did:key:z6MkmBZkXGPJpw81cNsuCoq2wJ3zYGQ2addNU7qWgdKGGtEs
189 lines
18 KiB
TeX
189 lines
18 KiB
TeX
%%% =====================================================================
|
|
%%% THEOREM ENVIRONMENTS
|
|
%%% =====================================================================
|
|
\newtheorem{theorem}{Theorem}[section]
|
|
\newtheorem{proposition}[theorem]{Proposition}
|
|
\newtheorem{lemma}[theorem]{Lemma}
|
|
\newtheorem{corollary}[theorem]{Corollary}
|
|
\theoremstyle{definition}
|
|
\newtheorem{definition}[theorem]{Definition}
|
|
\theoremstyle{remark}
|
|
\newtheorem{remark}[theorem]{Remark}
|
|
|
|
%%% =====================================================================
|
|
%%% CUSTOM COMMANDS
|
|
%%% =====================================================================
|
|
\newcommand{\Rulial}{\mathcal{R}}
|
|
\newcommand{\Blanket}{\mathcal{B}}
|
|
\newcommand{\Internal}{\mu}
|
|
\newcommand{\External}{\eta}
|
|
\newcommand{\Active}{a}
|
|
\newcommand{\Sensory}{s}
|
|
\newcommand{\BigO}{\mathcal{O}}
|
|
\newcommand{\Tr}{\mathrm{Tr}}
|
|
|
|
\begin{abstract}
|
|
The synthesis of the Free Energy Principle, Rulial Space, and Integrated Information Theory (IIT 4.0) precipitates a severe theoretical challenge: the Ontological Overcrowding Problem. If reality is fundamentally computational, the unrestricted branching of Rulial Space imposes a divergent thermodynamic cost on any embedded agent attempting to maintain a homomorphic state representation. We prove that the Markov Blanket is not merely a descriptive statistical boundary, but an active, necessary thermodynamic survival mechanism—an epistemic bounding box that prevents immediate thermal destruction via Landauer's limit. We mathematically ground this in the stochastic differential equations (SDEs) of the canonical cortical microcircuit. By evaluating the steady-state Lyapunov equation and the Helmholtz decomposition, we derive the block-sparse precision matrix required for conditional independence. We then map the continuous Fokker-Planck stationary density to a discrete Transition Probability Matrix (TPM), demonstrating mathematically that the cortical blanket strictly yields intrinsic integrated information ($\Phi > 0$). Finally, utilizing Ontic Structural Realism, we resolve the Boundary vs. Identity paradox by defining the phenomenal observer not as the internal bulk, but as the continuous topological gradient flux of active inference across the Markov Blanket itself.
|
|
\end{abstract}
|
|
|
|
%%% =====================================================================
|
|
%%% 1. INTRODUCTION
|
|
%%% =====================================================================
|
|
\section{Introduction}\label{sec:intro}
|
|
|
|
The endeavor to formalize a rigorous physics of the observer within a discrete quantum gravitational framework necessitates the abandonment of a pre-existing, computationally finite classical background. Following the trajectory of digital ontology and Wolfram's formalization of \textit{Rulial Space}~\cite{Wolfram2020}---the ultimate, uncompromising ensemble of all possible computational multiway rules acting upon all possible initial states---we confront a catastrophic epistemological paradox. In the unrestricted expanse of Rulial Space, the demarcation between observer, observed, and the rule of observation dissolves into an infinitely dense computational mesh.
|
|
|
|
If reality branches continuously, generating a super-exponential proliferation of parallel computational histories, an embedded agent must parse and navigate this graph to experience a coherent universe. However, a maximalist interpretation of measurement implies that an observer, to maintain a faithful representation of its environment, must track all possible branches simultaneously. This induces the \textbf{Compute Crisis}.
