{"id":"9ce7a0e2-5449-4843-b94e-5f11bc985b33","arxiv_id":"1908.09737","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Under the decoherence framework, a Wigner's friend's memory is rewritten by external interference, and the Frauchiger-Renner and Brukner no-go arguments no longer produce contradictions.","lead":"The paper applies the standard decoherence picture of quantum measurement to Wigner's friend thought experiments, arguing that external interference measurements rewrite the friend's memory. It then shows that recent no-go theorems lose their force and proposes a modified Bell-type test consistent with a single set of facts.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The FR resolution conditions on post-interference memory, not the original outcome used in the no-go chain; the claimed resolution of [2] is not established by Eq. (49).","rationale":"The reader correctly identified the decoherence postulate as the main interpretational premise, and I agree that the conclusions depend on it. However, there is a more specific internal issue in the Frauchiger-Renner section: the calculation of 'New fact 1' appears to condition on a different variable than the one used in the no-go chain. The original outcome v_a is a fact about the pre-interference branch, while Eq. (49) describes IA's memory after EA's interference, which is a mixture of original branches. Since the paper's central claim that the contradictions in [2] disappear rests on this specific calculation, the claim is not established as written. The modified CHSH proposal in Sec. V remains a valuable and testable contribution, so a conditional verdict is appropriate, but the condition should now include correcting the FR derivation to condition on the original outcomes used in the paradox.","tokens_in":32145,"tokens_out":35632,"duration_ms":378681,"concrete_test":"Recompute the conditional probability used in the FR chain from the pre-EA state rather than from Eq. (48)/(49). Take Eq. (46), project on the original v_A pointer branch (or on the joint v_A and h_B branch implied by -A), apply the EA interference unitary of Eq. (29) only to A, and evaluate EB's +/- probabilities on B. If the result is P(+B)=1/2 for the joint branch, rather than the 5/6 of Eq. (49), then Eq. (49) conditions on the post-interference memory, not the original outcome, and the claimed resolution of the Frauchiger-Renner paradox is not supported.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The most load-bearing step is the replacement of Fact 1 by 'New fact 1' in Sec. IV B. The paradox chain uses original outcomes: Fact 3a says EA=-A implies IB=h_B, Fact 2 says IB=h_B implies IA's original record was v_a, and Fact 1 was the claim that v_a implies EB=+_B. The paper instead derives from rho_3 (Eq. 49) a probabilistic statement conditioned on IA's memory record evaluated after EA's interference, and uses this to reject Fact 1. But for the original v_a branch, the pre-EA state (Eq. 46) has B=|+_B>; the EA unitary acts only on laboratory A, so that branch's B component remains |+_B>, and if one conditions further on IB=h_B (which the -A branch supplies), B is |h_B>, so EB sees + or - with probability 1/2. Neither equals the 5/6-1/6 split of Eq. (49), because Eq. (49) traces over EA's outcome and apparatus, mixing the original h_a and v_a branches into the same post-EA pointer state. Thus 'New fact 1' changes the conditioning variable; it does not show that the certainty claim used in [2] fails. The contradiction (P(-A,-B)=1/12 versus the agents' deduction P(-B|-A)=0) is therefore not resolved by the calculation given.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a decoherence-based framework in which a quantum measurement is not completed until an uncontrolled environment, with exactly orthogonal branch states, monitors the system-apparatus-observer composite. It applies this framework to Wigner's friend experiments, deriving that an external interference measurement changes the internal agent's memory record in a calculable way (Eq. (26)), and it uses this effect to argue that the Frauchiger-Renner and Brukner no-go theorems are no longer valid. The paper also proposes a modified version of Brukner's experiment in which the final four-agent CHSH value becomes S=1/√2, and it supports the decoherence model with numerical simulations of a GOE environment.","tokens_in":32437,"tokens_out":6835,"duration_ms":79097,"significance":"The manuscript contains explicit, largely reproducible calculations: the reduced-state computation leading to Eq. (26), the four-agent state in Eq. (48), and the modified CHSH expectations in