The whole is greater than the sum of the parts: on the possibility of purely statistical interpretations of quantum theory Joseph Emerson, Dmitry Serbin, Chris Sutherland, Victor Veitch Institute for Quantum Computing, 200 University Ave West, Waterloo, ON, Canada (Dated: July 16, 2018) The Pusey-Barrett-Rudolph theorem (PBR) claims to rule out the possibility of a purely sta- tistical interpretation of the quantum state under an assumption of how to represent independent operations in any hidden variable model. We show that PBR's assumption of independence encodes an assumption of local causality, which is already known to conflict with the predictions of quantum theory via Bell-type inequalities. We devise a weaker formulation of independence within a general hidden variable model that is empirically indistinguishable from the PBR assumption in situations where certain hidden variables are inaccessible. Under this weaker principle we are able to construct an explicit hidden variable model that is purely statistical and also reproduces the quantum predic- tions. Our results suggest that the assumption of a purely statistical interpretation is actually an innocent bystander in the PBR argument, rather than the driving force behind their contradiction. \What is proved by the impossibility proofs inference is characterized as an implication of their result is lack of imagination."[1] [6, 7]. Because the PBR assumption of preparation in- { John S. Bell dependence involves an assumption of locality, it is this assumption that should be regarded with utmost suspi- Perhaps the most long-standing open question in the cion; readers that consider locality assumptions natural foundations of quantum mechanics is whether pure quan- in the hidden variable framework should consult this ob- tum states can be understood as states of incomplete scure paper: Ref. [4]. Indeed the first contribution of this knowledge about some underlying physical reality. That paper is to show that the PBR argument does not imply such an interpretation was possible, or even necessary, the impossibility of a purely statistical interpretation of was central to Einstein's view of quantum theory: indeed, quantum states, but instead demonstrates the impossibil- the EPR argument aimed to demonstrate that quantum ity of local causality within these interpretations. There states must be considered incomplete because of an in- is, however, a positive upside to the PBR argument, or compatibility between an assumption of locality and the rather our analysis of it: we obtain novel insight into the assumption that quantum states are complete a descrip- structure of hidden variable models that are consistent tion of the physical states [2]. Bell resolved that tension with the predictions of quantum theory. in the way that \Einstein would have liked least" [3] by It is important to recognize that the PBR scenario is demonstrating that the assumption of hidden variables conceptually dual to the EPR-Bell scenario [4]: in the (and hence incompleteness) was also incompatible with EPR-Bell set-up the states are entangled and the mea- an assumption of locality [4]. In this way, EPR's idea surements are independent, whereas in the PBR set-up that the properties of entangled states could be applied the states are independent and measurements are entan- to disprove a particular type of interpretation was instead gled. Consequently it is natural to conjecture that the generalized by Bell into an insight about a violation of same kind of \locality" assumption can account for the locality in a broad spectrum of possible interpretations. conflict with quantum predictions in either scenario. In Recently, Pusey, Barrett and Rudolph (PBR) have de- the derivation of Bell's theorem, it is widely accepted that vised a new impossibility proof showing that the pre- the assumption of \local causality" plays a critical role arXiv:1312.1345v1 [quant-ph] 4 Dec 2013 dictions of quantum theory are incompatible with the and, if one accepts the hidden variables framework, then following two assumptions [5]: (1) quantum states are it is this assumption that is violated by quantum me- purely statistical, in a sense we will make precise, and chanics. Below we first show that the PBR assumption (2) a particular independence assumption about how in- of \preparation independence" implicitly encodes the as- dependent quantum preparations must be represented in sumption of local causality. We then propose a weaker the hidden variables framework. The PBR argument con- notion of independence that is appropriate in the frame- siders a situation where each subsystem of a composite work of hidden variables models: this notion, termed \lo- system is prepared independently, giving a product state cal independence", is consistent with the no-signalling preparation, after which the subsystems are brought to- principle, and remarkably, is empirically indistinguish- gether and then subjected to an entangling measurement. able from preparation independence when certain hid- From the resulting conflict with the quantum predictions den variables are assumed to be inaccessible under lo- they infer that it is the assumption that quantum states cal measurements. We then construct an explicit hidden are purely statistical which is at fault and hence should variable model for the PBR scenario to show that it is be eliminated as a possibility. In subsequent work this indeed possible to have a purely statistical hidden vari- 2 able interpretation of quantum states that reproduces the tions to which the PBR theorem applies. quantum predictions. PBR argument.