Quantum Measurement Problem

← Encyclopedia

Quantum mechanics carries two rules for how a state changes, and where they overlap they contradict each other. Most of the time a system evolves by the Schrödinger equation — smooth, deterministic, reversible, linear. The moment a measurement is made, the textbook prescribes something else: an abrupt, irreversible, probabilistic jump to a single definite value, what von Neumann called collapse. The first is the law; the second is a stipulation bolted on at the exact point where the law would otherwise predict an absurdity. That seam is the measurement problem, and a century after the theory’s founding no consensus has closed it — not for want of answers, but because every answer demands a price most physicists decline to call settled.

The trouble follows from linearity alone. An apparatus that reliably reads “up” for a system that is up and “down” for one that is down, fed a superposition of the two, returns — the Schrödinger equation being additive — a superposition of “apparatus reads up” and “apparatus reads down,” the reading variable now entangled with the system. Yet no one has ever seen a pointer rest on two values at once. The premise that turns this into paradox is the eigenstate–eigenvalue link, the rule that an observable has a definite value only when the state is an eigenstate of it; the superposition is no such eigenstate, so on the link the apparatus has no definite reading. John Bell compressed the impasse to a sentence: “Either the wavefunction, as given by the Schrödinger equation, is not everything, or it is not right.”

That sentence is precise enough to serve as a map. The three main families of response each accept one horn or reject the framing entirely. The first adds something the formalism omits so that the wavefunction is not the whole story. The second rewrites the dynamics so collapse becomes a genuine physical event. The third says both horns are wrong: the wavefunction is complete and the equation is right, and the appearance of a single outcome is perspectival rather than absolute. Every viable position on the measurement problem can be located on that map, and the Stanford Encyclopedia of Philosophy’s entry on philosophical issues in quantum theory confirms that “among the major lines of approach, none is favored in a straightforward way by the empirical evidence.”

The Cat and the Cut

Schrödinger built the cruelty of it into a single image. In 1935, writing against the reigning Copenhagen view rather than in its defense, he imagined a cat sealed in a steel chamber with a Geiger counter, a trace of radioactive substance, and a relay that, should a single atom decay, releases a hammer onto a flask of acid. After an hour the atom has an even chance of having decayed; linear evolution entangles the cat with it, and the formalism, taken at its word, places the animal in a superposition of alive and dead — a case Schrödinger called “quite ridiculous,” and meant to. The cat does nothing the mathematics of a single particle does not already do; it merely carries the superposition up from the microscopic scale, where interference is confirmed, to the scale of an everyday object, where no one has met it.

What turns prediction into number is a third ingredient, the Born rule: the probability of an outcome equals the squared modulus of its amplitude. Max Born introduced it in July 1926 — only a footnote added in proof corrected his main text to the square of the amplitude, the form that later won him a share of the 1954 Nobel Prize. The rule says what the probabilities are; it is silent on why a single outcome is realized at all rather than the superposition simply persisting, and it remains a separate postulate with “no generally accepted derivation.”

Where, exactly, does the first rule give way to the second? Standard practice draws a line — the Heisenberg cut — between a quantum system and a classical apparatus that registers results. Von Neumann’s decisive contribution, in his 1932 Mathematical Foundations of Quantum Mechanics, was to show that the line will not stay put: it can be drawn between system and apparatus, or moved outward to where light strikes the retina, or pushed further into the nervous system. He took this not as a physical fact but as “a somewhat arbitrary division of the world into an observing part and an observed part.” Bell’s rule of thumb — “when in doubt enlarge the quantum system” — is what physicists actually do: push the boundary out until further inclusions stop changing the answer. It is a recipe, not a resolution. This is von Neumann’s chain — the unbounded regress of systems that could, in principle, be brought inside the quantum description — alongside which he distinguished what he called process 1 (the discontinuous projection upon measurement) from process 2 (the continuous unitary evolution). The existence of process 1 as a distinct postulate is, in a single phrase, what the measurement problem is about.

The Interpretive Map

The orthodox response, for decades, was to invoke the collapse and ask no further. The Copenhagen interpretation is associated above all with Bohr and Heisenberg in the late 1920s, but the Stanford Encyclopedia makes clear that “until Heisenberg coined the term in 1955, there was no unitary Copenhagen interpretation.” Bohr’s own position was that the formalism is a tool for coordinating experience rather than a description of an underlying reality, and — contrary to popular compression — he never accepted wavefunction collapse, viewing the state as purely symbolic. Heisenberg’s version did emphasize observer-induced collapse; the two are frequently conflated. On either reading, Copenhagen never specifies what counts as a measurement or where the cut falls. It is the incumbent attitude of most working physicists, but not a solution.

