Objective Collapse Models

Speculative If the wave function really does collapse — if the Born rule is dynamics rather than bookkeeping — then there exists a universal noise field with a rate and a length, and both are numbers an instrument can chase. That is the entire prize, and the conditional form is the honest one: no experiment has seen the field, and section 7 says what would.

Established The magnitude is best stated as parameter space deleted per year, because that is the only quantity this programme has so far produced. At a correlation length of 10⁻⁷ m, the strongest bound on the continuous-spontaneous-localisation rate quoted in the field’s 2022 review was λ < 5.2 × 10⁻¹³ s⁻¹, set by an underground germanium measurement published in 2021. In March 2026 a xenon dark-matter detector published λ/r_C² < 3.8 × 10⁻³ s⁻¹m⁻², which is λ < 3.8 × 10⁻¹⁷ s⁻¹ at that length: four orders of magnitude in five years, and the rate Ghirardi, Rimini and Weber originally proposed now sits 9.4 standard deviations from the data.

Frontier Who collects this is unusual, because almost nobody is paid to do it. The bounds above come from a dark-matter detector, a neutrinoless-double-beta-decay germanium array, a gravitational-wave technology demonstrator and a cryogenic bolometer — four instruments built for other questions, whose collapse limits cost a re-analysis of data already on disk rather than a facility. The gain accrues to the foundations of physics and to four detector collaborations’ publication lists. There is no product at the end of it, and this page does not pretend otherwise.

Handwave The condition, stated plainly: this upside is real only if the collapse noise is white, Markovian and mass-proportional — which is precisely the version being killed. Coloured and dissipative extensions restore free parameters at least as fast as experiment removes them, and whether that is legitimate model-building or an indefinite retreat is the argument section 5 is about.

Objective collapse models are not an interpretation of quantum mechanics. They are a different theory. The Schrödinger equation is modified by a stochastic, non-linear term that localises the wave function in position at a rate that grows with the mass of the system; microscopic systems behave as quantum mechanics says because that term is negligible for them, and macroscopic ones have definite positions because for them it is not. The single outcome of a measurement becomes a physical event rather than a change of description, and the price of that is a set of predictions ordinary quantum mechanics does not make.

Established Three models carry the literature, and they differ in what they blame. The Ghirardi–Rimini–Weber (GRW) model and its continuous descendant, continuous spontaneous localisation (CSL), postulate a classical noise field of unstated origin with two free parameters: a collapse rate λ and a correlation length r_C. The Diósi–Penrose (DP) model blames gravity — a superposition of two mass distributions curves spacetime in two ways, and the resulting ambiguity decays the superposition — leaving one free parameter, a short-distance cutoff R₀ on the Newtonian self-energy. Speculative None of them derives its parameters from anything. They are inputs, and that is the deepest structural weakness in the field.

Established What makes this an experimental science rather than a philosophy seminar is that the same added term produces two effects that are hard to hide. First, it destroys interference, so an interferometer holding a heavy enough object in superposition for long enough will lose fringes where quantum mechanics says it should not. Second, and far more usefully, it is a stochastic force: it heats matter, jiggles it, and — because a jiggled charge radiates — makes ordinary matter emit a faint X-ray spectrum with no source. The second effect is why the tightest limits on this page come from detectors sitting in the dark under a mountain rather than from interferometers.

Frontier The framing this page tests is that objective collapse is a parameter space being closed from below by experiment, and it survives with one large qualification. The closing is real, measured, and running faster than the field’s own 2013 review expected. The qualification is that what is being closed is the smallest version of the space — white noise, Markovian, mass-proportional — and each model has generalisations with extra parameters that the same data barely touches. Speculative Whether those generalisations are physics or bookkeeping is not a question an instrument can settle, and it is the question this subject will live or die on.

Established The strongest channel is spontaneous radiation, and its record since 2021 is the cleanest quantitative story in this brief. A collapse noise field accelerates charges; accelerated charges radiate; so a block of shielded matter deep underground should glow faintly in the X-ray band with no radioactive explanation. An experiment at the Gran Sasso National Laboratory (LNGS), published in 2021, measured that emission with a germanium detector and found none, setting a lower bound on the effective size of the nuclear mass density about three orders of magnitude above previous bounds and ruling out the natural, parameter-free version of the Diósi–Penrose model. Established The result carries unusual weight because Lajos Diósi, who proposed the model, is among the authors of the measurement that killed its parameter-free form.

Established The germanium channel then improved twice more, both times on data taken for other physics. The Majorana Demonstrator — a 44 kg array of point-contact germanium detectors, 30 kg of it enriched in ⁷⁶Ge, built to look for neutrinoless double-beta decay — searched 19–100 keV in a 37.5 kg-year enriched exposure collected between December 2015 and November 2019. It found no such radiation, set the most stringent white-CSL limit then available with a factor of 40 to 100 improvement in sensitivity over comparable searches, and improved the Diósi–Penrose lower bound by almost an order of magnitude at 95% confidence. Frontier The collaboration tested mass-proportional and non-mass-proportional CSL under both quasi-free-electron and coherent-nuclear emission assumptions — four cases rather than one, which is how to report a bound that depends on a modelling choice.

