What does it mean to "observe" a quantum system?

1. The problem in plain language

Imagine an electron in a superposition of two possible positions — spin-up and spin-down, say, or here and there. Quantum mechanics describes this state with a wavefunction that contains both possibilities at once. Until someone measures it, the wavefunction evolves smoothly, deterministically, according to the Schrödinger equation. Two paths, both real, both contributing.

Then someone measures it. The screen lights up at one location, not two. The detector clicks once, on one side, not both. The wavefunction appears to have "collapsed" from the two-possibility superposition to a single definite outcome. This collapse, often called the measurement postulate, is added to quantum mechanics by hand. It does not follow from the Schrödinger equation. It is an extra rule, applied at the moment of measurement, that no one has been able to derive from anything more fundamental.

This is the measurement problem. Stated as crisply as I can:

The Schrödinger equation says quantum systems evolve into superpositions and never out of them. Experience says we never observe superpositions; we observe definite outcomes. What process takes us from one to the other — and why?

Every interpretation of quantum mechanics is an answer to that question. None of the answers is uncontroversial. The fact that physics has worked beautifully for a century with this hole at its centre is a measure of how powerful the framework is, and also of how comfortable physicists have become looking past the hole.

2. The von Neumann chain — where does measurement end?

In 1932 John von Neumann formalised what happens during a measurement. The system to be measured starts in a superposition. The measuring device starts in some "ready" state. They interact, and the result is a joint superposition: now the system-plus-device is in superposition. The pointer of the dial is in superposition of "pointing left" and "pointing right." The device has not collapsed anything; it has joined the system in being uncertain.

If a second device measures the first, the same thing happens: now system-plus-device-plus-second-device is in superposition. Add a third measurement, a fourth, the experimenter's eye, the optic nerve, the visual cortex. The chain just keeps growing. Nothing in the Schrödinger equation tells it to stop. The whole world — the universe of quantum mechanical objects participating in the measurement — is now in a giant superposition in which one branch contains a left-pointing dial and a "left-perceiving" brain, and the other branch contains a right-pointing dial and a "right-perceiving" brain.

So where does the collapse happen? Von Neumann's own answer was uncomfortable: somewhere, he said, the chain must terminate, and the only place we have any independent reason to suspect a special process is at the boundary of consciousness. The collapse happens when a conscious observer becomes aware of the result. Wigner, later, agreed. This view came to be called the von Neumann-Wigner interpretation, and it is the most direct historical link between the measurement problem and the question of what consciousness is.

Most physicists reject this view today. It feels mystical, and it places consciousness in a privileged role that physics is generally allergic to. But the von Neumann argument itself — that the chain has no internal cut-off and something must end it — has never been answered. What has been added since are other places one could put the cut-off, and other accounts of what happens at the cut-off. The cut-off remains.

3. Decoherence — the technical answer

The most successful modern response is the decoherence programme worked out by Zurek, Joos, Zeh, and Schlosshauer over the last forty years (the same programme the quantum-classical-line page walks through). The story:

As the measurement chain grows, each new system entangles with the previous one. By the time the chain reaches the experimenter, the original superposition is entangled with everything in the room — the dial, the photons that bounced off the dial, the air molecules they hit, the experimenter's eye, the optic nerve, the visual cortex. The information about which branch the system is on has been smeared into the environment so completely that no laboratory could ever recover it.

What this means: the two branches of the superposition, considered locally, no longer interfere. The mathematical structure of quantum mechanics says they are still both there, in some sense; but operationally, in any experiment anyone can do, the world looks classical from the moment the first photon escapes the apparatus. There is, locally, a single definite outcome. The two branches do not vanish — they decouple. Each remains real in its own branch, with no way to interact with the other.

Decoherence is real, well-confirmed, and explains essentially everything about why we never observe macroscopic superpositions in our daily life. It solves what philosophers call the practical measurement problem — why the world looks classical to us, why dials always point one way, why cats are never simultaneously alive and dead.

It does not solve the philosophical measurement problem — why we only experience one branch. The other branch is still there in the equations. Decoherence tells us why we cannot detect it. It does not tell us why we are not in it.

4. Wigner's friend and its modern variants

The cleanest sharpening of the philosophical problem is Eugene Wigner's 1961 thought experiment, called Wigner's friend.

