{"id":"4b54dfe8-5c25-44bb-9475-24bd78b479c7","arxiv_id":"2501.12870","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A violation of the temporal Bell inequality in the Zych et al. two-switch protocol can be read as temporal nonlocality only under an output-impacting retrocausal interpretation of indefinite causal orders.","lead":"Indefinite causal orders, as realized in two entangled quantum switches, can, on one specific retrocausal interpretation, produce correlations that violate Adlam's principle of temporal locality. The proposed test is not fully model-independent, and the paper carefully maps the assumptions needed to extract temporal nonlocality from the protocol.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central inference depends on an unargued stipulation that retrocausal influences are output-impacting; input-impacting retrocausality, which is what known cyclic causal models of the switch deliver, does not violate Adlam's temporal locality.","rationale":"The reader's conditional verdict is appropriate. My stress-test converges on the second half of the reader's weakest assumption: output-impacting retrocausality. The author is transparent about model-dependence, but the paper's abstract and Section 4.3 claim a violation is 'shown' from ICOs; in fact the violation follows only after an unargued choice among empirically indistinguishable retrocausal models. This is not an internal inconsistency, but it is a load-bearing gap. The concrete test is feasible in principle: one can take the existing cyclic causal model and evaluate Adlam's condition; if it fails to violate Eq. (1), the central claim reduces to a definitional variant. The paper's proposal to weaken Adlam's definition is interesting but not settled, so no upgrade to ACCEPT. No reason to reject: the conditional analysis is sound and valuable. Thus the verdict remains unchanged.","tokens_in":21637,"tokens_out":4340,"duration_ms":46284,"concrete_test":"Implement the Vilasini–Renner cyclic causal model for the quantum switch (their Corollary 6.3) and compute the joint probabilities P(i,j,k|a,b,z) for the two-switch setup of Zych et al. with the parties Ai, Bi treated as definite spacetime events. Check whether these probabilities satisfy Adlam's Eq. (1) when λ(t_a), λ(t_b) are the states immediately prior to the respective operations. If they do, the known cyclic model yields no temporal nonlocality. Then attempt to construct an explicit output-impacting retrocausal model that reproduces the process-matrix predictions of Eq. (6) for all local operations; if no such model can be written down, the paper's inference to temporal nonlocality is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central inference depends on an unargued stipulation that the retrocausal influences in a cyclic reading of the quantum switch are output-impacting. Section 4.3 correctly notes that the Vilasini–Renner cyclic model of an ICO produces feedback loops in which each party's quantum output feeds the other party's input; that is input-impacting retrocausality, and as the author admits it does not violate Eq. (1), because the retrocausal influence is included in the pre-measurement state λ(t_a). The output-impacting variant is introduced only as 'conceivable' (footnote 8), with no worked example, no proof of consistency with the process-matrix probabilities, and no precedent in the retrocausal literature. Thus the conclusion that the protocol can violate Adlam's temporal locality is not derived from the ICO; it is assumed by choosing one of two empirically equivalent retrocausal models. The suggested modification of Adlam's definition from 'immediately prior' to 'prior' would make input-impacting retrocausality count as temporal nonlocality, but that is a change of the principle under test and is explicitly left open. The violation of A6 alone (Section 4.3, point 6) also does not force 'both directions of time'; in scenarios 1 and 2 the notion of temporal locality is inapplicable rather than violated. So the strongest defensible claim is conditional, and the condition is not independently supported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a protocol based on two entangled quantum switches (Zych et al. 2019) as a test of Adlam's principle of temporal locality. It argues that a violation of the temporal Bell inequality derived by Zych et al., under assumptions A1–A4, forces a choice between rejecting A5 (free-evolution/control correlations) and rejecting A6 (classical order). The paper contends that rejecting A6 can be read as implying retrocausal influences and, if those influences are 'output-impacting,' a violation of Adlam's temporal locality, without the realism assumptions of Leggett-Garg or temporal CHSH tests. It then reviews three possible spacetime embeddings of the quantum switch and concludes that only a 'proper ICO with a cyclic structure' yields temporal nonlocality, making the test model-dependent. A variant of Adlam's definition that would count input-impacting retrocausality is proposed but left open.","tokens_in":21828,"tokens_out":7032,"duration_ms":66978,"significance":"If the conditional claim were established, the paper would provide a genuine conceptual bridge between indefinite causal orders and temporal nonlocality, and it would sharpen the distinction between violating and merely making inapplicable a notion like temporal locality. The paper's strengths are its transparency and its careful taxonomy: it explicitly states at the end of Section 4.3 that the inference is 'highly model-dependent,' and Section 4.4 gives three distinct embeddings and identifies which one would support the conclusion. It also correctly notes that the Vilasini–Renner cyclic model is input-impacting and therefore does not violate Eq. (1). The main limitation is that the load-bearing notion