Decoherence is often described as the process that explains the transformation of the quantum world into the classical one. A system interacts with its environment, interference is suppressed, stable states become classical, and familiar definite reality supposedly appears before us.
But this formulation imperceptibly combines several entirely different assertions.
Decoherence does indeed allow us to calculate the disappearance of locally observable interference effects. Yet this does not establish that it explains:
- why the world is initially divided into this particular system and its environment;
- why this particular system should be considered rather than another decomposition of the whole;
- why one possible outcome becomes actual;
- whether there is even a single definite version of reality outside a relation to an observer.
To understand the limitations of decoherence, we must examine closely not only the result of the calculation but also the premises from which it begins.
First, the World Must Be Divided
The formal description of decoherence begins with a decomposition of the state space:
$$
\mathcal H=\mathcal H_S\otimes\mathcal H_E,
$$
where $S$ denotes the system under investigation and $E$ its environment.
Suppose the system is in a superposition:
$$
|\psi_S\rangle=\sum_i c_i|s_i\rangle,
$$
and the environment is in state $|e_0\rangle$. Interaction produces an entangled state:
$$
\left(\sum_i c_i|s_i\rangle\right)|e_0\rangle
\longrightarrow
\sum_i c_i|s_i\rangle|e_i\rangle.
$$
If the environment states $|e_i\rangle$ become approximately orthogonal, interference between the components $|s_i\rangle$ ceases to be accessible in local measurements of the system.
To obtain a local description, a partial trace over the environment is taken:
$$
\rho_S=\operatorname{Tr}_E |\Psi_{SE}\rangle\langle\Psi_{SE}|.
$$
As a result, the off-diagonal elements of the reduced density matrix become very small:
$$
\rho_S\approx \sum_i |c_i|^2 |s_i\rangle\langle s_i|.
$$
It is this result that is commonly described as the emergence of a classical mixture.
But a fundamental problem is already hidden in the first line: who determined what counts as the system and what counts as its environment, and on what grounds?
The environment does not choose the system. First, the researcher constructing the model singles out certain degrees of freedom as the object of interest and declares the rest to be the environment. Only then can one specify the interaction, perform the partial trace, and speak of a stable basis selected by the environment.
Zurek himself acknowledged the fundamental character of this difficulty: if we are given only the Hilbert space of “everything” and the total Hamiltonian, it is far from obvious how to identify the systems themselves. He later even formulated an additional premise: “The Universe consists of systems.” This is not a result of decoherence, but a condition of its application. See Zurek’s works on decoherence and the existential interpretation and on the quantum origins of the classical.
The Division Is Not Necessarily Absolute
In laboratory practice, choosing the system usually appears natural. The system is an atom, a photon, a resonator, a measuring apparatus, or another object that can be prepared, controlled, and measured. The environment consists of uncontrolled degrees of freedom: thermal radiation, air molecules, substrate vibrations, and so on.
This choice is by no means arbitrary. It is conditioned by spatial locality, the structure of interactions, the capabilities of instruments, and the available set of observables. But being “physically motivated” does not mean being “uniquely derived from the universal quantum state.”
The same collection of degrees of freedom may admit different decompositions into subsystems. Paolo Zanardi and his coauthors have shown that the structure of subsystems itself depends on which observables and operations are regarded as physically accessible: in this sense, the concept of a composite system is relative to the chosen algebra of operations. The authors explicitly connect decomposition into subsystems with operationally accessible resources.
Consequently, decoherence does not operate on a neutrally given “object in itself.” It operates on an object that has already been distinguished in a particular way.
More precisely:
Decoherence does not create the boundary between a system and the world. It calculates the consequences of interaction after that boundary has already been introduced.
What Actually Disappears
The common expression “decoherence destroys superposition” is misleading.
Under unitary evolution, the global state
$$
|\Psi_{SE}\rangle=\sum_i c_i|s_i\rangle|e_i\rangle
$$
does not transform into one of the terms in the sum. It remains a pure entangled state. No component formally disappears.
What disappears are the interference terms accessible to the system within the reduced description. We obtain their suppression because we consider only $S$ and do not take correlations with $E$ into account.
The partial trace is a perfectly legitimate mathematical operation. It yields all probabilities for the outcomes of local measurements on the system. Yet it is not a physical selection of one outcome. The transition from a global pure state to a local density matrix shows why interference is practically impossible to detect, but it does not show why an observer encounters precisely one outcome.
The reduced density matrix has the form of a mixture, but it is what is called an improper mixture: it arises from entanglement, not from the fact that the system is already in one definite state unknown to us.
It is therefore necessary to distinguish two questions:
- Why is interference between macroscopic alternatives not locally observed?
- Why does one definite alternative become actual?
Decoherence answers the first question well within a specified model. The second remains the problem of definite outcomes. This limitation is discussed in detail in Schlosshauer’s reviews of decoherence and the measurement problem and Colanero’s discussion of decoherence and definite outcomes.
What Einselection Selects
The concept of environment-induced superselection, or einselection, maintains that interaction with the environment selects stable pointer states, states that persist better than others during interaction.
But here too the level of explanation must be specified precisely.
The environment can select a stable basis only after the following have been specified:
- the system;
- the environment;
- their tensor-product decomposition;
- the interaction Hamiltonian;
- the set of observables attributed to the system.