Wigner's friend is in a sealed laboratory performing a quantum measurement. From the friend's point of view, she observes a definite outcome — the system collapsed, the result is recorded, the experiment is over. From Wigner's point of view, outside the sealed lab, no measurement has happened yet. As far as Wigner can tell, the friend-plus-system is now in a giant superposition: "friend observed up" superposed with "friend observed down." If Wigner could perform a sufficiently clever interference experiment on the whole lab (impossible in practice, possible in principle), he could in fact verify that the friend was in superposition all along.

So who is right? Did the wavefunction collapse for the friend, or didn't it? If both answers are right — collapsed for the friend, uncollapsed for Wigner — then the wavefunction is not a description of an objective state of affairs. It is observer-relative. And if it is observer-relative, what does that mean? Are there two facts of the matter? Are facts themselves observer-dependent?

For decades Wigner's friend was treated as a paradox without a resolution — an interesting thought experiment that physicists could afford to set aside. Then in 2018, Daniela Frauchiger and Renato Renner published a paper showing that if you take quantum mechanics seriously and assume it applies to all observers including conscious ones, you can construct a scenario where four agents reasoning consistently from their respective wavefunctions arrive at contradictory predictions about a single experimental outcome. The contradiction is not subtle. It is a flat, demonstrable impossibility within the standard framework.

The Frauchiger-Renner result has not been overturned. It implies that at least one of the following commonsense assumptions about quantum mechanics has to be abandoned:

• Quantum mechanics applies universally, to all observers, including ones who are themselves observed.
• Different agents reasoning correctly about a shared situation cannot reach contradictory conclusions.
• An observation that has happened has a single definite outcome.

Pick at least one to drop. Which one you drop largely determines which interpretation of quantum mechanics you favour. None of the choices is comfortable. The Frauchiger-Renner paradox is the modern face of Wigner's friend, and it has sharpened the measurement problem rather than blunting it.

5. So what is actually proposed as the mechanism?

This is the question every reader of the double-slit experiment eventually arrives at: physically, what happens at the moment of measurement? The textbook answer — "the wavefunction collapses" — is a postulate, not a mechanism. It tells you what changes; it does not tell you how, or why, or where. The interpretations of quantum mechanics differ precisely because each offers a different account of what is happening between the smooth Schrödinger evolution and the single observed outcome. A reader's guide to the main contenders:

Two layers to keep separate

Before listing the mechanisms it helps to factor the question. Why does interference disappear when which-path information exists? has an empirically settled answer: decoherence. When a quantum system becomes entangled with any environmental degree of freedom that records which path it took — a detector, an apparatus, a photon scattered off the slit edge, even a single air molecule that interacts with the particle — the off-diagonal terms of the density matrix decay exponentially fast and interference vanishes. The trigger for "observation" in the double-slit experiment is not a human eye. It is any interaction sufficient to register which-path information anywhere, even in principle. No consciousness required, no mystery required. This part is mainstream physics. (For the longer version see §3 above and the quantum-classical-line page.)

The deeper question — why does a single outcome occur rather than a continuing superposition or a mixture of all outcomes? — is the measurement problem proper, and this is where mechanisms diverge. The proposals below each give a different answer to that deeper question.

1. Copenhagen (Bohr, Heisenberg) — collapse as a primitive postulate

The textbook account. The Schrödinger equation governs the system between measurements; at the moment of measurement, the system "collapses" into a definite outcome with probabilities given by the Born rule. No mechanism is proposed for the collapse itself. It is added by hand. The measurement device is treated as classical and outside the quantum description. This is the position most physicists are trained to operate inside, and it is the position that leaves the deepest part of the problem unanswered. Copenhagen is not so much a mechanism as a refusal to ask the mechanism question.

2. Many-Worlds (Everett, Deutsch) — there is no collapse

The wavefunction never reduces. The Schrödinger equation is exact and universal. When a measurement occurs the apparatus becomes entangled with the system, the experimenter becomes entangled with the apparatus, the laboratory becomes entangled with the experimenter, and the global wavefunction now contains every possible outcome as a separate branch. Each branch contains a version of the experimenter who observed a particular result. The "single outcome" you remember is what one branch experiences from inside itself. There is no mechanism for collapse because nothing collapses. Deutsch's argument that quantum computers must literally compute across these branches (covered on the quantum-computing primer) is the most aggressive application of this view. Strength: mathematically the cleanest. Cost: commits you to an ontology of branching universes you cannot directly access.

