{
  "id": "many-worlds-interpretation",
  "title": "Many-Worlds Interpretation",
  "lang": "en",
  "source_lang": "en",
  "summary": "The many-worlds interpretation treats quantum mechanics as universally unitary, with distinct observed outcomes realized in effectively independent branches of the quantum state.",
  "category": "science",
  "subcategory": "physics",
  "type": "concept",
  "aliases": [
    "Everett interpretation",
    "Everettian quantum mechanics",
    "relative-state interpretation"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/many-worlds-interpretation/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/many-worlds-interpretation/",
      "json": "https://aiwiki.page/zh/many-worlds-interpretation.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The many-worlds interpretation is an interpretation of quantum mechanics in which the quantum state of an isolated system always evolves according to the theory’s usual deterministic dynamics, without a physical collapse during measurement. When a measurement has several possible outcomes, the measuring apparatus and its observers become correlated with different components of the wave function. Each effectively independent component contains a definite recorded outcome, although the complete state retains all of them.\n\nIts central purpose is to address the measurement problem: how the mathematical superpositions permitted by quantum theory relate to the definite results encountered in experiments. Instead of introducing a special rule that selects one result and removes the others, the interpretation treats measuring devices, observers, and their environments as quantum systems. Modern accounts use quantum decoherence to explain why different outcome records cease to interfere in practice.\n\n“Many worlds” names a family of related positions rather than a uniquely specified theory. Everettians differ over what constitutes a world, how probability should be understood when every outcome occurs, and whether the universal quantum state alone provides a complete physical ontology. The interpretation generally reproduces the ordinary predictions of quantum mechanics when supplemented by the Born rule, but its explanation of that rule and its account of observed reality remain subjects of foundational debate."
    },
    {
      "heading": "Historical origins and development",
      "text": "The interpretation originated in the work of Hugh Everett III, whose doctoral research at Princeton University was supervised by John Archibald Wheeler. Everett sought a formulation of quantum mechanics applicable to an entire closed system, potentially including the universe. Conventional measurement rules appeared problematic in this setting: if a measurement requires an external observer or a division between quantum objects and classical apparatus, it is unclear how those rules apply when every observer and apparatus is already included in the system being described.\n\nEverett’s paper “‘Relative State’ Formulation of Quantum Mechanics” appeared in Reviews of Modern Physics in 1957. Its central proposal was to retain the linear quantum dynamics for all physical processes, including observation. A subsystem does not generally possess an independent pure state when it is entangled with another subsystem. Nevertheless, a state can be assigned to it relative to a specified state of the other subsystem. Applied to measurement, this means that each observer record is correlated with a corresponding state of the measured object. Everett argued that observers described within the theory would register consistent sequences of definite results without a collapse of the total state.\n\nThe early formulation did not contain the mature environmental account of branching used by many later Everettians. Everett discussed correlations, records, and the statistical behavior of long measurement sequences, but modern decoherence theory developed subsequently. His work initially attracted limited attention. Wheeler helped bring it into publication and discussed it with physicists associated with Niels Bohr, but it did not displace established approaches to measurement.\n\nBryce DeWitt played a major role in popularizing the interpretation, especially through his 1970 article “Quantum Mechanics and Reality.” DeWitt and Neill Graham edited The Many-Worlds Interpretation of Quantum Mechanics in 1973, bringing together Everett’s work and related discussions. The vivid language of a universe splitting into worlds became closely associated with this presentation. That language is historically influential, but it is not a literal description accepted by every Everettian.\n\nA second major phase followed the development of decoherence. H. Dieter Zeh’s 1970 work emphasized the importance of environmental interactions, and Wojciech Zurek developed accounts of environment-selected stable states during the 1980s and afterward. David Deutsch’s 1999 decision-theoretic argument and subsequent work by David Wallace, Simon Saunders, and others renewed debate about probability. Consequently, contemporary Everettian quantum mechanics combines an older commitment to universal quantum dynamics with later accounts of emergent classical structure and rational expectation."