Established Then, in March 2026, a dark-matter detector took the channel two further orders of magnitude and excluded the values the theory was born with. XENONnT analysed low-energy electronic recoils from 1 to 140 keV in a 1.16 tonne-year first science run, using a model that for the first time accounts for cancellation between the opposing electron and proton charges in the xenon atom — an atomic-physics correction, not a counting improvement. The result was no excess and new world-leading limits: λ/r_C² < 3.8 × 10⁻³ s⁻¹m⁻² at 90% confidence, which is λ < 3.8 × 10⁻¹⁷ s⁻¹ at r_C = 10⁻⁷ m, a factor of about 120 beyond the previous best, and R₀ > 4.5 × 10⁻¹⁰ m for Diósi–Penrose, a factor of five. Established The GRW parameter choice sits 9.4 standard deviations from the best fit and is excluded for the first time. The paper is careful about scope: this is the most stringent constraint for r_C below about 10⁻⁵ m, and a wider range remains open for coloured or dissipative variants.

Established The mechanical channel is weaker per experiment and much broader in reach, and it is entirely made of by-products. The 2022 Nature Physics review of non-interferometric tests puts the strongest bounds at r_C = 10⁻⁷ m as follows: cryogenic phonon excitation in CUORE at λ < 3.3 × 10⁻¹¹ s⁻¹; multilayer cantilevers at 2.0 × 10⁻¹⁰; LISA Pathfinder at 3.8 × 10⁻⁹; cold atoms at 5.1 × 10⁻⁸; and LIGO, the largest and least sensitive to this particular effect, at 1.0 × 10⁻⁵. Established A 2024 re-analysis of LISA Pathfinder’s kilogram test masses across all ten science operations tightened that to λ < 8.3 × 10⁻¹¹ s⁻¹ at the same correlation length, with R₀ > 285.5 fm for the DP model; a 2025 analysis of the same spacecraft’s angular motion improved the bound by about a factor of two for r_C between roughly 10^−5.5 and 10^−3.5 m. A dead spacecraft is still producing limits on the measurement problem, which is a fair summary of how this field is funded.

Established The interferometric channel moved in January 2026 for the first time in seven years, and by a factor of seven in mass. The previous record was molecular interference beyond 25 kDa, set in 2019. A Talbot–Lau interferometer with three ultraviolet gratings of 133 nm period spaced 0.983 m apart has now interfered sodium clusters of more than 7,000 atoms at masses above 170 kDa, roughly 8 nm across, at de Broglie wavelengths of 10 to 22 femtometres, with a fringe visibility of 0.10 ± 0.01 at 172 kDa. Established The reported macroscopicity is µ = 15.5, about an order of magnitude past the previous record, corresponding to excluding minimal macrorealistic modifications with a classicalisation time τ_e ≥ 2.84 × 10¹⁵ s. Frontier In absolute mass this is 2.9 × 10⁻²² kg, which is still five orders of magnitude short of where the decisive experiment in section 7 has to be run.

Frontier Theory supplies the other side of the vice, and it is doing more work than most summaries admit. A collapse model that is too weak does not do its job: the standard requirement is that a single-layer graphene disk of radius 10 µm — about the smallest thing a human eye resolves — must collapse within 0.01 s, about the eye’s time resolution. That assumption is what puts a floor under the CSL parameter space, and it is an assumption rather than a measurement. Frontier The same reasoning applied to Diósi–Penrose in 2024 produced a genuinely awkward result for the model: the collapse of a human-eye-resolution graphene sheet is about two orders of magnitude slower than the observation time, so DP does not guarantee classicality for objects you can see, and requiring that it should implies R₀ of at most about 10⁻⁴ m. Established Set against the measured floor of 4.5 × 10⁻¹⁰ m, the entire surviving Diósi–Penrose model now lives in a window of about five and a half orders of magnitude in one parameter, bounded below by an X-ray measurement and above by the requirement that the model do the job it was invented for.

Frontier Two further theory results constrain the models from inside. Dissipative CSL — the variant that lets the noise field have a finite temperature so that objects stop heating forever — was tested in 2024 against three levitated-optomechanics datasets, and the entire dissipative parameter space is excluded for collapse-field temperatures below 6 × 10⁻⁹ K, with the Adler parameter proposal excluded at every temperature examined and the GRW proposal excluded for T_β at or below 10⁻⁵ K. Speculative Separately, a 2025 thermodynamic analysis finds that the original DP model induces unbounded heating, consistent with the second law only if the noise field is assigned infinite temperature, and that only the dissipative extension at low friction thermalises properly. These are not experimental exclusions, but they are the kind of internal result that closes a model without an instrument.

Speculative And the record of what is not established should be stated as flatly as the record of what is. There has never been a positive collapse signal in any channel. No model derives λ, r_C or R₀ from a deeper theory; they are fitted to nothing and bounded from both sides. Speculative Until 2025 there was no relativistic collapse model at all that avoided microcausality violation, infinite energy production, spontaneous particle creation from the vacuum and tachyonic behaviour all at once; a construction using local field collapse operators with non-Markovian Lorentz-invariant noise correlations and a normal-ordering prescription now claims a finite energy rate without vacuum particle production. Handwave It is one paper, one year old, and nobody has yet extracted an observable from it.