Many-Worlds and Boltzmann brains — an uncomfortable consequence

Hugh Everett's 1957 dissertation proposed the no-collapse interpretation as a way of making quantum mechanics internally consistent without the measurement postulate. The interpretation has a corollary that has bothered physicists ever since: if the wavefunction never reduces, every quantum fluctuation that could in principle produce a momentarily-conscious-thing produces one in some branch. The most-discussed special case is the Boltzmann brain — a self-aware structure that arises from a random thermal fluctuation in a high-entropy environment (the late, heat-death universe, on most cosmological models) and exists just long enough to have one momentary experience before dissipating back into noise. The math is unfavourable: by some estimates the number of Boltzmann brains the universe will eventually produce by random fluctuation vastly exceeds the number of ordinary brains that evolved on planets. If most conscious experiences in the multiverse are Boltzmann-brain experiences, the troubling question is whether your experience is one of them.

The Boltzmann brain paradox cuts in two directions. As an internal consistency challenge for cosmology, it suggests that any theory in which most observers are random thermal fluctuations is in trouble — we are presumably typical observers, and we observe a low-entropy environment full of structured causal history, not the noise field a Boltzmann brain would observe. Sean Carroll, Andreas Albrecht, and others have used the paradox to argue against certain inflationary and multiverse cosmologies on these grounds. As a consciousness question, it sharpens the issue of whether any structure with the right internal organisation is conscious, or whether consciousness requires a substrate with a causal history of the right kind. The Penrose-Hameroff Orch-OR proposal, the Faggin-D'Ariano framework, and Integrated Information Theory all give different answers; none has been settled.

The 2025 Wolpert-Scharnhorst-Rovelli paper on Boltzmann brains, covered in detail on the Rovelli companion essay section 12, sharpens the analysis further by demonstrating that physics alone cannot decide whether memories track a real past without making additional assumptions. The Boltzmann brain problem is structurally similar: physics alone cannot decide whether a given experience belongs to a substrate with a real causal history or to a thermal fluctuation that just appeared. The choice between the two has to be made on grounds physics does not supply.

Are Boltzmann brains conscious? The honest answer: every consciousness-theoretic framework gives a different answer, and the question may not be empirically decidable. Production-model frameworks that locate consciousness in information integration (IIT) tend to say yes — if the integration is the same, the experience is the same, regardless of how the integration arose. Substrate-history frameworks (which include parts of the receiver-model thread the trilogy engages) tend to say less clearly — if consciousness requires the substrate to be coupled to the field through a history of causally-connected pattern-building, then a momentary thermal fluctuation may instantiate the structural conditions for awareness without instantiating the deeper field-coupling that makes awareness be of-something. The trilogy's wager is closer to the substrate-history reading. The framework permits the Boltzmann brain to be conscious; it does not require it; and it makes the Boltzmann brain's relationship to the field a separate question from its instantaneous internal structure.

3. Pilot-wave / de Broglie-Bohm — particles always have definite positions

Particles never go into superposition; they always have definite trajectories. The wavefunction is a real physical field (the pilot wave) that guides them, evolving according to the Schrödinger equation. The apparent randomness of measurement outcomes is due to ignorance of the particle's actual initial position, not to any intrinsic indeterminacy. Measurement reveals what was always there; no collapse occurs because no superposition of particle positions ever existed in the first place. Strength: deterministic, restores ordinary realism. Cost: explicitly non-local (the pilot wave responds instantaneously to distant changes), and the picture for fields rather than particles is much harder to write down cleanly.

4. Objective-collapse models — the wavefunction reduces spontaneously

The wavefunction does physically collapse, but not because anyone observes it. Collapse is a built-in stochastic feature of the dynamics. The main variants:

GRW (Ghirardi-Rimini-Weber, 1986) — every particle has a tiny intrinsic probability per unit time of spontaneously localizing. The rate is tuned to be negligible for a single particle (one localization roughly every 10¹⁶ seconds) but enormous for macroscopic objects (which contain ~10²³ particles and therefore localize essentially instantaneously). This explains why single quantum systems hold their superposition while macroscopic apparatuses do not. Variants include CSL (continuous spontaneous localization).

Penrose-Diósi (gravity-induced collapse) — Roger Penrose's proposal: collapse is triggered by gravity. When a quantum system enters a superposition involving sufficiently different mass-energy distributions, the gravitational self-energy difference between the branches exceeds a threshold and the system spontaneously localizes. Time-to-collapse is roughly ℏ/E_G, where E_G is the gravitational self-energy difference. This is the only mechanism on the list that is directly testable with current technology, and experiments at increasing mass scales (Aspelmeyer's group, MAQRO mission proposals, levitated nanospheres) are actively narrowing the parameter space. So far, no objective-collapse signal has been detected; bounds keep tightening.