Eqs. (69)-(72). The S=1/√2 result is a parameter-free prediction of the modified protocol, and the numerical work on environment size and chaos is a concrete contribution to the decoherence-based modeling of Wigner's friend scenarios. If the framework's central postulate is accepted, the paper offers a coherent way to compute memory changes under external interference and gives predictions that differ from real-collapse models. However, the claimed resolution of the Frauchiger-Renner theorem is not established, because the argument replaces the conditioning event used in the original no-go chain with a different, later-time condition.","major_comments":[{"comment":"The replacement of 'Fact 1' by 'New fact 1' changes the conditioning variable. In the original Frauchiger-Renner chain, Fact 1 is a statement about the original outcome v_a obtained by IA before EA acts; it says that if IA obtains v_a, then a later measurement of laboratory B in the {|+>,|−>} basis will yield +_B. The calculation leading to Eq. (49) instead conditions on IA's memory record after EA's interference, with EA's outcome and apparatus traced out. These are different events: before EA acts, the original v_a branch has B=|+_B> as shown in Eq. (46), and since the EA unitary acts only on laboratory A, that branch's B component remains |+_B> until EB acts. If one further conditions on the h_B supplied by Fact 3a, the B state is |h_B>, so EB sees + or − with probability 1/2. The 5/6-1/6 split of Eq. (49) arises only after tracing over EA's outcome and apparatus, which mixes the original h_a and v_a branches into the same post-EA pointer state. Thus 'New fact 1' invalidates a different proposition from the one used in the no-go chain; the contradiction P(-A,-B)=1/12 versus P(-B|-A)=0 is not resolved by the calculation given.","section":"Sec. IV B, Eq. (49)"},{"comment":"The conclusion that joint truth values can be assigned to all four agents' outcomes is inferred from the single value S=1/√2<2. For two-party, two-setting, two-outcome correlations, existence of a joint distribution for the four observables is guaranteed by Fine's theorem only when all CHSH inequalities (with all sign choices) hold, not by a single CHSH combination being below 2. The four expectation values computed here do satisfy all these inequalities, so the conclusion is likely correct, but the paper should invoke Fine's theorem explicitly; as written, the inference from one CHSH value to joint assignability is incomplete.","section":"Sec. V B, Eqs. (69)-(72)"},{"comment":"The framework's central postulate—that a measurement is completed only when an uncontrolled environment whose two branch states are exactly orthogonal monitors the system-apparatus-observer composite—is asserted and used as the basis for all later conclusions. The paper is transparent that its conclusions are conditional on this postulate, and it correctly notes that real-collapse theories would give different predictions. This is not an internal inconsistency, but it means the paper does not resolve the Frauchiger-Renner or Brukner paradoxes within quantum theory alone; it shows only that they do not arise under one particular extra postulate. The conclusions section should state this limitation more prominently, since the abstract's phrasing could be read as a stronger claim.","section":"Sec. II B and Table III"}],"minor_comments":[{"comment":"There are numerous typographical errors, including 'recenly', 'relizations', 'suposse', 'Copenhaguen', 'publised', 'objetivity', and 'diamons'. These should be corrected before publication.","section":"Abstract and throughout"},{"comment":"The inequality in Eq. (63) is attributed to 'Claude-Horne-Shimony-Holt'; the correct name is Clauser-Horne-Shimony-Holt.","section":"Sec. V A"},{"comment":"The caption reads 'The number of qbits of both environment is N=6'; this should be 'both environments are N=6'.","section":"Fig. 5 caption"},{"comment":"In the sentence introducing New Fact 1, 'the results are different is the decoherence framework is not taken into account' should read 'if the decoherence framework is not taken into account'.","section":"Sec. IV B, page 20"}],"recommendation":"major_revision","confidential_remarks":"The main obstacle to acceptance is the conditioning-variable issue in the Frauchiger-Renner discussion. If the author can either repair that argument or explicitly limit the claim to a conditional consistency result that does not address the original no-go chain, the paper could be suitable for publication after revision. The Brukner section is more convincing and could be the basis for a stronger paper if separated from the FR resolution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the modified Brukner-style test is the real contribution: the authors construct a protocol where the four agents' memory records satisfy S=1/√2, a concrete and falsifiable prediction. The calculation is straightforward and appears correct. Second, the memory-change result in Eq. (26) — an external interference measurement shifts the friend's record from 1/2–1/2 to 1/4–3/4 — is clean and new. Those two pieces are worth keeping.