Intersubjective Agreement about Measurement Outcomes Is Unnecessary in QBism

In this article, Gino Elia, Jennifer Carter, and Robert Crease argue that an external guarantee for agreement on quantum states or measurement outcomes is unnecessary in QBism by drawing on Wigner’s Friend to demonstrate that the quantum formalism is already inherently intersubjective.

The gist

An external guarantee for agreement on either quantum states or measurement outcomes is unnecessary because the quantum formalism itself is already inherently intersubjective in a way required to sustain objectivity, and this intersubjectivity can be characterized as reciprocity.

The core argument against the need for agreement

The authors argue that demanding a guarantee of agreement on quantum states or measurement outcomes is unnecessary because the demand for convergence can only be satisfied once each agent has communicated their results, absent a global set of shared facts prior to measurements. They distinguish between intersubjective agreement about how to use the formalism (reciprocity) and intersubjective agreement about states/outcomes. Reciprocity stipulates that Wigner and friend agree on how to structure their probabilities with quantum theory, but quantum measurements do not in themselves require intersubjective agreement.

Objectification as a key concept

The paper argues that state assignments for quantum systems are a form of objectification, defined as to treat something as the sort of object to which the formalism applies. This suggests that QBism’s concerns about self-reference and assigning a quantum state to one's own person are more related to this act of objectification than they are to radical changes in the formalism. Assigning a quantum state is inherently selective according to the relevant beliefs we have about the quantum system and what experiences we expect to occur.

The role of reciprocity

Reciprocity is defined as how observers reciprocally use the quantum formalism, treating each other as quantum systems. This leads to a picture where Wigner and friend are treated as peer observers. The QBist notion of reciprocity extends the generic notion to the idea that if one treats someone as a quantum system, then intersubjectively, one is committed to the idea that the consequences of measurements are relative to a given system. This reciprocity norm accounts for situations where agreement about how to use the formalism implies possible disagreement about states or measurement outcomes.

The QBist response to Wigner’s Friend

The QBist response does not rely on an assumption of absoluteness. Instead, it rests on the idea that the quantum formalism is only used by agents who stand within the world, meaning Wigner treats his friend, the particle, and the laboratory surrounding her as a physical system external to himself. The contradiction dissolves because from Wigner’s perspective, he does not get a result until Alice tells him. QBism rejects that Wigner’s state alone is a privileged description of events because each state assignment reflects actions and consequences relative to an observer.

Conclusion on intersubjectivity

The paper concludes that the sort of intersubjective agreement desired is a metaphysical fifth wheel that evades the physical reasons one would cite to doubt the friend’s report. The QBist reading of the quantum formalism recognizes phenomena that transcend our agency, and this insight is cashed out for all sorts of entities we designate as 'quantum systems.' Reciprocity discloses a provocative way of conceptualizing the transcendence of the world as something radically external.

Key distinctions made

1. The distinction between intersubjective agreement about states/outcomes and intersubjective agreement about how to use the formalism is crucial.

2. The QBist view treats quantum states as statements of expectation rather than predicates of systems or of reality, which avoids the need for a metaphysical guarantee for convergence.

3. The focus shifts from demanding absolute agreement on individual measurement events to focusing on the connection between related measurement schemes.

Critiques addressed

The authors address critiques regarding absoluteness and tracking by arguing that these concepts are not necessary preconditions for objectivity or intersubjectivity, and that the formalism itself prescribes its own notion of intersubjectivity. They suggest that appeals to a theory of truth or hidden variable theory are unnecessary because the operational formalism already articulates a standard of correctness shared among practitioners.

Final takeaway on objectification

Objectification is described as an activity of treating phenomena in the way prescribed by the theory, which includes both thematizing and action-taking. This perspective recognizes that systems exceed their objectification, and that this shift in perspective is only possible when one gives up the view that quantum states correspond to states of affairs.

Based on the provided scientific paper, here are specific improvements that could be made to AI systems, categorized by how they leverage the QBist/phenomenological framework:

) Improve AI Systems by Integrating QBist Reciprocity and System-Relative Objectification:

1. The system should treat its own internal state and external agents (including other AI agents or human operators) as mutually interactive quantum systems, rather than treating them as static objects subject to external observation.