5. QBism (Quantum Bayesianism) — the wavefunction is personal belief

Christopher Fuchs, David Mermin, and others: the wavefunction is not an objective state of the world. It is a personal degree of belief held by an agent about the outcomes of experiments she might perform. "Collapse" is what happens when she learns the result — she updates her beliefs, just as a card player updates her estimate after seeing the next card. There is no physical mechanism because nothing physical collapses; only an agent's information state changes. Strength: removes the measurement problem entirely by reframing what the wavefunction is about. Cost: requires accepting that quantum mechanics is fundamentally a theory of agents' bets rather than of the world.

6. Relational quantum mechanics (Rovelli) — states are relations between systems

Carlo Rovelli's interpretation, developed in close company with his work on time. There are no observer-independent quantum states. Every quantum description is a description of one system relative to another. When system A "measures" system B, the result is the establishment of a particular relational fact between A and B; from a third system C's point of view, A-plus-B may still be in superposition until C interacts with the pair. Different observers can each have a complete and correct quantum description of the same system, and these descriptions need not match. "Collapse" is the establishment of a relational fact, not a global event. Strength: handles Wigner's friend cleanly (the friend and Wigner simply have different relations to the system, both correct). Cost: requires giving up the intuition of a single observer-independent reality. Closely allied with Rovelli's broader project on time (see the Rovelli companion essay).

7. Transactional interpretation (Cramer) and the Two-State Vector Formalism

John Cramer's transactional interpretation and the time-symmetric Two-State Vector Formalism (Aharonov, Vaidman) treat measurement as a handshake across time. An emitter sends an "offer wave" forward in time; an absorber sends a "confirmation wave" backward in time; when the two match, a transaction completes and an event is recorded. The collapse is the closing of the handshake. This naturally accommodates retrocausal phenomena like the delayed-choice quantum eraser (see simulation-hypothesis page, entry #E): the past is not a fixed ledger but a log of transactions, some pending until all parties have been processed. Strength: makes time-symmetric experimental results look natural rather than paradoxical. Cost: requires comfort with retrocausal language even if the underlying physics is reformulable without it.

8. Consciousness-causes-collapse (von Neumann-Wigner) — the chain ends in mind

Already discussed in §2, but worth restating here as a mechanism proposal among the others. The von Neumann argument that the measurement chain has no internal stopping point and must terminate somewhere is correct on its own terms; the proposal is that the terminus is consciousness itself. Mind is what collapses the wavefunction. Mainstream physics has largely set this aside as too mystical, but it has never been refuted, and serious contemporary defenders (Henry Stapp, Richard Conn Henry, and parts of David Chalmers's project under some readings) keep it alive. It is the proposal closest to what the trilogy's receiver model points back at — with the important difference that the receiver model treats consciousness as fundamental rather than as a special process attached to certain biological systems.

Where each interpretation puts the mechanism

The honest summary

There is no consensus. The eight options above are not eight competing theories with different predictions; they are eight different stories told about the same mathematics. They all reproduce the experimental record. They disagree about what the mathematics is describing. The only one of the eight that makes different experimental predictions in the directly testable regime is objective collapse (GRW, Penrose-Diósi), and experiments to discriminate it from the others are actively running and so far have not separated it. Everywhere else, the choice is interpretive — a metaphysical commitment, not an empirical decision.

For the trilogy's purposes, the most useful framing is that the receiver model is compatible with most of these interpretations and is most natural with three: relational (states are observer-system relations — see the Rovelli companion essay), QBism (the wavefunction is what an agent knows about its couplings), and Wheeler's observer-participatory framing (no phenomenon is a real phenomenon until observed — see simulation-hypothesis entry #8). It is friendlier to the von Neumann-Wigner reading than mainstream Copenhagen is, but does not require the strong claim that consciousness is the cause of collapse — only that consciousness is constitutive of the world the collapse occurs in. The trilogy's wager is the same one Wheeler made, less polemically: the observer was never an awkward intruder in physics. The observer was the substrate physics was describing all along.

→ For the trilogy's full statement of the receiver-model position, see the Synthesis (the integrated argument), The hard problem, re-stated (how the receiver model dissolves the explanatory gap by abandoning the assumption that the universe started non-conscious), and Information as the foundation (the IIT + Wheeler + Tegmark synthesis the receiver-model field draws on). The next section below sketches the receiver-model resolution at this page's level; the linked essays carry it the full way down.