In other words, einselection selects stable states within an already selected system. It does not explain why this particular portion of reality is a system.
The formula can be expressed very briefly:
The subject distinguishes the system; interaction selects the stable states of the distinguished system.
This does not mean that the subject arbitrarily invents physical outcomes. Experimental constraints are real, and calculated decoherence rates are tested. But the very scheme of relevance, what we regard as the object and what we disregard as the environment, is not produced by a calculation that already presupposes that scheme.
Decoherence Explains Things, But Not What Is Attributed to It
The statement “decoherence explains nothing” is justified if the issue is the origin of a single definite reality. Taken literally, however, it would be too broad.
Decoherence genuinely explains and predicts many things:
- the dependence of interference visibility on interaction with the environment;
- extremely short coherence times for macroscopic superpositions;
- the stability of some states relative to others;
- the practical irreversibility of the dispersion of phase information;
- the effectiveness of classical descriptions for locally accessible observables.
Experiments allow the degree of decoherence to be controlled, weakened, and interference to be partially restored. Therefore, this is a real physical effect, not merely a verbal interpretation. A modern review of the formalism and experiments is provided by Schlosshauer in Quantum Decoherence.
The substitution occurs elsewhere: a calculation of the suppression of interference is declared to explain why one actual world exists with one outcome.
The correct formulation is:
Given a previously specified decomposition into a system and its environment, decoherence calculates how entanglement makes interference between certain states locally inaccessible and produces a stable effective description.
The incorrect formulation is:
Decoherence transforms quantum indeterminacy into one classical fact.
The second does not follow from the first.
What Follows for the Universe
If we consider the Universe as a whole, the situation becomes even sharper. By definition, the Universe has no external physical environment. To apply the mechanism of decoherence, it must be divided internally into subsystems, each of which acts as an environment for another.
This may be a productive method of calculation. But an internal, relative decomposition cannot imperceptibly be turned into proof that the Universe as a whole possesses one definite state.
Still less can it be claimed that decoherence, by itself, establishes the actualization of one global alternative. In a unitary description, the global state does not collapse.
Thus the question:
Does a single, fully definite version of the Universe exist without a subject?
receives no answer from decoherence theory alone.
But the opposite answer would be equally unjustified:
Without a subject, the Universe is one continuous superposition.
Superposition is always defined relative to a basis, distinguished states, and a particular decomposition of the state space. Therefore, one cannot first eliminate every condition of distinction and then continue describing reality using categories that already presuppose such distinctions.
Where the Subject Enters
None of this entails that human consciousness magically creates matter or physically causes the wave function to collapse. That would be a separate ontological hypothesis for which decoherence provides insufficient grounds.
A more restrained, yet fundamental conclusion follows:
Every definiteness accessible to us arises within a relation in which something has been distinguished as a system, some property as measurable, and some outcome as registered.
The subject matters here primarily not as a supernatural physical force but as a condition of distinction. The subject determines the operational setup: what counts as an object, which interaction is used to investigate it, which degrees of freedom are included, and which are excluded from the description.
It is possible that the structure of the world itself restricts and stabilizes the permissible ways of making such distinctions. That is precisely why different observers may agree on their results and discover one stable public world. But agreement among outcomes does not yet prove that outside every relation the world was already divided into the same objects and possessed the same set of actual values.
Ontological Caution
We have no empirical access to material reality completely removed from relations of measurement, interaction, and distinction. As soon as something becomes accessible to description, it is already represented:
- in a particular basis;
- through a particular set of observables;
- as a definite system;
- in relation to other systems;
- within a particular registration procedure.
Therefore, the assertion that outside the subject there exists a single predetermined version of reality is not an observation but an ontological reconstruction.
However, the assertion that outside the subject there exists a universal superposition of all possibilities is also an ontological reconstruction.
Different interpretations of quantum mechanics fill this gap in different ways. Some postulate real collapse, others a universal wave function and branching, still others the relativity of states, and others hidden variables. Decoherence can be used within these interpretations, but does not automatically choose among them.
Conclusion
The main limitation of decoherence is not that its calculations are erroneous. It is the gap between what is calculated and what is often attributed to those calculations.
Decoherence does not derive the system from an undistinguished whole. It presupposes a system.
It does not destroy the global superposition. It suppresses locally accessible interference.
It does not select a unique outcome. It produces a reduced description that is practically indistinguishable from a statistical mixture.
It does not prove that a ready-made classical reality exists outside the subject. It shows the stability of particular descriptions within an already established relation among system, environment, and observation.
The strictest conclusion is therefore neither:
Consciousness creates the Universe.
Nor:
Without consciousness, the Universe necessarily remains in superposition.
But:
We observe a single definite reality only within a relation that includes distinguishing the system, a mode of interaction, and the registration of an outcome. Whether a single fully definite version of the Universe exists outside every such relation is unknown to us. Decoherence does not eliminate this unknown.
Perhaps reality does not simply consist of predetermined entities that subsequently interact. Perhaps definiteness itself arises only within relations. Decoherence does not prove this ontology, but neither does it permit the question to be closed by appealing to the supposedly already explained “transition from quantum to classical.”
