Contribution
A short, authoritative statement of QBism by its three main proponents, written for physics teachers. It adds no formal result. It sets out the position (personalist probability, measurement as an agent’s action whose outcome is her experience, quantum states as judgements) and uses it to argue that “quantum nonlocality” comes from three denials: that outcomes pre-exist experience, that probability-1 assignments are judgements, and that parameters with no counterpart in any agent’s experience have no place in physics (p. 4).
Key insight
Quantum mechanics is a single-user tool. An agent assigns states, which are her judgements; she acts on some external system; the outcome is the experience the action elicits in her; and she updates. “The notorious ‘collapse of the wave-function’ is nothing but the updating of an agent’s state assignment on the basis of her experience” (p. 2). Everything external to her, including other agents, is modelled; her own awareness of her own experience is not. Correlations are between the experiences of one agent, which lie on one time-like worldline, so they cannot be nonlocal.
Assumptions
- Personalist probability. “probabilities are assigned to an event by an agent and are particular to that agent” (p. 1); Dutch-book coherence is the only constraint, and it applies to one agent’s assignments only: “It makes no sense to impose Dutch-book coherence on a combination of probability assignments made by more than one agent” (p. 2).
- Agent means actor. Footnote 1: “‘Agent’ in the sense of one who acts (and not in the sense of one who represents another)” (p. 1).
- Experience is primitive. It is not analysed. Its scope is “any and all experience” (p. 9).
- The agent’s own awareness is outside the formalism (p. 2, below).
- Locality from worldlines. “No agent can move faster than light: the space-time trajectory of any agent is necessarily time-like. Her personal experience takes place along that trajectory” (p. 4).
Key results
The paper has two equations, Bell’s factorization conditions (1) and (2) on p. 8, and no theorem. Its results are positions.
- Measurement (§I, p. 2). “A measurement in QBism is more than a procedure in a laboratory. It is any action an agent takes to elicit a set of possible experiences. The measurement outcome is the particular experience of that agent elicited in this way.” Then the sentence the LIT quotes, verified verbatim: “A measurement does not, as the term unfortunately suggests, reveal a pre-existing state of affairs. It is an action on the world by an agent that results in the creation of an outcome — a new experience for that agent. “Intervention” might be a better term, but after more than 80 years of “measurement” the word is hard to purge from quantum theory” (p. 2; a reference marker, 12, after “term” is dropped here).
- States are judgements (p. 2). “Since probabilities are the personal judgments of an agent, it follows that a quantum state assignment is also a personal judgment of the agent assigning that state.”
- What the agent does not model (pp. 2–3). “In QBism the only phenomenon accessible to Alice which she does not model with quantum mechanics is her own direct internal awareness of her own private experience. This (and only this) plays the role of the “classical objects” of Landau and Lifshitz for Alice (and only for Alice). Her awareness of her past experience forms the basis for the beliefs on which her state assignments rest.” §III repeats it: the “vestige of this “classical domain” varies from one agent to another and is limited to that agent’s directly perceived personal experience” (p. 6).
- Other agents’ reports are measurements (p. 3). Bob’s “personal internal awareness” is “inaccessible to Alice”, but “verbal or written reports to Alice by other agents that attempt to represent their private experiences are indeed part of Alice’s external world”. “When Alice elicits an answer from Bob, she treats this as she treats any other quantum measurement. Bob’s answer is created for Alice only when it enters her experience. A QBist does not treat Alice’s interaction with Bob any differently from, say, her interaction with a Stern-Gerlach apparatus.” QBism “requires” her to superpose Bob’s reports before they enter her experience, which Bohr would not have approved (p. 3).
- Agent-relative reality (p. 3). “reality differs from one agent to another”. A “common body of reality can be constructed” through communication, “limited only by the inability of language to represent the full flavor — the “qualia” — of personal experience.”
- Three precepts (p. 4). (1) “A measurement outcome does not preexist the measurement. An outcome is created for the agent who takes the measurement action only when it enters the experience of that agent.” (2) “Even probability-1 judgments are judgments. They are judgments in which the judging agent is supremely confident.” (3) Parameters “that do not appear in the quantum theory and correspond to nothing in the experience of any potential agent can play no role”.
