Joint Realism in Quantum & Information Science

• Joint realism is a doctrine exploring whether observables possess simultaneous, measurement-independent values in both quantum and classical contexts.
• The concept emphasizes contextuality, showing that assigning definite properties across incompatible measurements contradicts standard quantum mechanics.
• In information theory, joint realism requires exact reproduction of joint laws, impacting lossy source coding and the design of reconstruction methods.

Joint realism is a family of doctrines about whether multiple physical quantities can be treated as simultaneously real. In quantum foundations, its standard classical form is the claim that observables possess jointly well-defined, measurement-independent values, including for incompatible quantities; several recent treatments instead formulate “joint reality” operationally, through invariance under sequential unrevealed measurements, or geometrically, through the existence of simultaneous value assignments for two observables. Outside foundations, the same phrase has acquired a distinct information-theoretic meaning: in lossy source coding with side information, joint realism is the exact constraint $P_{Y^n,Z^n}\equiv p_{X,Z}^{\otimes n}$, requiring the reconstruction and side information together to have the same law as source and side information (Karakostas, 2012, Caetano et al., 2024, Hamdi et al., 20 Jul 2025).

1. Classical joint realism and its quantum target

In the foundational literature, the classical form of joint realism is tied to three assumptions emphasized in philosophical analyses of quantum mechanics: the Separability Principle, the definite values principle, and non-contextuality. Karakostas formulates the classical picture as one in which spatiotemporally separated subsystems have individually well-defined states, every physical quantity has a definite value at each time, and a property is possessed independently of how it is measured. In this setting, strong joint realism is the view that all observables have their values simultaneously and independently of measurement context (Karakostas, 2012).

The same package appears in Bell- and EPR-style discussions under nearby labels such as value definiteness, counterfactual definiteness, or simultaneous elements of reality. Zwirn defines a realist theory as one “where it is meaningful to assign a property to a system independently of whether the measurement of this property is carried out,” and reconstructs EPR reasoning as a move from perfect predictability to the claim that incompatible observables would have to be “simultaneous elements of reality.” Dieks likewise treats “joint reality” as the assumption that two observables can possess simultaneous definite values, even when quantum theory does not allow them to be jointly measured (Zwirn, 2018, Hall et al., 2019).

Several papers therefore use “joint realism” not as a generic synonym for realism, but as a specific thesis about simultaneous definiteness. A narrower but closely related formalization appears in recent work on Bell scenarios: a locally realistic theory is taken to provide one common sample space $(\Omega,\mu)$ from which the outcomes of $Q,R,S,T$ are all simultaneously defined, so that each $\omega\in\Omega$ carries values $Q(\omega),R(\omega),S(\omega),T(\omega)$ and the observed pairwise distributions arise as marginals. That formulation is mathematically close to a global joint distribution over all relevant outcomes (Fullwood, 26 Feb 2025).

2. Why standard quantum mechanics rejects the classical form

A recurring conclusion across the cited literature is that standard quantum mechanics does not support unrestricted, noncontextual joint realism. Karakostas states the point explicitly: “One cannot assign, in a consistent manner, definite sharp values to all quantum mechanical observables pertaining to a microscopic object, in particular to pairs of incompatible observables, independently of the measurement context actually specified,” and more generally “it is not possible, not even in principle, to assign to a quantum system non-contextual properties corresponding to all possible measurements” (Karakostas, 2012).

Contextuality supplies the decisive obstruction. Karakostas appeals directly to Kochen–Specker-type constraints, including the pattern

$[A,B]=0=[A,C],\qquad [B,C]\neq 0,$

to argue that even when an observable $A$ is compatible with each of two others, its value can depend on whether it is embedded in the $A$-$B$ or $A$-$(\Omega,\mu)$0 commuting context. Dieks presents the same lesson as a general consequence of contextuality: there is no globally valid, non-contextual value assignment for all observables, and hence no unrestricted observer-independent joint definiteness (Karakostas, 2012, Dieks, 2019).

This rejection is often illustrated with entangled states. Karakostas writes the spin-singlet as

$(\Omega,\mu)$1

and insists that “no spin component of either particle $(\Omega,\mu)$2 or particle $(\Omega,\mu)$3 exists in an actual form, possessing occurrent spin properties.” The singlet state is said to represent “the entanglement, the inseparable correlation of potentialities,” not “a catalogue of actual pre-existing values” for all spin observables. Zwirn makes the same point in EPR language: it is “not possible to consider that the value of the spin along any axis is already fixed as soon as $(\Omega,\mu)$4 and $(\Omega,\mu)$5 separate and before any measurement” (Karakostas, 2012, Zwirn, 2018).

What survives is typically a restricted claim. Karakostas allows jointly determinate properties within a single Boolean context of compatible observables; Dieks allows jointly real properties within one perspective or hyperplane; both deny any global valuation spanning incompatible contexts. This suggests a replacement of unrestricted joint realism by context-relative joint definiteness, not by simple anti-realism (Karakostas, 2012, Dieks, 2019).