of output-impacting retrocausality is introduced only as 'conceivable' (footnote 8), with no worked example, consistency proof, or precedent, so the paper establishes a conditional possibility rather than a derivation of temporal nonlocality from ICOs.","major_comments":[{"comment":"The central inference to temporal nonlocality is not derived from the indefinite causal order; it is assumed by stipulating an 'output-impacting' retrocausal model. The paper states that the Vilasini–Renner cyclic causal model of the quantum switch produces feedback loops in which each party's quantum output affects the other's input, and that this input-impacting model does not violate Eq. (1) because the retrocausal influence is already contained in λ(t_a). The output-impacting variant is introduced only as 'conceivable and could be posited with the same experimental predictions,' with no worked example, no proof of consistency with process-matrix probabilities, and no precedent in the retrocausal literature. Consequently the abstract's claim that the paper 'shows how Adlam's principle of temporal locality can be violated' is stronger than what the body supports; the body supports only a conditional claim. This gap is load-bearing because without output-impacting retrocausality the protocol does not violate Eq. (1).","section":"Section 4.3 (paragraph beginning 'One sees, however...' and footnote 8)"},{"comment":"The argument from violation of assumption A6 to 'causal influences happening in both directions of time' is valid only for scenario (3), where the quantum switch is embedded as a definite cyclic relativistic causal structure among well-localized events. The paper itself acknowledges in Section 4.4 that in scenario (1) (superposition of worldlines in a fixed spacetime) and scenario (2) (metaphysically indefinite spacetime) the events are not well-defined, so the notion of temporal locality is inapplicable rather than violated. This means that a violation of the temporal Bell inequality does not by itself establish temporal nonlocality; it requires a prior, model-dependent commitment to scenario (3), including the claim that the operations A_i and B_i actually take place as well-localized measurements. The paper says the latter is 'merely assumed' in Section 4.3, but that assumption is precisely what is needed for Eq. (1) and Eq. (2) to be applicable.","section":"Section 4.4, scenarios (1)–(2), and Section 4.3, point (6)"},{"comment":"The proposed revision of Adlam's definition from 'immediately prior' to 'prior' is explicitly left open. If this revision is adopted, input-impacting retrocausality would count as a violation of the modified principle, but then the protocol would not be testing Adlam's original principle, so the paper's stated target would shift. If the revision is not adopted, the claim that the protocol can violate Adlam's temporal locality remains hostage to the unsupported output-impacting retrocausal model. The paper needs to take a definite position and argue for it; as written, the central claim is ambiguous between testing Adlam's principle and testing a close variant.","section":"Section 4.3, final two paragraphs"}],"minor_comments":[{"comment":"The control state in the caption is written as '1/2 [|0_c⟩ + |1_c⟩]', which lacks the correct normalization and should be '1/√2 (|0_c⟩ + |1_c⟩)' as in Eq. (4).","section":"Section 4.1, Fig. 3 caption"},{"comment":"The parenthetical for the case 'B_i ⪯ A_i' repeats '(A_i is in the past causal cone of B_i)' from the previous clause; it should say that B_i is in the past causal cone of A_i or that A_i is in the future causal cone of B_i.","section":"Section 4.2, assumption A6"},{"comment":"The phrase 'A model-independent guarantee that at least [A6] is violated via (output-impacting) retrocausal influences' is unclear, since A6 is an assumption in the Zych et al. derivation rather than a statement about retrocausality; consider rephrasing to say that the guarantee concerns the spatiotemporal embedding of the switch as a cyclic structure with output-impacting influences.","section":"Section 4.4, first paragraph"},{"comment":"The phrase 'the corresponding temporal Bell inequality' in the description of the Zych setup is not defined until Section 4.2; a forward reference to the inequality's derivation would help the reader.","section":"Section 1, last paragraph of the introduction"},{"comment":"The conclusion says the protocol 'could imply temporal nonlocality modulo model-dependent assumptions,' which is accurate, but it should be aligned with the stronger wording in the abstract; the abstract should carry the same conditional qualification.","section":"Section 6, Conclusion"}],"recommendation":"major_revision","confidential_remarks":"The paper is a philosophy-of-science contribution that is transparent about its own limitations. The main issue is the gap between the conditional claim supported by the body and the stronger claim in the abstract and conclusion, together with reliance on an unconstructed 'output-impacting' retrocausal model. This is fixable by revising the claims and either providing a concrete model or explicitly framing the result as a conditional possibility. The paper is within scope for a philosophy of physics journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Laurie Letertre's paper is a rare thing in this area: a genuinely clear conditional analysis. It takes Zych et al.'s two-switch temporal Bell protocol and asks what it would take to infer a violation of Adlam's temporal locality from the violation of the inequality. The answer, it argues, is a specific reading of ICOs as definite cyclic relativistic causal structures with output-impacting retrocausal influences. The paper is admirably transparent about its own model-dependence: it flags assumptions A5 and A6, discusses three distinct spacetime embeddings, and explicitly admits that the test is not model-independent. The distinction between input-impacting and output-impacting retrocausality is a useful conceptual contribution that I haven't seen drawn sharply before.