    },
    {
      "heading": "Universal dynamics and the measurement process",
      "text": "The basic dynamical commitment is that every isolated physical system obeys the Schrödinger equation, or an appropriate relativistic generalization. In ordinary nonrelativistic notation,\n\n\\[i\\hbar\\frac{\\partial}{\\partial t}|\\Psi(t)\\rangle = \\hat H|\\Psi(t)\\rangle,\\]\n\nwhere \\(\\hbar\\) is the reduced Planck constant and \\(\\hat H\\) is the Hamiltonian operator. Evolution generated by a suitable self-adjoint Hamiltonian is unitary: it preserves inner products and the total norm of the state. The interpretation denies that measurement introduces an additional, genuinely nonunitary physical process.\n\nA simple model illustrates the consequences. Suppose a system has two orthogonal states, \\(|0\\rangle\\) and \\(|1\\rangle\\), and an apparatus begins in a ready state \\(|A_{\\mathrm r}\\rangle\\). An ideal measuring interaction correlates each system state with a distinguishable apparatus record:\n\n\\[|0\\rangle|A_{\\mathrm r}\\rangle\\longrightarrow |0\\rangle|A_0\\rangle, \\qquad |1\\rangle|A_{\\mathrm r}\\rangle\\longrightarrow |1\\rangle|A_1\\rangle.\\]\n\nBecause the dynamics is linear, an initial superposition evolves as\n\n\\[(\\alpha|0\\rangle+\\beta|1\\rangle)|A_{\\mathrm r}\\rangle \\longrightarrow \\alpha|0\\rangle|A_0\\rangle+ \\beta|1\\rangle|A_1\\rangle,\\]\n\nwith \\(|\\alpha|^2+|\\beta|^2=1\\). Standard collapse prescriptions replace this final superposition with one selected term. Everettian accounts retain the complete final state.\n\nIncluding an observer extends the same correlation structure:\n\n\\[|\\Psi\\rangle = \\alpha|0\\rangle|A_0\\rangle|O_0\\rangle + \\beta|1\\rangle|A_1\\rangle|O_1\\rangle.\\]\n\nHere \\(|O_0\\rangle\\) represents an observer with a record of outcome 0, and \\(|O_1\\rangle\\) an observer with a record of outcome 1. Neither component describes an observer reporting an indefinite pointer reading. Rather, the complete state contains distinct observer–apparatus correlations. This distinction between an indefinite global state and definite relative records is the core of Everett’s response to the measurement problem.\n\nThe mathematics uses a Hilbert space for the combined system, commonly constructed as a tensor product of subsystem spaces. Entanglement prevents the final state from being represented as a single product of independent states for the object, apparatus, and observer. Relative states express the conditional correlations within this entangled whole.\n\nThis model is deliberately idealized. Real instruments have imperfect resolution, finite-temperature components, noise, and many environmental interactions. Measurements also include continuous outcomes and generalized measurement procedures, not merely two orthogonal alternatives. None of those extensions changes the fundamental Everettian proposal: whatever quantum dynamics describes the interaction also describes the production of records, without a separate law activated by observation."
