Part I: Foundations

Thermodynamic Foundations

Driven Nonlinear Systems and the Emergence of Structure

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 because they channel the imposed gradients efficiently. Standard bifurcation theory supplies the mechanism: as a driving parameter exceeds a critical threshold, the uniform state loses stability through a bifurcation (pitchfork, Hopf, saddle-node, or a period-doubling cascade), giving rise to structured alternatives. The result is a system with multiple metastable attractors $\{\mathcal{A}_i\}$ exhibiting:

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)

Among possible structured states, those that persist tend to be those that efficiently dissipate the imposed gradients (2013). Not teleology—differential persistence. This is the thermodynamic foundation for organized structure: not forbidden but enabled.

Empirical Grounding

Bénard convection cells, the canonical laboratory case. Heat a thin fluid layer from below and behavior turns sharply on a threshold. Below Rayleigh number $\text{Ra}_c \approx 1708$ , heat moves by conduction alone: a uniform, unstructured state. Above it, spontaneous symmetry breaking produces hexagonal convection cells that transport heat more efficiently than conduction can. Exactly the predicted structure — a bifurcation at critical driving, multiple equivalent attractors (cells rotate either way), path-dependent selection.

Lipid bilayer self-assembly, the boundary case taken up in the next section. Critical micelle concentration for phospholipids is $\sim 10^{-10}$ M; bilayer formation is entropically driven, releasing ordered water from hydrophobic surfaces; bilayers spontaneously close into vesicles with no free edges; and the membrane sustains a $\sim$ 70 mV difference across 5 nm — a field of $\sim 10^7$ V/m. Arising spontaneously, creating an inside and an outside, actively maintained, enabling gradients that would otherwise equilibrate.

The Viability Manifold

One structure recurs 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 is central to normativity: trajectories that remain within $\viable$ are, in a precise sense, “good” for the system; trajectories approaching the boundary $\partial\viable$ are “bad.” Note what this is and is not. Distance and direction relative to $\partial\viable$ are an objective fact about the system — some trajectories preserve it, others destroy it. That is not yet preference, experience, or morality. It is the control problem from which those may later develop.

The construction is Aubin's viability theory (1991) with stochasticity added: a state is viable iff some trajectory from it remains in $\viable$ indefinitely, the viability kernel is the largest subset with that property, and in controlled systems viability requires the control to point inward at boundaries. The phenomenological hook, developed later, is that the felt sense of threat corresponds to proximity to $\partial\viable$ .

Boundary Formation

Among the dissipative structures that emerge, one class matters most: 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 enough amphiphilic molecules and energy, membranes form spontaneously—a low-free-energy configuration. Once formed, they separate internal chemical concentrations from external, enable maintenance of ion gradients and pH differences, provide a substrate for embedded machinery (channels, pumps, receptors), and must be actively maintained against degradation.

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

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