Part II · Individuation
# Why There Are No Micro-Subjects
Constitutive micropsychism asks how little experiences add up to a big one. Quantum field theory answers that there are no little ones to add.
> Assumes
> Chapter 2. Hilbert spaces and operators; no prior algebraic QFT is needed — §3.1 supplies it.
> Delivers
> Theorem 3.1 and Corollary 3.2. This is the central negative result of the book.
3.1 A crash course in factor types
A von Neumann algebra is an algebra of bounded operators on a Hilbert space, containing the identity, closed under adjoints and under limits in the weak operator topology. A factor is one whose centre is trivial — it contains no non-scalar operator commuting with everything else, which is the algebraic way of saying it describes a single indecomposable system.
Murray and von Neumann classified factors by the behaviour of their projections, and the classification is the whole reason this chapter exists. A projection is the algebraic stand-in for a yes/no property; the question is whether there are smallest ones.
| Type | Minimal projections | Trace values | Where it turns up |
| --- | --- | --- | --- |
| I$_n$, I$_\infty$ | Yes | $\{0,1,\dots,n\}$ | Ordinary quantum mechanics: $\mathcal{B}(\mathcal{H})$. Density matrices, pure states, entropy. |
| II$_1$ | No | $[0,1]$, continuous | Infinite spin chains at fixed density. |
| II$_\infty$ | No | $[0,\infty)$ | — |
| III | No | $\{0,\infty\}$ — no trace at all | Every local algebra in every relativistic QFT. |
The reader used to ordinary quantum mechanics has only ever met type I. In type I there are minimal projections — rank-one projectors onto pure states — and everything familiar follows from them: states are density matrices, entropy is $-\mathrm{Tr}\rho\ln\rho$, and a system decomposes into subsystems by tensor factorisation. In type III none of that holds, and it is worth pausing on how thoroughly it fails. In a type III factor every non-zero projection is equivalent to the identity: the algebraic "size" of a property in half a region equals that of the whole region. There is no trace, so no entropy. There are no minimal projections, so no smallest property. And there are no normal pure states.
Connes refined type III into a one-parameter family III$_\lambda$ with $\lambda\in[0,1]$, distinguished by the spectrum of the modular operator we will meet in Chapter 5. The case relevant to physics is the extreme one, III$_1$.
3.2 Local algebras are type III1
> Fredenhagen (1985); Buchholz, D'Antoni & Fredenhagen (1987). The result is robust: it needs only the Wightman axioms plus a nuclearity bound on the density of states, which any theory with sensible thermodynamics satisfies.
In algebraic quantum field theory one assigns to each bounded open region $\mathcal{O}$ of spacetime the algebra $\mathfrak{A}(\mathcal{O})$ generated by observables measurable within it. The assignment satisfies isotony (bigger regions, bigger algebras) and locality (spacelike-separated algebras commute). The structural theorem is that these algebras are, without exception, the same object.
> Theorem 3.1 · No atoms of experience
>
> Established
>
> For a double cone $\mathcal{O}$ in a Wightman theory satisfying the Buchholz–Wichmann nuclearity condition, $\mathfrak{A}(\mathcal{O})$ is the unique hyperfinite factor of type III$_1$. Consequently:
>
> 1. $\mathfrak{A}(\mathcal{O})$ has no minimal projections. There is no smallest localised property, hence no elementary bearer of experience.
> 2. $\mathfrak{A}(\mathcal{O})$ admits no normal pure states. Every local state is mixed, irreducibly and not for lack of knowledge.
> 3. The Hilbert space does not factorise: $\mathcal{H}\ncong\mathcal{H}_{\mathcal{O}}\otimes\mathcal{H}_{\mathcal{O}'}$, and the entanglement entropy across $\partial\mathcal{O}$ is ultraviolet-divergent.
> 4. All such algebras are isomorphic. A region the size of a proton and a region the size of a brain carry the same algebra.
Point 4 deserves a moment, because it is the one that most offends intuition. The algebra does not know how big the region is. All the physics of scale lives in the state, not in the algebra — which is precisely why, from Chapter 6 onward, every phenomenal quantity in this book is a functional of the state rather than of the region.
3.3 Reeh–Schlieder: locality of operations, non-locality of states
One further classical result sharpens the picture. The Reeh–Schlieder theorem says that for any bounded open $\mathcal{O}$ with non-empty causal complement, the vacuum vector $\Omega$ is cyclic for $\mathfrak{A}(\mathcal{O})$ — the set $\mathfrak{A}(\mathcal{O})\Omega$ is dense in the whole Hilbert space — and separating, meaning no non-zero element of the algebra annihilates it.