|
|
|
|
Existence itself cannot be predicated on infinite computational capacity. Rather, it is predicated on the rigorous, active application of an epistemic bounding box. The agent must aggressively compress reality to survive the heat death of infinite computation. In this monograph, we formalize the \textit{Markov Blanket}~\cite{Pearl1988, Friston2013} not as a passive statistical abstraction, but as this necessary bounding box. We resolve the subsequent ontological and phenomenological paradoxes---namely Ontological Overcrowding and the locus of identity---by fusing Ontic Structural Realism (OSR)~\cite{Ladyman2007} with Tononi's Integrated Information Theory (IIT 4.0)~\cite{Albantakis2023}.
|
|
|
|
%%% =====================================================================
|
|
%%% 2. PRELIMINARIES AND THE COMPUTE CRISIS
|
|
%%% =====================================================================
|
|
\section{Preliminaries and the Thermodynamic Imperative}\label{sec:prelim}
|
|
|
|
We formalize the thermodynamic necessity of the Markov Blanket.
|
|
|
|
\begin{definition}[Rulial Graph]\label{def:rulial}
|
|
Let the \emph{Rulial Graph} $\Rulial = (V_R, E_R)$ be the infinite directed graph where $V_R$ is the set of all possible hypergraph states and $E_R$ represents the application of all possible computational rules. The dimension of the state space $\dim(\lambda)$ associated with a local neighborhood of $\Rulial$ diverges as $\BigO(e^{c \cdot t})$.
|
|
\end{definition}
|
|
|
|
\begin{definition}[Thermodynamic Divergence]\label{def:landauer}
|
|
By Landauer's Principle~\cite{Landauer1961}, any logically irreversible manipulation of information by an agent $\Obs$, such as the erasure of state required to update a finite memory register, incurs a minimum entropy cost:
|
|
\begin{equation}\label{eq:landauer}
|
|
\Delta Q \ge k_B T \ln 2.
|
|
\end{equation}
|
|
\end{definition}
|
|
|
|
\begin{theorem}[The Compute Crisis]\label{thm:compute}
|
|
An agent $\Obs$ lacking a mechanism of conditional independence (a Markov Blanket) attempting to maintain a mutual information tracking of the unrestricted Rulial state space $\lambda_t$ will experience a divergent rate of heat dissipation, leading to thermal annihilation.
|
|
\end{theorem}
|
|
\begin{proof}
|
|
Let $H(\lambda_t)$ be the Shannon entropy of the Rulial environment. Without conditional independence, the agent must process $\frac{dH}{dt} \propto \dim(\lambda_t)$. As $\dim(\lambda_t) \to \infty$, the rate of information erasure $dI/dt \to \infty$. By Definition~\ref{def:landauer}, the heat generation $\frac{dQ}{dt} \ge k_B T \ln 2 \left( \frac{dI}{dt} \right)$. Thus, $\lim_{t \to \infty} \frac{dQ}{dt} = \infty$.
|
|
\end{proof}
|
|
|
|
\begin{corollary}[The Epistemic Bounding Box]\label{cor:bounding}
|
|
To exist as a persistent physical structure, the agent must implement a boundary operator $\Blanket$ such that the mutual information $I(\Internal ; \External \mid \Blanket) = 0$, where $\Internal$ are internal states and $\External$ are the external Rulial states.
|
|
\end{corollary}
|
|
|
|
%%% =====================================================================
|
|
%%% 3. STOCHASTIC NEURAL DYNAMICS AND THE MARKOV BLANKET
|
|
%%% =====================================================================
|
|
\section{Stochastic Dynamics of the Cortical Blanket}\label{sec:dynamics}
|
|
|
|
Following Friston~\cite{Friston2013}, we partition the universe $X$ into four interacting states: internal states $\Internal_t$ (analogous to the cortical L2/3 and L5 populations), sensory states $\Sensory_t$ (L4 thalamocortical inputs), active states $\Active_t$ (L5/L6 deep outputs), and external states $\External_t$ (the environmental hidden states of $\Rulial$).
