What Decoherence Does Not Solve

Decoherence explains suppression of interference in reduced states and helps identify robust records. In a unitary system–environment model, it does not itself replace a superposition of correlated branches by one selected branch. This distinction separates a calculable physical mechanism from the further questions of outcome interpretation and probability.

Helpful background. Decoherence gives the reduced-state mechanism; proper and improper mixtures explain why a density matrix does not determine its preparation history.

Local interference and the joint state

Take normalized system amplitudes, orthonormal $|0\rangle,|1\rangle$, and a pure initially uncorrelated environment. A controlled unitary can act as

$$\left( c_0\lvert0\rangle + c_1\lvert1\rangle \right) \lvert E_{\mathrm{ready}}\rangle \longmapsto c_0\lvert0\rangle\lvert E_0\rangle + c_1\lvert1\rangle\lvert E_1\rangle.$$

The reduced state of the system is

$$\rho_S ={} |c_0|^2 \lvert0\rangle\langle0\rvert + |c_1|^2 \lvert1\rangle\langle1\rvert + c_0c_1^* \langle E_1|E_0\rangle \lvert0\rangle\langle1\rvert + c_0^*c_1 \langle E_0|E_1\rangle \lvert1\rangle\langle0\rvert.$$

If the environmental states are nearly orthogonal,

$$\langle E_1|E_0\rangle\approx0,$$

then

$$\rho_S \approx |c_0|^2 \lvert0\rangle\langle0\rvert + |c_1|^2 \lvert1\rangle\langle1\rvert.$$

If the overlap is exactly zero, system-only measurements have exactly the classical-mixture statistics with weights $|c_0|^2$ and $|c_1|^2$. For a nonzero overlap, the largest possible error in any system-only event probability made by discarding the cross terms is $|c_0c_1\langle E_1|E_0\rangle|$.

But the global state in the unitary model is still

$$\lvert\Psi\rangle = c_0\lvert0\rangle\lvert E_0\rangle + c_1\lvert1\rangle\lvert E_1\rangle.$$

The diagonal reduced state is therefore not automatically the statement that exactly one branch has been selected. It is a statement about local statistics after environmental degrees of freedom are ignored.

Diagonal Is Not the Same as Selected

A reduced density matrix can be diagonal for several reasons:

• a source actually sampled one preparation from a classical list;
• a measurement outcome was recorded but not revealed to the observer using the state;
• a subsystem is entangled with an environment that has been traced out;
• a model has already inserted a collapse or conditioning rule;
• the chosen basis diagonalizes $\rho$ at a particular time without representing a stable record basis.

These cases can produce the same matrix but have different physical descriptions. These possibilities are distinguished by their preparation and joint correlations; see Proper and Improper Mixtures.

In the pure entangling model above, the reduced state is an improper mixture when both branches have nonzero amplitudes and distinguishable environment states. Other decohering processes can involve an initially mixed environment; no universal entanglement claim is needed for the outcome distinction.

No Unique Outcome from Decoherence Alone

In a purely unitary system-apparatus-environment model, the measurement-like evolution has the schematic form

$$\sum_a c_a \lvert a\rangle_S \lvert A_{\mathrm{ready}}\rangle \lvert E_{\mathrm{ready}}\rangle \longmapsto \sum_a c_a \lvert a\rangle_S \lvert A_a\rangle \lvert E_a\rangle.$$

Environmental decoherence suppresses interference between branches in the reduced system–apparatus state. In this ideal model, tracing out the orthogonal system states already makes the apparatus alone diagonal, even before environmental decoherence. The environmental interaction additionally makes interference unavailable to someone controlling the system and apparatus but not the environment.

It does not add, by itself, a nonunitary replacement

$$\sum_a c_a \lvert a\rangle_S \lvert A_a\rangle \lvert E_a\rangle \longrightarrow \lvert a_0\rangle_S \lvert A_{a_0}\rangle \lvert E_{a_0}\rangle$$

for one particular $a_0$. That is a different claim. In standard calculations, such a transition is represented by conditioning on an observed record or by a state-update rule after an outcome is known. In objective-collapse theories, it is represented by modified dynamics. In Everettian accounts, all branches remain in the global state and the interpretation supplies a branch-relative account of experience and probability. In Bohmian mechanics, the actual configuration adds further structure.

The Core Formalism boundary page What Measurement Formalism Does Not Settle states the same distinction from the measurement-rule side.

