Preparation Independence Postulate

• Preparation Independence Postulate (PIP) is a foundational principle asserting that independently prepared quantum systems have statistically independent ontic states.
• It enforces the factorization of joint ontic probability distributions, playing a key role in ψ-ontology debates and the PBR theorem.
• Recent studies examine weakened versions and counter-examples, highlighting implications for quantum locality, separability, and epistemic interpretations.

The Preparation Independence Postulate (PIP) is a core assumption in the foundations of quantum mechanics, particularly in ψ-ontology no-go theorems such as the Pusey–Barrett–Rudolph (PBR) theorem. PIP asserts that independently prepared quantum subsystems possess independently sampled values of underlying ontic variables: mathematically, it requires the factorization of the joint ontic probability distribution for independently prepared product states. The ontological model framework formalizes quantum theory in terms of measurable spaces of hidden variables (ontic states) and their associated probability measures. PIP sits at the heart of debates about the ontic vs. epistemic character of the quantum state and is tightly connected, both in form and in operational consequences, to postulates of locality, symmetry, and separability. Recent research has scrutinized the necessity, physical justification, and possible weakenings of PIP, examining its mathematical role in the transmission of overlap, no-signalling conditions, and the possibility of ψ-epistemic models.

1. Formal Definition and Role in Ontological Frameworks

In ontological (hidden-variable) models of quantum mechanics, each quantum preparation $|\psi\rangle$ induces a probability measure $\mu_{|\psi\rangle}(\lambda)$ over an ontic space $\Lambda$, and measurement responses are provided by conditional probability functions. The Preparation Independence Postulate is defined for composite systems $A,B$ each prepared in pure quantum states $|\psi_A\rangle$, $|\psi_B\rangle$. PIP demands that the joint probability distribution over the combined ontic space $\Lambda_A\times\Lambda_B$ factorizes:

\begin{equation} p(\lambda_A, \lambda_B \mid \psi_A, \psi_B) = p(\lambda_A \mid \psi_A)\;p(\lambda_B \mid \psi_B) \end{equation}

This property extends naturally to $N$-partite systems for arbitrary numbers of independent preparations. In the PBR theorem, PIP acts as the sole assumption beyond the standard ontological model framework and is critical for establishing ψ-onticity: if the distributions for single-system states overlap, PIP transmits that overlap into the joint system, leading to a contradiction with quantum exclusion measurements (Mansfield, 2014, Mansfield, 2014, Sen, 2017).

2. Structural Analogy to Bell-Type Locality

Preparation Independence displays a precise formal analogy to Bell locality. In Bell scenarios, factorizability of measurement outcomes for independently chosen settings is given by

\begin{equation} p(a, b \mid m_A, m_B, \lambda) = p(a \mid m_A, \lambda)\;p(b \mid m_B, \lambda) \end{equation}

There exists a direct 'dictionary' mapping between preparation choices and measurement choices, ontic states and measurement outcomes, and between PIP and Bell locality. Both act as 'no-conspiracies' assumptions in their respective domains: PIP for preparation procedures, Bell locality for measurement settings (Mansfield, 2014). This connection underpins deeper relationships between nonlocality, contextuality, and composition rules in quantum theory.

3. Weakenings and Alternative Independence Assumptions

Recent analyses demonstrate that PIP is stronger than minimally required by physical principles and may even exclude plausible classical correlations. Two natural weakenings are studied:

• Classical-correlation independence: Once a common cause variable $\lambda_c$ is fixed, the local ontic state distributions factorize, but global correlations are allowed via hidden variables,

$$p(\lambda_A, \lambda_B \mid \psi_A, \psi_B, \lambda_c) = p(\lambda_A \mid \psi_A, \lambda_c) p(\lambda_B \mid \psi_B, \lambda_c)$$

• Subsystem (no-preparation-signalling) condition: Marginals over a subsystem's ontic variable do not depend on the other subsystems’ preparation,

$$p(\lambda_A \mid \psi_A, \psi_B) = p(\lambda_A \mid \psi_A)$$

and analogously for $B$.