Frontier The live positions here are unusually well separated, and several are held by people who agree about every measurement and nothing else. They are set out as positions with holders rather than as a consensus with a fringe.

Speculative Position one: white Markovian CSL is still alive, just not where it started. The X-ray bound scales as λ/r_C², so it bites hardest at short correlation lengths and leaves the long-r_C region to mechanical experiments that are six orders of magnitude weaker in λ. Established The originally proposed GRW point is excluded and the Adler proposal has been excluded for years. Speculative What survives is a band nobody proposed, which is a real position and a much weaker one than the theory started with.

Speculative Position two: the noise is coloured, and the white-noise exclusions therefore do not apply. Non-Markovian CSL introduces a noise spectrum with a cutoff frequency, and the spontaneous-radiation bound — which depends on the noise having power at the X-ray frequencies where the emission is looked for — weakens sharply if the spectrum is cut off below them. Frontier The XENONnT paper says so itself: a wider parameter range remains to be probed for coloured or dissipative extensions. Handwave The version of this position that treats a cutoff as a free parameter to be set wherever the latest bound fell is not a research programme, and the field is not always careful about which version it is defending.

Speculative Position three: collapse falls out of string theory, and is coloured for a reason rather than for convenience. A 2026 paper proposes that in a scenario where cosmic acceleration comes from instant folded strings and their decay products, collapse is driven by stochastic fluctuations of the Newtonian potential, producing temporally coloured noise and a model its authors describe as significantly less constrained by existing experiments than standard Diósi–Penrose. Handwave It makes no numerical prediction for a rate or a length, so it currently adds a mechanism rather than a test.

Frontier Position four is an attribution dispute with experimental consequences: what exactly did Gran Sasso kill? Diósi’s dynamical model predicts spontaneous radiation and is the thing the germanium and xenon measurements constrain. Established Penrose’s own proposal is a decay-rate criterion rather than a noise dynamics, and does not by itself predict spontaneous radiation at all — a point made by the LNGS group’s own leader when discussing the result. Frontier So the headline “gravity-related collapse ruled out” is right about one model and premature about a family, and the people running the experiment say so more carefully than the coverage of it does.

Frontier Position five is where this brief touches Quantum Gravity and runs the opposite way. A 2024 result argues that the Diósi–Penrose model — a model of classical gravity — can itself entangle the mechanical degrees of freedom of two separated particles, below a critical separation set by its one free parameter, with the entanglement surviving more than a day for reasonable experimental parameters. Speculative If that holds, detecting gravitationally induced entanglement would not establish that gravity is quantised; it would, however, be a sharp test of DP. Frontier Both halves are contested. One paper argues classical gravity cannot mediate entanglement at all and that the witness therefore stands; another argues that collapse-based gravity models are fundamentally non-local, that the entanglement in them comes from a hidden monitoring process with quantum features rather than from gravity, and that the general witness theorem is vindicated rather than broken. The dispute is live, it is between named groups, and this page takes no side.

Speculative Position six: the whole programme is unnecessary. Environmental decoherence plus a no-collapse ontology — many-worlds, or pilot-wave — reproduces every observation without adding a field, and adds no parameters to be excluded. Frontier The counter is that decoherence explains the absence of interference and not the occurrence of a single outcome, which is the thing collapse models were built to deliver. Handwave No experiment in this brief distinguishes these positions, and none of the proposed ones would.

Handwave Position seven, and the one that would end the field without a measurement: the parameter space has no bottom. On this reading every exclusion is answered by a generalisation — colour, dissipation, non-mass-proportionality, a relativistic smearing — and the programme degenerates into a sequence of models each tuned just past the last null. Frontier The honest rebuttal is not rhetorical but technical, and it exists: a 2026 proposal to bound λ from Coulomb-mediated squeezing of two charged nanospheres reports bounds comparable to X-ray emission that are explicitly robust against coloured-noise extensions. A test that survives the escape hatch is worth more to this field than another decade on a test that does not.

Frontier The binding bottleneck is not an instrument. It is that the models generalise faster than the instruments exclude. Minimal CSL has two free parameters. Allowing the noise to be coloured adds a cutoff frequency. Allowing it to dissipate adds a temperature. Dropping mass proportionality adds a scaling exponent. Established Each of those extensions was introduced for a defensible physical reason, and each of them also happens to move the model out of the region the last experiment closed. Handwave There is no accepted rule in this field for when an extension is legitimate and when it is retreat, and until there is, the phrase “collapse models are excluded” will never have a stable referent.

