Part I: Foundations

Thermodynamic Foundations

Introduction

0:00 / 0:00

Thermodynamic Foundations

Driven Nonlinear Systems and the Emergence of Structure

Existing Theory

The thermodynamic foundations here draw on several established theoretical frameworks:

Prigogine’s dissipative structures (1977 Nobel Prize): Systems far from equilibrium spontaneously develop organized patterns that dissipate energy more efficiently than uniform states. My treatment of “Generic Structure Formation” formalizes Prigogine’s core insight.
Friston’s Free Energy Principle (2006–present): Self-organizing systems minimize variational free energy, which bounds surprise. The viability manifold $\viable$ corresponds to regions of low expected free energy under the system’s generative model.
Autopoiesis (Maturana \& Varela, 1973): Living systems are self-producing networks that maintain their organization through continuous material turnover. The “boundary formation” section formalizes the autopoietic insight that life is organizationally closed but thermodynamically open.
England’s dissipation-driven adaptation (2013): Driven systems are biased toward configurations that absorb and dissipate work from external fields. The “Dissipative Selection” proposition extends this to selection among structured attractors.

Consider a physical system $\mathcal{S}$ described by a state vector $\mathbf{x} \in \R^n$ evolving according to dynamics:

\frac{d\mathbf{x}}{dt} = \mathbf{f}(\mathbf{x}, t) + \bm{\eta}(t)

where $\mathbf{f}: \R^n \times \R \to \R^n$ is a generally nonlinear vector field and $\bm{\eta}(t)$ represents stochastic forcing with specified statistics.

Such a system is far from equilibrium when three conditions hold: (a) a sustained gradient—continuous influx of free energy, matter, or information preventing relaxation to thermodynamic equilibrium; (b) dissipation—continuous entropy export to the environment; and (c) nonlinearity—dynamics $\mathbf{f}$ containing terms of order $\geq 2$ .

Such systems generically develop dissipative structures—organized patterns that persist precisely because they efficiently channel the imposed gradients. This can be made precise. Let $\mathcal{S}$ be a far-from-equilibrium system with dynamics admitting a Lyapunov-like functional $\mathcal{L}: \R^n \to \R$ such that:

\frac{d\mathcal{L}}{dt} = -\sigma(\mathbf{x}) + J(\mathbf{x})

where $\sigma(\mathbf{x}) \geq 0$ is the entropy production rate and $J(\mathbf{x})$ is the free energy flux from external driving. Then for sufficiently strong driving ( $J > J_c$ for some critical threshold $J_c$ ), the system generically admits multiple metastable attractors ${\mathcal{A}_i}$ with:

Structured internal organization (reduced entropy relative to uniform distribution)
Finite basins of attraction with measurable barriers
History-dependent selection among attractors (path dependence)
Spontaneous symmetry breaking (selection of one among equivalent configurations)

Proof. [Proof sketch] The proof follows from bifurcation theory for dissipative systems. As the driving parameter exceeds

J_c

, the uniform/equilibrium state loses stability through a bifurcation (typically pitchfork, Hopf, or saddle-node), giving rise to structured alternatives. The multiplicity of attractors follows from the broken symmetry; the barriers from the existence of separatrices in the deterministic skeleton; path dependence from noise-driven selection among equivalent states. □

Empirical Grounding

Bénard Convection Cells: The canonical laboratory demonstration of dissipative structure formation.

When a thin layer of fluid is heated from below:

For $\Delta T < \Delta T_c$ (Rayleigh number $\text{Ra} < \text{Ra}_c \approx 1708$ ): Heat transfers by conduction only. Uniform, unstructured state.
For $\Delta T > \Delta T_c$ : Spontaneous symmetry breaking produces hexagonal convection cells. The fluid self-organizes into a pattern that transports heat more efficiently than conduction alone.

This is precisely the predicted structure: a bifurcation at critical driving ( $J_c$ ), multiple equivalent attractors (cells can rotate clockwise or counterclockwise), and path-dependent selection.

Future Empirical Work

Quantitative validation: Measure entropy production rates $\sigma$ in Bénard cells at various $\text{Ra}$ values. Verify that $\sigma_{\text{structured}} > \sigma_{\text{uniform}}$ for $\text{Ra} > \text{Ra}_c$ , confirming dissipative selection.

Parameters to measure: Critical Rayleigh number, entropy production above/below transition, correlation between cell size and $\Delta T$ .