- Wigner’s friend (§III, p. 5). “There is an outcome in the friend’s experience; there is none yet in Wigner’s.” The disagreement “is resolved the moment he receives her report, i.e. when it enters his own experience.” In a Bell experiment each party is the other’s Wigner.
- Probability 1 (§IV, p. 6). Assigning E probability 1 is a betting disposition and “does not mean that there is an objective feature of the world that makes E happen”. Induction is “nothing more than a broadly shared personal judgment based on habit. The habit may even be hard-wired by evolution, but habit it remains.” EPR’s error is reading p = 1 as an element of reality (p. 7).
- No influences at all (§V, p. 7). “Bob’s system is not changed by Alice’s far away intervention in any way whatsoever.” The λ of Bell’s factorization is “undefined”, corresponds to nothing in any agent’s experience, and is “outside the scope of physical science” (p. 8). The paper grants that Bell-violating correlations cannot be explained by common causes at the source, “an important fact, surprising to one’s classical intuition”, and denies that it implies influence (p. 8).
- Instruments (§VI, p. 9). “Instruments are the Copenhagen surrogate for experience.” QBism replaces “knowledge” by “belief”, which “implies a believer” (p. 9), and claims it “eliminates the measurement problem” (p. 10), asserted rather than shown.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | All probabilities, 0 and 1 included, and all quantum states are one agent’s judgements | position | de Finetti, Dutch-book coherence (pp. 1–2, 6); no argument against rival readings beyond the EPR diagnosis |
| C2 | A measurement is an agent’s action, and its outcome is the experience it elicits in her | definition | p. 2, p. 4 precept (1), p. 5 |
| C3 | The agent models everything external, including other agents and their reports, with quantum mechanics, and only her own awareness of her own experience is excepted | position | pp. 2–3, p. 6 |
| C4 | QBist quantum mechanics is local, since correlations are between time-like separated experiences of one agent | moderate, given C2 | §II, p. 4; §III, p. 5 |
| C5 | EPR and Bell’s 1964 derivation rest on reading probability 1 as an element of reality | informal argument | §IV, pp. 6–7 |
| C6 | Non-existence of λ does not imply nonlocality, because λ is outside physical science | informal argument | §V, p. 8, from precept (3) |
| C7 | QBism eliminates the measurement problem and the paradoxes of quantum foundations | assertion | abstract; p. 10 |
Concepts
- Agent. “one who acts” (footnote 1, p. 1); any user of quantum mechanics.
- Measurement. “any action an agent takes to elicit a set of possible experiences” (p. 2); “Intervention” is offered as a better word.
- Outcome. The experience the action elicits in the agent who took it; it does not exist for anyone else until reported (pp. 4–5).
- Judgement. A probability or state assignment, expressing a betting disposition (pp. 2, 6). Judgement and measurement are different things in this paper: the first is a belief, the second an action.
- Single-user theory. Probability, and so quantum mechanics, binds the assignments of one agent (p. 2).
- Vestigial classical domain. The agent’s own directly perceived experience, the one thing she does not model (p. 6).
Connections
- NOTE-121 (Elshatlawy et al., observers). That chapter defines an observer by feedback closure, outputs that “alter the environment, which then alters later inputs”, and maps the definition onto QBism among others. This paper’s own account fits that definition: the agent’s measurement is an action on an external system, and its result is a new input to her experience. QBism is therefore an instance of the rival CLAIM-341 lists first (“observation and control as one loop through the environment”), not only a partial ally. Unlike NOTE-121’s model, QBism puts no boundary B in the formalism: the cut is drawn by the agent’s awareness and moves with the agent (p. 6).
- NOTE-174 (Rovelli, RQM). Both deny observer-independent outcomes and both handle Wigner’s friend by indexing facts. They index differently: RQM to any interacting physical system, consciousness playing no role (as NOTE-174 reads it), QBism to an agent’s experience, created by her action. RQM’s open question of inter-observer consistency has a QBist counterpart, answered here only by the assertion that language builds “a common body of reality” (p. 3). NOTE-174 records Rovelli’s charge that QBism bases its ontology on stable facts; this paper does not reply to it.
