated with each $n$, in which $t$ represents the knowledge at sequential time $n$ about the temporal relations between present and past events.

arXiv:1601.00680•Jan 4, 2016•1 citation
The measurement problem in quantum mechanics resists resolution within traditional substance ontologies. This paper proposes the Kantian-Bayesian Flux (KB-Flux) formalism, an operationalist framework that reinterprets physical laws not as descriptions of an observer-independent reality, but as necessary data-compression and belief-updating protocols. By mapping the quantum wave function () to the Bayesian belief state () of an observer operating under the Free Energy Principle, we structurally align quantum superposition with latent epistemic uncertainty and measurement collapse with discrete Bayesian updates. This framework formally navigates quantum no-go theorems, synthesizes Relational Quantum Mechanics (RQM) with Active Inference, delineates its departures from Quantum Bayesianism (QBism), and yields heuristic predictions for the architectural evolution of Artificial General Intelligence and neurophysics. For a broader, accessible exploration of this framework and its implications for the nature of reality and mind, follow the Substack publication: The Shum Briefing.
[Author’s Note (April 2026)]Supersession Notice: This 2025 paper explores the early KB-Flux framework. It has been structurally overhauled and formally superseded by “The Light & Flux (L&F) Framework (2026)”. While the 2025 version relies on cognitive metaphors—treating the wave function as a “generative model” —the 2026 L&F framework provides a rigorous mathematical proof deriving complex Hilbert space from finite-inference constraints. It replaces “phenomenal grammar” with the Simplex Projection (𐍀), grounded in metabolic and thermodynamic mandates. Researchers seeking the formal Non-Derivability Principle and a stratified three-tier ontology should refer to the 2026 version. Original AbstractReaders are advised to treat the present paper strictly as a historical record of the framework’s initial development. For the current, accurate formulation, please refer to the new publication.The measurement problem—why quantum superpositions yield definite outcomes upon observation—has persisted as the central interpretative challenge of quantum mechanics for nearly a century. This paper proposes that the problem dissolves when viewed through the joint lens of Kantian epistemology and predictive processing (Bayesian brain) theory. Drawing on Kant’s distinction between phenomena (experience) and noumena (reality-in-itself), we argue that the quantum wave function (Ψ) does not represent a physical field in the noumenal world, but rather the generative model of a prediction-generating cognitive agent. In this framework, "superposition" represents a state of uninitialized probabilistic expectation (Latent Space), while "collapse" is reinterpreted as a Bayesian belief update—the discrete transition from a probability distribution to a definite value, necessitated by the information-processing constraints of the observer. By treating the observer not as a passive witness but as an active inference engine, this framework: Naturalizes the collapse postulate: Measurement is the update of an internal model, not a physical discontinuity in nature. Resolves nested-observer paradoxes: Wigner's Friend is resolved by treating reality as a distributed system with eventual consistency, where distinct agents hold distinct local cache states. Grounds the Born Rule: We argue that the squared-amplitude probability rule (P=∣Ψ∣2P=|\Psi|^2P=∣Ψ∣2) serves as a transcendental constraint (akin to Gleason’s Theorem) required for the internal consistency of any probabilistic agent. This interpretation suggests that the "weirdness" of quantum mechanics reflects the necessary architectural constraints of any finite mind attempting to model an inaccessible reality, effectively treating physics as the "User Interface" of reality.
[Author’s Note]This paper is retained as a historical record. The informal ‘Kantian-Bayesian’ framework presented here has been entirely superseded by the Light & Flux (L&F) framework, which corrects the temporal substrate circularity, introduces a three-tier formal ontology, and provides a rigorous deductive derivation of complex Hilbert space from inferential constraints. See [DOI of L&F V5]. For a broader, accessible exploration of this framework and its implications for the nature of reality and mind, follow the Substack publication: Mind & Machine. Original AbstractModern interpretations of Quantum Mechanics, particularly Quantum Bayesianism (QBism) and the Free Energy Principle, increasingly suggest that physical reality is not an observer-independent ontology but a generative model constructed to minimize predictive error. This paper proposes a formal synthesis between this "Physics-as-Interface" paradigm and the phenomenology of Eastern Zen philosophy. We argue that the Zen concept of "Subtle Impermanence" (Anicca) is mathematically equivalent to a Latent Markovian Substrate, while the "illusion of constancy" maps rigorously to the Belief State (q(x)) maintained by the observer to facilitate inference under partial observability. By framing the Wave Function not as a physical object but as a computational strategy to mitigate Variational Free Energy, we demonstrate that the laws of physics are essentially stability protocols encoded by consciousness. This synthesis moves beyond metaphorical comparisons, offering a rigorous, information-theoretic grounding for the convergence of Eastern phenomenology and Western computational mechanics. Key Theoretical Contributions: The Markovian Noumena (Hardware): Reinterpreting Kant’s Noumena as a Latent Markov Process (x_t)—a generative substrate of discrete, conditionally independent moments that lacks intrinsic object permanence. The Non-Markovian Interface (Software): Defining conscious perception as a Belief State (q(x)q(x)q(x)) that acts as an "Active Buffer." This interface imposes a non-Markovian constraint (memory/history) onto the latent data to construct a navigable user interface. Wave Function as Belief State: Aligning with QBism, we formally equate the quantum wave function (Ψ) with the observer's internal belief state, where "collapse" is redefined as a standard Bayesian belief update minimizing prediction error. Free Energy & Complexity: Utilizing Friston’s Free Energy Principle, we argue that "Causality," "Time," and "Objects" are optimal compression algorithms derived to minimize the Complexity Cost (KL Divergence) of the internal model. The Thermodynamics of Time: Proposing that the "Arrow of Time" is an artifact of Inference Asymmetry, where predicting a divergent future requires higher computational entropy than retrodicting a convergent past.