    },
    {
      "heading": "Decoherence and the emergence of classical branches",
      "text": "The measurement equations alone do not explain why one decomposition of a state should correspond to familiar macroscopic alternatives. The same vector can be expanded in many different bases, including bases whose elements are superpositions of apparently distinct pointer readings. This is the preferred-basis problem. Modern Everettian accounts address it through the dynamics of interactions among systems, measuring apparatuses, and environments.\n\nAn apparatus is never perfectly isolated. Scattered photons, surrounding gas molecules, thermal radiation, and internal degrees of freedom continually acquire information about its condition. A more realistic post-measurement state therefore has the form\n\n\\[|\\Psi\\rangle = \\alpha|0,A_0\\rangle|E_0\\rangle + \\beta|1,A_1\\rangle|E_1\\rangle,\\]\n\nwhere \\(|E_0\\rangle\\) and \\(|E_1\\rangle\\) are environmental states correlated with the outcomes. Even individually weak interactions can make these environmental states nearly orthogonal when many particles participate.\n\nFor an observer who does not control all environmental degrees of freedom, the relevant description is the reduced density matrix, obtained by tracing out the environment:\n\n\\[\\rho_{SA} = \\operatorname{Tr}_E|\\Psi\\rangle\\langle\\Psi|.\\]\n\nIts off-diagonal terms contain factors such as \\(\\langle E_1|E_0\\rangle\\). When this overlap becomes extremely small, the terms responsible for observable interference between the records become correspondingly suppressed. Local experiments then behave approximately as though the apparatus had an ordinary statistical mixture of definite readings.\n\nThe distinction between an effective mixture and a fundamental collapse is essential. Decoherence does not remove components from the total state, and it does not select one outcome. The reduced state is an improper mixture arising from entanglement, not necessarily a statement that the entire system really occupies just one component. Everettian interpretations take the retained components seriously; other interpretations use the same decoherence calculations while adding a different account of outcomes.\n\nEnvironmental interactions also favor stable, approximately classical variables. Apparatus states that resist disruption while leaving persistent records are often called pointer states. Their form depends on the interaction Hamiltonian: position-like states are especially important because many ordinary interactions depend strongly on spatial separation. Related research under the name quantum Darwinism examines how information about selected observables becomes redundantly recorded in many environmental fragments.\n\nBranching is therefore normally understood as an approximate, continuous physical process rather than an instantaneous cosmic duplication. Decoherence rates depend on the object, environment, temperature, and spatial separation involved. Macroscopic alternatives can decohere extremely rapidly, while carefully protected microscopic systems can preserve coherence for experimentally useful periods. The boundary is dynamical, not a universal dividing line between objects of different sizes."
    },
    {
      "heading": "What constitutes a world?",
      "text": "The word “world” can suggest a precisely defined additional object, but many contemporary Everettians use it for an emergent structure within the universal state. A world is then a dynamically stable pattern of approximately classical objects, records, and observers. Its status is compared to that of other higher-level structures in physics: a fluid vortex is not an extra fundamental ingredient added to molecular dynamics, yet it can be a real and useful physical pattern.\n\nOn this account, the fundamental description does not contain an exact integer specifying how many worlds exist. Branch decompositions depend partly on the resolution used to distinguish macroscopic alternatives. A measuring device may record one of two displayed values while its environment develops an enormous variety of microscopic correlations. Counting every environmentally distinguished component would produce a different answer from counting only visibly different records. Because decoherence is approximate, neither count is generally a uniquely privileged inventory of fundamental entities.\n\nThis is one reason that “the universe splits into two whenever a choice is made” is misleading. Human decisions do not have a special role in the dynamics, and branching is not restricted to deliberate measurements. Ordinary interactions can spread distinguishable records without anyone observing them. Conversely, a coherent superposition in a controlled experiment need not already correspond to distinct classical worlds merely because it can be written as a sum of terms.\n\nThe familiar Schrödinger’s cat example illustrates both the proposal and its interpretive burden. After a hypothetical quantum-triggered event becomes correlated with a cat and its environment, one decohered component contains a living cat and another a dead cat. The full state does not describe a single classical cat possessing contradictory properties. Nevertheless, explaining why the mathematical components support ordinary objects and experiences requires an account of emergence, not simply the assertion that every term in any expansion is a world.\n\nEverettian ontologies differ on what the quantum state represents. Some regard the wave function as a physical field on a high-dimensional configuration space. Others emphasize the structure of entanglement, local physical quantities, or patterns recoverable in ordinary space. These positions agree about rejecting collapse but differ about the relation between the formalism and material objects.\n\nDescriptions in terms of histories offer another route. The consistent-histories approach studies sequences of alternatives whose interference is negligible enough to support probability assignments. It can provide mathematical tools for characterizing quasi-classical branching, although consistent histories is not inherently a many-worlds interpretation. Different compatible coarse-grainings can describe the same underlying state, reinforcing the view that worlds need not form a unique, sharply bounded set."