Decoherence Does Not Derive the Born Rule by Itself

The reduced state after decoherence has weights $|c_a|^2$:

$$\rho_S \approx \sum_a |c_a|^2 \lvert a\rangle\langle a\rvert.$$

The Born rule already gives $p_a=|c_a|^2$ for a pointer-basis measurement before the coherences are suppressed. Decoherence neither creates that probability rule nor derives it merely by leaving diagonal entries unchanged. Its additional contribution is dynamical stability and reduced interference between the alternatives.

Different foundational programs try to connect decoherence with probability in different ways. Some Everettian approaches combine decoherence with decision-theoretic, symmetry-based, or typicality arguments. Operational accounts take the Born rule as part of the empirical probability calculus. Collapse theories modify the dynamics and must still match Born-rule statistics. Hidden-variable theories add variables and an equilibrium distribution.

Those are not the same claim as environmental suppression of interference. Decoherence makes the probability calculus effective for records; it is not a standalone replacement for the probability postulate.

Decoherence Does Not Identify a Universal Preferred Basis

Decoherence selects structures relative to dynamics. A pointer basis or pointer subspace depends on:

• the system Hamiltonian;
• the system-environment interaction;
• the initial state and temperature of relevant environments;
• the coarse graining used to describe macroscopic records;
• the timescale on which robustness is assessed;
• the observables accessible to observers or apparatuses.

There is no single basis that decoherence selects for every system in every situation.

For example, environmental scattering often monitors coarse position. A dephasing bath coupled to $Z$ monitors $Z$-basis alternatives. A lossy optical mode may favor coherent-state-like wavepackets. Degenerate monitoring may select subspaces rather than individual vectors. These are all legitimate pointer structures in their own models.

This is why diagonalizing the instantaneous $\rho_S$ is not enough. The preferred structure is dynamical and operational, not merely spectral.

Recoherence requires the right control

Unitary dynamics can be reversible even after environment states become orthogonal. The inverse operation, however, must act on the relevant joint correlations. An environment-only unitary cannot map two distinct record states to one common ready state: it preserves their inner product. More generally, any trace-preserving operation performed only on the environment leaves the unconditioned system state unchanged.

A two-qubit example makes the distinction concrete. A controlled-NOT acting from system to environment gives

$$|+\rangle_S|0\rangle_E \longmapsto\frac{|00\rangle+|11\rangle}{\sqrt2}.$$

Applying the same joint gate again restores $|+\rangle_S|0\rangle_E$. Resetting only the environment to $|0\rangle$ instead gives $(I_S/2)\otimes|0\rangle\langle0|_E$: the system remains mixed. Measuring the environment in the $X$ basis produces conditional system states $|+\rangle$ or $|-\rangle$; selecting a record or applying a record-dependent correction to the system can recover interference. Averaging over unread records cannot.

Echoes cancel particular reversible phase evolutions, and finite environments can produce revivals, but neither guarantees reversal of arbitrary noise. In an uncontrolled macroscopic environment, recovering all relevant correlations is ordinarily infeasible. This practical irreversibility is not an additional exact prohibition in unitary quantum mechanics.

Decoherence Does Not Replace Detector Modeling

Decoherence is often part of a detector model, but it is not the whole model. A laboratory measurement also involves:

• coupling the microscopic system to an apparatus variable;
• amplification from microscopic differences to robust macroscopic records;
• detector efficiency and dark counts;
• thresholds, bins, and data-selection rules;
• calibration against known inputs;
• environmental monitoring of the apparatus;
• a conditioning rule for selected records.

The same decohering apparatus may be represented by different effective POVMs or instruments depending on resolution, inefficiency, and readout electronics. The open-system physics helps explain why records persist, but the actual measurement model still requires apparatus-specific assumptions. See Von Neumann Measurement Model and Selective and Nonselective Measurements.

Redundant records add accessibility

Quantum Darwinism asks when many environmental fragments independently reveal the same pointer information. That is an additional property of particular environments, not a consequence of every dephasing process. It strengthens the explanation of agreement among observers without changing the distinction between a joint superposition and a selected outcome.

Exercises

1. Trace and compare. Starting from

$$\lvert\Psi\rangle = c_0\lvert0\rangle\lvert E_0\rangle + c_1\lvert1\rangle\lvert E_1\rangle,$$

derive $\rho_S$ and state what happens when $\langle E_1|E_0\rangle=0$. Does that condition alone select one branch?

