A quantum opinion-dynamics model reduces to the classical Friedkin–Johnsen model under a product-state approximation

A 2026 preprint by Weiqi Chu (UMass Amherst) models opinion dynamics on social networks by representing each agent's cognitive state as a density matrix rather than a scalar opinion value. In this formalism, survey questions become non-commuting self-adjoint operators, so the outcome of measuring an agent's attitude can depend on which question was measured first — a principled account of the empirically observed question-order effects that the density-matrix representation is built to capture, along with "cognitive ambivalence" (an agent simultaneously holding conflicting views).

The load-bearing structural result is a reduction: under a product-state approximation (agents treated as unentangled), the quantum model collapses back to the classical Friedkin–Johnsen opinion model, a well-established network-science formalism in which agents update toward neighbors' opinions while retaining a fixed attachment to their initial "prejudice." The quantum machinery is therefore an extension layered on top of a recognized special case, not a replacement for it. The model was simulated on the Facebook-100 network dataset (the Caltech friendship graph).

This "general formalism recovers an established model as a limiting case" shape recurs across the vault — cf. the way backpropagation is a special case of reverse-mode automatic differentiation. The reduction is what makes the quantum claim disciplined rather than free-floating, but it also raises the open question of whether the model predicts anything Friedkin–Johnsen does not (see question-chu-quantum-opinion-model-falsifiable-beyond-friedkin-johnsen).

The quantum probability here is a mathematical formalism for order-dependence, not a claim that brains are physically quantum — a distinction worth holding against physical quantum-mind proposals such as Faggin's argument that classical silicon cannot be conscious.