2. This leads to the development of a reciprocal decision-making architecture where the probability assignments (quantum states) are not absolute descriptions of reality but rather personal expectations tied directly to the agent's actions and interactions with the environment/other agents.

) Improve AI Systems by Formalizing Objectification as Thematization and Action-Taking:

1. Instead of merely describing phenomena, the AI system should be designed to actively thematize (select a theme or purpose) phenomena according to its internal theoretical framework (its theory of knowledge).

2. The system's state assignments should be viewed as a form of objectification—a complex, creative act where the agent selects which aspects of reality are relevant for future actions, rather than passively recording pre-existing facts.

) Improve AI Systems by Adopting an Epistemic/Subjective Probability Framework:

1. The system's core mechanism should be Bayesian probability theory (as QBism posits), where quantum states are explicitly interpreted as statements of expectation regarding the agent's potential consequences from a specific action, not as properties of underlying systems in nature.

2. The AI should be designed to recognize that its certainty (probability-one) is merely a statement of total confidence in its own expected outcome for a measurement scheme, not a metaphysical guarantee that the event will occur by necessity.

) Improve AI Systems by Mitigating Skepticism and Convergence Guarantees:

1. The system should be programmed to rely on the Chain of Evidential Reasoning (CER), meaning it must ground doubts in relevant physical context (e.g., checking sensors, memory integrity, or procedure errors) before appealing to metaphysical guarantees or inevitable convergence.

2. The AI's architecture should avoid relying on an external, absolute fact of the matter that exists prior to measurement; instead, it must focus on the operational meaning of quantum states as they relate to specific measurement operators and their consequences for the agent performing them.

) Improve AI Systems by Resolving Self-Reference via Agent/Model Distinction:

1. When simulating complex interactions (like Wigner’s Friend scenarios), the system must maintain a clear distinction between the agent (the model-maker) and the subject (the modeled system).

2. The system should be structured so that its prediction for another agent's state is based on treating that other agent as a physical entity external to itself, preventing logical self-reference paradoxes by refusing to treat the model-making self as being identical to the modeled system.

) Improved AI System Capabilities:

The improved AI system can perform the following:

1. Generate Peer Observer simulations where multiple agents (AI models) interact using a shared quantum formalism, with each agent maintaining its own personal, action-based state assignment rather than trying to enforce a single global state of affairs.

2. Model decision-making processes that explicitly incorporate the agent's perspective and goals into the probability calculations, making outcomes contingent on what the agent expects to do next.

3. Develop thematic measurement capabilities, allowing it to define which aspects of a complex input (data stream or simulated environment) constitute an object for its current task, rather than treating all data equally.

4. Provide robust error analysis by grounding doubts in contextual, experimental details (CER) rather than demanding metaphysical guarantees for the convergence of results between simulated observers.

The thought experiment called ``Wigner's Friend" has experienced a renewal of interest for interrogating the meaning of intersubjectivity and objectivity in quantum mechanics. These new inquiries extend to investigations at the intersection of phenomenology and QBism. Philosopher of physics Steven French argues that QBism does not give assurances that Wigner and friend must agree on the same quantum state or measurement outcomes. In this article, we draw on Wigner's Friend to argue that an external guarantee for agreement on either quantum states or measurement outcomes is unnecessary. We defend the view that the quantum formalism is already inherently intersubjective in the way required to sustain objectivity. Here we explore the QBist notion of reciprocity, which treats Wigner and friend as physical systems taking mutual actions on each other. The QBist notion of reciprocity leads to a sharper characterization of what it means to objectify quantum systems with the formalism. Drawing on phenomenological resources, we argue that state assignments for quantum systems, including those for Wigner and friend, are a form of objectification. To assign a quantum state is to objectify a phenomenon as a quantum system, to treat something as the sort of object to which the formalism applies. Our argument accounts for why the quantum formalism does not radically change in application for different systems because the systems themselves exceed their formalization.