The remaining responses sort along Bell’s dilemma. One family says the wavefunction is not everything and adds to it. Bohmian mechanics — de Broglie’s 1927 proposal, completed by Bohm in 1952 — restores definite particle positions guided by a wavefunction that never collapses. Pointers always point because particles always have positions, and what looks like collapse is only the effective collapse of a subsystem’s conditional wavefunction. The theory reproduces every standard prediction; its price is explicit nonlocality, which Bell paid willingly. A second family says the wavefunction is not right and rewrites the dynamics so collapse becomes a real physical event. The Ghirardi–Rimini–Weber theory of 1986 has each particle’s wavefunction suffer rare spontaneous localizations, negligible for a lone particle but amplifying with particle number, so that a macroscopic pointer in superposition collapses almost instantly while an atom stays coherent for ages — the cut made dynamical, written into the law rather than the observer. These are not mere interpretations: they predict faint emissions standard quantum mechanics forbids, and so can be tested. The gravitationally induced version of Penrose and Diósi, that a superposition of two mass distributions is unstable, was in its simplest form already bounded out by a 2021 experiment — the same null result that constrains the objective-reduction half of Orch OR.

A third family rejects the framing. The many-worlds interpretation, Everett’s, takes the Schrödinger equation as the whole law and drops collapse: every term in the post-measurement superposition is equally real, each a branch, and the single outcome an observer reports is a perspectival fact within one of them. It buys a clean dynamics at the cost of the probability problem — if all outcomes occur, what the Born weights mean is no longer obvious — and it has its own entry. Two nearer relatives relocate the difficulty into the act of describing. Relational quantum mechanics holds that a variable’s value exists only relative to the system it interacts with, never absolutely, and that “subjects, or agents play no special role” — any physical system can serve as observer. QBism takes the quantum state to encode a single agent’s expectations about their own future experiences, so that “collapse” is only that agent updating their beliefs.

Decoherence and Einselection

Decoherence is the piece most often mistaken for the solution. When a system interacts with the countless degrees of freedom in its environment, it becomes entangled with them, interference between the branches is suppressed, and the system’s description goes diagonal in a stable, preferred basis. This is genuine and powerful: it explains why macroscopic objects look localized, why a dust grain loses coherence after a microsecond in air, and why the basis in which the world appears classical is environmentally selected rather than arbitrary.

The technical name for this environmental selection of a preferred basis is einselection — environment-induced superselection — developed in detail by Wojciech Zurek beginning in the 1980s and consolidated in his 2003 Reviews of Modern Physics survey. Zurek’s treatment shows that the environment continuously monitors certain observables of the system, and the states that survive this monitoring — the pointer states — are those that “can retain correlations with the rest of the Universe in spite of the environment.” Einselection imposes, in Zurek’s phrase, “an effective ban on the vast majority of the Hilbert space, eliminating especially the flagrantly non-local ‘Schrödinger cat’ states.” The preferred pointer basis emerges from the physics of system-environment coupling rather than being stipulated by hand — this is what decoherence resolves, and it is a genuine advance.

It dissolves the preferred-basis problem. It does not deliver a single outcome. Folding in the environment yields a larger superposition, not one result rather than the rest. The Stanford Encyclopedia entry is exact: decoherence “does explain why we do not observe superpositions of measurement results,” but “does not explain why we do observe measurement results in the first place.” Einselection replaces quantum entanglement between apparatus and system with the structure of classical correlation; it does not select one term of that correlation as the actual one. Decoherence is a tool every interpretation uses — Bohmians use it to explain the effective collapse of the conditional wavefunction; Everettians use it to define the branches; GRW uses it to understand why collapses are practically instantaneous for large systems — but it is not a tool that retires the interpretive question.

Wigner’s Friend, Extended

One response stands apart for the door it seems to open, and the popular version of it is wrong twice over. The idea that the chain of unmeasured superposition terminates only when it reaches a conscious mind, whose awareness precipitates the collapse, has a real lineage — but it is not von Neumann’s. He showed the cut was movable; he did not claim mind moves it. The conjecture that consciousness does the collapsing belongs to Wigner, who proposed it in his 1961 “Remarks on the Mind-Body Question” and sharpened the cat into the puzzle of Wigner’s friend: if a colleague performs a measurement inside a sealed laboratory, the outside observer (Wigner) can, in principle, treat the whole laboratory as an unobserved quantum system, still in superposition, while the friend has a definite outcome. The familiar phrase “von Neumann–Wigner interpretation” conflates a theorem with a speculation; the Encyclopedia, following Wallace, calls the attribution to von Neumann “the opposite of his view.” Then Wigner abandoned even the speculation, calling it close to solipsism in 1982 and crediting H. Dieter Zeh’s decoherence work with convincing him out of it.