Frontier The second bottleneck is the interferometric mass gap, and it is five orders of magnitude wide. The heaviest object ever brought to interfere with itself is a sodium cluster at about 2.9 × 10⁻²² kg. The decisive experiment in section 7 wants a particle between 10⁻¹⁷ and 10⁻¹⁴ kg. Established Every enabling result in levitated optomechanics — ground-state cooling of nanoparticles, ultralow-damping magnetic levitation — addresses one side of that gap, and none of them closes it. Frontier Getting heavier makes superposition harder, which is why this gap has moved by a factor of seven in seven years.

Frontier The third is environmental isolation, and it has specific numbers attached. A published blueprint for distinguishing collapse from decoherence in a 50 nm, 10⁻¹⁷ kg dielectric nanosphere asks for pressure below 10⁻¹⁵ mbar, temperature below 5 K, a trap frequency of 10⁵ Hz, a zero-point width of 7 × 10⁻¹³ m, and a coherence time exceeding 10 s against an environmental decoherence baseline of about 3 × 10⁻³ s⁻¹ with an expected CSL signal near 3 × 10⁻⁴ s⁻¹. Speculative The signal is an order of magnitude below the noise floor the same document specifies, which is the honest statement of how hard this is.

Frontier The fourth is that the X-ray bound depends on an atomic-physics calculation as much as on a count rate. XENONnT’s factor of 120 came partly from exposure and partly from being the first analysis to account for cancellation between opposing electron and proton charges in the emitted spectrum. Frontier That is a modelling input, and modelling inputs get revised: a recalculation of atomic correlation effects in the emitted spectrum is exactly the kind of result that can move a world-leading limit without any new data.A bound resting on a spectral calculation is weaker than one resting on a null count, and this one rests on both.

Speculative The fifth is theoretical and may be terminal: there is still no relativistic collapse model anyone has extracted an observable from. Without one there is no cosmological test, no high-energy test, and no principled statement about what the noise field is made of. Handwave A theory that only exists in the non-relativistic limit is not yet a candidate for fundamental physics, however good its bounds look.

Established This brief waits on capabilities rather than on results other briefs produce, and the capabilities belong to other fields entirely. Ultra-low-background underground detectors at tonne scale, built and operated by the dark-matter and neutrinoless-double-beta-decay communities, supply every strong bound in section 3. Levitated optomechanics and matter-wave interferometry, developed for Precision Quantum Sensing and for inertial metrology, supply the apparatus the decisive experiment needs. Frontier Neither community is trying to settle the measurement problem, which is exactly why their nulls are worth more than a partisan’s.

Frontier One dependency is genuinely unusual: this field needs better atomic structure calculations. The conversion from a collapse rate to a predicted X-ray spectrum runs through the internal charge structure of the target atom, and the strongest bound in the brief moved by two orders of magnitude partly because somebody did that conversion more carefully. Speculative A field whose world-leading limit can shift on a theory calculation has a dependency most experimental programmes do not.

Speculative The dependency that nothing supplies is a reason for the parameters. Trace dynamics, gravitational self-energy and now a string-theoretic mechanism have each been offered as an underlying origin for the collapse noise; none of them yields a number. Handwave Until something does, every exclusion plot in this subject is a map of a region whose interesting point nobody can mark.

Frontier Traffic with the rest of this map runs mostly outward. Quantum Gravity already carries the Gran Sasso falsification as its single cleanest result and treats gravity-mediated entanglement as one sharply posed question inside a gravity programme; this page owns the inverse, and supplies that brief with the reason its entanglement test may not prove what it is advertised to prove. Speculative Consciousness Research inherits the constraint on gravity-related reduction; Fault-Tolerant Quantum Computing inherits, at present very loosely, a mass-dependent ceiling on how large a coherent register can be. Nothing on this map blocks this brief.

Frontier Every position in section 4 has a stated test, and they rank by what a result would settle rather than by how hard they are.

Speculative Tier one, and the decisive experiment for this brief: matter-wave interferometry with a levitated nanoparticle between 10⁻¹⁷ and 10⁻¹⁴ kg. This is the only route that can return a positive result rather than a tighter limit, and the only one that separates collapse from environmental decoherence by its shape rather than by its size. Frontier The published discriminants are specific: CSL-induced decoherence saturates once the superposition separation exceeds the correlation length, while environmental decoherence keeps growing as the square of the separation; and the collapse rate scales as the square of the mass. Speculative A blueprint using a 50 nm, 10⁻¹⁷ kg sphere with a conditional displacement gate driven by an internal two-level defect targets coherence beyond 10 s and, on a null, bounds λ at the 10⁻²² to 10⁻²¹ s⁻¹ level. Handwave No such apparatus exists, and the mass gap to it from the January 2026 interference record is five orders of magnitude.

Established Tier two is already running and has just delivered: spontaneous X-ray emission, pushed with bigger detectors and better spectra. The standard statement of what this channel needed next was a further decade in λ; XENONnT delivered two decades in March 2026 and excluded the GRW point at 9.4 standard deviations, so the near-term version of that test has already been run. Frontier What remains is exposure and energy threshold: the VIP collaboration’s broad-energy germanium detector at LNGS has used a denoising autoencoder and a convolutional classifier to push its minimum detectable energy from about 20 keV to about 10 keV, with a 14% signal-to-background improvement and 1.0 to 1.5 keV resolution at the 46 keV line. Speculative The channel’s ceiling is structural rather than technical: because the bound scales as λ/r_C², more exposure closes short correlation lengths and does almost nothing to the long-r_C region where the surviving parameter space now lives.