\n\nThe soft spot is exactly where the stress-test note puts it. The Vilasini-Renner cyclic model of the quantum switch, which the paper relies on for the cyclic reading, delivers feedback loops in which each party's quantum output feeds the other's input. That is input-impacting, and as the author correctly notes, it does not violate Eq. (1) because the retrocausal influence is already included in the pre-measurement state \\lambda(t_a). The output-impacting variant is introduced only as 'conceivable' in footnote 8, with no worked example, no consistency proof with the process-matrix probabilities, and no precedent in the retrocausal literature. So the conclusion that the protocol can violate Adlam's temporal locality is not derived from the ICO; it is assumed by choosing one of two empirically equivalent retrocausal models. The suggested modification of Adlam's definition from 'immediately prior' to 'prior' would make input-impacting retrocausality count, but that is a change of the principle under test, and the author leaves it open.\n\nThe paper does not pretend otherwise. It explicitly says the inference is model-dependent, and it works carefully through what would need to be true for a model-independent test. That is honest, but it means the central claim is conditional on a premise that has no independent support. For a philosophy-of-physics paper this is acceptable; it is a contribution to the conceptual geography rather than a proof. For an experiment or a mathematical derivation it would be a problem.\n\nThis is a paper for philosophers and foundations types, not for experimentalists. It deserves a serious referee — it already has one, since it is forthcoming in BJPS — and I would send it to review without hesitation. I'd bring it to a reading group if the group's interest includes temporal nonlocality or ICOs, because the output-impacting assumption is worth arguing about. I'd cite it for the input/output-impacting distinction, while being careful not to cite it as evidence of temporal nonlocality.","headline":"Careful conditional analysis of how indefinite causal orders might violate Adlam temporal locality, but the central inference rests on an unargued stipulation of output-impacting retrocausality.","tokens_in":22428,"tokens_out":2870,"would_cite":true,"duration_ms":27660,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A protocol built on two entangled quantum switches can violate temporal locality, provided the switch is read as a genuine cyclic causal loop whose backward influence changes earlier measurement outcomes.","keywords":["temporal nonlocality","indefinite causal order","quantum switch","process matrix formalism","temporal Bell inequality","retrocausality","causal nonseparability","temporal locality"],"falsifier":"Show that a flat-spacetime implementation of the two-switch protocol involves four mutually exclusive event candidates instead of two definite operation events—for example with the event-count discrimination test discussed in the paper—and the cyclic retrocausal reading is ruled out. Alternatively, exhibit a complete model of the switch whose only backward influences are input-impacting; such a model keeps the temporal-locality equalities intact and falsifies the inference.","tokens_in":21300,"feed_emoji":"🔄","tokens_out":7399,"duration_ms":74387,"temperature":0.7,"pith_summary":"The paper tries to show that a particular quantum process with no definite causal order—the quantum switch—can be used to test, and in one reading violate, the principle that all influences on a measurement outcome pass through the state of the world just before the measurement. The route runs through a known protocol with two entangled quantum switches whose correlations violate a temporal Bell inequality. Standard temporal tests need extra realism assumptions; this protocol avoids those, but the paper argues the inference to temporal nonlocality is model-dependent: it requires interpreting the switch as a definite but cyclic causal structure whose backward influences change earlier measurement outcomes. If the argument holds, temporal nonlocality becomes a live consequence of indefinite causal orders, and may point to a common retrocausal explanation of ordinary Bell nonlocality.","feed_headline":"Indefinite causal order can yield temporal nonlocality","feed_subtitle":"A two-switch Bell test, read as a cyclic retrocausal loop, forces the backward influence to hit measurement outcomes.","key_machinery":"The load-bearing object is the quantum switch, a process in which a control qubit in a superposition determines whether operation A precedes B or B precedes A, leaving the order itself indefinite. The protocol couples two spacelikely separated quantum switches through a common massive control, so the final target state is entangled and violates a Bell inequality under a set of explicit assumptions; the relevant step is rejecting the classical-order assumption. The paper connects that rejection to a proof that any causal structure compatible with an indefinite-order process must be cyclic, and then interprets the cyclic structure as retrocausal influence between well-localized events. Temporal locality fails precisely when the backward influence reaches the earlier party's measurement outcome rather than its input.","core_discovery":"The central claim is that a violation of the temporal Bell inequality in the two-switch protocol can be read as a violation of temporal locality, provided the indefinite causal order is not dismissed as a superposition of worldlines in a fixed spacetime. The paper argues