The thought-experiment did not retire with Wigner’s recantation. Two decades of work on extended Wigner’s-friend scenarios has turned the setup into a precision instrument for testing interpretations. Daniela Frauchiger and Renato Renner published a theorem in 2018 showing that if quantum theory applies universally to systems of any complexity — including observers themselves — then three seemingly natural assumptions cannot all hold simultaneously: that quantum theory is universally valid; that agents’ reasoning is classically consistent; and that each measurement has a unique outcome for each agent. In their construction, when two pairs of observers apply quantum theory to each other’s measurements, their conclusions become mutually inconsistent: “one agent, upon observing a particular measurement outcome, must conclude that another agent has predicted the opposite outcome with certainty.” The theorem does not refute quantum mechanics; it establishes that at least one of those three commitments must go, and different interpretations each sacrifice a different one.

In 2020, Kok-Wei Bong and collaborators published both a theoretical strengthening and a proof-of-principle experiment. Their “Local Friendliness” framework derives inequalities analogous to Bell’s from three operational assumptions: no superdeterminism, locality, and the absoluteness of observed events — the requirement “that every observed event exists absolutely, not relatively.” They showed that quantum correlations violate these inequalities. The experimental result means that if quantum evolution is controllable at the scale of an observer, at least one of those three assumptions must be false. This is a strictly stronger constraint than Bell’s theorem: Bell ruled out local hidden variables; Local Friendliness additionally constrains whether observed events have observer-independent existence. QBism and relational quantum mechanics give up the third assumption explicitly; many-worlds gives up the second; Bohmian mechanics preserves all three by restricting the claim of universal controllability. No interpretation escapes without a cost.

The 2022 Nobel Prize in Physics, awarded to Aspect, Clauser, and Zeilinger for experiments establishing violations of Bell inequalities, bears on this landscape in one key way: it closed the major loopholes in Bell tests, confirming that whatever quantum mechanics describes, it is not a local realistic theory in the classical sense — the framework that most naive collapse pictures quietly presuppose.

A 2011 poll at a foundations conference found two of thirty-three physicists willing to grant the observer a distinguished physical role — about six percent. The objections are familiar: the vagueness of which systems count as conscious enough to collapse anything, the dualism required, the awkwardness that the universe would have stayed in superposition until a mind first existed to collapse it. The view keeps serious defenders in Henry Stapp, and David Chalmers treats it as a live possibility; reported faithfully, it is a small minority position, recanted by its own namesake, and not what the equations require — no mainstream approach grants conscious observers any special physical role. The Orch OR proposal of Penrose and Hameroff, which locates objective collapse in quantum processes within neural microtubules, belongs to this broader tradition of consciousness-adjacent collapse theories and has its own entry.

What the Problem Is, and Is Not

It is no accident that this problem became the largest on-ramp for “the observer creates reality” in popular culture. “Observer,” in the formalism, means any system that registers a result, any interaction that decoheres — no mind required; the mystical reading swaps in “conscious mind,” then “mind shapes reality,” and that equivocation is the whole trick. The older cosmologies held that observer and observed form one fabric, that to know a thing is to take part in it — and the hermetic ear catches something here, in the one corner of mainstream physics where the act of measurement will not factor out of the account of what is. That resonance is the archive’s to mark, not a claim the physics makes: the gap the formalism leaves is observer-shaped, but it is not a place the mind is given power to fill.

What remains is the rarest kind of problem a mature science can hold: a theory, in Ghirardi’s phrase, “excellent in telling us everything about what we observe, but [meeting] serious difficulties in telling us what there is.” The extended Wigner’s-friend theorems of the last decade have sharpened this from a philosophical curiosity into a set of provable constraints: no interpretation that holds all the obvious commitments simultaneously survives. Complete in its predictions and perfect in its arithmetic, the theory cannot say, without choosing among incompatible metaphysical commitments, what happens when a single atom decays and a single needle swings. The needle settles on one mark. The mathematics never tells it to.

→ Related: Quantum Entanglement · Many Worlds Interpretation · Orch Or · Hard Problem Of Consciousness · Multiverse