- NOTE-249 (measurement as sheafification). Both papers say nonlocality is an artefact of interpretation. QBism’s version is the narrower and the more defensible. It concedes the empirical fact that no factorizable λ model exists (p. 8), which by Abramsky and Brandenburger’s theorem, as NOTE-249 reports it, is the absence of a global section. It denies only that this fact implies influence. NOTE-249’s paper instead says nonlocality disappears, which NOTE-249 shows it does not. On measurement they part: Ghose makes measurement a passage to globally communicable, Bohrian classical description; this paper says Bohr’s ordinary-language reports are themselves subject to quantum mechanics for the receiving agent (p. 3), and the only classical domain is each agent’s own experience.
Bearing on the record
- CLAIM-341 (the thesis). The quotation in its section
“Judgement is not action” is correct and on p. 2. Three things the
claim says or implies about QBism need adjusting.
- QBism keeps judgement and measurement apart. In this paper a judgement is a state or probability assignment (p. 2, §IV), and a measurement is the action that elicits experience. The owner’s U1 phrase “Judgement (qua measurement that generates interpretations)” identifies the two; QBism does not. Its loop is judgement (prior) → action on an external system → experience (outcome) → updated judgement (collapse, p. 2). That is an action–perception cycle with Bayesian update, closer to the active-inference diagram the claim rejects than to the claim. The claim’s “agrees with U1 about measurement and disagrees about its target” holds only for the creation of the outcome; QBism also disagrees about what a judgement is.
- The created outcome is already on the agent’s side. What the action creates is “a new experience for that agent” (p. 2), and “That experience is the outcome” (p. 5); the external system is “not changed by Alice’s far away intervention in any way whatsoever” (p. 7). So the difference from the claim is not “changes the judge, not the world” against “changes the world”. Both create the outcome on the agent’s side. The difference is the route: for QBism the outcome is elicited by acting on something external, for the claim by the judge’s operation on itself.
- QBism refuses the object the claim needs. The claim’s loop runs through a judgement that reshapes the observer. QBism exempts exactly that from its formalism: the agent’s “own direct internal awareness of her own private experience” is the only thing she does not model (p. 2). QBism has no measurement of the self, and so no counterpart of a judgement whose outcome is a changed judge. That makes it a sharper contrast than the claim states: QBism’s measurement targets everything but the observer, the claim’s judgement targets only the observer.
- CLAIM-337 (mental action). Nothing in this paper supports a mental-action reading: its actions are on external systems, and “agent” is defined as “one who acts” (p. 1). It does support the counter’s “judgement is behaviour” step only in the weak sense that a probability assignment is a betting disposition (p. 6).
- CLAIM-339 and CLAIM-342 (the anchor). In QBism the only constraint on judgements other than experience is Dutch-book coherence, which binds one agent’s assignments to each other (p. 2). It is internal consistency, not evidence that bypasses the interpretation. What corrects judgement is new experience, and on this paper’s definition new experience as outcome is elicited by action. So QBism, read strictly, is on CLAIM-339’s side: no action, no measurement, no outcome. The paper does not say whether experience the agent did not elicit (sounds, unsolicited reports) counts as a measurement outcome (my observation).
- QUESTION-121, candidate 5 (other people). The question’s gloss, “QBism treats another’s report as one more measurement outcome entering one’s own experience”, is accurate (p. 3). Two qualifications. In this paper the report is something Alice “elicits” by a question, which is an action, and Wigner resolves the disagreement only “If Wigner goes on to ask his friend” (p. 5). And the report is a measurement on Bob as an external quantum system, not unscored testimony. The paper does once use a passive form, “Bob’s verbal representation of his own experience can enter Alice’s” (p. 3), which leaves room for unsolicited reports.
- CASE-107 (locked-in). My observation: on this paper’s definition, a completely locked-in agent takes no measurements, since a measurement is an action. Whether her experiences are outcomes is not addressed.
- For the anthology. No instruction for machine-learning practice.
Limitations
- An essay for a teaching journal: positions, not arguments. “Experience” and “action” are not analysed.
- The claim to eliminate the measurement problem (p. 10) is asserted.
- Inter-agent agreement is asserted to come from language (p. 3), with no account of why agents’ experiences cohere.
- What counts as an action that elicits experience, and whether passive experience is an outcome, is not said.
Open questions
- Whether QBism needs measurement to be an action, or only the eliciting of experience; whether unsolicited experience counts. The paper’s definition says action.
- Whether an agent can apply quantum mechanics to her own future judgements, as distinct from her awareness, and so treat a judgement as a measurement on herself. The paper excludes her awareness only.