6. What is an "observer"?

The word observer is doing an extraordinary amount of work in quantum mechanics, and the equations don't tell us what it means. Several candidates have been proposed for what physically distinguishes an observer from any other system:

• Any irreversible amplification. A photon hits a photographic plate and triggers a chemical reaction that cannot be reversed. The reaction is the "measurement." No consciousness required — just thermodynamic irreversibility. (Mainstream view, but has trouble with situations where reversibility is in principle possible but practically not.)
• A specific complexity threshold. The Tegmark proposals: above some level of internal information processing, a system becomes capable of "observing." Vague on where the threshold sits.
• Consciousness. The von Neumann-Wigner view. Consciousness is what collapses wavefunctions. Out of fashion but not refuted.
• No special process at all. The Everett / Many Worlds view. There is no collapse. All branches are real. The "observer" is just another quantum system that becomes entangled with the rest. What we call "observation" is the branch we happen to be in.
• Information-relative reality. The Relational / QBist views. There are no observer-independent facts. Every observer has their own consistent account; "the world" is the consensus across observers.

The honest answer is: physics has not decided. Each option carries philosophical costs. None has been ruled out by experiment. The dispute is not at the level of equations — the equations of quantum mechanics are agreed — but at the level of what the equations are about. And the conscious observer keeps showing up in the conversation, however much physics tries to remove them, because no one has been able to formulate a notion of "measurement" that does not, somewhere, refer to information being registered by something that knows it has registered the information.

7. The measurement problem and the receiver model

Here is where the trilogy takes its position. The receiver model the books are built around says: consciousness is not produced by the brain. Consciousness is a field-property of the universe that brains are configured to receive, decode, and localise. On this view, the universe is conscious-substrate-by-default, and brains are the particular biological architecture that allows one localised consciousness to inhabit one body in one timeline.

The receiver model does not solve the measurement problem in the technical sense. But it dissolves something the measurement problem has been struggling with for ninety years: the awkward, persistent, surprising centrality of the observer in a theory that was supposed to describe a world independent of observation.

If consciousness is fundamental — if it is part of what reality is, not what reality produces — then the entanglement of physical systems with conscious observers is not a peculiar add-on to physics. It is the natural consequence of physical systems interacting with the substrate they were never separate from. Wigner's friend stops being paradoxical: of course wavefunctions are observer-relative, because reality itself is partly constituted by observation. Frauchiger-Renner's three assumptions can be re-evaluated: the assumption that physical observation is a neutral, third-person process that simply registers facts is the one that gives way.

None of this is conclusive. The receiver model is a candidate ontology, not a proof. But the measurement problem has been waiting for ninety years for someone to take seriously the idea that the observer was never an awkward intruder in physics — that the observer was the substrate physics had been describing all along, looking back at itself. The trilogy treats that idea as the only one consistent with both the data and the experience of being a consciousness reading these words.

8. What this means for the trilogy

Three specific touchpoints, where the measurement problem appears explicitly in the books:

• Anima's late chapters — the moment José writes in his folder that the patient with the IED premonition was not "predicting" the future. The patient's consciousness was already coupled to the field that would in due course render the explosion locally observable. The terminology is not Bell's; the structure is. What looks like premonition from inside one branch of a measurement chain looks like ordinary causation from outside it.
• Numen's entanglement scenes between Alex and Alma — two consciousnesses, on two substrates, sharing a single non-local correlate. The hybrid case for the receiver model. Wigner's friend reframed: two friends, one biological and one computational, finding that the wavefunction is the same for both of them because they are the same kind of observer.
• Limen's field cosmology — the trilogy's companion volume lays out the receiver model as direct argument. The wavefunction is not collapsed by observation; observation is what the field looks like when one of its localised modes registers information about another. Every measurement is the field knowing itself, partially, through one of its localised receivers. Limen is where the framework Anima and Numen dramatize is set down as ontology.

The honest summary, again, is the one the site keeps converging on. The measurement problem has been the centre of foundations of physics for ninety years, and no interpretation has cleanly resolved it because every interpretation has had to either deny consciousness, or marginalise consciousness, or relegate consciousness to a special category in a theory that was supposed to be impersonal. The trilogy proposes a different move: take consciousness as fundamental, and let the measurement problem dissolve into a description of how a fundamentally conscious universe registers information about itself. The receiver model is the proposal. The books are the demonstration. The physics is the floor under both.

This page is part of the Reading companion essays. For the entanglement architecture, see Entanglement at every scale; for where the quantum-classical boundary actually sits, the quantum-classical-line page; for the synthesis, The Evidence.