    },
    {
      "heading": "Probability and the Born rule",
      "text": "Probability is one of the interpretation’s most contested issues. In conventional presentations, the Born rule assigns probability \\(p_i=|c_i|^2\\) to outcome \\(i\\) in a normalized state \\(\\sum_i c_i|i\\rangle\\). In many-worlds quantum mechanics, every nonzero outcome component persists. The problem is therefore not merely to calculate squared amplitudes but to explain why they should govern expectations when no unique outcome is selected globally.\n\nSimply counting branches does not provide an adequate rule. Branch number is not generally well defined, and further environmental interactions can subdivide a component without changing the original measurement’s squared amplitudes. For example, a state with weights \\(0.9\\) and \\(0.1\\) should not become a fifty–fifty experiment merely because its macroscopic description contains two outcome labels. Any viable Everettian probability account must explain why Hilbert-space weight, rather than an arbitrary count of descriptions, matters.\n\nOne line of argument concerns repeated measurements. For \\(N\\) identically prepared two-outcome systems, the total squared weight of components containing \\(k\\) occurrences of one result is\n\n\\[W(k)=\\binom Nk p^k(1-p)^{N-k},\\]\n\nwhere \\(p=|\\alpha|^2\\). As \\(N\\) increases, most of the total weight lies on histories whose relative frequencies \\(k/N\\) are close to \\(p\\). This establishes a mathematical concentration result. It does not by itself settle the conceptual issue, because histories with anomalous frequencies still exist, and treating squared weight as the relevant measure is part of what requires justification.\n\nA second approach uses decision theory. Deutsch, Wallace, and others analyze an agent’s preferences over quantum experiments with rewards attached to their outcomes. Under proposed rationality conditions, together with assumptions about the physical equivalence of certain experiments, the agent’s preferences can be represented as maximizing Born-weighted expected utility. The aim is not to identify a hidden random event, but to show why rational expectations and choices in a branching universe should have the same numerical form as ordinary quantum probabilities.\n\nCritics question whether these rationality assumptions already incorporate the substance of the desired probability measure. They also distinguish a theorem about preferences from a physical explanation of observed frequencies. Defenders answer that probability in other deterministic settings likewise requires a connection between physical structure, uncertainty, and rational expectation.\n\nOther proposals appeal to symmetry, including Zurek’s environment-assisted invariance, or to self-locating uncertainty about which recorded outcome an observer occupies. These programs overlap in some respects but are not interchangeable. No single derivation has ended the dispute. Everettian quantum mechanics has several developed accounts of Born weights, while the necessity and interpretation of their assumptions remain open to examination."
    },
    {
      "heading": "Observers, uncertainty, and personal identity",
      "text": "Observers have no special dynamical status in the many-worlds interpretation. A person, laboratory notebook, computer memory, or photographic detector is a physical system that can become correlated with a measurement outcome. Conscious awareness does not cause branching, and the interpretation requires no additional law specifying when a sufficiently complex observer appears.\n\nThe claim that observers register definite outcomes is relational. In the state\n\n\\[\\alpha|0\\rangle|O_0\\rangle+\\beta|1\\rangle|O_1\\rangle,\\]\n\nthe record represented by \\(|O_0\\rangle\\) is correlated with outcome 0, and the record represented by \\(|O_1\\rangle\\) with outcome 1. An observer component does not ordinarily have access to the alternative component’s memories. After decoherence, interference between the records is negligible for accessible operations, so each sequence of memories supports an apparently continuous classical history.\n\nThis raises a question about anticipation before a measurement. If both future records will exist, what does it mean to be uncertain about the result? Some Everettians describe genuine uncertainty about which future experience will be encountered. Others maintain that the complete physical future is known in principle but that uncertainty can concern an observer’s location within it. Still others emphasize how an agent should distribute concern among future successors, without construing the situation as ordinary ignorance about a single future fact.\n\nA particularly clear case of self-locating uncertainty occurs in an idealized interval after a measurement has produced distinct apparatus records but before an observer has inspected them. Different observer copies may have the same memories while being correlated with different external results. They can know the full experimental arrangement yet lack knowledge of their own branch-relative situation. Whether such uncertainty fully grounds the Born rule depends on further principles about how branch weights determine credence.\n\nPersonal identity introduces a related philosophical problem. If one observer state develops into several successor states, ordinary one-to-one assumptions about persistence may no longer apply. Physical continuity and psychological continuity can hold between an earlier person and multiple later records. The quantum formalism specifies correlations and dynamics; it does not, by itself, settle every metaphysical question about whether these successors should be called the same person.\n\nThe Wigner’s friend scenario sharpens the distinction between internal and external descriptions. An observer inside an ideally isolated laboratory records a definite result, while an external experimenter may describe the entire laboratory as a coherent superposition. Everettian accounts regard these as compatible descriptions at different relational levels. Restoring measurable interference would require control over the relevant records and their environmental correlations. A global interference experiment need not preserve the friend’s original memory as an undisturbed, independently accessible record."