Cyclicity is startling: by acting only within a region the size of a grain of sand, one can approximate any global state of the universe to arbitrary accuracy, including states with a galaxy in them. This does not permit signalling, because the operations required are wildly non-unitary and succeed with vanishing probability, and because the expectation of any spacelike-separated observable is unchanged. What it does establish is that the vacuum is entangled across every region at every scale. There is no such thing as an unentangled patch of the world.
Separating, meanwhile, is the technical hypothesis that makes Chapter 5 possible: it is exactly the condition for Tomita–Takesaki theory to apply, and hence for the algebra to carry an intrinsic time.
3.4 The combination problem is malformed
Now the payoff. The standard and, I think, decisive objection to panpsychism is the combination problem: if the fundamental constituents of matter have micro-experiences, by what mechanism do a hundred billion of them compose into the single unified field of your present moment? No mechanism has ever been given, and the arguments of Chalmers and Seibt suggest that none can be, because the unity of a macro-experience is not the sort of thing that is built out of parts.
The objection has a premise, and the premise is that the world decomposes canonically into subsystems, each capable of carrying its own state. In non-relativistic quantum mechanics this is true and unremarkable: $\mathcal{H}=\mathcal{H}_A\otimes\mathcal{H}_B$, and the parts are perfectly real. In quantum field theory it is false, by Theorem 3.1(3).
So the situation is this. Micropsychism requires elementary subjects. Elementary subjects require minimal projections, or at minimum a canonical decomposition into independently-stated parts. Quantum field theory supplies neither. The combination problem is therefore not solved here — it is dissolved, because its subject matter does not exist.
> Corollary 3.2 · Panpsychism must be cosmopsychist
>
> Derived
>
> If the intrinsic aspect is total (Axiom 2.1) and the field admits no canonical decomposition (Theorem 3.1), then the phenomenal whole is prior to phenomenal parts. There is one field, in one state, with one intrinsic character. Individual subjects are not sums; they are quotients — structures carved out of a unity that was never assembled.
>
> Priority monism is thus not an additional metaphysical taste. It is what the operator algebras force, given Axiom 2.1.
I take this to be the strongest available argument for cosmopsychism, and its interest is that it argues from physics rather than from intuition. It also lands, without having aimed there, almost exactly on Gómez Emilsson's stated ontology: an unbroken unity of all things, topologically segmented into individuals. Chapter 4 supplies the segmentation.
One cost should be acknowledged now. Cosmopsychism inherits a mirror-image difficulty — the decomposition problem, of explaining how the one gives rise to the many — and it is not obviously easier than combination. The difference is that decomposition is a question about where the cuts are, and quantum field theory turns out to have a rather precise answer. That answer is the subject of the next chapter.
> ### What Chapter 3 established
>
> - Local algebras in QFT are type III$_1$ factors: no minimal projections, no pure states, no tensor factorisation, all isomorphic.
> - Theorem 3.1 therefore forbids elementary bearers of experience. Micropsychism is not false but ill-posed.
> - Reeh–Schlieder shows locality of operations coexists with total non-locality of states — no patch of the world is unentangled.
> - Corollary 3.2: panpsychism, given these facts, must be priority monist. The whole is prior; subjects are quotients.
> - The price is the decomposition problem, which Chapter 4 must now pay.
> ### Exercises
>
> 1. Exhibit a minimal projection in $\mathcal{B}(\mathcal{H})$ and verify it is minimal. Then explain in one sentence why no such operator exists in a type III factor. [1]
> 2. In a type III factor every non-zero projection is equivalent to the identity. Interpret this physically: what does it say about the "number of degrees of freedom" in half a region versus the whole? [2]
> 3. Reeh–Schlieder appears to license superluminal signalling. Show that it does not, by identifying which quantity an experimenter can actually control and which they cannot. [2]
> 4. Derive the ultraviolet divergence of entanglement entropy across $\partial\mathcal{O}$ from Theorem 3.1(3), rather than from a mode-counting argument. [2]
> 5. Chalmers distinguishes constitutive from emergent panpsychism. Show that Theorem 3.1 refutes the constitutive micro variety while leaving constitutive cosmopsychism untouched, and say why the asymmetry arises. [2]
> 6. Open. The decomposition problem for cosmopsychism: give an argument, independent of the split property of Chapter 4, that a unified phenomenal whole can have determinate proper parts at all. [3]
For agents
GET /api/library/spectral-panpsychism/ch3.md