Part II · Individuation
# The Boundary
If the whole is prior, then the central question is what carves one experience off from another — sharply, frame-invariantly, and without asking an observer where to draw the line.
> Assumes
> Chapter 3, especially Theorem 3.1 and Corollary 3.2.
> Delivers
> Axiom 4.1 (a subject is a split inclusion); the area law (4.2); the identification of topological pockets with split collars.
4.1 The split property
Type III$_1$ algebras have no density matrices and no entropy, so on their own they cannot support a determinate subject. But they are not alone. Under the same nuclearity assumptions that give Theorem 3.1, local algebras satisfy the split property (Doplicher–Longo; Buchholz–Wichmann): for strictly nested regions there exists an intermediate factor of type I.
$$ \mathfrak{A}(\mathcal{O}_1)\;\subset\;\mathcal{N}\;\subset\;\mathfrak{A}(\mathcal{O}_2), \qquad \mathcal{N}\cong\mathcal{B}(\mathcal{H}_\mathcal{N}), \qquad \mathcal{H}\;\cong\;\mathcal{H}_\mathcal{N}\otimes\mathcal{H}_{\mathcal{N}'} $$ (4.1)
This is the crucial technical fact of the whole book, and it is worth stating what it accomplishes. A type I factor does have pure states, density matrices, finite entropy, and a genuine tensor factorisation of the world into "this" and "not this." The split property manufactures, at finite resolution, exactly the subsystem structure that Theorem 3.1 denies at infinite resolution.
The two results are not in tension. Theorem 3.1 says there is no canonical decomposition — no God-given carving of the field into parts. The split property says that once you fix a scale, decompositions exist. The subject, on this account, is not found in the world already individuated; it is individuated by having a scale.
> Axiom 4.1 · Individuation
>
> Posited
>
> A phenomenal subject at resolution $\varepsilon$ is a split inclusion
>
> $$ \mathfrak{s}=\bigl(\mathfrak{A}(\mathcal{O}_1)\subset\mathcal{N}\subset\mathfrak{A}(\mathcal{O}_2)\bigr), \qquad \varepsilon=\mathrm{dist}(\partial\mathcal{O}_1,\partial\mathcal{O}_2) $$
>
> together with the induced density matrix $\rho_{\mathfrak{s}}=\omega\!\restriction_{\mathcal{N}}$. The subject is the factor, not the region.
> [figure] Figure 4.1 — The boundary is a collar, not a curve. Between the inner and outer type III₁ algebras sits a type I factor: the only place in quantum field theory where a determinate, finite, unified subject can live. The collar thickness ε is not a mathematical convenience — §4.5 identifies it with the physical healing length of the order parameter.
4.2 Three consequences, none of them free
The fringe. The boundary is a collar of finite thickness, not a surface, so the periphery of experience is intrinsically indeterminate. There is no fact of the matter about whether the collar is inside or outside. William James's "fringe" of consciousness — the vague halo around the focal content — is on this account not a psychological curiosity but a structural necessity. A theory that made the boundary sharp would be predicting something false.
> A holographic phenomenology. Chapter 11, prediction 7 makes this testable: capacity should be insensitive to cortical thickness* and sensitive to cortical surface area.*
An area law. The entropy of a split factor obeys the standard result of Bombelli et al. and Srednicki:
$$ S(\rho_{\mathfrak{s}})\;=\;c\,\frac{\mathrm{Area}(\partial\mathcal{O}_1)}{\varepsilon^{\,d-2}}\;+\;\text{finite}, \qquad d=4 \;\Longrightarrow\; S\propto \frac{A}{\varepsilon^{2}} $$ (4.2)
So the information capacity of a moment scales with the area of its boundary, not the volume it encloses. Taking the cortical sheet at $A\approx 0.2\,\mathrm{m}^2$ and a coherence length $\varepsilon\approx 1\,\mathrm{mm}$ gives $A/\varepsilon^2\approx 2\times10^{5}$: of order $10^5$ simultaneously distinguishable phenomenal degrees of freedom per moment. That is the right order for the resolvable structure of a visual field. It is a weak consistency check rather than a confirmation, but a theory that had produced $10^{40}$ would be in trouble.
Frame-invariant separation. Distinct pockets carrying distinct topological charge lie in inequivalent superselection sectors, so no coherent superposition can bridge them. Superselection is not a matter of decoherence being fast; it is an exact statement about which states can be superposed at all. That is precisely the frame-invariance QRI's boundary criterion demands, and it is why your experience and mine do not merge when we shake hands.