Solution

The projector is

$$\begin{aligned} \lvert\Psi\rangle\langle\Psi\rvert ={}& |c_0|^2 \lvert0\rangle\langle0\rvert \otimes \lvert E_0\rangle\langle E_0\rvert \\ &+ |c_1|^2 \lvert1\rangle\langle1\rvert \otimes \lvert E_1\rangle\langle E_1\rvert \\ &+ c_0c_1^* \lvert0\rangle\langle1\rvert \otimes \lvert E_0\rangle\langle E_1\rvert \\ &+ c_0^*c_1 \lvert1\rangle\langle0\rvert \otimes \lvert E_1\rangle\langle E_0\rvert. \end{aligned}$$

Tracing over $E$ gives

$$\rho_S ={} |c_0|^2 \lvert0\rangle\langle0\rvert + |c_1|^2 \lvert1\rangle\langle1\rvert + c_0c_1^* \langle E_1|E_0\rangle \lvert0\rangle\langle1\rvert + c_0^*c_1 \langle E_0|E_1\rangle \lvert1\rangle\langle0\rvert.$$

If $\langle E_1|E_0\rangle=0$, the reduced state is diagonal:

$$\rho_S = |c_0|^2 \lvert0\rangle\langle0\rvert + |c_1|^2 \lvert1\rangle\langle1\rvert.$$

This condition removes local interference between the two alternatives. It does not, by itself, replace the global state by one branch.

2. Proper or improper. A reduced state after decoherence is

$$\rho_S = 0.7\lvert0\rangle\langle0\rvert + 0.3\lvert1\rangle\langle1\rvert.$$

Explain why this matrix alone does not tell you whether the mixture is proper or improper.

Solution

The same density matrix can arise from a source that classically prepares $\lvert0\rangle$ with probability $0.7$ and $\lvert1\rangle$ with probability $0.3$, or from tracing out an environment entangled with the system. The matrix determines system-only statistics, but the proper/improper distinction depends on the larger physical description: preparation records, environmental correlations, and which degrees of freedom are accessible.

3. Born-rule boundary. Why does the appearance of weights $|c_a|^2$ in a diagonal reduced state not, by itself, derive the Born rule?

Solution

Decoherence suppresses off-diagonal terms and leaves diagonal weights $|c_a|^2$ in the pointer basis. Those weights already give the corresponding Born probabilities before decoherence. But the dynamical fact that off-diagonal terms are small is not the same as a derivation of why the diagonal weights must be interpreted as probabilities. That interpretive or postulate-level step is supplied by the broader framework being used.

4. Reset versus reversal. The system and environment are in $(|00\rangle+|11\rangle)/\sqrt2$. Compare (a) discarding and resetting only the environment, and (b) applying a controlled-NOT from system to environment. What is the final system state in each case?

Solution

An environment reset is the local trace-preserving map $\tau_E\mapsto|0\rangle\langle0|\operatorname{Tr}\tau_E$. It produces $(I_S/2)\otimes|0\rangle\langle0|_E$, so the system remains $I/2$.

The joint gate maps $|00\rangle$ to $|00\rangle$ and $|11\rangle$ to $|10\rangle$. It therefore produces $|+\rangle_S|0\rangle_E$, restoring the initial pure system state. The distinction is access to joint correlations, not merely removal of the visible environmental record.

5. Classify a claim. A popular explanation says, “The environment observes the apparatus, so the wavefunction collapses into one definite outcome.” Rewrite this claim in a way that is precise and interpretation-neutral.

Solution

A precise version is: interaction with the environment correlates different apparatus pointer states with nearly orthogonal environmental states, so interference between pointer alternatives becomes locally inaccessible and the reduced apparatus state is approximately diagonal in the pointer basis. Whether this corresponds to a literal collapse into one outcome depends on the measurement postulate, conditioning rule, or interpretation being used.

References

• J. S. Bell, Speakable and Unspeakable in Quantum Mechanics, 2nd ed., Cambridge University Press (2004).
• M. Schlosshauer, Decoherence and the Quantum-to-Classical Transition, Springer (2007). Publisher.
• M. Schlosshauer, “Quantum decoherence,” Physics Reports 831, 1–57 (2019). Author manuscript.
• W. H. Zurek, “Decoherence, einselection, and the quantum origins of the classical,” Reviews of Modern Physics 75, 715–775 (2003). DOI.