\n\nThe paper is honest about its interpretational starting point: the decoherence postulate is stated plainly, and the authors admit that real-collapse theories would give different predictions. The numerics with GOE environments are a nice illustration, though the claim that a chaotic interaction is necessary is supported only by finite-size simulation; the paper itself flags this as a conjecture.\n\nThe soft spot is the Frauchiger-Renner section. The authors replace Fact 1 with “New fact 1” based on ρ3 in Eq. (49). But that ρ3 is obtained after tracing out EA's apparatus and environment, so the conditioning variable is IA's memory record after EA's interference — not the original v_a that appears in the FR chain. In the original v_a branch, the B state before EB is |+⟩_B, and EA's unitary acts only on A, so EB still sees + with certainty. The 5/6–1/6 split is a different conditional probability, mixing branches via the trace. As written, the paper does not establish that the FR contradiction disappears. This is a load-bearing flaw, not a typo.\n\nAlso, the inference from S<2 to joint assignability of truth values needs Fine's theorem or an explicit local model; the paper just asserts it. The abstract overstates “rules out all inconsistencies” relative to the careful limitations in Sec. VI.\n\nWho benefits: people working on Wigner's friend experiments, decoherence, and the foundations of measurement. The Brukner modification and Eq. (26) are worth citing; the FR argument needs rethinking, not just polishing. I would send it to peer review with a referee who will press on the conditioning issue, because the paper has enough substance to justify referee time even if the final version looks quite different.","headline":"A serious decoherence-based reply to Wigner's friend no-go theorems with a clean falsifiable prediction (S=1/√2), but the FR resolution rests on a conditioning slip that needs fixing.","tokens_in":32973,"tokens_out":15316,"would_cite":true,"duration_ms":151456,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that in Wigner-friend experiments an external interference measurement rewrites the inner agent's memory in a computable way, and that accounting for this change removes the known contradictions.","keywords":["Wigner's friend","decoherence","pointer states","environment-induced collapse","observer-independent facts","CHSH inequality","quantum measurement problem","quantum machines"],"falsifier":"Run the paper's three-stage protocol with roughly ten environment qubits per laboratory, an initial 50/50 horizontal/vertical state, and the external pre-measurement at $\\theta=\\pi/8$, making the external pre-measurement faster than the internal environment's correlation time. If the central claim is right, reading the friend's memory after the external interference over many runs gives horizontal with probability $1/4$ and vertical with $3/4$; finding the pre-interference $1/2$/$1/2$ statistics would falsify the framework.","tokens_in":31927,"feed_emoji":"⚛️","tokens_out":8587,"duration_ms":86054,"temperature":0.7,"pith_summary":"This paper applies the decoherence interpretation of quantum measurements to Wigner-friend experiments, arguing that a definite outcome appears only after an uncontrolled environment with orthogonal branch states monitors the system, apparatus, and observer. Within this framework every step remains unitary, and the apparent collapse is what the observer sees after tracing out that environment. The central new consequence is that the external interference measurement changes the friend's memory record in a calculable way: for an initial 50/50 horizontal/vertical state and external angle $\\theta=\\pi/8$, the record becomes $1/4$ horizontal and $3/4$ vertical instead of $1/2$ each. Taking these changes into account, the extended two-laboratory no-go arguments no longer force contradictory claims, and the modified four-agent CHSH test gives $S=1/\\sqrt{2}$, compatible with joint truth values for all agents' outcomes. The paper explicitly leaves open whether the no-go theorems fail in general, treating their main statements as