3. EPR, Bell, local realism, and the geometry of two-observable reality

The EPR argument made joint realism philosophically salient by claiming that two noncommuting observables could both be elements of reality when distant perfect predictability and local causality are combined. Recent analyses reconstruct that argument in different ways. One line, due to Graft, dissents from orthodox Bell-test reasoning by denying that separated local measurements directly sample the entangled joint distribution $(\Omega,\mu)$6; it instead proposes that genuinely separated measurements access only local marginals and therefore should be modeled context by context. That proposal is explicitly weaker than Bell/Fine-style global joint realism, because it does not posit one grand sample space over incompatible settings (Graft, 2013).

A more standard formal route is the diagrammatic one. In the categorical formulation of local realism, local realism is equivalent to the existence of a common extension $(\Omega,\mu)$7 making a Bell-square diagram commute. Realism is encoded by the existence of simultaneous values for $(\Omega,\mu)$8 on $(\Omega,\mu)$9, while locality is encoded by equalities such as

$Q,R,S,T$0

which say that the local outcome does not depend on the distant setting. From this common-extension assumption, the usual CHSH inequality follows (Fullwood, 26 Feb 2025).

A complementary approach studies only two observables $Q,R,S,T$1 and $Q,R,S,T$2. Cabello and collaborators start from the hidden-frequency identity

$Q,R,S,T$3

and derive geometric no-go results for joint reality from the positivity of the joint counts $Q,R,S,T$4. Under locality this yields device-independent steering inequalities; under “operational completeness” it yields no-go results without locality or noncontextuality. For qubits, the geometry implies that any two noncommuting projective observables are incompatible with locality under joint reality, and likewise incompatible with operational completeness under joint reality (Hall et al., 2019).

Experimental work has also tested stronger realist programs that retain pre-existing properties while abandoning locality. Leggett-type non-local realistic theories assume that each emitted pair belongs to a subensemble with definite polarization properties obeying Malus’ law. The tested inequality,

$Q,R,S,T$5

was violated in the reported two-photon experiment, leading to the conclusion that a broad class of non-local realistic theories is incompatible with the observed correlations. The result does not eliminate every conceivable realism, but it does rule out an important family of theories that preserve an intuitively classical pre-existing polarization structure (0704.2529).

4. Operational and information-theoretic reformulations

A distinct modern development defines realism through the effect of unrevealed measurements. In the Bilobran–Angelo framework, a single observable $Q,R,S,T$6 is real for $Q,R,S,T$7 when

$Q,R,S,T$8

Building on this, recent work introduces a criterion of joint reality for two observables $Q,R,S,T$9 and $\omega\in\Omega$0: $\omega\in\Omega$1 with $\omega\in\Omega$2. The associated joint irreality is

$\omega\in\Omega$3

Under this criterion, quantum mechanics “generally prevents non-commuting observables from having joint elements of reality.” For maximally incompatible observables, $\omega\in\Omega$4; for the maximally mixed state, $\omega\in\Omega$5 for all $\omega\in\Omega$6 (Caetano et al., 2024).

A broader GPT-based treatment defines realism theory-independently. A property $\omega\in\Omega$7 is real in state $\omega\in\Omega$8 iff an unrevealed measurement of $\omega\in\Omega$9 leaves unchanged the probabilities of all properties $Q(\omega),R(\omega),S(\omega),T(\omega)$0: $Q(\omega),R(\omega),S(\omega),T(\omega)$1 The same paper remarks that if joint probabilities $Q(\omega),R(\omega),S(\omega),T(\omega)$2 exist for all $Q(\omega),R(\omega),S(\omega),T(\omega)$3, independent of measurement order, then this realism criterion follows. It also introduces robustness- and Kullback–Leibler-based irrealism quantifiers, but it does not itself define a separate multi-property “joint realism” measure (Fucci et al., 2024).

For continuous variables, Freire and Angelo extend the same operational idea by coarse-graining position and momentum. The irreality of $Q(\omega),R(\omega),S(\omega),T(\omega)$4 is

$Q(\omega),R(\omega),S(\omega),T(\omega)$5

and for Gaussian states they obtain

$Q(\omega),R(\omega),S(\omega),T(\omega)$6

hence

$Q(\omega),R(\omega),S(\omega),T(\omega)$7

Their conclusion is explicit: one can never prepare a pure state for which position and momentum are simultaneous elements of reality in this framework (1904.02490).

An indirect but experimentally oriented line studies the emergence of single-observable realism under weak monitoring. For Werner states, local monitoring with

$Q(\omega),R(\omega),S(\omega),T(\omega)$8

yields realism gain exactly equal to weak discord suppression,

$Q(\omega),R(\omega),S(\omega),T(\omega)$9

while the tradeoff

$[A,B]=0=[A,C],\qquad [B,C]\neq 0,$0

shows that full realism for incompatible observables cannot generally coexist. The paper is explicit that this is only an indirect contribution to joint realism, not a direct criterion of simultaneous value assignment (Lustosa et al., 19 Feb 2025).