that if the process is embedded as a definite but cyclic relativistic causal structure among well-localized events, then the absence of any acyclic order forces retrocausal influences between the operations. Those influences violate temporal locality only when they are output-impacting—when they change the outcome of the earlier measurement. The paper states plainly that this last step is model-dependent, and that other embeddings either make temporal locality inapplicable (simulations and metaphysically indefinite spacetimes) or fail to violate it (input-impacting retrocausality). It also contends that because the same retrocausal mechanism could account for ordinary Bell nonlocality, the distinction between spatial and temporal nonlocality may not mark a difference in physical underpinning.","pith_inferences":["If the model-dependence is resolved in favor of the cyclic reading, temporal locality becomes empirically testable in flat spacetime without gravitational switches, making the temporal/spatial distinction operational rather than metaphysical.","The same logic could be applied to other causally nonseparable processes: any process whose only compatible causal structure is cyclic would be a candidate witness for temporal nonlocality, not just the quantum switch.","A decisive experiment could use the event-count discrimination between two-event and four-event embeddings; showing four events would reclassify apparent violations as inapplicability rather than violation of temporal locality."],"forward_implications":["If the two-switch protocol is realized and its Bell inequality violated, the classical-order assumption must be dropped, forcing either retrocausal influence or a violation of the free-evolution assumption.","Under the cyclic retrocausal reading, that violation is a genuine violation of temporal locality, without the realism commitments carried by Leggett-Garg or temporal CHSH tests.","The same retrocausal mechanism that would produce temporal nonlocality can also account for standard Bell nonlocality, so the two phenomena may share one physical explanation.","A modest broadening of the definition—from a state immediately prior to a measurement to a state merely prior to it—would let input-impacting retrocausal models also count as temporally nonlocal."],"supporting_citations":[{"why":"Supplies the definition of temporal locality that the protocol is designed to test.","marker":"Adlam [2018]"},{"why":"Provides the two-switch protocol and the temporal Bell inequality whose violation drives the argument.","marker":"Zych et al. [2019]"},{"why":"Makes explicit the free-evolution assumption and adapts the protocol to flat spacetime.","marker":"Debski et al. [2023]"},{"why":"Proves that indefinite causal order forces a cyclic causal structure, the basis for the retrocausal reading.","marker":"Vilasini and Renner [unpublished]"},{"why":"Introduces the quantum switch as the concrete indefinite-causal-order process.","marker":"Chiribella et al. [2013]"},{"why":"Sets out the process matrix formalism in which causally nonseparable processes are defined.","marker":"Oreshkov et al. [2012]"},{"why":"Distinguishes simulated from proper indefinite causal orders and offers an event-count discrimination test.","marker":"Paunković and Vojinović [2020]"},{"why":"Argues that indefinite causal order is frame-independent and arises either from worldline or metric superpositions, shaping the three embedding scenarios.","marker":"de la Hamette et al. [unpublishedb]"}],"fun_headline_variants":["Indefinite causal order can force temporal nonlocality","Retrocausal loops from indefinite order break locality","Temporal nonlocality from causal cycles is model-dependent","Cyclic causal order yields temporal nonlocality","ICO makes temporal nonlocality a retrocausal effect"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on treating the quantum switch as a real loop of influence between well-defined events, with backward effects that alter the earlier measurement outcome rather than only the input that leads to it; if either part fails, the Bell violation no longer demonstrates temporal nonlocality.","fun_headline_variants_meta":{"raw":{"variants":["Indefinite causal order can force temporal nonlocality","Retrocausal loops from indefinite order break locality","Temporal nonlocality from causal cycles is model-dependent","Cyclic causal order yields temporal nonlocality","ICO makes temporal nonlocality a retrocausal effect"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000775,"raw_usage":{"total_tokens":3432,"prompt_tokens":950,"completion_tokens":2482,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":566,"completion_tokens_details":{"reasoning_tokens":2405}},"tokens_in":566,"tokens_out":2482,"duration_ms":22114,"temperature":1.0,"reasoning_tokens":2405,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T16:42:08.204165+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Show that a flat-spacetime implementation of the two-switch protocol involves four mutually exclusive event candidates instead of two definite operation events—for example with the event-count discrimination test discussed in the paper—and the cyclic retrocausal reading is ruled out. Alternatively, exhibit a complete model of the switch whose only backward influences are input-impacting; such a model keeps the temporal-locality equalities intact and falsifies the inference.","supporting_citations":[{"cited_title":"[2018]: ‘Spooky Action at a Temporal Distance’, Entropy, 20(1), pp","cited_arxiv_id":null,"evidence_quote":"Supplies the definition of temporal locality that the protocol is designed to test."},{"cited_title":"and Brukner,ˇC","cited_arxiv_id":null,"evidence_quote":"Sets out the process matrix formalism in which causally nonseparable processes are defined."}],"review_version":1}