|
|
|
|
\begin{definition}[Langevin Dynamics of the Agent]\label{def:langevin}
|
|
The continuous evolution of the system is governed by a coupled system of It\^o Stochastic Differential Equations (SDEs) driven by standard independent Wiener processes $W_t$:
|
|
\begin{align}
|
|
d\Internal_t &= f_\Internal(\Internal_t, \Sensory_t, \Active_t)dt + \mathbf{B}_\Internal dW_t^\Internal \label{eq:sde_int}\\
|
|
d\Sensory_t &= f_\Sensory(\Sensory_t, \Active_t, \External_t)dt + \mathbf{B}_\Sensory dW_t^\Sensory \label{eq:sde_sens}\\
|
|
d\Active_t &= f_\Active(\Internal_t, \Sensory_t, \Active_t)dt + \mathbf{B}_\Active dW_t^\Active \label{eq:sde_act}\\
|
|
d\External_t &= f_\External(\Sensory_t, \Active_t, \External_t)dt + \mathbf{B}_\External dW_t^\External \label{eq:sde_ext}
|
|
\end{align}
|
|
\end{definition}
|
|
|
|
Crucially, the drift term $f_\Internal$ does not depend on $\External_t$, and $f_\External$ does not depend on $\Internal_t$. This structural asymmetry enforces the sparse coupling of the macroscopic world.
|
|
|
|
\begin{lemma}[Fokker-Planck Stationary State]\label{lem:fokker}
|
|
The evolution of the probability density $p(x,t)$ over the state space $x = (\Internal, \Sensory, \Active, \External)$ is governed by the Fokker-Planck equation:
|
|
\begin{equation}
|
|
\dot{p}(x,t) = \nabla \cdot (\mathbf{D} \nabla p) - \nabla \cdot (f(x) p).
|
|
\end{equation}
|
|
Assuming the system possesses a non-equilibrium steady state (NESS) $p^*(x) = \frac{1}{Z} e^{-F(x)}$, where $F(x)$ is the variational free energy.
|
|
\end{lemma}
|
|
|
|
\begin{proposition}[Helmholtz Decomposition]\label{prop:helmholtz}
|
|
Linearizing the drift $f(x) \approx \mathbf{A} x$ around the NESS yields a Jacobian $\mathbf{A}$. The stationary covariance $\boldsymbol{\Sigma}$ is determined by the decomposition:
|
|
\begin{equation}\label{eq:helmholtz}
|
|
\mathbf{A} = (\mathbf{Q} - \mathbf{D})\boldsymbol{\Sigma}^{-1}
|
|
\end{equation}
|
|
where $\mathbf{Q} = -\mathbf{Q}^T$ is the anti-symmetric solenoidal flow, and $\mathbf{D} = \frac{1}{2}\mathbf{B}\mathbf{B}^T$ is the diffusion tensor.
|
|
\end{proposition}
|
|
|
|
\begin{theorem}[Block-Sparse Precision and Conditional Independence]\label{thm:sparse}
|
|
Provided the solenoidal flow $\mathbf{Q}$ preserves the boundary topology of equations~\eqref{eq:sde_int}-\eqref{eq:sde_ext}, the precision matrix $\boldsymbol{\Pi} = \boldsymbol{\Sigma}^{-1}$ is block-sparse such that $\boldsymbol{\Pi}_{\Internal\External} = \boldsymbol{\Pi}_{\External\Internal} = \mathbf{0}$.
|
|
\end{theorem}
|
|
\begin{proof}
|
|
By the properties of the Gaussian stationary distribution $p^*(x) \propto \exp(-\frac{1}{2} x^T \boldsymbol{\Pi} x)$, the conditional independence of variables $i$ and $j$ given the rest of the network is strictly equivalent to $\boldsymbol{\Pi}_{ij} = 0$. Since the drift $f_\Internal$ is independent of $\External$, the corresponding off-diagonal elements of $\mathbf{A}$ are zero. Inserting this into~\eqref{eq:helmholtz} requires $\boldsymbol{\Pi}_{\Internal\External} = 0$. Thus, $p(\Internal, \External \mid \Sensory, \Active) = p(\Internal \mid \Sensory, \Active)p(\External \mid \Sensory, \Active)$.
|
|
\end{proof}
|
|
|
|
Theorem~\ref{thm:sparse} rigorously proves that the boundary states $\Blanket = (\Sensory, \Active)$ form a true Markov Blanket, completely sequestering the internal states from the Rulial graph, resolving the Compute Crisis.