5. Revised realist programs after the failure of the classical version

Once unrestricted joint realism is denied, several papers replace it with revised realist programs rather than with instrumentalism. Karakostas’s proposal is contextual realism: there is a mind-independent, nonseparable reality, but empirical reality is disclosed only through a context, a Boolean frame of co-measurable observables, effective disentanglement, and a Heisenberg cut. Objectivity survives because, “given a particular experimental context, concrete objects (structures of reality) have well-defined properties independently of our knowledge of them” (Karakostas, 2012).

Dieks develops a closely related perspectival realism for unitary quantum mechanics. Properties are real, but in entangled and relativistic situations they may be hyperplane-dependent and relational rather than monadic and globally jointly assignable. He therefore rejects non-perspectival joint reality while preserving definite outcomes within a single perspective. This is a reformulation, not an elimination, of realism (Dieks, 2019).

Other approaches revise realism more radically. Zwirn’s modified realism in Convivial Solipsism keeps “something” external to consciousness but rejects physical collapse, observer-independent unique outcomes, and any absolute catalogue of simultaneous values for incompatible observables. Fuchs’s participatory realism also rejects pre-existing observer-independent outcomes, including probability-1 “elements of reality,” while insisting that such views are “far from instances of instrumentalism or antirealism” and concern “the very nature of reality” (Zwirn, 2018, Fuchs, 2016).

A different kind of revision separates realism from ontology. Dorato and Laudisa argue that one can be realist about quantum mechanics without realism about the wave function, allowing realism about Bohmian particles, GRW flashes, or matter density while remaining instrumentalist about $[A,B]=0=[A,C],\qquad [B,C]\neq 0,$1. A later duality-based proposal extends this separation: realism should attach to a theory’s full formal structure, while ontology is tied to empirical contexts such as subsystem decompositions or near-classical limits. This suggests that whatever survives of “joint realism” may be structural or context-bound rather than a commitment to one unique ontology (Dorato et al., 2014, Dawid et al., 18 Dec 2025).

The most direct constructive attempt to recover objective joint realism is the many-copy proposal. There, the desired joint probabilities are required to have the form

$[A,B]=0=[A,C],\qquad [B,C]\neq 0,$2

with positive $[A,B]=0=[A,C],\qquad [B,C]\neq 0,$3, but the author argues that this is impossible in a single-copy Hilbert space for sharp incompatible observables. The proposed solution is to enlarge the ontology to many copies, assign different incompatible observables to different copies so they commute, and invoke weak inter-copy interactions that produce effective collapse to a single identical pointer state. The proposal is explicitly meant to be experimentally distinguishable through incomplete collapse in sequential measurements or deviations from the single-copy Born rule (Bednorz, 2018).

6. Joint realism in source coding and perceptual communication

In information theory, “joint realism” has a different and sharply formal meaning. For lossy source coding with side information $[A,B]=0=[A,C],\qquad [B,C]\neq 0,$4, marginal realism requires only

$[A,B]=0=[A,C],\qquad [B,C]\neq 0,$5

whereas joint realism requires

$[A,B]=0=[A,C],\qquad [B,C]\neq 0,$6

Near-perfect joint realism relaxes this to

$[A,B]=0=[A,C],\qquad [B,C]\neq 0,$7

The paper proves that perfect and near-perfect joint realism are asymptotically equivalent up to closure, so the exact matching formulation can be used without loss of generality (Hamdi et al., 20 Jul 2025).

The operational consequence is that joint realism is strictly stronger than matching the output law alone: it constrains the coupling with the actual side information. For side information available at both encoder and decoder, the achievable region is characterized by

$[A,B]=0=[A,C],\qquad [B,C]\neq 0,$8

with

$[A,B]=0=[A,C],\qquad [B,C]\neq 0,$9

For decoder-only side information the same rate bounds appear, but with the additional structural constraint $A$0. The paper’s central structural observation is the absence, under joint realism, of the entropy bonus $A$1 that appears under marginal realism: side information remains useful through conditional mutual information terms, but “does not seem to act as a source of common randomness” (Hamdi et al., 20 Jul 2025).

This coding-theoretic usage should be distinguished from the more general engineering use of realism as perceptual plausibility. In diffusion-aided joint source–channel coding for wireless image transmission, “high realism” refers to visually plausible reconstructions obtained by conditioning a pretrained Stable Diffusion model on noisy channel symbols, multimodal features, and SNR. That work is about perceptual realism, not about simultaneous value definiteness or exact joint-law matching, although it shares the broader shift from distortion fidelity toward realism-oriented objectives (Yang et al., 2024).

Across these literatures, joint realism is therefore not a single doctrine but a structured family of ideas. In quantum foundations it usually names the thesis of simultaneous, measurement-independent definiteness and is largely rejected in its classical, noncontextual form; what replaces it is contextual, perspectival, participatory, structural, or explicitly many-copy realism. In operational approaches it becomes a criterion of state invariance under unrevealed measurements, often yielding quantitative measures of joint irreality. In information theory it denotes exact reproduction of the joint source–side-information law. The continuity across these uses lies less in a shared formalism than in a shared question: what, exactly, must be preserved for multiple properties, outcomes, or signals to count as jointly real?