Speculative Tier three is the one test designed to be immune to the field’s escape hatch. A 2026 proposal bounds the collapse rate by detecting steady-state Coulomb-mediated reduction of thermal variance in the differential motional mode of two charged nanospheres, reaching λ < 2.09 × 10⁻¹⁴ s⁻¹ at r_C = 10⁻⁷ m for realistic parameters. Speculative Its value is not the number, which is weaker than XENONnT’s, but that the bound is explicitly robust against coloured-noise extensions and surpasses bulk-heating constraints. A test that cannot be evaded by adding a cutoff frequency is the single most useful thing an experimentalist could build for this subject right now.

Speculative Tier four is free fall in orbit, because the gap is coherence time and gravity is what limits it on the ground. The MAQRO Pathfinder white paper of December 2025 proposes optically levitated silica nanoparticles up to about 10⁹ atomic mass units in low Earth orbit at 600 km, with free-fall times of at least 10 s and a target above 100 s, vibration below 10⁻⁹ g, on a 5 to 10 kg ride-share or a 150 kg standalone platform drawing about 30 W. Frontier It has been submitted to the UK Space Agency’s Frontiers 2035 call and carries no budget or launch date, which is the accurate status of every space-based collapse test.

Frontier Tier five tests a model against itself rather than against an instrument, and it has already excluded more than most experiments. Requiring the model to collapse a visible object — a 10 µm graphene disk within 0.01 s — bounds the space from below, and applied to Diósi–Penrose it yields R₀ of at most about 10⁻⁴ m against a measured floor of 4.5 × 10⁻¹⁰ m. Speculative The same style of argument applied to dissipative CSL excluded its entire parameter space below a collapse-field temperature of 6 × 10⁻⁹ K. Consistency requirements are cheap and they are doing real work here.

Speculative Tier six is the experiment that belongs to a neighbouring brief and would land here as well. If the Diósi–Penrose dynamics really does entangle two separated masses below a critical distance set by R₀, then the gravity-mediated entanglement apparatus described in Quantum Gravity is simultaneously a DP test, and a null at the predicted separation would constrain the model where no X-ray detector can reach. Handwave Whether the prediction is real is exactly what the 2025 and 2026 exchange in section 4 is arguing about, so this tier currently has two possible meanings and no apparatus.

Established And the experiments that are not tests of collapse and are regularly reported as though they were. Ground-state cooling of a levitated nanoparticle, ultralow-damping magnetic levitation, and decoherence measurements in superconducting circuits are measurements of environmental decoherence at extraordinary precision. Speculative They are prerequisites for tier one and they are not evidence about collapse, because the quantity they measure is the background the collapse signal has to be separated from.

Frontier There is no engineering payoff here and the page will not manufacture one. This is prerequisite science, and every requirement below is a requirement for an instrument that would answer a question rather than build anything — which is why section 10 records a funding condition as a formal requirement.

Speculative The instrument specification for the decisive test, taken from the published blueprint rather than gestured at. A dielectric nanosphere of 50 nm radius and 10⁻¹⁷ kg, about 6 × 10¹⁰ atomic mass units, held in a high-frequency optical dipole trap at 10⁵ Hz with a zero-point width of 7 × 10⁻¹³ m; a conditional displacement gate driven by an internal two-level defect to open the superposition; residual pressure below 10⁻¹⁵ mbar; environment below 5 K; coherence held past 10 s; and a Bayesian inference protocol to separate collapse-induced excess decoherence from calibrated noise. Handwave Each of those is at or beyond the current state of the art individually, and the experiment needs all of them at once.

Established The underground half of the programme has an engineering specification too, and it is already built. A 44 kg point-contact germanium array with 30 kg enriched in ⁷⁶Ge, resolving 19 to 100 keV over a 37.5 kg-year exposure. A 1.16 tonne-year liquid-xenon time projection chamber resolving 1 to 140 keV. A broad-energy germanium detector at LNGS reaching 10 keV with machine-learned pulse-shape discrimination at 95% classification accuracy. Frontier Not one of these was built to test quantum mechanics, and all of them now do; the space platform in section 7 is the lightest specification of the three and the only one nobody has funded.

Established The adjacent fields supplying the evidence on this page are not foundations-of-physics fields, and that is the point. Dark-matter direct detection, neutrinoless double-beta decay, cryogenic bolometry and space-based gravitational-wave technology supply every strong bound in section 3. Frontier None of those communities is trying to solve the measurement problem, and their limits are worth more for it.

Frontier Within this map the most important adjacency is Quantum Gravity, and the two pages are deliberately inverse. That brief treats gravity-mediated entanglement as one sharply posed yes/no question inside a gravity programme, and carries the Gran Sasso falsification as its cleanest single result. This page starts from the falsification and asks what remains of the parameter space around it. Frontier The seam between them is the 2024 claim that a classical-gravity collapse model predicts the very entanglement that experiment is meant to witness — which, if it holds, makes that brief’s decisive test partly a test of this brief’s subject. The entanglement module covers the phenomenon itself.

