The Measurement Problem & Interpretations
Copenhagen, many-worlds, Bohm, collapse models and QBism.
The Schrödinger equation is linear and deterministic; measurements produce single, random outcomes. Reconciling the two is the measurement problem. Every interpretation of quantum mechanics is a different way of paying for it, and some of the bills can be checked in the lab.
What you’ll learn
- Formulate the measurement problem precisely with the von Neumann measurement scheme and Maudlin's trilemma.
- Explain what decoherence achieves (suppressed interference, preferred pointer states) and what it does not (a single outcome).
- Compare Copenhagen, Everett, de Broglie–Bohm, objective-collapse and QBist accounts by what each adds, denies or reinterprets.
- Derive the equivariance of in Bohmian mechanics and the amplification mechanism of GRW collapse.
- State the Pusey–Barrett–Rudolph theorem and sketch its proof.
The problem in one line of algebra
Treat the measuring device as a quantum system too, as von Neumann did. A pointer starts in a ready state , and a unitary interaction correlates it with the observable being measured.
A good measurement: the pointer reading records the eigenstate .
Linearity then forces a superposition of different readings, not one reading.
Locally the system looks like a mixture with Born weights. But it is an improper mixture: the global state is still the pure superposition above, and nothing in it says which occurred.
What decoherence does, and does not, do
Real pointers are not isolated. Each of environmental degrees of freedom (photons, phonons, gas molecules) picks up a little information about the pointer, and those partial records compound. This is decoherence.
Interference between readings and is weighted by the overlap of the corresponding environment states.
If each environmental qubit is nudged by a small angle depending on the reading, the overlaps multiply.
Even feeble records kill interference once . With and , the coherence factor is already below ; for a macroscopic pointer, is astronomically larger.
Which pointer basis gets singled out is decided by the interaction itself: the states that are left alone by the environment and redundantly copied into it (Zurek's einselection and quantum Darwinism). Decoherence therefore explains why we never see interference between macroscopic readings, and why every observer agrees about them. What it does not explain is why one reading, rather than all of them, is actual.
After decoherence, system and pointer are described by , a diagonal density matrix. Why does this not, by itself, solve the measurement problem?
- Because the off-diagonal terms never vanish exactly
- Because the global state is still a pure superposition of all readings; the diagonal matrix is an improper mixture and does not say which outcome occurred
- Because decoherence violates unitarity
- Because the diagonal weights differ from the Born probabilities
Show the answer
B. Because the global state is still a pure superposition of all readings; the diagonal matrix is an improper mixture and does not say which outcome occurred
Decoherence is ordinary unitary entanglement with the environment. Tracing out the environment gives a diagonal reduced state, but the whole is still in the superposition . To get one outcome you still need collapse, extra variables, branching, or a reinterpretation of .
A map of interpretations
| Interpretation | What it changes | Dynamics | Where the Born rule comes from |
|---|---|---|---|
| Copenhagen / textbook | Keeps a classical description of the apparatus; is a tool for predicting outcomes | Unitary between measurements, collapse at measurement; where the "cut" lies is left unspecified | Postulated |
| Everett / many worlds (1957) | Denies single outcomes: every branch is realized | Unitary, always | Contested: decision theory (Deutsch 1999; Wallace 2012) or envariance (Zurek) |
| de Broglie–Bohm (1927, 1952) | Denies completeness: particles also have definite positions | Schrödinger evolution plus a deterministic, nonlocal guidance equation | Quantum equilibrium , preserved by the dynamics |
| Objective collapse (GRW 1986, CSL) | Denies linearity: spontaneous localizations occur | Stochastic, nonlinear modification of the Schrödinger equation | Built into the collapse law; empirically testable |
| QBism (Fuchs, Mermin, Schack) | Reinterprets as an agent's degrees of belief about her future experiences | Unitary for the agent's predictions; "collapse" is Bayesian updating | A normative consistency condition on the agent's probabilities |
Two mechanisms in detail
The configuration moves along the velocity field defined by the wavefunction: the guidance equation.
The Schrödinger equation implies this continuity equation, since is the usual probability current.
Any ensemble density of configurations carried by the same velocity field obeys the same equation (Liouville).
Both and solve one linear transport equation, so if they agree at one time they agree at all times: this is equivariance. Assume this "quantum equilibrium" initially (Dürr, Goldstein & Zanghì, 1992, argue that it is typical) and the theory reproduces the Born rule for positions, and therefore for every measurement, because outcomes are ultimately recorded in pointer positions.