    },
    {
      "heading": "Relations to other interpretations and locality",
      "text": "Many-worlds differs from textbook collapse accounts chiefly in what it says happens during measurement, not in the usual mathematical calculation of outcome frequencies. The label Copenhagen interpretation covers several historically distinct positions, often involving a classical description of experimental arrangements or limits on assigning properties outside measurement contexts. Everettian quantum mechanics instead attempts to describe apparatuses and observers within one universal quantum framework.\n\nBohmian mechanics also retains a wave function evolving according to the Schrödinger equation, but supplements it with an actual configuration, commonly particle positions, guided by that wave function. In a measurement, only one configuration corresponds to the apparatus’s actual displayed result. The unoccupied wave-function components can remain in the dynamics, but they do not have the same ontological role as Everettian branches. Thus, retaining an uncollapsed wave function is not by itself sufficient to make an interpretation many-worlds.\n\nObjective-collapse theories alter the dynamics so that sufficiently separated or macroscopic superpositions are physically reduced. The Ghirardi–Rimini–Weber model, introduced in 1986, and continuous spontaneous localization models are prominent examples. Unlike an interpretation that leaves unitary dynamics intact, such models can predict departures from standard quantum theory. Their parameters determine when and how strongly localization occurs.\n\nLocality requires especially careful terminology. Everettian dynamics can avoid a physical collapse propagating instantaneously across space. In relativistic quantum field theory, spacelike separated local observables satisfy appropriate commutation relations, and ordinary quantum predictions do not allow controllable faster-than-light communication. Nevertheless, entangled states display correlations incompatible with the locally causal, single-outcome framework constrained by Bell’s theorem.\n\nWhether many-worlds should therefore be called “local” depends on the definition being used. The absence of superluminal signaling does not establish Bell local causality, and the existence of local interaction laws does not make the global entangled state separable into independent local states. Everettian accounts explain Bell experiments through correlations among branch-relative outcomes rather than a unique distant outcome selected by collapse. That response changes the single-outcome picture but does not erase the empirical content of Bell inequalities.\n\nThese distinctions also matter for special relativity. A literal picture in which the whole universe splits simultaneously would require an unexplained global simultaneity structure. Modern formulations instead focus on local interactions, spreading correlations, and compatible descriptions on spacetime regions or hypersurfaces. Extending those ideas to interacting quantum field theories is possible at the level of unitary dynamics, while precise definitions of relativistic worlds and branching remain matters of interpretation."
    },
    {
      "heading": "Experimental status and possible discrimination",
      "text": "Experiments confirm superposition, entanglement, interference, and decoherence across a wide range of physical systems. These results establish the phenomena that many-worlds seeks to interpret, but they do not by themselves establish that every decohered outcome is physically realized. An experiment that shows an atom occupying a coherent superposition tests quantum theory; it does not directly count worlds or reveal an additional inhabited universe.\n\nInterference experiments are particularly important because they demonstrate why collapse cannot simply be inserted wherever a system appears complicated. If all information distinguishing two alternatives remains coherently controlled, their amplitudes can interfere. Experiments with atoms, molecules, superconducting circuits, and mechanical devices progressively extend the domain in which quantum coherence can be studied. The significance of a result depends on the size, duration, and structure of the superposition, not just on an object’s mass or on a loosely defined claim of macroscopicity.\n\nSuch tests can constrain specific objective-collapse models. Depending on the model, proposed signatures include loss of interference, additional momentum diffusion, heating, or spontaneous radiation. Ordinary environmental decoherence can mimic some consequences of collapse, so experiments require careful modeling of conventional noise. A null result excludes particular parameter ranges rather than proving the many-worlds interpretation.\n\nThe reverse possibility is conceptually clear: convincing evidence for a fundamental departure from unitary dynamics would challenge the standard Everettian framework. However, identifying such a departure would require distinguishing it from uncontrolled environmental interactions and other errors. The relevant comparison would be between quantitative dynamical models, not between the imagery of “one world” and “many worlds.”\n\nThere is ordinarily no experiment that distinguishes many-worlds from every other interpretation reproducing the same quantum probabilities. Bohmian mechanics in quantum equilibrium, for example, agrees with standard quantum statistics in its usual domain. Other noncollapse approaches can also share the same operational predictions. Where interpretations are empirically equivalent, their comparison involves explanatory structure, ontological commitments, and the treatment of probability, rather than a presently available decisive measurement.\n\nDecohered branches are not ordinarily accessible destinations. Recombining alternatives requires reversing or otherwise controlling the correlations that suppressed interference, including records dispersed through the environment. For a laboratory qubit this may be feasible; for macroscopically different histories involving many uncontrolled degrees of freedom it is overwhelmingly impractical. Claims of communication with alternate personal histories, or direct travel between such histories, do not follow from the interpretation."