Speculative Three further neighbours inherit something concrete. Precision Quantum Sensing builds the levitated platforms this subject needs and gains nothing from it in return. Fault-Tolerant Quantum Computing would inherit a hard ceiling on coherent register size if collapse were real, though at the rates now excluded the ceiling is far above anything being built. Consciousness Research and Integrated Information Theory sit downstream of the gravity-related-reduction result through the orchestrated-objective-reduction proposal, which is handled in section 15. Handwave Claims that collapse models supply a mechanism for consciousness run in the opposite direction from the evidence and are treated here as a misconception rather than as an adjacency.

Frontier One adjacency is methodological rather than physical. Neutrino Astronomy, Laboratory Astrophysics and Gravitational Wave Engineering share this subject’s economics: rare-event searches in instruments whose primary science is something else. That is the business model of this entire brief.

Established The institutional fact that explains this field’s productivity is that its strongest results cost almost nothing at the margin. XENONnT exists to find dark matter. The Majorana Demonstrator existed to find neutrinoless double-beta decay. CUORE is a bolometer array for the same purpose. LISA Pathfinder was a technology demonstrator for a gravitational-wave observatory. Every world-leading collapse bound in section 3 is a re-analysis of data taken for another question, at the marginal cost of a re-analysis rather than a facility.

Frontier That arrangement is efficient and fragile. The bounds improve when dark-matter and double-beta-decay programmes are funded and not otherwise, and no agency anywhere is deciding how much collapse-model sensitivity to buy. Speculative If the dark-matter programme winds down after a run of nulls, the strongest constraint channel on objective collapse winds down with it, for reasons having nothing to do with quantum foundations.

Frontier The positive-signal experiment has the opposite problem: it is nobody’s by-product. A levitated nanoparticle interferometer at 10⁻¹⁷ kg with 10 s of coherence at 10⁻¹⁵ mbar has no other customer, will publish nothing for years, and can fail in a way that yields no paper. Speculative The funding that exists for it is small, specialised and philanthropic or national-agency-exploratory rather than mainstream: a foundational-questions institute co-announcing the XENONnT result, a UK Space Agency horizon call receiving the MAQRO Pathfinder white paper. This is exactly the instrument class that patient foundational money exists to buy and that short-cycle competitive funding cannot.

Speculative There is one institutional requirement this field genuinely lacks, and it is not money. Nobody has written down a rule for which model extensions count. Colour, dissipation, non-mass-proportionality and relativistic smearing are each defensible and each conveniently relocates the model past the latest null. Handwave A pre-registered convention — agreed before the next result, naming which variants are considered live and what would retire them — would cost nothing and would be worth more to this subject’s credibility than another order of magnitude in λ.

Frontier The hazards here are epistemic rather than physical, and the largest is a specific one: this is the field most exposed to the temptation of an unfalsifiable retreat. A theory whose free parameters can always be extended has a standing invitation to survive any measurement, and a community that accepts each extension without saying what would count as defeat converts a falsifiable programme into an unfalsifiable one without anybody deciding to. Speculative The field’s own best defence against this is to privilege tests like the Coulomb-squeezing proposal that are explicitly robust against the extensions.

Frontier The second is the consciousness connection, which is the most publicly visible thing about this subject and the least supported by it. The orchestrated-objective-reduction proposal ties conscious moments to gravity-related state reduction in neuronal microtubules, so a bound on gravity-related collapse is a bound on part of that proposal. Established The Gran Sasso group, who ran the measurement, concluded that almost none of the scenarios remain plausible, while also noting explicitly that Penrose’s own version does not predict spontaneous radiation and so remains untested. Handwave Reporting that as “physics disproves quantum consciousness” overstates what a null on one dynamical model can do, and reporting it as “Orch-OR survives” understates it.

Established Third, and to the field’s credit: every result in this brief is a null, and the people publishing them include the people whose models they constrain. The originator of the Diósi–Penrose model is an author on the measurement that ruled out its parameter-free form. Interest running that hard against a finding is the strongest evidential signal available in any field, and it should be said out loud.

Speculative Fourth, the allocation question, which deserves better than a slogan. This work competes for detector time, cryogenic facilities and theorists with dark-matter searches and quantum information. The defensible answer is not that the measurement problem matters more, but that the marginal cost is near zero for the underground half and shared instrumentation for the levitated half. Handwave That answer stops working the moment somebody proposes a dedicated facility, and nobody has had to make the real argument yet.

Speculative If a collapse signal were found, the immediate consequence is a new fundamental field — universal, coupled to mass, with a measured rate — and the honest statement about what follows from that is that nobody knows. The base rate for foundational discoveries producing capabilities is poor and slow, and the honest expectation is decades of nothing.