Each particle independently suffers a spontaneous localization at Poisson rate : the wavefunction is multiplied by a Gaussian of width centered at a random point (chosen with Born weights) and renormalized. These are GRW's original parameter values.
A pointer in a superposition of two positions more than apart collapses as soon as any one of its particles is hit.
A single particle waits about s (hundreds of millions of years) between hits, so microscopic interference is untouched; a macroscopic pointer localizes within about s. The price is that energy is not exactly conserved: the predicted tiny heating and spontaneous radiation are what experiments look for.
No-go theorems and open questions
| No-go theorem | What any underlying "realist" account must accept |
|---|---|
| Bell (1964) | Nonlocality: hidden variables that reproduce quantum correlations cannot be local |
| Kochen–Specker (1967) | Contextuality: measured values cannot all be assigned independently of the measurement context |
| Pusey–Barrett–Rudolph (2012) | -ontology: under preparation independence, the wavefunction is part of reality |
| Frauchiger–Renner (2018); Bong et al. (2020) | Wigner's-friend scenarios: no theory satisfies all of a set of plausible assumptions about agents who are themselves quantum systems |
Key takeaways
- The measurement problem: completeness, linear dynamics and single outcomes cannot all hold (Maudlin's trilemma), as the von Neumann scheme makes explicit.
- Decoherence suppresses interference exponentially in the number of environmental records and selects pointer states, but it leaves an improper mixture, not a single outcome.
- Interpretations differ in what they add (Bohm: positions; GRW: stochastic collapse), deny (Everett: single outcomes) or reinterpret (QBism: as belief).
- Collapse models make testable predictions, and experiments keep shrinking their parameter space.
- No-go theorems (Bell, Kochen–Specker, PBR, Wigner's-friend theorems) sharply constrain any realist account of quantum mechanics.
Check yourself
In the interactive lesson these are scored with your confidence, and what you miss comes back for review.
In terms of Maudlin's trilemma, which claim does de Broglie–Bohm theory reject?
- Linearity of the wavefunction's evolution
- Single definite outcomes
- Completeness of the wavefunction
- None of them
Show the answer
C. Completeness of the wavefunction
Bohmian mechanics keeps the Schrödinger equation and definite outcomes, but adds particle positions: the wavefunction is not the complete description.
In GRW theory, why do macroscopic superpositions collapse quickly while single particles essentially never do?
- Large objects have longer de Broglie wavelengths
- The localization width grows with mass
- Gravity is stronger for large objects
- The collapse rate of a rigid body scales with its number of particles,
Show the answer
D. The collapse rate of a rigid body scales with its number of particles,
One hit on any constituent localizes the whole entangled pointer, so the effective rate is : about for , against for a single particle.
What does decoherence explain?
- Which outcome of a measurement occurs
- Why interference between macroscopically distinct pointer states is unobservable, and which pointer basis is preferred
- Why the Born rule involves squared amplitudes
- Why quantum mechanics is nonlocal
Show the answer
B. Why interference between macroscopically distinct pointer states is unobservable, and which pointer basis is preferred
Decoherence, the entanglement of the pointer with its environment, suppresses interference exponentially and selects robust pointer states. It does not choose an outcome.
What does the Pusey–Barrett–Rudolph theorem conclude?
- Under preparation independence, distinct pure states correspond to disjoint distributions of physical states, so is ontic
- Hidden-variable theories must be nonlocal
- Quantum mechanics is incomplete
- The many-worlds interpretation is correct
Show the answer
A. Under preparation independence, distinct pure states correspond to disjoint distributions of physical states, so is ontic
PBR rules out "-epistemic" models, in which two different pure states could arise from the same physical state, provided independent preparations yield independent physical states.
Further reading
- M. Schlosshauer, "Decoherence, the measurement problem, and interpretations of quantum mechanics", Reviews of Modern Physics 76 (2005).
- H. Everett III, "'Relative state' formulation of quantum mechanics", Reviews of Modern Physics 29 (1957).
- D. Bohm, "A suggested interpretation of the quantum theory in terms of 'hidden' variables, I and II", Physical Review 85 (1952).
- G. C. Ghirardi, A. Rimini & T. Weber, "Unified dynamics for microscopic and macroscopic systems", Physical Review D 34 (1986).
- M. F. Pusey, J. Barrett & T. Rudolph, "On the reality of the quantum state", Nature Physics 8 (2012).