These weakened conditions permit arbitrary correlations in joint ontic states provided no superluminal preparation signalling occurs. Explicit ψ-epistemic models exist that satisfy these weakenings and evade the PBR contradiction while reproducing all quantum measurement statistics for the relevant exclusion measurements (Mansfield, 2014, Mansfield, 2014).

4. Counter-Examples and ψ-Epistemic Models Violating PIP

Under the subsystem condition or classical-correlation independence, ψ-epistemic ontological models can be constructed that reproduce all quantum predictions yet violate the full factorization required by PIP. A typical construction involves joint distributions in which overlap regions between distinct single-system ontic supports are correlated such that, for composite preparations, the problematic overlap enabling the PBR contradiction vanishes. The generalized Brans model provides an explicit example of a maximally ψ-epistemic hidden-variable theory in arbitrary dimension, achieving a degree of epistemicity $\Omega=1$ for all pairs, something forbidden by PIP in dimensions $d\geq3$ (Sen, 2017). Such models also inform protocols for low-cost classical simulation of quantum channels via correlated ontic variables.

Model Type	PIP Satisfied	ψ-Epistemicity	Classical Corr. Allowed
PBR-ontic model	Yes	No	No
Counter-example model	No	Yes	Yes
Generalized Brans	No	Maximally Yes	Yes

5. Approximate ψ-Ontology Under Minimal Conditions and Asymptotics

When only the subsystem condition and additional symmetry assumptions hold, the ψ-ontology result holds approximately. The joint ontic-preparation distribution for an ensemble of $n$ devices, selected randomly and measured in symmetric ways, can be approximated by mixtures of product distributions (finite de Finetti theorem). The epistemic overlap is then bounded as:

$$\omega(\psi, \varphi) \leq \min\left(\frac{4\,m\,|Λ|^2}{n}, \frac{2\,m(m-1)}{n}\right)$$

where $m$ is the number of measured systems and $n$ the total number. As $n\to\infty$ (with fixed $m$), these bounds vanish, recovering the strict PBR scenario (Mansfield, 2014). This demonstrates that even with relaxed independence, only contrived toy models permit nontrivial epistemic overlaps in the infinite-sample limit, clarifying the functional role of separability in ψ-ontology theorems.

6. Elimination of PIP in ψ-Onticity Theorems

Recent work demonstrates that PIP is, in fact, auxiliary for deriving single-system ψ-onticity if joint ψ-onticity for product states is already established. Specifically, once all product states are proven ψ-ontic, the tensor-product structure of quantum mechanics enforces ψ-onticity at the single-system level. No PIP or additional independence assumptions are required: the state-labeling representation theorem forces unique labels for subsystem preparations (Gao, 25 Jan 2026). This closes persistent loopholes allowing correlated single-system overlaps in ψ-epistemic models and removes reliance on strong, physically contentious auxiliary assumptions.

7. Current Limitations and Open Directions

The necessity and physical justification of PIP remain under extensive investigation. Models violating PIP while reproducing quantum statistics raise foundational questions about locality, separability, and the compositional structure of ontic space—especially in quantum field theoretic and generalized probabilistic theories where the tensor-product structure is not primitive (Myrvold, 2018). Open problems include finding minimal independence criteria that restore the PBR contradiction, generalizing ψ-ontology results to multipartite and field-theoretic settings under weakened assumptions, and clarifying the operational implications of global correlations in preparation procedures.

In summary, the Preparation Independence Postulate operationalizes the factorization of ontic statistics for independently prepared quantum systems. Its central mathematical function is the transmission of overlap necessary for ψ-ontology no-go results; its analogy to Bell locality highlights deep foundational links; its weakening to no-preparation-signalling reframes debates about locality, realism, and the epistemic status of quantum states; and its recent elimination as a necessary assumption points toward a more structural understanding of quantum reality (Mansfield, 2014, Mansfield, 2014, Gao, 25 Jan 2026, Sen, 2017, Myrvold, 2018).