Speculative The one consequence that is concrete and calculable is a ceiling. If collapse is real at any rate, there is a maximum mass-times-time over which a quantum superposition can be maintained, and therefore a hard upper bound on the size of a coherent quantum computation that no amount of error correction removes, because the noise is postulated to be fundamental rather than environmental. Established At the rates now excluded that ceiling sits far above anything anyone is building, so this is a consequence of the theory rather than a present engineering constraint on Fault-Tolerant Quantum Computing.

Speculative A second calculable consequence is that energy is not conserved, at a rate the experiments bound. Collapse noise heats matter; the original Diósi–Penrose dynamics heats without bound unless the noise field is assigned infinite temperature, and the dissipative repair thermalises only at low friction. Handwave A universe with a universal heating term is a different cosmological object from one without, and no cosmological consequence of collapse has been worked out to the point of a prediction.

Speculative And the outcome that is more likely than either: nothing is found, and the value is what was eliminated. If the programme ends with the whole accessible parameter space closed, the measurement problem returns to being a question about interpretation rather than about dynamics, and a century of proposals that promised an experimental resolution will have been tested and found empty. Frontier That is a real civilisational result and an unpopular one: it is the case where the most useful thing a research programme produced was a clean account of what is not the case.

These horizons track instrument sensitivity and parameter-space closure, because there is no engineering programme here to track:

• 10 yr: Frontier Expect the underground channel to keep closing short correlation lengths as xenon and germanium exposures grow and thresholds fall toward 10 keV, with the next factor of ten in λ arriving from exposure rather than from ideas. Speculative Expect the first serious attempts at levitated-nanoparticle interferometry at masses still short of 10⁻¹⁷ kg, reporting decoherence budgets rather than collapse limits. Speculative Expect the Coulomb-squeezing and rotational-noise routes to produce their first real data, and to matter more than their raw numbers suggest because they are robust against coloured-noise extensions.
• 25 yr: Speculative This is the window in which the decisive experiment plausibly runs: a 10⁻¹⁷ kg particle in superposition for 10 s, on the ground or in orbit. Speculative A positive result — excess decoherence saturating with separation and scaling as the square of the mass — would be the first new fundamental field in decades and would immediately reopen every question in section 4. Speculative A null at those parameters closes minimal CSL essentially completely and moves the whole argument onto the extensions.
• 50 yr: Handwave On the theory side, the plausible development is consolidation: a relativistic model with extractable observables, or the recognition that the non-relativistic restriction is structural rather than temporary. Speculative If the accessible parameter space is closed and the extensions are still being generated, the subject ends sociologically rather than empirically, and that is the modal outcome on this horizon.
• 100 / 250+ yr: Handwave Beyond useful forecasting. The defensible statement is about shape rather than date: a theory whose free parameters are bounded from both sides either gets measured or gets abandoned, and this one has been bounded from both sides for less than a decade. Handwave Any specific claim about what a confirmed collapse field would be used for is an assertion, and this page does not make one.

• Depends on No typed dependency edge to another brief on this map, and the absence is informative rather than an omission. This subject waits on instruments and on one theoretical result, none of which any brief here produces. Quantum Gravity is the nearest page and supplies context rather than a prerequisite: it carries the same Gran Sasso falsification as one result among many in a gravity programme, where this page treats it as the anchor of a parameter space. Precision Quantum Sensing develops the levitated hardware but is not blocking, because that hardware is being built for metrology whether or not anyone tests collapse with it.
• Requires (not on this map) Six conditions, in the order of the tokens above. A derivation of the rate. λ, r_C and R₀ are fitted to nothing; every exclusion plot in this subject is a map with no marked target, and trace dynamics, gravitational self-energy and the 2026 string-theoretic mechanism have each been offered as an origin without yielding a number. The interferometer. A matter-wave experiment on a particle of 10⁻¹⁷ to 10⁻¹⁴ kg — ten femtograms to ten nanograms — held coherent past 10 s below 10⁻¹⁵ mbar and 5 K; the record today is 2.9 × 10⁻²² kg, five orders short at the light end. A relativistic model. The 2025 construction using local field collapse operators and non-Markovian Lorentz-invariant noise avoids infinite energy production and vacuum particle creation, but nobody has extracted an observable from it, so there is still no cosmological or high-energy test. The detectors. Tonne-scale radiopure xenon and enriched germanium with thresholds near 10 keV, which exist only because dark-matter and double-beta-decay programmes pay for them. The platform. A small satellite giving 10 to 100 s of free fall at 10⁻⁹ g, proposed in December 2025 with no budget and no launch date. The convention. An agreement, written before the next result, on which model extensions are live and what would retire them — the cheapest requirement on this list and the one whose absence does the most damage.
• Enables A positive result would put a hard, error-correction-proof ceiling on the size of a coherent quantum computation, which is the one concrete consequence for Fault-Tolerant Quantum Computing, and would settle part of the gravity-related-reduction question that Consciousness Research inherits. A null across the accessible space enables nothing and eliminates a great deal, which is the more likely and the less discussed outcome. No typed enabling edge is claimed, because a field that has never been detected enables nothing on a schedule.
• Adjacent Dark-matter direct detection, neutrinoless double-beta decay, cryogenic bolometry, levitated optomechanics and matter-wave interferometry supply every number on this page. Within this map: Quantum Gravity for the gravitational half of the question, Precision Quantum Sensing for the apparatus, Neutrino Astronomy and Laboratory Astrophysics for the rare-event-search economics, Quantum Materials for the low-background fabrication, and the entanglement module for the phenomenon at the centre of the dispute in section 4.