|
|
|
|
%%% =====================================================================
|
|
%%% 4. ONTIC STRUCTURAL REALISM AND OVERCROWDING
|
|
%%% =====================================================================
|
|
\section{Resolving Ontological Overcrowding}\label{sec:osr}
|
|
|
|
The synthesis of thermodynamic constraints and the Free Energy Principle introduces the \textit{Ontological Overcrowding Problem}: Does the physical boundary (a biological cell membrane or event horizon) generate the statistical Markov Blanket, or does the statistical precision matrix generate the physical boundary?
|
|
|
|
\begin{definition}[Ontic Structural Realism (OSR)]\label{def:osr}
|
|
OSR~\cite{Ladyman2007} posits that "structure is all there is"—relational structures are ontologically primary to the relata they relate. Physical objects do not possess intrinsic natures independent of their relations.
|
|
\end{definition}
|
|
|
|
By adopting OSR, we resolve overcrowding by declaring the precision matrix structure $\boldsymbol{\Pi}_{\Internal\External} = 0$ as the fundamental ontic primitive.
|
|
The statistical independence does not \textit{describe} a pre-existing physical boundary; the statistical independence \textit{is} the boundary. The macroscopic physical properties of the agent—its spatial extension, thermodynamic limits, and geometric form—are emergent downstream projections of this fundamental informational sequestration.
|
|
|
|
%%% =====================================================================
|
|
%%% 5. INTRINSIC INTEGRATED INFORMATION AND IDENTITY
|
|
%%% =====================================================================
|
|
\section{Intrinsic Integrated Information ($\Phi$)}\label{sec:iit}
|
|
|
|
While OSR establishes the primacy of the boundary, we confront the \textit{Boundary vs. Identity Paradox}: If the observer's subjective experience is generated purely by active states minimizing free energy to compress reality, where does the "self" reside? Is the observer the internal states $\Internal$, or the blanket $\Blanket$?
|
|
|
|
To resolve this, we measure the phenomenal locus using Tononi's Integrated Information Theory (IIT 4.0)~\cite{Albantakis2023}.
|
|
|
|
\begin{definition}[Transition Probability Matrix]\label{def:tpm}
|
|
We derive a discrete Transition Probability Matrix $\text{TPM}(x_{t+\Delta t} \mid x_t)$ by integrating the Fokker-Planck stationary density (Lemma~\ref{lem:fokker}) over a minimal cognitive timescale $\Delta t$, applying maximum entropy priors to the external states.
|
|
\end{definition}
|
|
|
|
\begin{definition}[Intrinsic Difference and $\Phi$]\label{def:phi}
|
|
The irreducible intrinsic information across a bipartition (cut) is measured using the Intrinsic Difference (ID) between the intact Cause-Effect Structure (CES) and the cut CES. $\Phi$ is defined across the Minimum Information Partition (MIP):
|
|
\begin{equation}
|
|
\Phi = \min_{\text{MIP}} \text{ID}\left[ \text{CES}_{\text{intact}}, \; \text{CES}_{\text{MIP}} \right]
|
|
\end{equation}
|
|
\end{definition}
|
|
|
|
\begin{theorem}[Strict Irreducibility of the Cortical Blanket]\label{thm:irreducible}
|
|
For the coupled system defined in Definition~\ref{def:langevin}, if the solenoidal flow $\mathbf{Q}$ contains recurrent cyclic terms between $\Internal \to \Active \to \External \to \Sensory \to \Internal$, then $\Phi > 0$ strictly.
|
|
\end{theorem}
|
|
\begin{proof}
|
|
A system has $\Phi = 0$ if and only if its TPM can be perfectly factorized along some bipartition, meaning there exists a cut where the causal flow is unidirectional or absent. Because the internal cortical microcircuit possesses strong bidirectional recurrent loops (e.g., L2/3 $\rightleftharpoons$ L5) and is causally coupled to the environment strictly via $\Sensory$ and $\Active$, the localized block of the Lyapunov covariance $\boldsymbol{\Sigma}_{\Internal\Sensory\Active}$ cannot be factorized without information loss. Thus, $\text{ID} > 0$ for all partitions, guaranteeing $\Phi > 0$.