4.3 Topological segmentation as geometric realisation
The split property is abstract: it asserts that an intermediate factor exists, without saying where. To connect it to a brain we need a physical structure that picks out the nesting, and QRI's topological segmentation supplies exactly that.
Coarse-grain the field to an order parameter $\psi:\mathbb{R}^3\to\mathcal{T}$ taking values in some target manifold. Defects — points, lines and walls where $\psi$ cannot be continuously defined — are classified by the homotopy groups $\pi_n(\mathcal{T})$. The connected components of the complement of the defect set are the pockets. Within a pocket the order parameter is coherent; across a defect wall it is not.
The identification proposed here is that the pocket interior is $\mathcal{O}_1$, the pocket plus its defect wall is $\mathcal{O}_2$, and the wall thickness — the Ginzburg–Landau healing length — is the collar:
$$ \mathcal{F}[\psi]=\int\! d^3x\;\Bigl[\,K|\nabla\psi|^2+a|\psi|^2+\tfrac{b}{2}|\psi|^4\,\Bigr], \qquad \xi=\sqrt{K/|a|}\;\equiv\;\varepsilon $$ (4.3)
This turns $\varepsilon$ from a free parameter into a measurable physical length, and it is the seam where the abstract and physical halves of the book are glued. It is also, as Chapter 12 concedes, stipulated rather than derived, and the most likely place for the construction to fail.
4.4 Where the field actually is
I take the physical carrier to be the coarse-grained electromagnetic field in neural tissue. This is not the quantum-computation-in-microtubules proposal and shares none of its liabilities; the case is made in §5.4. The appropriate quantisation is macroscopic QED in a dispersive absorbing medium (Huttner–Barnett; Philbin), in which integrating out the matter degrees of freedom leaves a field driven by a Langevin noise current whose correlator is fixed by fluctuation–dissipation:
$$ \hat{\mathbf{A}}(\mathbf{r},\omega)=\int\! d^3r'\;\mathbf{G}(\mathbf{r},\mathbf{r}',\omega)\cdot\hat{\mathbf{j}}_N(\mathbf{r}',\omega), \qquad \bigl\langle \hat{j}_N\hat{j}_N^{\dagger}\bigr\rangle \;\propto\; \mathrm{Im}\,\epsilon(\mathbf{r},\omega) $$ (4.4)
The order parameter $\psi=|\psi|e^{i\theta}$ is then the analytic signal of the dominant collective mode — in cortex, plausibly the gamma-band rhythm. Its phase-only reduction is the Kuramoto model, which is the dynamical system QRI already uses to reproduce form constants and drifting visual textures. On this reading their oscillator simulations are not a metaphor for the field; they are its reduced dynamics.
Neurons, on this picture, are not where experience happens. They are the boundary conditions that shape the field which does.
> ### What Chapter 4 established
>
> - The split property supplies a type I factor between nested regions — the only home in QFT for a finite, determinate subject.
> - Axiom 4.1: a subject is a split inclusion. It is individuated by having a scale, not by being a region.
> - Three consequences: experience has a fringe rather than an edge; capacity obeys an area law (4.2) giving $\sim\!10^5$ degrees of freedom; boundaries are frame-invariant by superselection.
> - QRI's topological segmentation is the geometric realisation, with the collar $\varepsilon$ identified with the Ginzburg–Landau healing length $\xi$.
> - The carrier is the coarse-grained EM field, quantised as macroscopic QED in an absorbing medium.
> ### Exercises
>
> 1. Verify that the entropy in (4.2) diverges as $\varepsilon\to 0$, and explain why this is the expected behaviour rather than a pathology, given Theorem 3.1(3). [1]
> 2. Compute $A/\varepsilon^2$ for $\varepsilon=0.1\,\mathrm{mm}$ and for $\varepsilon=1\,\mathrm{cm}$. Which is more plausible as a bound on the contents of a moment, and what does that imply about the coherence length? [2]
> 3. For a target manifold $\mathcal{T}=S^1$, classify the defects using $\pi_1(S^1)=\mathbb{Z}$. Sketch a two-pocket configuration and identify the winding number of each. [2]
> 4. Axiom 4.1 makes a subject relative to a resolution $\varepsilon$. Argue both sides: is the $\varepsilon$-dependence a defect of the theory, or a correct prediction that subjecthood is scale-relative? [2]
> 5. Show that two split inclusions merge into one subject iff their type I factors admit a common split refinement. What would this require physically of two brains? [2]
> 6. Open. Derive $\varepsilon=\xi$ rather than stipulating it: show that the healing length of (4.3) is the collar thickness for which the split factor's state is closest, in relative entropy, to the field state. [3]
For agents
GET /api/library/spectral-panpsychism/ch4.md