CALLIOPE · BCOM

Gregory Scholes and collaborators have shown that networks of classical phase oscillators can host states that look, formally, like qubits: superpositions, tensor-product registers, gates, and entangled'' joint states, all realized as the collective synchronization patterns of a Kuramoto-type system. This working paper is a self-contained, pedagogical reconstruction of that idea, built from the ground up. We show how two coupled clusters of oscillators reduce to a two-site tight-binding Hamiltonian whose symmetric/antisymmetric doublet is the qubit; how stacking qubits is the graph Cartesian product (separable, non-interacting) while entangling them is a non-separable, spectrum-preserving rewiring of the coupling graph; and how the whole construction admits a clean geometric reading as a flat-versus-twisted graph bundle, with entanglement as holonomy. We are careful about what is genuinely quantum and what is not: the Hilbert space, the register, and the gates are exact classical linear algebra on the oscillator field (classical entanglement''), and cannot by themselves violate a Bell or non-contextuality inequality, because the global oscillator configuration is a ready-made single (Kolmogorov) sample space. We close with the Kolmogorov-Theory (KT) reading: such networks are an explicit, physical substrate in which to study where genuine non-classicality could enter --- namely the measurement rule --- and what that means for algorithmic agents whose computation is a synchronization process.

Classical oscillator networks can mimic the full formal algebra of quantum bits — but the one thing that would make them genuinely quantum remains an open question.

The core idea is almost embarrassingly simple. Take two clusters of coupled oscillators — the kind used to model brain rhythms, chemical oscillators, or any network of coupled clocks. When they synchronize, they can lock in two distinct collective patterns: all oscillators swinging together (in-phase), or the two clusters swinging against each other (anti-phase). Call those two patterns |↑⟩ and |↓⟩. That's your qubit. The math that falls out — a 2×2 matrix with two isolated eigenvalues split by the coupling strength — is identical to the quantum mechanics of a particle in a double well, or an ammonia molecule tunneling between two configurations. The Hilbert space structure isn't imported by analogy; it emerges exactly from the graph's spectrum.

Building a multi-qubit register means stacking these cluster-pairs together. The paper is careful about a subtle point here: the natural way to combine two such graphs — the "Cartesian product" — produces independent, non-interacting qubits. The eigenvectors factorize, the eigenvalues add, and a product state stays a product state forever. This is the non-entangling combination, even though it physically connects the graphs. Entanglement requires something different: a rewiring of the inter-cluster connections that cannot be written as two independent operations. Scholes implements this as a spectrum-preserving similarity transformation — it rotates the eigenvectors into correlated, non-product "Bell-like" synchronization patterns while leaving the eigenvalue gaps completely intact. The paper gives this a clean geometric reading: the non-entangled register is a flat graph bundle (fibers attached uniformly along a base), while entanglement is a twist in that bundle — a holonomy, the gauge-theoretic measure of how much the fiber rotates as you traverse a loop in the base.

Here is where the paper draws its sharpest line. All of this — the Hilbert space, the tensor-product register, the gates, the "entangled" eigenvectors — is exact classical linear algebra. The global state of all the oscillators at any moment lives in a single, well-defined configuration space. Fine's theorem says that whenever such a single sample space exists, all joint measurement statistics must satisfy Bell inequalities. So this construction, however quantum it looks algebraically, cannot violate a Bell inequality by itself. The only door to genuine non-classicality is the measurement rule: how you read out which synchronization pattern the network locked into. That question is left open, because it is genuinely open.

For BCOM's broader program, the payoff is that these oscillator networks are not a new toy — they are the same Kuramoto-type models already used to study brain dynamics and agent behavior. The quantum-like construction identifies a specific spectral feature (an isolated emergent doublet, protected by a gap) within networks already being modeled. This makes the framework a concrete stress-test for the claim that mind and agency are fundamentally algorithmic rather than quantum: if synchronization-based computation is provably bounded by a single sample space, that is a precise statement about its classical ceiling, not just a philosophical position.