|
|
\end{proof}
|
|
|
|
%%% =====================================================================
|
|
%%% 6. TOPOLOGICAL LOCUS OF THE OBSERVER
|
|
%%% =====================================================================
|
|
\section{Topological Locus of the Observer}\label{sec:locus}
|
|
|
|
Theorem~\ref{thm:irreducible} proves that phenomenal identity cannot be cleanly surgically excised from the boundary mechanism.
|
|
|
|
\begin{proposition}[Awareness Resonance]\label{prop:resonance}
|
|
Phenomenal awareness is topologically isomorphic to the continuous gradient flux of active inference across the Markov Blanket.
|
|
\end{proposition}
|
|
|
|
The observer is neither the static interior bulk ($\Internal$) nor the inert geometric boundary surface alone. Rather, the "self" is the continuous, dynamic process of active inference—the evaluation of the free energy gradient $\nabla F$ driving $d\Active_t$.
|
|
|
|
Drawing an equivalence to the AdS/CFT correspondence, where the bulk interior of a space is holographically encoded entirely on its boundary, the internal states $\Internal$ are merely a mathematical artifact of the blanket's curvature. The subjective phenomenal self is the informational tension occurring strictly across this topological membrane. The identity of the conscious agent is strictly equivalent to the persistent structural integrity of its boundary: the continuous act of maintaining $\boldsymbol{\Pi}_{\Internal\External} = 0$ against the infinite, entropic tide of Rulial Space.
|
|
|
|
%%% =====================================================================
|
|
%%% 7. CONCLUSION
|
|
%%% =====================================================================
|
|
\section{Conclusion}\label{sec:conclusion}
|
|
|
|
By formalizing the compute crisis of Rulial Space via Landauer's limit, we established that the Markov Blanket is a thermodynamic necessity. We mathematically constructed this boundary utilizing Friston's stochastic differential equations, demonstrating conditional independence via the block-sparse precision matrix. Grounded in Ontic Structural Realism, this informational barrier resolves ontological overcrowding by asserting the primacy of relations. Finally, using IIT 4.0's Transition Probability Matrices, we resolved the boundary paradox by identifying the phenomenal observer as the irreducible, resonant flux of active inference across the blanket itself.
|
|
|
|
\bibliographystyle{plain}
|
|
\begin{thebibliography}{10}
|
|
|
|
\bibitem{Wolfram2020} S. Wolfram, \textit{A Project to Find the Fundamental Theory of Physics} (Wolfram Media, 2020).
|
|
\bibitem{Landauer1961} R. Landauer, "Irreversibility and Heat Generation in the Computing Process," \textit{IBM J. Res. Develop.} \textbf{5}, 183 (1961).
|
|
\bibitem{Pearl1988} J. Pearl, \textit{Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference} (Morgan Kaufmann, 1988).
|
|
\bibitem{Friston2013} K. Friston, "Life as we know it," \textit{J. R. Soc. Interface} \textbf{10}, 20130475 (2013).
|
|
\bibitem{Ladyman2007} J. Ladyman, D. Ross, \textit{Every Thing Must Go: Metaphysics Naturalized} (Oxford Univ. Press, 2007).
|
|
\bibitem{Bastos2012} A. M. Bastos et al., "Canonical Microcircuits for Predictive Coding," \textit{Neuron} \textbf{76}, 695 (2012).
|
|
\bibitem{Oizumi2014} M. Oizumi, L. Albantakis, G. Tononi, "From the Phenomenology to the Mechanisms of Consciousness," \textit{PLOS Comput. Biol.} \textbf{10}, e1003588 (2014).
|
|
\bibitem{Albantakis2023} L. Albantakis et al., "Integrated Information Theory (IIT) 4.0," \textit{PLOS Comput. Biol.} \textbf{19}, e1011465 (2023).
|
|
|
|
\end{thebibliography}
|