    },
    {
      "heading": "Quantum cosmology, information, and computation",
      "text": "Universal quantum dynamics has particular appeal in cosmology, where there is no external laboratory from which the universe can be measured. An Everettian approach treats observers and their records as internal parts of the cosmic quantum state. It therefore avoids requiring a physically external observer to trigger the emergence of a cosmological outcome. This was part of the motivation behind applying relative-state ideas to closed systems from the outset.\n\nThe extension is not straightforward. In canonical approaches to quantum gravity, the quantum state may satisfy constraints rather than an ordinary Schrödinger equation with an external time parameter. The Wheeler–DeWitt equation is a prominent example. Recovering time, classical geometry, and branching histories then becomes part of a larger problem. The interpretation does not supply a completed theory of quantum gravity, although its treatment of correlations can inform attempts to describe observations within such a theory.\n\nCosmological fluctuations provide another setting. Models of cosmic inflation commonly describe early perturbations quantum mechanically and later use them as the seeds of approximately classical density variations. Decoherence helps explain why certain correlations can be treated classically. An Everettian account associates different realized perturbation patterns with different components of the quantum state. Other interpretations can use the same calculations, so agreement between observations and a fluctuation spectrum does not uniquely select many-worlds.\n\nEverettian branching should also be distinguished from the cosmological multiverse associated with some models of eternal inflation. Inflationary domains are regions arising from particular cosmological dynamics; Everettian branches are components or histories of a quantum state. A model can contain both kinds of multiplicity, either one, or neither. Combining them creates additional questions about probability measures and observer selection rather than making the two concepts identical.\n\nIn quantum computing, Deutsch has argued that Everettian ideas offer an illuminating account of quantum information processing. A quantum algorithm manipulates amplitudes in a large state space, and interference can enhance desired outcomes while suppressing others. Nevertheless, it is misleading to describe a quantum computer simply as ordinary computers in separate worlds, each performing an independently readable calculation. The amplitudes must remain coherently related, and a final measurement does not reveal every computational component.\n\nThe practical theory of quantum information does not require commitment to many-worlds. Error correction, entanglement-assisted protocols, and algorithms can be formulated operationally and used by researchers with different interpretations. Their foundational relevance lies in making the physical consequences of coherence, correlations, records, and information access precise—the same features on which modern accounts of Everettian branching depend."
    }
  ],
  "links": [
    "quantum-mechanics",
    "wave-function",
    "measurement-problem",
    "quantum-decoherence",
    "born-rule",
    "niels-bohr",
    "schrodinger-equation",
    "planck-constant",
    "hamiltonian-operator",
    "quantum-superposition",
    "hilbert-space",
    "tensor-product",
    "quantum-entanglement",
    "photon",
    "density-matrix",
    "interference",
    "decision-theory",
    "expected-value",
    "copenhagen-interpretation",
    "bohmian-mechanics",
    "objective-collapse-theory",
    "bells-theorem",
    "special-relativity",
    "quantum-field-theory",
    "cosmology",
    "quantum-gravity"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}