Established “Objective collapse is an interpretation of quantum mechanics.” It is not. Interpretations agree on every prediction; collapse models predict effects — loss of interference at high mass, spontaneous heating, spontaneous X-ray emission — that standard quantum mechanics does not. Established That is why this subject has exclusion plots and the many-worlds debate does not.

Established “The Diósi–Penrose model has been ruled out.” The parameter-free version was ruled out by the Gran Sasso measurement published in 2021, and the bound has since moved to R₀ > 4.5 × 10⁻¹⁰ m. Frontier The model with R₀ as a free parameter survives in a window running up to about 10⁻⁴ m, where it stops collapsing visible objects fast enough to be doing its job. Speculative Five and a half orders of magnitude is a much weaker position than a parameter-free theory and it is not nothing.

Frontier “Gran Sasso tested Penrose’s idea.” It tested Diósi’s dynamical implementation of it. Established Penrose’s own proposal is a decay-rate criterion rather than a stochastic dynamics and does not by itself predict spontaneous radiation, a distinction the experimental group makes in its own public statements about the result. Speculative The clean falsification everyone cites is a falsification of the version that was specific enough to be falsified, which is how this usually goes.

Established “GRW has been ruled out.” As of March 2026 the specific rate and correlation length Ghirardi, Rimini and Weber proposed are excluded, at 9.4 standard deviations from the best fit, by a xenon dark-matter detector. Speculative The CSL family is not ruled out: the surviving white-noise band sits at correlation lengths above about 10⁻⁵ m, where the bounds come from mechanical experiments six orders of magnitude weaker in λ. A model is not its founders’ favourite parameter values, and reporting the exclusion of one as the exclusion of the other is the commonest error made about this result.

Frontier “Gravitationally induced entanglement would prove gravity is quantum.” A 2024 result argues that the Diósi–Penrose model, which is a model of classical gravity, itself produces entanglement between separated masses below a critical distance, with the entanglement surviving more than a day for reasonable parameters. Frontier A second paper argues that classical gravity cannot mediate entanglement and the witness stands; a third argues that collapse models are fundamentally non-local, that their entanglement comes from a hidden monitoring process rather than from gravity, and that the witness theorem survives for that reason. Speculative Anyone who tells you the entanglement experiment settles the graviton is skipping an argument that is still running.

Frontier “Decoherence already solved this.” Environmental decoherence explains why interference disappears. It does not explain why one outcome occurs, because the reduced density matrix it produces is a description of ignorance about which branch, not a statement that there is one. Speculative Whether that gap needs a new field or a new ontology is the oldest live disagreement in the subject, and no experiment on this page distinguishes the two.

Speculative “Collapse experiments are experiments about consciousness.” Two of them touch the orchestrated-objective-reduction proposal, and only because that proposal invokes gravity-related state reduction. Established The Gran Sasso group concluded that almost none of the Orch-OR scenarios remain plausible after their measurement, while stating that modified gravity-related models could still support it and that the work is at an early stage. Handwave Everything beyond that — and there is a great deal of it — is assertion.

Speculative “Collapse models violate energy conservation, so they are already dead.” They do violate it, by a heating term the experiments bound, and the violation is small enough to have survived every measurement so far. Frontier The sharper version of the objection is thermodynamic: the original Diósi–Penrose dynamics heats without bound and is consistent with the second law only under an infinite-temperature noise field, while the dissipative repair thermalises only at low friction — and the dissipative variant’s entire parameter space is excluded below a collapse-field temperature of 6 × 10⁻⁹ K. The objection is real; it is not a one-line dismissal.

Handwave “The models can always be saved, so the programme is unfalsifiable.” This is the strongest objection in circulation and it is half right. Frontier The half that is wrong is that not every test is evadable: the Coulomb-mediated-squeezing route is explicitly robust against coloured-noise extensions, the macro-objectification requirement binds every variant that claims to explain definite outcomes, and the dissipative extension has had its whole parameter space closed below a stated temperature. Speculative The half that is right is that the field has no agreed rule for which extensions count, which is why section 10 records writing one as an institutional requirement.

Frontier And one correction this page makes to the summary of this subject that is still in general circulation. The anchor usually cited is the Gran Sasso germanium exclusion together with bounds from LISA Pathfinder and the gravitational-wave detectors, with a further decade in λ from spontaneous X-ray emission named as the decisive next step. Established The anchor is correct and out of date: the germanium channel was improved by the Majorana Demonstrator in 2022 and then by XENONnT in March 2026, which pushed the X-ray bound about two decades rather than one and excluded the GRW parameter choice. Speculative That part of the decisive test has therefore already been run, which is why this brief nominates the levitated-nanoparticle interferometer instead: it is the only remaining route that can return a positive result.