|Suppose there is a preparation device The considerations of EPR and Bell, originally con- that gives out two quantum statesp independently, either cerned with identifying how entanglement imposes sur- j0i or j+i such that j h0j+i j = 1= 2. prising constraints on possible interpretations of quan- Now suppose that the ontological model underlying tum theory, eventually inspired others to recognize the quantum mechanics is such that µ0(λ) and µ+(λ) overlap practical relevance of entanglement as a novel physical on a set of non-zero measure . The states j0i and j+i resource within the quantum formalism, for example, for span a 2-dimensional Hilbert space H0. Composition of achieving the advantages of quantum information. Simi- two such systems results in a Hilbert space H = H0 ⊗H0 larly we hope that our analysis of the PBR \no go" the- with j0i ⊗ j0i ; j0i ⊗ j+i ; j+i ⊗ j0i, and j+i ⊗ j+i as a ba- orem, in addition to clarifying the foundational question sis. Now introduce an entangled measurement M whose of which interpretations of quantum theory are even pos- eigenstates fjξ00i ; jξ0+i ; jξ+0i ; jξ++ig form an orthonor- sible, will also inspire further insight into the unique re- mal basis for H such that hξabjabi = 0 for a; b 2 f0; +g. sources and practical capabilities of quantum theory. To reach a contradiction with quantum mechanics an ad- Ontological models and hidden variables.| In Bell's ditional assumption about the nature of the ontological ontological model framework the possible physical prop- model is required: erties of a system are identified with a space Λ. A prepa- Definition (Preparation Independence). The composi- ration j i is associated with a measure (probability den- tion of independently prepared pure states j i and jφi sity) µ (λ) over Λ and each realization of the preparation results in probability distributions that satisfy results in a physical state λ 2 Λ sampled from µ (λ). A measurement operator E , e.g. drawn from a POVM k µ ,φ(λ1; λ2) = µ (λ1)µφ(λ2): (1) E = fEkg, is associated with a response function ξk(λ) which specifies the conditional probability of obtaining Here λi is a set of arbitrary \local" variables associ- outcome k given physical state λ. The conditional prob- ated with subsystem i. The implicit structure associated abilities ξk(λ) and probabilities µ (λ) are subject to the with the preparation independence assumption is that following two conditions: (i) For all ontic states, some a complete account of the composition of the two in- P outcome always occurs 8λ 2 Λ; k ξk(λ) = 1, and (ii) dependent preparations is prescribed by the state space the Born rule statistics must be reproduced via the law Λ = Λ × Λ . R LHV 1 2 of total probability Pr(kj j i) = Λ ξk(λ)µ (λ)dλ. The first outcome is orthogonal to j0i⊗j0i, thus quan- Thus we associate an ontological model for quantum tum theory predicts that this outcome has probability mechanics with the triplet (Λ; fµg; fξg). This framework zero when the quantum state prepared was j0i⊗j0i. Sim- is very general. For example, it can accommodate the ilarly for the other three states. Assuming the measure- orthodox interpretation of von Neumann and Dirac in ment device reads its outcome solely on the ontic states which quantum states are assumed to give a complete prepared by each system, with probability 2 the mea- specification of the physical reality - that is, the denial surement device is uncertain which of the four quantum of hidden variables: this interpretation is realized if we states was prepared. For example, outcome 1 might occur identify Λ with the (projective) Hilbert space and each even if the quantum state was j0i ⊗ j0i. This contradicts probability measure µ with a Dirac δ-function on Λ the predictions of quantum mechanics. PBR infer (incor- [8]. The ontological models framework also accommo- rectly, as we explain below) that this contradiction rules dates hybrid models in which the quantum state is both out the possibility of a purely statistical model. physical and statistical { indeed the deBroglie-Bohm in- The problem with PBR.| Here we show that the PBR terpretation fits in this category. Such hybrid interpreta- assumption of preparation independence relies on an as- tions can be sufficient to explain the randomness of ex- sumption of \local causality" which is the likely source of perimental outcomes because pure quantum states pro- the conflict they obtain with quantum theory.