[Author’s Note]This paper introduces the Light & Flux (L&F) framework, which formally supersedes the earlier "KB-Flux" formalism. The previous ontological mappings have been entirely replaced by the rigorous three-tier architecture and the Non-Derivability Principle presented herein. For a broader, accessible exploration of this framework and its implications for the nature of reality and mind, follow the Substack publication: Mind & Machine. AbstractQuantum mechanics is not an empirical physical law; it is the geometric theorem of coherent inference under fundamental uncertainty. The explanatory deadlock of modern quantum foundations stems from a flat ontology that falsely presumes classicality as the default state of reality. This paper introduces the Light & Flux (L&F) framework, a three-tier formal architecture demonstrating that incomputable complex Hilbert space is the universal inferential baseline. The macroscopic classical world, conversely, is a metabolically constrained, lossy compression rendered by finite biological agents. We formalize this irreversible compression via the Simplex Projection and establish the Non-Derivability Principle: demanding a unitary dynamical account for a specific classical measurement outcome is a mathematical category error. Finally, we map the precise deductive gap between pure inferential logic and generalized probabilistic theories (GPTs), confronting the Gödelian limits of our own cognitive codec.
The measurement problem in quantum mechanics challenges our understanding of reality, demanding explanations beyond both Copenhagen's "collapse" and the Many-Worlds' ontological multiplicity. This paper introduces and formalizes the Ze framework as a novel synthesis of Relational Quantum Mechanics (RQM) and the Active Inference paradigm from theoretical neuroscience. Ze posits that quantum states are relational, defined by the posterior beliefs of interacting generative models engaged in variational free energy minimization. Within this framework, quantum superposition is formalized as high compatibility (ℐ ≈ 1) between competing models, characterized by low free-energy conflict (ΔF < θ). Conversely, the transition to a localized state—the physical correlate of "collapse"—is reconceived not as a metaphysical event but as an optimization-driven phase transition. This occurs when model conflict exceeds a critical threshold (ΔF > θ), a process objectively driven by interactions like which-path marking. We demonstrate that matter-wave interferometry with complex molecules provides a direct experimental testbed for these principles, where which-path information and quantum erasure actively manipulate ΔF. Extending the isomorphism, we propose that transitions in human cognition—from focused wakefulness to dreaming and psychedelic states—are governed by analogous shifts in the brain's inferential threshold (θ). Thus, Ze offers a unified, testable architecture bridging quantum foundations, statistical physics, and the neuroscience of consciousness.
This paper provides an epistemic reformulation of quantum mechanics (QM) in terms of inference consistency requirements of objective Bayesianism, which include the principle of maximum entropy under physical constraints. Physical laws themselves are understood in terms of inference and physical consistency requirements. Strong complementarity - that different observers may "live" in separate Hilbert spaces - follows as a consequence, which resolves the firewall paradox. Other clues pointing to this reformulation are analyzed. The reformulation, with the addition of novel transition probability arithmetic, resolves the measurement problem completely, thereby eliminating charge of subjectivity of measurements from quantum mechanics. An illusion of collapse comes from Bayesian updates by observer's continuous outcome data. Spacetime is to be understood in epistemic sense, instead of existing independently of an observer, following spirits of black hole complementarity. Dark matter and dark energy pop up directly as entropic tug-of-war in the reformulation.
The general view is that all fundamental physical laws should be formulated within the framework given by quantum mechanics (QM). In a sense, QM therefore has the character of a metaphysical theory. Consequently, if it is possible to derive QM from more basic principles, these principles should be of general, philosophical nature. Here, we derive the formalism of QM from well-motivated epistemic principles. A key assumption is that a physical theory that relies on entities or distinctions that are unknowable in principle gives rise to wrong predictions. First, an epistemic formalism is developed, using concepts like knowledge and potential knowledge, identifying a physical state SSS with the potential knowledge of the physical world. It is demonstrated that QM emerges from this formalism. However, Hilbert spaces, wave functions and probabilities are defined in certain well-defined observational contexts only. This means that the epistemic formalism is broader than QM. In the fundamental layer of description, the physical state SSS is a subset of a state space S={Z}\mathcal{S}=\{Z\}S={Z}, such that SSS always contains many elements ZZZ. These elements correspond to unattainable states of complete knowledge of the world. The evolution of SSS cannot be determined in terms of the individual evolution of the elements ZZZ, unlike the evolution of an ensemble in classical phase space. The evolution of SSS is described in terms of sequential time n∈Nn\in \mathbf{\mathbb{N}}n∈N, which is updated according to n→n+1n\rightarrow n+1n→n+1 each time an event occurs, each time potential knowledge changes. Sequential time nnn can be separated from relational time ttt, which describes distances between events in space-time. There is an entire space-time associated with each nnn, in which ttt represents the knowledge at sequential time nnn about the temporal relations between present and past events.