conjectures outside this framework.","feed_headline":"External interference rewrites the friend's memory record","feed_subtitle":"Decoherence predicts a Wigner friend's 50-50 memory becomes 1:3 after outside measurement, dissolving the paradoxes.","key_machinery":"The load-bearing object is the environment: a set of qubits continuously coupled to the measuring apparatus through a chaotic interaction Hamiltonian. Its two pointer-branch states become orthogonal, $|\\langle\\varepsilon_1(t)|\\varepsilon_2(t)\\rangle|^2\\sim 0$, which fixes the preferred basis and makes the global state's triorthogonal decomposition unique. Tracing out this environment converts the superposition into the mixed state the agent experiences, and tracing it out again after an external interference measurement produces the altered memory record of Eq. (26). The framework also depends on an exactly known apparatus-environment Hamiltonian, a known environmental initial state, and precisely timed external measurements.","core_discovery":"The central claim is that the decoherence interpretation supplies a univocal protocol for Wigner-friend experiments: after a measurement the real global state is an entangled superposition of the system, the apparatus, and an environment whose two branch states are orthogonal, and the observer's definite outcome is the mixed state obtained by tracing out that environment. Because the external agent's interference acts on the whole laboratory, it also changes the pointer-state probabilities that constitute the inner agent's memory. Equation (26) gives the new record for the standard one-photon case: tracing out the inner environment and the external apparatus and environment leaves $C_{hh}=(2-\\sin 4\\theta)/4$ and $C_{vv}=(2+\\sin 4\\theta)/4$, so at $\\theta=\\pi/8$ the friend's remembered statistics change from $1/2$/$1/2$ to $1/4$/$3/4$. Applying the same accounting to the extended two-laboratory protocol replaces the certainty of the original 'fact 1' with a probability split of $5/6$ versus $1/6$, thereby blocking the no-go conclusion. In the observer-independent-facts setup, the original version is said not to produce well-defined outcomes under this framework at all, while the modified version yields $S=1/\\sqrt{2}<2$, allowing joint truth values for the four agents' memories.","pith_inferences":["If this framework is right, the 'measurement problem' shifts partly into engineering: what counts as definite is determined by how many environment qubits each agent carries, so a few-qubit quantum machine with roughly ten environment qubits should already exhibit all Wigner-friend effects without any conscious observer.","The $1:3$ memory shift is a sharp discriminator: a real-collapse theory would leave the friend's record at $1:1$ after later interference, so an ensemble experiment comparing memory-read statistics before and after the external measurement could distinguish the two approaches.","A direct experimental next step would be to map the full $\\sin 4\\theta$ curve of Eq. (26) across many angles; a mismatch would immediately identify where the model's assumptions fail.","The requirement of a chaotic apparatus-environment interaction suggests that integrable environmental couplings will not produce definite outcomes; testing this with engineered non-chaotic environments could isolate the role of chaos in measurement."],"forward_implications":["The friend's memory record is dynamical: an outside interference measurement at angle $\\theta$ changes the recorded outcome distribution from $1/2$/$1/2$ to $(2-\\sin 4\\theta)/4$ versus $(2+\\sin 4\\theta)/4$, so identical machines in identical runs can honestly report different memories depending on when the memory is read.","The original extended no-go reasoning fails: conditioned on the internal agent's vertical outcome, the final external measurement yields $+_B$ not with certainty but with probability $5/6$, and $-A$ no longer implies $+_B$, so the joint outcome with probability $1/12$ is not ruled out.","The original observer-independent-facts test does not engage the decoherence framework because its initial state is only a pre-measurement; no definite outcomes exist until the environments act, so the CHSH violation concerns correlations of laboratory states, not observer facts.","In the modified four-agent protocol, applying CHSH to the final memory records gives $S=1/\\sqrt{2}$, below 2, so joint truth values for all four agents' outcomes are consistent with the framework.","Wigner-friend interference on large, human-scale laboratories becomes practically impossible: the external pre-measurement must be