Relational and Agent-Centered Readings

Relational quantum mechanics and QBism both resist treating every quantum-state assignment as an observer-independent inventory of intrinsic properties. They differ in what an observer is, what an outcome means, and what quantum probabilities refer to. This comparison focuses on those differences; their full accounts belong to Relational Quantum Mechanics and QBism.

Relative events and personal probabilities

In Rovelli’s relational proposal, the relevant observer may be any interacting physical system. A variable takes a value relative to another system in an interaction. The description of a larger system can remain quantum relative to a third system. “Relative” here concerns physical relations; it does not mean that a conscious mind manufactures a freely chosen world.

In the QBist account of Fuchs, Mermin, and Schack, a quantum state expresses an agent’s probability judgments for the consequences of that agent’s actions. The Born rule is a normative constraint on those judgments. Outcomes are experiences for the agent. This is a stronger agent-centered commitment than the ordinary observation that different experimenters may possess different data.

Both approaches need an account of how stable records and communication support agreement. Neither permits choosing an outcome merely by assigning it high probability.

One calculation, different readings

Alice measures $Z$ on half of a Bell pair,

$$\lvert\Phi^+\rangle=\frac{\lvert00\rangle+\lvert11\rangle}{\sqrt2}.$$

Before learning her result, Bob assigns $\rho_B=I/2$. Given Alice’s result $a$, he can predict the matching result of a later $Z$ measurement using $\rho_{B|a}=\lvert a\rangle\langle a\rvert$. Summing over Alice’s outcomes restores $I/2$.

An ordinary operational account treats these as probabilities conditioned on different available records. A relational account can further index the occurrence of an event to the interacting systems. A QBist account interprets the probabilities themselves as the assigning agent’s commitments. The identical density-matrix algebra does not prove any one of those philosophical readings.

In particular, Alice’s update does not change Bob’s locally accessible unconditioned statistics. “State depends on observer” must not be silently replaced by “a controllable distant physical signal occurred.”

Where a genuine disagreement can arise

A useful comparison asks whether each account asserts unique absolute events, relative events, personal outcomes, or branch-relative records. It then asks what rule allows statements made in one description to enter another. Friend scenarios expose this issue because the alternative external measurement may coherently mix the internal record states.

RQM and QBism are not one general “subjective interpretation.” They assign different roles to physical interaction and to an agent. Nor should interpretive language replace the mathematical specification of preparations, interventions, and accessible records. When a proposal claims a novel empirical consequence, it must supply probability assignments that differ for the same operational protocol; otherwise the distinction remains interpretive.

Exercise

Alice and Bob use the same density matrix and the same POVM but disagree about what the state represents. Can that disagreement alone change their predicted outcome frequencies?

Solution

No. Both compute $p(a)=\operatorname{tr}(\rho E_a)$. Different meanings attached to $\rho$ do not alter this number. A predictive disagreement requires different states, effects, dynamics, or probability rules. The interpretive difference may still matter to explanation or ontology; empirical equivalence does not make those questions meaningless.

References

• Fuchs, C. A., Mermin, N. D., and Schack, R., “An introduction to QBism with an application to the locality of quantum mechanics,” American Journal of Physics 82, 749–754 (2014), arXiv:1311.5253 — QBism stated by its proponents.
• Rovelli, C., “Relational Quantum Mechanics,” International Journal of Theoretical Physics 35, 1637–1678 (1996), arXiv:quant-ph/9609002 — the original relational proposal.