The Free Energy Landscape

For systems amenable to such analysis, one can define an effective free energy functional:

\mathcal{F}[\mathbf{x}] = U[\mathbf{x}] - T \cdot S[\mathbf{x}] + \text{(non-equilibrium corrections)}

where $U$ captures internal energy, $S$ entropy, and $T$ an effective temperature. The dynamics can often be written as:

\frac{d\mathbf{x}}{dt} = -\Gamma \cdot \nabla_\mathbf{x} \mathcal{F}[\mathbf{x}] + \bm{\eta}(t)

for some positive-definite mobility tensor $\Gamma$ . In this representation:

Local minima of $\mathcal{F}$ correspond to metastable attractors
Saddle points determine transition rates between attractors
The depth of minima relative to barriers determines persistence times

One structure within this landscape will recur throughout the book. For a self-maintaining system, the viability manifold $\viable \subset \R^n$ is the region of state space within which the system can persist indefinitely (or for times long relative to observation scales):

\viable = \left\{ \mathbf{x} \in \R^n : \E\left[\tau_{\text{exit}}(\mathbf{x})\right] > T_{\text{threshold}} \right\}

where $\tau_{\text{exit}}(\mathbf{x})$ is the first passage time to a dissolution state starting from $\mathbf{x}$ .

The viability manifold will play a central role in understanding normativity: trajectories that remain within $\viable$ are, in a precise sense, “good” for the system, while trajectories that approach the boundary $\partial\viable$ are “bad.”

Viability Theory

The viability manifold concept connects to Aubin’s viability theory (1991), which provides mathematical tools for analyzing systems that must satisfy state constraints over time. Key results:

A state is viable iff there exists at least one trajectory remaining in $\viable$ forever
The viability kernel is the largest subset from which viable trajectories exist
For controlled systems, viability requires the control to “point inward” at boundaries

I’ll add stochasticity and connect viability to phenomenology: the felt sense of threat corresponds to proximity to $\partial\viable$ .

Dissipative Structures and Selection

A crucial insight is that among the possible structured states, those that persist tend to be those that efficiently dissipate the imposed gradients. This is not teleological; it follows from differential persistence.

We can quantify this. The dissipation efficiency of a structured state $\mathcal{A}$ measures how much of the available entropy production the state actually channels:

\eta(\mathcal{A}) = \frac{\sigma(\mathcal{A})}{\sigma_{\max}}

where $\sigma(\mathcal{A})$ is the entropy production rate in state $\mathcal{A}$ and $\sigma_{\max}$ is the maximum possible entropy production given the imposed constraints. This quantity governs a selection principle: in the long-time limit, the probability measure over states concentrates on high-efficiency configurations:

\lim_{t \to \infty} \prob(\mathbf{x} \in \mathcal{A}) \propto \exp\left(\beta \cdot \eta(\mathcal{A})\right)

for some effective selection strength $\beta > 0$ depending on the noise level and barrier heights.

This provides the thermodynamic foundation for the emergence of organized structures: they are not thermodynamically forbidden but thermodynamically enabled—selected for by virtue of their gradient-channeling efficiency.

Boundary Formation

Among the dissipative structures that emerge, a particularly important class involves spatial or functional boundaries that separate an “inside” from an “outside.”

A boundary $\partial\Omega$ in a driven system is emergent if it satisfies four conditions:

It arises spontaneously from the dynamics (not imposed externally)
It creates a region $\Omega$ (the “inside”) with dynamics partially decoupled from the exterior
It is actively maintained by the system’s dissipative processes
It enables gradients across itself that would otherwise equilibrate

The canonical example is the lipid bilayer membrane in aqueous solution. Given appropriate concentrations of amphiphilic molecules and energy input, membranes form spontaneously because they represent a low-free-energy configuration. Once formed, they:

Separate internal chemical concentrations from external
Enable maintenance of ion gradients, pH differences, etc.
Provide a substrate for embedded machinery (channels, pumps, receptors)
Must be actively maintained against degradation

Empirical Grounding

Lipid Bilayer Self-Assembly: Spontaneous boundary formation from amphiphilic molecules.

Key thermodynamic facts:

Critical micelle concentration (CMC) for phospholipids: $\sim 10^{-10}$ M
Bilayer formation is entropically driven (releases ordered water from hydrophobic surfaces)
Once formed, bilayers spontaneously close into vesicles (no free edges)
Membrane maintains $\sim$ 70 mV potential difference across 5 nm $\Rightarrow$ field strength $\sim 10^7$ V/m

This exemplifies emergent boundary formation: arising spontaneously, creating inside/outside distinction, actively maintained, enabling gradients.

Historical Context

The recognition that membranes self-assemble was a key insight linking physics to biology:

1925: Gorter \& Grendel estimate bilayer structure from lipid/surface-area ratio
1935: Danielli \& Davson propose protein-lipid sandwich model
1972: Singer \& Nicolson’s fluid mosaic model (still current)
1970s–80s: Lipid vesicle (liposome) research shows spontaneous membrane formation

The membrane is the minimal instance of “self” in biology: a dissipative structure that creates the inside/outside distinction necessary for all subsequent organization.

Boundaries appear because they stabilize coarse-grained state variables. The emergence of bounded systems—entities with an inside and an outside—is a generic feature of driven nonlinear systems, not a special case requiring explanation.