completed faster than the internal environment's correlation time, so only small quantum machines can implement the experiment."],"supporting_citations":[{"why":"Supplies the original Wigner-friend thought experiment whose apparent paradoxes the framework reinterprets.","marker":"[1]"},{"why":"Supplies the four-agent consistency no-go theorem that the decoherence framework claims to neutralise by modifying the relevant agent fact.","marker":"[2]"},{"why":"Supplies the observer-independent-facts no-go theorem and CHSH setup that the paper argues is not a well-defined outcome scenario.","marker":"[3]"},{"why":"Supplies the experimental realisation of the observer-independent-facts test, which the framework says tests pre-outcome correlations rather than observer facts.","marker":"[4]"},{"why":"Supplies the decoherence and einselection machinery, including pointer states and the environment's role in fixing definite outcomes.","marker":"[5]"},{"why":"Introduces the pointer-basis stability criterion from which definite memory records are derived.","marker":"[6]"},{"why":"Supplies the real-collapse alternative whose different predictions are contrasted with the decoherence framework for future experiments.","marker":"[7]"}],"fun_headline_variants":["Decoherence flips friend's memory odds to 1:3","Wigner's friend memory changes to 1:3 under outside measurement","Friend's 50-50 record becomes 1:3 in decoherence framework","External interference rewrites friend's memory: 1:3 not 50-50","Decoherence predicts friend's memory shift to 1:3"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the postulate that a measurement is only complete when an uncontrollable environment, with two exactly orthogonal branch states, monitors the system-apparatus-observer composite; if that postulate fails, the predicted memory changes and the disappearance of the paradoxes do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Decoherence flips friend's memory odds to 1:3","Wigner's friend memory changes to 1:3 under outside measurement","Friend's 50-50 record becomes 1:3 in decoherence framework","External interference rewrites friend's memory: 1:3 not 50-50","Decoherence predicts friend's memory shift to 1:3"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000526,"raw_usage":{"total_tokens":2649,"prompt_tokens":1161,"completion_tokens":1488,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":777,"completion_tokens_details":{"reasoning_tokens":1387}},"tokens_in":777,"tokens_out":1488,"duration_ms":12103,"temperature":1.0,"reasoning_tokens":1387,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:03:10.720024+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the paper's three-stage protocol with roughly ten environment qubits per laboratory, an initial 50/50 horizontal/vertical state, and the external pre-measurement at $\\theta=\\pi/8$, making the external pre-measurement faster than the internal environment's correlation time. If the central claim is right, reading the friend's memory after the external interference over many runs gives horizontal with probability $1/4$ and vertical with $3/4$; finding the pre-interference $1/2$/$1/2$ statistics would falsify the framework.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the original Wigner-friend thought experiment whose apparent paradoxes the framework reinterprets."},{"cited_title":"Following the same spirit, agent EB chooses between A1 and B1, A1 =|h⟩b|Ah⟩b⟨h|b⟨Ah|b−|v⟩b|Av⟩b⟨v|b⟨Av|b, (61) B1 =|+⟩B|+⟩B−|−⟩B⟨−|B, (62) where|±⟩B = (|h⟩b|Ah⟩b±|v⟩b|Av⟩b)/ √ 2","cited_arxiv_id":null,"evidence_quote":"Supplies the four-agent consistency no-go theorem that the decoherence framework claims to neutralise by modifying the relevant agent fact."},{"cited_title":"(63), applied to |Ψ4⟩ leads to S = 1/ √ 2< 2","cited_arxiv_id":null,"evidence_quote":"Supplies the observer-independent-facts no-go theorem and CHSH setup that the paper argues is not a well-defined outcome scenario."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the experimental realisation of the observer-independent-facts test, which the framework says tests pre-outcome correlations rather than observer facts."},{"cited_title":"Frauchiger and R","cited_arxiv_id":null,"evidence_quote":"Supplies the decoherence and einselection machinery, including pointer states and the environment's role in fixing definite outcomes."},{"cited_title":"Brukner, A No-Go Theorem for Observer-Independent Facts , Entropy 20, 350 (2018)","cited_arxiv_id":null,"evidence_quote":"Introduces the pointer-basis stability criterion from which definite memory records are derived."}],"review_version":1}